{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999363016752659,
  "eval_steps": 500,
  "global_step": 7849,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 1.736907991191692,
      "learning_rate": 8.474576271186441e-09,
      "loss": 0.6756,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.037276039206706,
      "learning_rate": 1.6949152542372882e-08,
      "loss": 0.8562,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9584492301874183,
      "learning_rate": 2.5423728813559323e-08,
      "loss": 0.8069,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6159148861978476,
      "learning_rate": 3.3898305084745764e-08,
      "loss": 0.6642,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7507462262585467,
      "learning_rate": 4.23728813559322e-08,
      "loss": 0.6532,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.72856886019445,
      "learning_rate": 5.0847457627118645e-08,
      "loss": 0.643,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8661752453165121,
      "learning_rate": 5.932203389830508e-08,
      "loss": 0.7828,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7845271408013215,
      "learning_rate": 6.779661016949153e-08,
      "loss": 0.6582,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9253846188650432,
      "learning_rate": 7.627118644067796e-08,
      "loss": 0.8298,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9423391927911509,
      "learning_rate": 8.47457627118644e-08,
      "loss": 0.7816,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9276764473749466,
      "learning_rate": 9.322033898305084e-08,
      "loss": 0.837,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8090641850120723,
      "learning_rate": 1.0169491525423729e-07,
      "loss": 0.6383,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9330060837522742,
      "learning_rate": 1.1016949152542372e-07,
      "loss": 0.8522,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9741691900741056,
      "learning_rate": 1.1864406779661017e-07,
      "loss": 0.8672,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7229307581715625,
      "learning_rate": 1.271186440677966e-07,
      "loss": 0.6488,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7513066679665468,
      "learning_rate": 1.3559322033898305e-07,
      "loss": 0.6687,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.124278059287193,
      "learning_rate": 1.440677966101695e-07,
      "loss": 0.8082,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.452047425526202,
      "learning_rate": 1.5254237288135593e-07,
      "loss": 0.8645,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5986937931149383,
      "learning_rate": 1.6101694915254234e-07,
      "loss": 0.6787,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.964984784515822,
      "learning_rate": 1.694915254237288e-07,
      "loss": 0.8268,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9634178593301097,
      "learning_rate": 1.7796610169491524e-07,
      "loss": 0.8554,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8321764340988278,
      "learning_rate": 1.8644067796610168e-07,
      "loss": 0.6615,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0461313931003997,
      "learning_rate": 1.9491525423728814e-07,
      "loss": 0.8641,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6747467242769618,
      "learning_rate": 2.0338983050847458e-07,
      "loss": 0.6928,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5553017316746622,
      "learning_rate": 2.11864406779661e-07,
      "loss": 0.7425,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8900258047171405,
      "learning_rate": 2.2033898305084743e-07,
      "loss": 0.8241,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7904468737726056,
      "learning_rate": 2.288135593220339e-07,
      "loss": 0.7992,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7227275208881083,
      "learning_rate": 2.3728813559322033e-07,
      "loss": 0.6786,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6089708826434301,
      "learning_rate": 2.457627118644068e-07,
      "loss": 0.6599,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2838727573755087,
      "learning_rate": 2.542372881355932e-07,
      "loss": 0.866,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6500016352473748,
      "learning_rate": 2.6271186440677967e-07,
      "loss": 0.656,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7800906444768814,
      "learning_rate": 2.711864406779661e-07,
      "loss": 0.7132,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7238876681241084,
      "learning_rate": 2.796610169491525e-07,
      "loss": 0.6724,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5146284027965435,
      "learning_rate": 2.88135593220339e-07,
      "loss": 0.6049,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5963153470274938,
      "learning_rate": 2.966101694915254e-07,
      "loss": 0.6265,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8534980062117663,
      "learning_rate": 3.0508474576271186e-07,
      "loss": 0.8883,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7748900141364816,
      "learning_rate": 3.135593220338983e-07,
      "loss": 0.7677,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6702097228507513,
      "learning_rate": 3.220338983050847e-07,
      "loss": 0.6794,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.880199727289493,
      "learning_rate": 3.3050847457627117e-07,
      "loss": 0.8155,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4414042551540716,
      "learning_rate": 3.389830508474576e-07,
      "loss": 0.6662,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8226410905100863,
      "learning_rate": 3.4745762711864405e-07,
      "loss": 0.8292,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4265036111952265,
      "learning_rate": 3.559322033898305e-07,
      "loss": 0.735,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3513287066156798,
      "learning_rate": 3.644067796610169e-07,
      "loss": 0.5731,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4786209127441055,
      "learning_rate": 3.7288135593220336e-07,
      "loss": 0.6241,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8510598139820174,
      "learning_rate": 3.813559322033898e-07,
      "loss": 0.8815,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3290851439739941,
      "learning_rate": 3.898305084745763e-07,
      "loss": 0.6544,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.33582877345575,
      "learning_rate": 3.9830508474576267e-07,
      "loss": 0.6757,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.979031845647636,
      "learning_rate": 4.0677966101694916e-07,
      "loss": 0.8738,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.751990682888358,
      "learning_rate": 4.152542372881356e-07,
      "loss": 0.8014,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3604883945939532,
      "learning_rate": 4.23728813559322e-07,
      "loss": 0.6335,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6485861706131393,
      "learning_rate": 4.322033898305085e-07,
      "loss": 0.8294,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1751355376740993,
      "learning_rate": 4.4067796610169486e-07,
      "loss": 0.5926,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.12095195415341,
      "learning_rate": 4.4915254237288135e-07,
      "loss": 0.64,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4812998599977767,
      "learning_rate": 4.576271186440678e-07,
      "loss": 0.7446,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2057323995503466,
      "learning_rate": 4.661016949152542e-07,
      "loss": 0.6702,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5227887973019671,
      "learning_rate": 4.7457627118644066e-07,
      "loss": 0.7957,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.645979191935828,
      "learning_rate": 4.830508474576271e-07,
      "loss": 0.8409,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.06135731514654,
      "learning_rate": 4.915254237288136e-07,
      "loss": 0.6786,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9645039052788676,
      "learning_rate": 5e-07,
      "loss": 0.6491,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1322859683136068,
      "learning_rate": 5.084745762711864e-07,
      "loss": 0.6446,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5744537082289736,
      "learning_rate": 5.169491525423729e-07,
      "loss": 0.8238,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0371717244642593,
      "learning_rate": 5.254237288135593e-07,
      "loss": 0.619,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0200097922405438,
      "learning_rate": 5.338983050847457e-07,
      "loss": 0.7248,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9014095782184974,
      "learning_rate": 5.423728813559322e-07,
      "loss": 0.5905,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3748332438239352,
      "learning_rate": 5.508474576271186e-07,
      "loss": 0.7573,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.941859025394431,
      "learning_rate": 5.59322033898305e-07,
      "loss": 0.6675,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.542485954445273,
      "learning_rate": 5.677966101694916e-07,
      "loss": 0.8226,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8670841859736869,
      "learning_rate": 5.76271186440678e-07,
      "loss": 0.6528,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8206143637356134,
      "learning_rate": 5.847457627118644e-07,
      "loss": 0.6114,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1695599053335903,
      "learning_rate": 5.932203389830508e-07,
      "loss": 0.7393,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8807560514359539,
      "learning_rate": 6.016949152542372e-07,
      "loss": 0.6264,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8495158693418726,
      "learning_rate": 6.101694915254237e-07,
      "loss": 0.5833,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1037404169285423,
      "learning_rate": 6.186440677966102e-07,
      "loss": 0.8157,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3898605466952594,
      "learning_rate": 6.271186440677966e-07,
      "loss": 0.791,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1774056983193821,
      "learning_rate": 6.35593220338983e-07,
      "loss": 0.7602,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8878185503609648,
      "learning_rate": 6.440677966101694e-07,
      "loss": 0.6094,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9377796089897491,
      "learning_rate": 6.52542372881356e-07,
      "loss": 0.6189,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7735081300754656,
      "learning_rate": 6.610169491525423e-07,
      "loss": 0.5823,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3035050675411433,
      "learning_rate": 6.694915254237287e-07,
      "loss": 0.7418,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0666600791413023,
      "learning_rate": 6.779661016949152e-07,
      "loss": 0.7282,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.017823886818616,
      "learning_rate": 6.864406779661016e-07,
      "loss": 0.7493,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7547975133145556,
      "learning_rate": 6.949152542372881e-07,
      "loss": 0.5746,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.73184861994517,
      "learning_rate": 7.033898305084746e-07,
      "loss": 0.6166,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9582702160049402,
      "learning_rate": 7.11864406779661e-07,
      "loss": 0.6313,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1239899531200919,
      "learning_rate": 7.203389830508474e-07,
      "loss": 0.7616,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3986129979013686,
      "learning_rate": 7.288135593220338e-07,
      "loss": 0.8416,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0064870167909685,
      "learning_rate": 7.372881355932203e-07,
      "loss": 0.7437,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.029078360489064,
      "learning_rate": 7.457627118644067e-07,
      "loss": 0.6885,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0839861541694729,
      "learning_rate": 7.542372881355932e-07,
      "loss": 0.7679,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7358869048632126,
      "learning_rate": 7.627118644067796e-07,
      "loss": 0.5264,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2088648839893386,
      "learning_rate": 7.711864406779661e-07,
      "loss": 0.7689,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.120303884023188,
      "learning_rate": 7.796610169491526e-07,
      "loss": 0.8094,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9631871045216217,
      "learning_rate": 7.88135593220339e-07,
      "loss": 0.7234,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7794033238755568,
      "learning_rate": 7.966101694915253e-07,
      "loss": 0.7052,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9551524577761956,
      "learning_rate": 8.050847457627118e-07,
      "loss": 0.7059,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.241098614689768,
      "learning_rate": 8.135593220338983e-07,
      "loss": 0.7344,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.685452828600968,
      "learning_rate": 8.220338983050847e-07,
      "loss": 0.5852,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9908084317417779,
      "learning_rate": 8.305084745762712e-07,
      "loss": 0.7078,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7759864093838978,
      "learning_rate": 8.389830508474576e-07,
      "loss": 0.6286,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6588435616176912,
      "learning_rate": 8.47457627118644e-07,
      "loss": 0.6297,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0780622896941123,
      "learning_rate": 8.559322033898306e-07,
      "loss": 0.7916,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8634576635077581,
      "learning_rate": 8.64406779661017e-07,
      "loss": 0.6609,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9633133489846274,
      "learning_rate": 8.728813559322033e-07,
      "loss": 0.7608,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0763224900705421,
      "learning_rate": 8.813559322033897e-07,
      "loss": 0.7842,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6933564189256128,
      "learning_rate": 8.898305084745762e-07,
      "loss": 0.5805,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2201861803855387,
      "learning_rate": 8.983050847457627e-07,
      "loss": 0.7236,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9969497849781063,
      "learning_rate": 9.067796610169491e-07,
      "loss": 0.7544,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0977139267521236,
      "learning_rate": 9.152542372881356e-07,
      "loss": 0.7179,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8838634691331352,
      "learning_rate": 9.23728813559322e-07,
      "loss": 0.7207,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.748110567914877,
      "learning_rate": 9.322033898305083e-07,
      "loss": 0.6404,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9035080674620525,
      "learning_rate": 9.406779661016949e-07,
      "loss": 0.666,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8215729440307579,
      "learning_rate": 9.491525423728813e-07,
      "loss": 0.6738,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9985290224208793,
      "learning_rate": 9.576271186440678e-07,
      "loss": 0.7507,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8881217346231228,
      "learning_rate": 9.661016949152542e-07,
      "loss": 0.6597,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8387068607992167,
      "learning_rate": 9.745762711864406e-07,
      "loss": 0.5862,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7013763758333112,
      "learning_rate": 9.830508474576272e-07,
      "loss": 0.638,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9729094966631777,
      "learning_rate": 9.915254237288136e-07,
      "loss": 0.7219,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.075490651244943,
      "learning_rate": 1e-06,
      "loss": 0.7233,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4660095232612844,
      "learning_rate": 1.0084745762711863e-06,
      "loss": 0.7159,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6953319855373109,
      "learning_rate": 1.0169491525423727e-06,
      "loss": 0.6101,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7487617910462017,
      "learning_rate": 1.0254237288135591e-06,
      "loss": 0.5972,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9754826562644147,
      "learning_rate": 1.0338983050847457e-06,
      "loss": 0.7256,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7014813783630192,
      "learning_rate": 1.0423728813559323e-06,
      "loss": 0.5945,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8535104986573425,
      "learning_rate": 1.0508474576271187e-06,
      "loss": 0.6683,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9792311827984785,
      "learning_rate": 1.059322033898305e-06,
      "loss": 0.692,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7402035619731817,
      "learning_rate": 1.0677966101694915e-06,
      "loss": 0.597,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7054330763593275,
      "learning_rate": 1.0762711864406778e-06,
      "loss": 0.5649,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7752496570670936,
      "learning_rate": 1.0847457627118644e-06,
      "loss": 0.6389,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.708578374649769,
      "learning_rate": 1.0932203389830508e-06,
      "loss": 0.584,
      "step": 129
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7694352115175147,
      "learning_rate": 1.1016949152542372e-06,
      "loss": 0.6728,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1038397003062124,
      "learning_rate": 1.1101694915254236e-06,
      "loss": 0.7486,
      "step": 131
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2155941234125076,
      "learning_rate": 1.11864406779661e-06,
      "loss": 0.8189,
      "step": 132
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7971035604067273,
      "learning_rate": 1.1271186440677966e-06,
      "loss": 0.5827,
      "step": 133
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9358974996612655,
      "learning_rate": 1.1355932203389832e-06,
      "loss": 0.6304,
      "step": 134
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6799249401488097,
      "learning_rate": 1.1440677966101696e-06,
      "loss": 0.6041,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8728300170888112,
      "learning_rate": 1.152542372881356e-06,
      "loss": 0.7383,
      "step": 136
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9499540369375933,
      "learning_rate": 1.1610169491525423e-06,
      "loss": 0.718,
      "step": 137
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6769253124950282,
      "learning_rate": 1.1694915254237287e-06,
      "loss": 0.5817,
      "step": 138
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2204843539263122,
      "learning_rate": 1.177966101694915e-06,
      "loss": 0.743,
      "step": 139
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6716394240064161,
      "learning_rate": 1.1864406779661017e-06,
      "loss": 0.5908,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.015631981503019,
      "learning_rate": 1.194915254237288e-06,
      "loss": 0.6783,
      "step": 141
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0183453896762789,
      "learning_rate": 1.2033898305084745e-06,
      "loss": 0.6437,
      "step": 142
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0723305416299242,
      "learning_rate": 1.211864406779661e-06,
      "loss": 0.7592,
      "step": 143
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6677300249274252,
      "learning_rate": 1.2203389830508474e-06,
      "loss": 0.5473,
      "step": 144
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7642530848110866,
      "learning_rate": 1.2288135593220338e-06,
      "loss": 0.6643,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7785754946837703,
      "learning_rate": 1.2372881355932204e-06,
      "loss": 0.7155,
      "step": 146
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8377482285397575,
      "learning_rate": 1.2457627118644068e-06,
      "loss": 0.5985,
      "step": 147
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0977320492459255,
      "learning_rate": 1.2542372881355932e-06,
      "loss": 0.7161,
      "step": 148
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6518293469641255,
      "learning_rate": 1.2627118644067796e-06,
      "loss": 0.5659,
      "step": 149
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7253123451641916,
      "learning_rate": 1.271186440677966e-06,
      "loss": 0.6691,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7245782196375631,
      "learning_rate": 1.2796610169491523e-06,
      "loss": 0.5796,
      "step": 151
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8873047078263828,
      "learning_rate": 1.2881355932203387e-06,
      "loss": 0.7144,
      "step": 152
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7978648182453759,
      "learning_rate": 1.2966101694915255e-06,
      "loss": 0.6118,
      "step": 153
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9492892287716275,
      "learning_rate": 1.305084745762712e-06,
      "loss": 0.726,
      "step": 154
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6411207453563611,
      "learning_rate": 1.3135593220338983e-06,
      "loss": 0.6108,
      "step": 155
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6765600144090188,
      "learning_rate": 1.3220338983050847e-06,
      "loss": 0.5846,
      "step": 156
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.763246198843269,
      "learning_rate": 1.330508474576271e-06,
      "loss": 0.6339,
      "step": 157
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6774376546958141,
      "learning_rate": 1.3389830508474575e-06,
      "loss": 0.6263,
      "step": 158
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9304974601713167,
      "learning_rate": 1.347457627118644e-06,
      "loss": 0.7302,
      "step": 159
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7346087648293955,
      "learning_rate": 1.3559322033898304e-06,
      "loss": 0.6046,
      "step": 160
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7591533949697616,
      "learning_rate": 1.3644067796610168e-06,
      "loss": 0.6218,
      "step": 161
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3182181126014183,
      "learning_rate": 1.3728813559322032e-06,
      "loss": 0.699,
      "step": 162
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7867617731133156,
      "learning_rate": 1.3813559322033898e-06,
      "loss": 0.5479,
      "step": 163
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8341730596805433,
      "learning_rate": 1.3898305084745762e-06,
      "loss": 0.6217,
      "step": 164
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6068608848025123,
      "learning_rate": 1.3983050847457628e-06,
      "loss": 0.5667,
      "step": 165
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0938284867069135,
      "learning_rate": 1.4067796610169492e-06,
      "loss": 0.7484,
      "step": 166
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.134039413749055,
      "learning_rate": 1.4152542372881356e-06,
      "loss": 0.7122,
      "step": 167
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6571908554662265,
      "learning_rate": 1.423728813559322e-06,
      "loss": 0.5929,
      "step": 168
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6333718599518113,
      "learning_rate": 1.4322033898305083e-06,
      "loss": 0.5604,
      "step": 169
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8441410553098189,
      "learning_rate": 1.4406779661016947e-06,
      "loss": 0.5584,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7553081798861739,
      "learning_rate": 1.4491525423728813e-06,
      "loss": 0.6201,
      "step": 171
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9715502616767301,
      "learning_rate": 1.4576271186440677e-06,
      "loss": 0.7184,
      "step": 172
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0630557439864587,
      "learning_rate": 1.4661016949152543e-06,
      "loss": 0.7068,
      "step": 173
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7085120333954494,
      "learning_rate": 1.4745762711864407e-06,
      "loss": 0.5687,
      "step": 174
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.920567393855287,
      "learning_rate": 1.483050847457627e-06,
      "loss": 0.6649,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8859635417658249,
      "learning_rate": 1.4915254237288134e-06,
      "loss": 0.7284,
      "step": 176
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8575496337923804,
      "learning_rate": 1.5e-06,
      "loss": 0.6348,
      "step": 177
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6154629229842142,
      "learning_rate": 1.5084745762711864e-06,
      "loss": 0.5649,
      "step": 178
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7748967016331698,
      "learning_rate": 1.5169491525423728e-06,
      "loss": 0.6059,
      "step": 179
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6601826638948631,
      "learning_rate": 1.5254237288135592e-06,
      "loss": 0.6107,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7083304048584579,
      "learning_rate": 1.5338983050847456e-06,
      "loss": 0.6084,
      "step": 181
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0141184973912347,
      "learning_rate": 1.5423728813559322e-06,
      "loss": 0.727,
      "step": 182
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8866145473315817,
      "learning_rate": 1.5508474576271186e-06,
      "loss": 0.7106,
      "step": 183
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7883327930960949,
      "learning_rate": 1.5593220338983052e-06,
      "loss": 0.6055,
      "step": 184
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.001263325318397,
      "learning_rate": 1.5677966101694915e-06,
      "loss": 0.7681,
      "step": 185
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.886301112577374,
      "learning_rate": 1.576271186440678e-06,
      "loss": 0.7628,
      "step": 186
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7786724627559611,
      "learning_rate": 1.5847457627118643e-06,
      "loss": 0.5789,
      "step": 187
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.792192151552707,
      "learning_rate": 1.5932203389830507e-06,
      "loss": 0.6173,
      "step": 188
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6851675266587554,
      "learning_rate": 1.601694915254237e-06,
      "loss": 0.5854,
      "step": 189
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6235249334397138,
      "learning_rate": 1.6101694915254237e-06,
      "loss": 0.6144,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3399635605596842,
      "learning_rate": 1.61864406779661e-06,
      "loss": 0.6902,
      "step": 191
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9262852886186115,
      "learning_rate": 1.6271186440677967e-06,
      "loss": 0.7442,
      "step": 192
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8481980246517118,
      "learning_rate": 1.635593220338983e-06,
      "loss": 0.6338,
      "step": 193
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1255447307355642,
      "learning_rate": 1.6440677966101694e-06,
      "loss": 0.7652,
      "step": 194
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8908038467928799,
      "learning_rate": 1.6525423728813558e-06,
      "loss": 0.6901,
      "step": 195
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8222134161165471,
      "learning_rate": 1.6610169491525424e-06,
      "loss": 0.6373,
      "step": 196
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7326869515693809,
      "learning_rate": 1.6694915254237288e-06,
      "loss": 0.615,
      "step": 197
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7054126369518261,
      "learning_rate": 1.6779661016949152e-06,
      "loss": 0.5617,
      "step": 198
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2784394598959574,
      "learning_rate": 1.6864406779661016e-06,
      "loss": 0.65,
      "step": 199
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8791275870588131,
      "learning_rate": 1.694915254237288e-06,
      "loss": 0.6436,
      "step": 200
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9542073313260363,
      "learning_rate": 1.7033898305084743e-06,
      "loss": 0.7317,
      "step": 201
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9229840191225319,
      "learning_rate": 1.7118644067796611e-06,
      "loss": 0.6364,
      "step": 202
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9622714461397894,
      "learning_rate": 1.7203389830508475e-06,
      "loss": 0.6868,
      "step": 203
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5899307448507893,
      "learning_rate": 1.728813559322034e-06,
      "loss": 0.5548,
      "step": 204
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6242083877762877,
      "learning_rate": 1.7372881355932203e-06,
      "loss": 0.5873,
      "step": 205
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.692201534933359,
      "learning_rate": 1.7457627118644067e-06,
      "loss": 0.5528,
      "step": 206
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6975976844947165,
      "learning_rate": 1.754237288135593e-06,
      "loss": 0.61,
      "step": 207
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.896528926808541,
      "learning_rate": 1.7627118644067794e-06,
      "loss": 0.6469,
      "step": 208
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.173139683805913,
      "learning_rate": 1.771186440677966e-06,
      "loss": 0.6983,
      "step": 209
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7057066492934985,
      "learning_rate": 1.7796610169491524e-06,
      "loss": 0.562,
      "step": 210
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6674353468969882,
      "learning_rate": 1.7881355932203388e-06,
      "loss": 0.6048,
      "step": 211
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9005066283379761,
      "learning_rate": 1.7966101694915254e-06,
      "loss": 0.6698,
      "step": 212
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7900697075878133,
      "learning_rate": 1.8050847457627118e-06,
      "loss": 0.7254,
      "step": 213
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6709247827891486,
      "learning_rate": 1.8135593220338982e-06,
      "loss": 0.5656,
      "step": 214
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7225734968984099,
      "learning_rate": 1.8220338983050848e-06,
      "loss": 0.6251,
      "step": 215
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7523079531355961,
      "learning_rate": 1.8305084745762712e-06,
      "loss": 0.6351,
      "step": 216
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9032931881053967,
      "learning_rate": 1.8389830508474575e-06,
      "loss": 0.6848,
      "step": 217
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7338938725122566,
      "learning_rate": 1.847457627118644e-06,
      "loss": 0.5706,
      "step": 218
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8021844151100668,
      "learning_rate": 1.8559322033898303e-06,
      "loss": 0.5991,
      "step": 219
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7449679398020179,
      "learning_rate": 1.8644067796610167e-06,
      "loss": 0.6129,
      "step": 220
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8654100394073853,
      "learning_rate": 1.8728813559322033e-06,
      "loss": 0.653,
      "step": 221
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7538362027872093,
      "learning_rate": 1.8813559322033899e-06,
      "loss": 0.6084,
      "step": 222
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.728231635158702,
      "learning_rate": 1.8898305084745763e-06,
      "loss": 0.6296,
      "step": 223
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6250014532731726,
      "learning_rate": 1.8983050847457627e-06,
      "loss": 0.6077,
      "step": 224
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6077670298424928,
      "learning_rate": 1.906779661016949e-06,
      "loss": 0.6227,
      "step": 225
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5565700456470007,
      "learning_rate": 1.9152542372881356e-06,
      "loss": 0.4893,
      "step": 226
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6178364207463651,
      "learning_rate": 1.923728813559322e-06,
      "loss": 0.5292,
      "step": 227
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8163350391093931,
      "learning_rate": 1.9322033898305084e-06,
      "loss": 0.6135,
      "step": 228
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6997941270332168,
      "learning_rate": 1.9406779661016946e-06,
      "loss": 0.6565,
      "step": 229
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.100103558563823,
      "learning_rate": 1.949152542372881e-06,
      "loss": 0.7274,
      "step": 230
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9737007994959831,
      "learning_rate": 1.9576271186440678e-06,
      "loss": 0.7127,
      "step": 231
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8818286457797845,
      "learning_rate": 1.9661016949152544e-06,
      "loss": 0.5957,
      "step": 232
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7042373402430965,
      "learning_rate": 1.9745762711864405e-06,
      "loss": 0.6315,
      "step": 233
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8290814391329554,
      "learning_rate": 1.983050847457627e-06,
      "loss": 0.6889,
      "step": 234
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0850347895148524,
      "learning_rate": 1.9915254237288133e-06,
      "loss": 0.6217,
      "step": 235
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7561140413632217,
      "learning_rate": 2e-06,
      "loss": 0.5888,
      "step": 236
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8625860213461869,
      "learning_rate": 1.9999999148552087e-06,
      "loss": 0.7218,
      "step": 237
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8166646229523257,
      "learning_rate": 1.99999965942085e-06,
      "loss": 0.6284,
      "step": 238
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6551901369601575,
      "learning_rate": 1.9999992336969675e-06,
      "loss": 0.5826,
      "step": 239
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9304362470364699,
      "learning_rate": 1.9999986376836335e-06,
      "loss": 0.617,
      "step": 240
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0359386542519404,
      "learning_rate": 1.9999978713809487e-06,
      "loss": 0.693,
      "step": 241
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6159971874590597,
      "learning_rate": 1.999996934789045e-06,
      "loss": 0.5893,
      "step": 242
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7307011844542025,
      "learning_rate": 1.999995827908081e-06,
      "loss": 0.6368,
      "step": 243
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9208318783383151,
      "learning_rate": 1.9999945507382454e-06,
      "loss": 0.7098,
      "step": 244
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9179643292882264,
      "learning_rate": 1.9999931032797555e-06,
      "loss": 0.6841,
      "step": 245
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7855969182168681,
      "learning_rate": 1.999991485532858e-06,
      "loss": 0.5896,
      "step": 246
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8809749072365072,
      "learning_rate": 1.9999896974978285e-06,
      "loss": 0.6883,
      "step": 247
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9665855957156472,
      "learning_rate": 1.999987739174971e-06,
      "loss": 0.762,
      "step": 248
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7802559405751034,
      "learning_rate": 1.99998561056462e-06,
      "loss": 0.6096,
      "step": 249
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6257717038750468,
      "learning_rate": 1.9999833116671367e-06,
      "loss": 0.5582,
      "step": 250
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6304227616063485,
      "learning_rate": 1.9999808424829135e-06,
      "loss": 0.5243,
      "step": 251
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7272651060843149,
      "learning_rate": 1.9999782030123702e-06,
      "loss": 0.6365,
      "step": 252
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6841305216463638,
      "learning_rate": 1.9999753932559567e-06,
      "loss": 0.6012,
      "step": 253
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8512458896259056,
      "learning_rate": 1.9999724132141516e-06,
      "loss": 0.6715,
      "step": 254
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7166163865557734,
      "learning_rate": 1.999969262887462e-06,
      "loss": 0.5863,
      "step": 255
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6931728423039321,
      "learning_rate": 1.9999659422764248e-06,
      "loss": 0.5759,
      "step": 256
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8867546810799449,
      "learning_rate": 1.999962451381605e-06,
      "loss": 0.6563,
      "step": 257
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9755805820638891,
      "learning_rate": 1.999958790203597e-06,
      "loss": 0.6736,
      "step": 258
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.748505813485469,
      "learning_rate": 1.999954958743025e-06,
      "loss": 0.7157,
      "step": 259
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8280797001119953,
      "learning_rate": 1.9999509570005406e-06,
      "loss": 0.6616,
      "step": 260
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9041278479566665,
      "learning_rate": 1.999946784976826e-06,
      "loss": 0.7529,
      "step": 261
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9158660580121655,
      "learning_rate": 1.999942442672591e-06,
      "loss": 0.655,
      "step": 262
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6594797430670454,
      "learning_rate": 1.9999379300885754e-06,
      "loss": 0.5778,
      "step": 263
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8084013065232527,
      "learning_rate": 1.999933247225548e-06,
      "loss": 0.6182,
      "step": 264
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7741929925462311,
      "learning_rate": 1.999928394084305e-06,
      "loss": 0.5902,
      "step": 265
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6113369204995305,
      "learning_rate": 1.9999233706656743e-06,
      "loss": 0.5736,
      "step": 266
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8158108757664658,
      "learning_rate": 1.999918176970511e-06,
      "loss": 0.6617,
      "step": 267
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7170955866836686,
      "learning_rate": 1.9999128129996984e-06,
      "loss": 0.6386,
      "step": 268
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6456584365381373,
      "learning_rate": 1.999907278754151e-06,
      "loss": 0.5859,
      "step": 269
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6257160586247476,
      "learning_rate": 1.999901574234811e-06,
      "loss": 0.5918,
      "step": 270
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7624562451180734,
      "learning_rate": 1.99989569944265e-06,
      "loss": 0.6811,
      "step": 271
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2449455730842223,
      "learning_rate": 1.9998896543786683e-06,
      "loss": 0.6845,
      "step": 272
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0122328677163068,
      "learning_rate": 1.999883439043895e-06,
      "loss": 0.6639,
      "step": 273
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6669394614726304,
      "learning_rate": 1.999877053439389e-06,
      "loss": 0.6313,
      "step": 274
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8928051159400339,
      "learning_rate": 1.999870497566237e-06,
      "loss": 0.6152,
      "step": 275
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9021328838958339,
      "learning_rate": 1.9998637714255562e-06,
      "loss": 0.6512,
      "step": 276
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9101399433152626,
      "learning_rate": 1.9998568750184916e-06,
      "loss": 0.6087,
      "step": 277
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7449958707856078,
      "learning_rate": 1.9998498083462172e-06,
      "loss": 0.5713,
      "step": 278
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7639084341476106,
      "learning_rate": 1.9998425714099375e-06,
      "loss": 0.6,
      "step": 279
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7279828890510773,
      "learning_rate": 1.9998351642108836e-06,
      "loss": 0.5928,
      "step": 280
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.815161596122287,
      "learning_rate": 1.9998275867503177e-06,
      "loss": 0.5829,
      "step": 281
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6351697163457766,
      "learning_rate": 1.9998198390295297e-06,
      "loss": 0.6166,
      "step": 282
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6783285677768011,
      "learning_rate": 1.9998119210498397e-06,
      "loss": 0.5675,
      "step": 283
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7119826568632397,
      "learning_rate": 1.999803832812595e-06,
      "loss": 0.5777,
      "step": 284
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9578081686428986,
      "learning_rate": 1.9997955743191736e-06,
      "loss": 0.6802,
      "step": 285
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.970271384687729,
      "learning_rate": 1.9997871455709817e-06,
      "loss": 0.6425,
      "step": 286
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8391588999241391,
      "learning_rate": 1.9997785465694545e-06,
      "loss": 0.6482,
      "step": 287
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7559879265309987,
      "learning_rate": 1.9997697773160566e-06,
      "loss": 0.6013,
      "step": 288
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8002748589128438,
      "learning_rate": 1.999760837812281e-06,
      "loss": 0.6673,
      "step": 289
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6300829015929642,
      "learning_rate": 1.9997517280596503e-06,
      "loss": 0.5778,
      "step": 290
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7266115572516427,
      "learning_rate": 1.9997424480597157e-06,
      "loss": 0.6705,
      "step": 291
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5856628988568351,
      "learning_rate": 1.999732997814057e-06,
      "loss": 0.5574,
      "step": 292
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6912973568761449,
      "learning_rate": 1.9997233773242845e-06,
      "loss": 0.6121,
      "step": 293
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6981284546199165,
      "learning_rate": 1.9997135865920354e-06,
      "loss": 0.6535,
      "step": 294
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7437958073516342,
      "learning_rate": 1.999703625618978e-06,
      "loss": 0.5956,
      "step": 295
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7987457717931304,
      "learning_rate": 1.9996934944068075e-06,
      "loss": 0.6469,
      "step": 296
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8670990352336557,
      "learning_rate": 1.99968319295725e-06,
      "loss": 0.6559,
      "step": 297
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8145183156889081,
      "learning_rate": 1.9996727212720587e-06,
      "loss": 0.7407,
      "step": 298
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6949355368150785,
      "learning_rate": 1.999662079353018e-06,
      "loss": 0.6135,
      "step": 299
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7637559389237999,
      "learning_rate": 1.9996512672019395e-06,
      "loss": 0.6037,
      "step": 300
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7101951019587592,
      "learning_rate": 1.9996402848206647e-06,
      "loss": 0.569,
      "step": 301
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8554522757524007,
      "learning_rate": 1.9996291322110633e-06,
      "loss": 0.6624,
      "step": 302
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6989714697190335,
      "learning_rate": 1.999617809375035e-06,
      "loss": 0.5813,
      "step": 303
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9230558846204552,
      "learning_rate": 1.9996063163145075e-06,
      "loss": 0.6439,
      "step": 304
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7548869346421067,
      "learning_rate": 1.9995946530314383e-06,
      "loss": 0.6058,
      "step": 305
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6974381289402516,
      "learning_rate": 1.9995828195278133e-06,
      "loss": 0.5988,
      "step": 306
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9482477842978514,
      "learning_rate": 1.9995708158056473e-06,
      "loss": 0.753,
      "step": 307
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7553292869751534,
      "learning_rate": 1.9995586418669853e-06,
      "loss": 0.6253,
      "step": 308
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8287173499925506,
      "learning_rate": 1.9995462977139e-06,
      "loss": 0.6017,
      "step": 309
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8928406702802455,
      "learning_rate": 1.999533783348493e-06,
      "loss": 0.7325,
      "step": 310
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6410943924428769,
      "learning_rate": 1.999521098772896e-06,
      "loss": 0.5277,
      "step": 311
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.744168490567295,
      "learning_rate": 1.999508243989269e-06,
      "loss": 0.5779,
      "step": 312
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7628269145518654,
      "learning_rate": 1.9994952189998007e-06,
      "loss": 0.6934,
      "step": 313
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6178141324596883,
      "learning_rate": 1.9994820238067093e-06,
      "loss": 0.6109,
      "step": 314
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9101328667244376,
      "learning_rate": 1.9994686584122417e-06,
      "loss": 0.7364,
      "step": 315
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7133317665538189,
      "learning_rate": 1.999455122818674e-06,
      "loss": 0.6025,
      "step": 316
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.646645637586129,
      "learning_rate": 1.999441417028311e-06,
      "loss": 0.5914,
      "step": 317
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8781012727347236,
      "learning_rate": 1.999427541043487e-06,
      "loss": 0.7113,
      "step": 318
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7833710064149482,
      "learning_rate": 1.9994134948665646e-06,
      "loss": 0.6952,
      "step": 319
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7751873987941748,
      "learning_rate": 1.9993992784999358e-06,
      "loss": 0.6617,
      "step": 320
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6007609206877843,
      "learning_rate": 1.9993848919460212e-06,
      "loss": 0.5767,
      "step": 321
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8028915776237311,
      "learning_rate": 1.9993703352072714e-06,
      "loss": 0.5928,
      "step": 322
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9478064106321482,
      "learning_rate": 1.999355608286165e-06,
      "loss": 0.6444,
      "step": 323
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9097189962760596,
      "learning_rate": 1.9993407111852097e-06,
      "loss": 0.7153,
      "step": 324
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6934288937132419,
      "learning_rate": 1.9993256439069423e-06,
      "loss": 0.583,
      "step": 325
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6777138899758901,
      "learning_rate": 1.9993104064539287e-06,
      "loss": 0.5626,
      "step": 326
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6507862813625521,
      "learning_rate": 1.9992949988287636e-06,
      "loss": 0.6106,
      "step": 327
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7488861670223014,
      "learning_rate": 1.9992794210340705e-06,
      "loss": 0.6585,
      "step": 328
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7219018574477484,
      "learning_rate": 1.9992636730725025e-06,
      "loss": 0.5834,
      "step": 329
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6736314658646795,
      "learning_rate": 1.9992477549467415e-06,
      "loss": 0.5688,
      "step": 330
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7023867874955008,
      "learning_rate": 1.999231666659498e-06,
      "loss": 0.5607,
      "step": 331
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8030130857913237,
      "learning_rate": 1.999215408213511e-06,
      "loss": 0.5323,
      "step": 332
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9262149514699182,
      "learning_rate": 1.9991989796115502e-06,
      "loss": 0.6496,
      "step": 333
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8663654491781175,
      "learning_rate": 1.9991823808564125e-06,
      "loss": 0.6494,
      "step": 334
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0026044603291644,
      "learning_rate": 1.9991656119509246e-06,
      "loss": 0.7175,
      "step": 335
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.902878487920715,
      "learning_rate": 1.9991486728979426e-06,
      "loss": 0.6908,
      "step": 336
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8437281463568033,
      "learning_rate": 1.9991315637003504e-06,
      "loss": 0.7043,
      "step": 337
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.875589412073372,
      "learning_rate": 1.9991142843610624e-06,
      "loss": 0.6696,
      "step": 338
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0851819180964717,
      "learning_rate": 1.9990968348830198e-06,
      "loss": 0.709,
      "step": 339
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9044385211502225,
      "learning_rate": 1.999079215269195e-06,
      "loss": 0.6506,
      "step": 340
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7809235906964044,
      "learning_rate": 1.9990614255225878e-06,
      "loss": 0.608,
      "step": 341
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6095029885370832,
      "learning_rate": 1.9990434656462284e-06,
      "loss": 0.5945,
      "step": 342
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9220214106594393,
      "learning_rate": 1.9990253356431745e-06,
      "loss": 0.6963,
      "step": 343
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9001375257853677,
      "learning_rate": 1.999007035516514e-06,
      "loss": 0.6403,
      "step": 344
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8459662460180508,
      "learning_rate": 1.998988565269363e-06,
      "loss": 0.6638,
      "step": 345
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6129945305513942,
      "learning_rate": 1.9989699249048662e-06,
      "loss": 0.5613,
      "step": 346
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8336338793790357,
      "learning_rate": 1.9989511144261988e-06,
      "loss": 0.6887,
      "step": 347
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9154432557238324,
      "learning_rate": 1.9989321338365636e-06,
      "loss": 0.6035,
      "step": 348
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7792883337104146,
      "learning_rate": 1.9989129831391927e-06,
      "loss": 0.6286,
      "step": 349
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8489696176930003,
      "learning_rate": 1.9988936623373475e-06,
      "loss": 0.6918,
      "step": 350
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7414159012919416,
      "learning_rate": 1.9988741714343175e-06,
      "loss": 0.6978,
      "step": 351
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7886948019403772,
      "learning_rate": 1.998854510433423e-06,
      "loss": 0.6262,
      "step": 352
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8428040774724056,
      "learning_rate": 1.998834679338011e-06,
      "loss": 0.6843,
      "step": 353
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7611416632666679,
      "learning_rate": 1.9988146781514587e-06,
      "loss": 0.6253,
      "step": 354
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9635616594157547,
      "learning_rate": 1.998794506877173e-06,
      "loss": 0.5757,
      "step": 355
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.55843438226619,
      "learning_rate": 1.998774165518587e-06,
      "loss": 0.7075,
      "step": 356
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7477539747945171,
      "learning_rate": 1.9987536540791667e-06,
      "loss": 0.5852,
      "step": 357
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6554910573429403,
      "learning_rate": 1.998732972562404e-06,
      "loss": 0.5484,
      "step": 358
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8127809210005906,
      "learning_rate": 1.9987121209718203e-06,
      "loss": 0.534,
      "step": 359
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7529362415409898,
      "learning_rate": 1.9986910993109674e-06,
      "loss": 0.5677,
      "step": 360
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.853655342997557,
      "learning_rate": 1.9986699075834242e-06,
      "loss": 0.6316,
      "step": 361
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7094755895769688,
      "learning_rate": 1.9986485457928e-06,
      "loss": 0.6035,
      "step": 362
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.960450051591529,
      "learning_rate": 1.9986270139427323e-06,
      "loss": 0.6422,
      "step": 363
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6725458366668882,
      "learning_rate": 1.998605312036888e-06,
      "loss": 0.5842,
      "step": 364
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6659406004405898,
      "learning_rate": 1.998583440078962e-06,
      "loss": 0.5732,
      "step": 365
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9920451039700708,
      "learning_rate": 1.9985613980726794e-06,
      "loss": 0.6518,
      "step": 366
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1236637195135857,
      "learning_rate": 1.998539186021794e-06,
      "loss": 0.7254,
      "step": 367
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6801252339305343,
      "learning_rate": 1.998516803930087e-06,
      "loss": 0.5683,
      "step": 368
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8331883060473511,
      "learning_rate": 1.998494251801372e-06,
      "loss": 0.6656,
      "step": 369
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8255426690603981,
      "learning_rate": 1.9984715296394874e-06,
      "loss": 0.6337,
      "step": 370
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7515752677935769,
      "learning_rate": 1.9984486374483036e-06,
      "loss": 0.6105,
      "step": 371
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8471385911343636,
      "learning_rate": 1.9984255752317185e-06,
      "loss": 0.5816,
      "step": 372
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8592903638985805,
      "learning_rate": 1.9984023429936595e-06,
      "loss": 0.6064,
      "step": 373
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8795317440284671,
      "learning_rate": 1.9983789407380825e-06,
      "loss": 0.624,
      "step": 374
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7483281469460907,
      "learning_rate": 1.998355368468973e-06,
      "loss": 0.5929,
      "step": 375
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6753107671083313,
      "learning_rate": 1.9983316261903454e-06,
      "loss": 0.5904,
      "step": 376
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7599112145704452,
      "learning_rate": 1.998307713906242e-06,
      "loss": 0.6541,
      "step": 377
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9125710764193486,
      "learning_rate": 1.998283631620736e-06,
      "loss": 0.6487,
      "step": 378
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8136045737757747,
      "learning_rate": 1.998259379337927e-06,
      "loss": 0.6288,
      "step": 379
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.602431024946439,
      "learning_rate": 1.998234957061945e-06,
      "loss": 0.5567,
      "step": 380
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8766381742741123,
      "learning_rate": 1.99821036479695e-06,
      "loss": 0.7868,
      "step": 381
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8995830783826105,
      "learning_rate": 1.998185602547129e-06,
      "loss": 0.5891,
      "step": 382
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6788697854745167,
      "learning_rate": 1.998160670316699e-06,
      "loss": 0.5666,
      "step": 383
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.796738389057112,
      "learning_rate": 1.9981355681099053e-06,
      "loss": 0.7052,
      "step": 384
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7518308815659634,
      "learning_rate": 1.998110295931023e-06,
      "loss": 0.6227,
      "step": 385
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2820655771792429,
      "learning_rate": 1.9980848537843555e-06,
      "loss": 0.6452,
      "step": 386
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9415761880118673,
      "learning_rate": 1.9980592416742355e-06,
      "loss": 0.7437,
      "step": 387
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6111766231882224,
      "learning_rate": 1.9980334596050247e-06,
      "loss": 0.5343,
      "step": 388
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.650041383753845,
      "learning_rate": 1.9980075075811125e-06,
      "loss": 0.5456,
      "step": 389
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9342830809689981,
      "learning_rate": 1.9979813856069193e-06,
      "loss": 0.6755,
      "step": 390
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.830498412942376,
      "learning_rate": 1.997955093686893e-06,
      "loss": 0.6308,
      "step": 391
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8414585563356384,
      "learning_rate": 1.9979286318255104e-06,
      "loss": 0.6603,
      "step": 392
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7417964186877467,
      "learning_rate": 1.9979020000272787e-06,
      "loss": 0.6582,
      "step": 393
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6350388014950262,
      "learning_rate": 1.997875198296732e-06,
      "loss": 0.6192,
      "step": 394
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.727422189560833,
      "learning_rate": 1.9978482266384354e-06,
      "loss": 0.6194,
      "step": 395
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7538397104096738,
      "learning_rate": 1.997821085056981e-06,
      "loss": 0.6172,
      "step": 396
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8555191334261188,
      "learning_rate": 1.997793773556991e-06,
      "loss": 0.657,
      "step": 397
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7205520909245473,
      "learning_rate": 1.9977662921431167e-06,
      "loss": 0.6061,
      "step": 398
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7538479516126748,
      "learning_rate": 1.997738640820037e-06,
      "loss": 0.5984,
      "step": 399
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8029526196745616,
      "learning_rate": 1.9977108195924613e-06,
      "loss": 0.6469,
      "step": 400
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.12361740607408,
      "learning_rate": 1.9976828284651273e-06,
      "loss": 0.6364,
      "step": 401
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9998887207797958,
      "learning_rate": 1.9976546674428012e-06,
      "loss": 0.6408,
      "step": 402
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6331078684674015,
      "learning_rate": 1.9976263365302786e-06,
      "loss": 0.5904,
      "step": 403
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8439407709010107,
      "learning_rate": 1.9975978357323845e-06,
      "loss": 0.6771,
      "step": 404
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7898180393420955,
      "learning_rate": 1.9975691650539712e-06,
      "loss": 0.6306,
      "step": 405
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6638433144801531,
      "learning_rate": 1.9975403244999223e-06,
      "loss": 0.5796,
      "step": 406
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9641760611378809,
      "learning_rate": 1.9975113140751484e-06,
      "loss": 0.6647,
      "step": 407
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6976555198536526,
      "learning_rate": 1.9974821337845895e-06,
      "loss": 0.5524,
      "step": 408
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7870931942980063,
      "learning_rate": 1.9974527836332146e-06,
      "loss": 0.6556,
      "step": 409
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8607140365717414,
      "learning_rate": 1.9974232636260225e-06,
      "loss": 0.6384,
      "step": 410
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6205235370420497,
      "learning_rate": 1.9973935737680395e-06,
      "loss": 0.5491,
      "step": 411
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8029058689101558,
      "learning_rate": 1.997363714064322e-06,
      "loss": 0.5687,
      "step": 412
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8565690131809278,
      "learning_rate": 1.997333684519954e-06,
      "loss": 0.6499,
      "step": 413
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.872306166184901,
      "learning_rate": 1.99730348514005e-06,
      "loss": 0.6984,
      "step": 414
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7058095527728597,
      "learning_rate": 1.997273115929752e-06,
      "loss": 0.5688,
      "step": 415
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7386514309222343,
      "learning_rate": 1.997242576894232e-06,
      "loss": 0.5799,
      "step": 416
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7273716879003697,
      "learning_rate": 1.9972118680386903e-06,
      "loss": 0.6194,
      "step": 417
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7342382312263401,
      "learning_rate": 1.9971809893683565e-06,
      "loss": 0.5739,
      "step": 418
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7275605551295318,
      "learning_rate": 1.997149940888489e-06,
      "loss": 0.5337,
      "step": 419
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7393495700873955,
      "learning_rate": 1.9971187226043744e-06,
      "loss": 0.655,
      "step": 420
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8097193901619947,
      "learning_rate": 1.997087334521329e-06,
      "loss": 0.6812,
      "step": 421
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1007582266099434,
      "learning_rate": 1.997055776644699e-06,
      "loss": 0.7001,
      "step": 422
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6115835360113718,
      "learning_rate": 1.9970240489798567e-06,
      "loss": 0.5459,
      "step": 423
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6519049847636927,
      "learning_rate": 1.9969921515322057e-06,
      "loss": 0.54,
      "step": 424
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6863780538412851,
      "learning_rate": 1.9969600843071783e-06,
      "loss": 0.6256,
      "step": 425
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6162443703739764,
      "learning_rate": 1.996927847310235e-06,
      "loss": 0.5844,
      "step": 426
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7886252374418712,
      "learning_rate": 1.996895440546865e-06,
      "loss": 0.5857,
      "step": 427
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8084761217273592,
      "learning_rate": 1.996862864022587e-06,
      "loss": 0.6244,
      "step": 428
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7488517497681477,
      "learning_rate": 1.9968301177429482e-06,
      "loss": 0.6193,
      "step": 429
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7155175447410872,
      "learning_rate": 1.9967972017135257e-06,
      "loss": 0.6058,
      "step": 430
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9753882013330053,
      "learning_rate": 1.996764115939924e-06,
      "loss": 0.6374,
      "step": 431
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6084892181187914,
      "learning_rate": 1.996730860427778e-06,
      "loss": 0.5586,
      "step": 432
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9057634887775081,
      "learning_rate": 1.99669743518275e-06,
      "loss": 0.6766,
      "step": 433
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9674983190091517,
      "learning_rate": 1.9966638402105326e-06,
      "loss": 0.5723,
      "step": 434
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8598216942434391,
      "learning_rate": 1.996630075516846e-06,
      "loss": 0.6522,
      "step": 435
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7187819145132707,
      "learning_rate": 1.9965961411074406e-06,
      "loss": 0.5842,
      "step": 436
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.662830114474183,
      "learning_rate": 1.996562036988095e-06,
      "loss": 0.5486,
      "step": 437
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7206508711581904,
      "learning_rate": 1.9965277631646156e-06,
      "loss": 0.5915,
      "step": 438
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9050266889353334,
      "learning_rate": 1.996493319642841e-06,
      "loss": 0.6914,
      "step": 439
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.644419177040819,
      "learning_rate": 1.9964587064286353e-06,
      "loss": 0.6161,
      "step": 440
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8980599632794635,
      "learning_rate": 1.9964239235278922e-06,
      "loss": 0.6016,
      "step": 441
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8132854686340011,
      "learning_rate": 1.9963889709465365e-06,
      "loss": 0.6598,
      "step": 442
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7718078124800741,
      "learning_rate": 1.9963538486905186e-06,
      "loss": 0.641,
      "step": 443
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6522656866198393,
      "learning_rate": 1.996318556765821e-06,
      "loss": 0.5915,
      "step": 444
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8683128077530391,
      "learning_rate": 1.996283095178452e-06,
      "loss": 0.5898,
      "step": 445
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8293878684604463,
      "learning_rate": 1.9962474639344514e-06,
      "loss": 0.6358,
      "step": 446
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.773598182112365,
      "learning_rate": 1.9962116630398865e-06,
      "loss": 0.5926,
      "step": 447
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6860253798573333,
      "learning_rate": 1.996175692500854e-06,
      "loss": 0.5808,
      "step": 448
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9029446960124653,
      "learning_rate": 1.996139552323479e-06,
      "loss": 0.6698,
      "step": 449
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6595919887257407,
      "learning_rate": 1.996103242513916e-06,
      "loss": 0.5714,
      "step": 450
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6264165343534965,
      "learning_rate": 1.996066763078348e-06,
      "loss": 0.5299,
      "step": 451
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7175733174312324,
      "learning_rate": 1.9960301140229874e-06,
      "loss": 0.5333,
      "step": 452
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.873568294212464,
      "learning_rate": 1.995993295354075e-06,
      "loss": 0.6863,
      "step": 453
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8824102340737074,
      "learning_rate": 1.9959563070778805e-06,
      "loss": 0.6326,
      "step": 454
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6195609281265144,
      "learning_rate": 1.9959191492007025e-06,
      "loss": 0.6096,
      "step": 455
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8051827517846066,
      "learning_rate": 1.9958818217288687e-06,
      "loss": 0.6429,
      "step": 456
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7602985880173697,
      "learning_rate": 1.995844324668736e-06,
      "loss": 0.543,
      "step": 457
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6994694603792948,
      "learning_rate": 1.9958066580266894e-06,
      "loss": 0.649,
      "step": 458
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.769562412206393,
      "learning_rate": 1.9957688218091434e-06,
      "loss": 0.5918,
      "step": 459
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8607009036612931,
      "learning_rate": 1.9957308160225406e-06,
      "loss": 0.6612,
      "step": 460
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0990021737889297,
      "learning_rate": 1.9956926406733533e-06,
      "loss": 0.6467,
      "step": 461
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7054435098964321,
      "learning_rate": 1.9956542957680824e-06,
      "loss": 0.6062,
      "step": 462
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7565580420791708,
      "learning_rate": 1.9956157813132573e-06,
      "loss": 0.6155,
      "step": 463
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8254829621953097,
      "learning_rate": 1.995577097315437e-06,
      "loss": 0.6318,
      "step": 464
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8376409535744992,
      "learning_rate": 1.9955382437812093e-06,
      "loss": 0.6111,
      "step": 465
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6324610420016746,
      "learning_rate": 1.9954992207171895e-06,
      "loss": 0.5611,
      "step": 466
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7741536575285715,
      "learning_rate": 1.995460028130024e-06,
      "loss": 0.6892,
      "step": 467
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7010385432251874,
      "learning_rate": 1.995420666026386e-06,
      "loss": 0.5707,
      "step": 468
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8179419862545474,
      "learning_rate": 1.995381134412979e-06,
      "loss": 0.6544,
      "step": 469
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8209329876527347,
      "learning_rate": 1.9953414332965344e-06,
      "loss": 0.6304,
      "step": 470
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6602453497344377,
      "learning_rate": 1.9953015626838132e-06,
      "loss": 0.5876,
      "step": 471
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.700014033721996,
      "learning_rate": 1.995261522581605e-06,
      "loss": 0.6623,
      "step": 472
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7137584730827138,
      "learning_rate": 1.995221312996728e-06,
      "loss": 0.6087,
      "step": 473
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8008197649868769,
      "learning_rate": 1.995180933936029e-06,
      "loss": 0.6114,
      "step": 474
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7344816607657141,
      "learning_rate": 1.9951403854063857e-06,
      "loss": 0.5712,
      "step": 475
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0962893537156433,
      "learning_rate": 1.9950996674147013e-06,
      "loss": 0.6992,
      "step": 476
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7188700522057578,
      "learning_rate": 1.9950587799679107e-06,
      "loss": 0.5839,
      "step": 477
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9062586258796633,
      "learning_rate": 1.9950177230729765e-06,
      "loss": 0.6295,
      "step": 478
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9130385854423889,
      "learning_rate": 1.9949764967368897e-06,
      "loss": 0.6229,
      "step": 479
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8214759920337584,
      "learning_rate": 1.994935100966671e-06,
      "loss": 0.6747,
      "step": 480
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7628341427462815,
      "learning_rate": 1.99489353576937e-06,
      "loss": 0.6667,
      "step": 481
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5913864957882137,
      "learning_rate": 1.9948518011520646e-06,
      "loss": 0.5501,
      "step": 482
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8607712987666438,
      "learning_rate": 1.9948098971218615e-06,
      "loss": 0.6423,
      "step": 483
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.854437823294419,
      "learning_rate": 1.994767823685897e-06,
      "loss": 0.6678,
      "step": 484
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6329898822777572,
      "learning_rate": 1.9947255808513356e-06,
      "loss": 0.6115,
      "step": 485
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7071754119219292,
      "learning_rate": 1.994683168625371e-06,
      "loss": 0.5803,
      "step": 486
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8645829146535393,
      "learning_rate": 1.994640587015225e-06,
      "loss": 0.6927,
      "step": 487
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7001108300136843,
      "learning_rate": 1.994597836028149e-06,
      "loss": 0.6013,
      "step": 488
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7276133918104546,
      "learning_rate": 1.994554915671423e-06,
      "loss": 0.5975,
      "step": 489
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6693202431399882,
      "learning_rate": 1.9945118259523566e-06,
      "loss": 0.5793,
      "step": 490
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.886659854560941,
      "learning_rate": 1.9944685668782866e-06,
      "loss": 0.6687,
      "step": 491
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7786195211250581,
      "learning_rate": 1.9944251384565802e-06,
      "loss": 0.6558,
      "step": 492
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.755450925655243,
      "learning_rate": 1.9943815406946325e-06,
      "loss": 0.6361,
      "step": 493
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7175708601429536,
      "learning_rate": 1.9943377735998677e-06,
      "loss": 0.6066,
      "step": 494
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6737474261183992,
      "learning_rate": 1.9942938371797392e-06,
      "loss": 0.5348,
      "step": 495
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6726366108513238,
      "learning_rate": 1.9942497314417285e-06,
      "loss": 0.5691,
      "step": 496
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7766121790255995,
      "learning_rate": 1.9942054563933466e-06,
      "loss": 0.6386,
      "step": 497
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9466418524135742,
      "learning_rate": 1.994161012042133e-06,
      "loss": 0.6884,
      "step": 498
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6064936698573115,
      "learning_rate": 1.9941163983956564e-06,
      "loss": 0.5475,
      "step": 499
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.729347407140469,
      "learning_rate": 1.9940716154615137e-06,
      "loss": 0.5673,
      "step": 500
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7925009390349311,
      "learning_rate": 1.994026663247331e-06,
      "loss": 0.5988,
      "step": 501
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8799582084510409,
      "learning_rate": 1.9939815417607632e-06,
      "loss": 0.6289,
      "step": 502
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.630126424146249,
      "learning_rate": 1.993936251009494e-06,
      "loss": 0.6375,
      "step": 503
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7668011395566893,
      "learning_rate": 1.993890791001236e-06,
      "loss": 0.6055,
      "step": 504
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9127365570320746,
      "learning_rate": 1.9938451617437306e-06,
      "loss": 0.7055,
      "step": 505
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8686573791774603,
      "learning_rate": 1.9937993632447483e-06,
      "loss": 0.6904,
      "step": 506
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7641158662166746,
      "learning_rate": 1.9937533955120872e-06,
      "loss": 0.5906,
      "step": 507
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7095443133999298,
      "learning_rate": 1.9937072585535758e-06,
      "loss": 0.5743,
      "step": 508
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8136751763390865,
      "learning_rate": 1.993660952377071e-06,
      "loss": 0.5829,
      "step": 509
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8441615515146447,
      "learning_rate": 1.9936144769904575e-06,
      "loss": 0.6955,
      "step": 510
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.69454392770871,
      "learning_rate": 1.9935678324016504e-06,
      "loss": 0.5893,
      "step": 511
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9138403095149903,
      "learning_rate": 1.993521018618592e-06,
      "loss": 0.6462,
      "step": 512
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6488234500048476,
      "learning_rate": 1.9934740356492546e-06,
      "loss": 0.5746,
      "step": 513
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7232167794691058,
      "learning_rate": 1.993426883501639e-06,
      "loss": 0.5837,
      "step": 514
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8914086635543118,
      "learning_rate": 1.9933795621837744e-06,
      "loss": 0.6579,
      "step": 515
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7827369727963219,
      "learning_rate": 1.9933320717037192e-06,
      "loss": 0.568,
      "step": 516
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8330877192635204,
      "learning_rate": 1.993284412069561e-06,
      "loss": 0.6593,
      "step": 517
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9670235134915235,
      "learning_rate": 1.9932365832894156e-06,
      "loss": 0.6997,
      "step": 518
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8633228944132165,
      "learning_rate": 1.993188585371427e-06,
      "loss": 0.5734,
      "step": 519
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6430274093038407,
      "learning_rate": 1.9931404183237693e-06,
      "loss": 0.6233,
      "step": 520
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6957965916265959,
      "learning_rate": 1.9930920821546452e-06,
      "loss": 0.5427,
      "step": 521
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.662545927951576,
      "learning_rate": 1.993043576872285e-06,
      "loss": 0.6064,
      "step": 522
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6526612293277911,
      "learning_rate": 1.99299490248495e-06,
      "loss": 0.5615,
      "step": 523
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5086238833618104,
      "learning_rate": 1.992946059000927e-06,
      "loss": 0.665,
      "step": 524
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7675668011819149,
      "learning_rate": 1.992897046428535e-06,
      "loss": 0.5994,
      "step": 525
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9130233856179439,
      "learning_rate": 1.99284786477612e-06,
      "loss": 0.6728,
      "step": 526
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8210887431824286,
      "learning_rate": 1.992798514052057e-06,
      "loss": 0.6813,
      "step": 527
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7416502243640162,
      "learning_rate": 1.99274899426475e-06,
      "loss": 0.5835,
      "step": 528
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7002805608352144,
      "learning_rate": 1.9926993054226317e-06,
      "loss": 0.6252,
      "step": 529
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7409014266039685,
      "learning_rate": 1.9926494475341632e-06,
      "loss": 0.6191,
      "step": 530
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9939483676049348,
      "learning_rate": 1.9925994206078353e-06,
      "loss": 0.5919,
      "step": 531
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7711549899709258,
      "learning_rate": 1.992549224652167e-06,
      "loss": 0.6566,
      "step": 532
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7965055630250588,
      "learning_rate": 1.992498859675706e-06,
      "loss": 0.6962,
      "step": 533
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8250135235957727,
      "learning_rate": 1.992448325687029e-06,
      "loss": 0.6099,
      "step": 534
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6374817796334638,
      "learning_rate": 1.9923976226947413e-06,
      "loss": 0.5362,
      "step": 535
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7866966560468455,
      "learning_rate": 1.9923467507074774e-06,
      "loss": 0.5872,
      "step": 536
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8780860175483405,
      "learning_rate": 1.9922957097339e-06,
      "loss": 0.6597,
      "step": 537
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8232361897762414,
      "learning_rate": 1.9922444997827007e-06,
      "loss": 0.6831,
      "step": 538
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9518994100733643,
      "learning_rate": 1.9921931208626003e-06,
      "loss": 0.5798,
      "step": 539
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8439863343606367,
      "learning_rate": 1.992141572982348e-06,
      "loss": 0.5875,
      "step": 540
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6539230806691111,
      "learning_rate": 1.992089856150722e-06,
      "loss": 0.557,
      "step": 541
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8024210266163878,
      "learning_rate": 1.992037970376529e-06,
      "loss": 0.6311,
      "step": 542
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6851560582765146,
      "learning_rate": 1.9919859156686046e-06,
      "loss": 0.5768,
      "step": 543
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6728905386475592,
      "learning_rate": 1.991933692035813e-06,
      "loss": 0.5454,
      "step": 544
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8297912590440232,
      "learning_rate": 1.9918812994870477e-06,
      "loss": 0.6528,
      "step": 545
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.722362985156497,
      "learning_rate": 1.99182873803123e-06,
      "loss": 0.588,
      "step": 546
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6833773640554377,
      "learning_rate": 1.9917760076773117e-06,
      "loss": 0.534,
      "step": 547
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6494228908847088,
      "learning_rate": 1.9917231084342716e-06,
      "loss": 0.5509,
      "step": 548
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6020082727644316,
      "learning_rate": 1.9916700403111174e-06,
      "loss": 0.5622,
      "step": 549
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7402075214920953,
      "learning_rate": 1.9916168033168862e-06,
      "loss": 0.6502,
      "step": 550
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7932825698733326,
      "learning_rate": 1.9915633974606447e-06,
      "loss": 0.6707,
      "step": 551
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9320208296213243,
      "learning_rate": 1.991509822751486e-06,
      "loss": 0.6215,
      "step": 552
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6570217784161347,
      "learning_rate": 1.9914560791985343e-06,
      "loss": 0.5781,
      "step": 553
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7455204330314638,
      "learning_rate": 1.9914021668109414e-06,
      "loss": 0.5999,
      "step": 554
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6728673335636264,
      "learning_rate": 1.9913480855978877e-06,
      "loss": 0.58,
      "step": 555
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2052907452724413,
      "learning_rate": 1.991293835568583e-06,
      "loss": 0.7481,
      "step": 556
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6256460585619213,
      "learning_rate": 1.991239416732265e-06,
      "loss": 0.601,
      "step": 557
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7632164316726977,
      "learning_rate": 1.9911848290982008e-06,
      "loss": 0.6731,
      "step": 558
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8504271689212162,
      "learning_rate": 1.9911300726756863e-06,
      "loss": 0.6465,
      "step": 559
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7030363463851032,
      "learning_rate": 1.9910751474740467e-06,
      "loss": 0.5653,
      "step": 560
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6219413514707076,
      "learning_rate": 1.9910200535026335e-06,
      "loss": 0.5529,
      "step": 561
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.874478506488717,
      "learning_rate": 1.99096479077083e-06,
      "loss": 0.6317,
      "step": 562
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.753319092066656,
      "learning_rate": 1.9909093592880467e-06,
      "loss": 0.619,
      "step": 563
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6603021049409944,
      "learning_rate": 1.9908537590637222e-06,
      "loss": 0.5789,
      "step": 564
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9341903408844926,
      "learning_rate": 1.9907979901073256e-06,
      "loss": 0.6835,
      "step": 565
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9158173897189834,
      "learning_rate": 1.990742052428353e-06,
      "loss": 0.7234,
      "step": 566
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6994294970537547,
      "learning_rate": 1.9906859460363304e-06,
      "loss": 0.6211,
      "step": 567
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.766674955504057,
      "learning_rate": 1.9906296709408125e-06,
      "loss": 0.6073,
      "step": 568
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8361008042379616,
      "learning_rate": 1.9905732271513818e-06,
      "loss": 0.5853,
      "step": 569
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8318469677219228,
      "learning_rate": 1.99051661467765e-06,
      "loss": 0.6818,
      "step": 570
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7019197979384322,
      "learning_rate": 1.990459833529258e-06,
      "loss": 0.5677,
      "step": 571
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8514044815844871,
      "learning_rate": 1.9904028837158746e-06,
      "loss": 0.5864,
      "step": 572
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9651689292138219,
      "learning_rate": 1.9903457652471982e-06,
      "loss": 0.6982,
      "step": 573
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8455242221546466,
      "learning_rate": 1.9902884781329555e-06,
      "loss": 0.6439,
      "step": 574
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6561544674849393,
      "learning_rate": 1.9902310223829014e-06,
      "loss": 0.6085,
      "step": 575
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.937057215300035,
      "learning_rate": 1.9901733980068206e-06,
      "loss": 0.712,
      "step": 576
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6849258021895515,
      "learning_rate": 1.9901156050145257e-06,
      "loss": 0.5343,
      "step": 577
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6278893872899395,
      "learning_rate": 1.990057643415858e-06,
      "loss": 0.5769,
      "step": 578
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.780980702399632,
      "learning_rate": 1.9899995132206883e-06,
      "loss": 0.655,
      "step": 579
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7935012714201848,
      "learning_rate": 1.989941214438915e-06,
      "loss": 0.6604,
      "step": 580
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7947200362013553,
      "learning_rate": 1.989882747080466e-06,
      "loss": 0.7058,
      "step": 581
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.15899467874248,
      "learning_rate": 1.989824111155298e-06,
      "loss": 0.6747,
      "step": 582
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9260353981555233,
      "learning_rate": 1.989765306673395e-06,
      "loss": 0.7168,
      "step": 583
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7619693256660294,
      "learning_rate": 1.9897063336447727e-06,
      "loss": 0.5069,
      "step": 584
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8632482255427164,
      "learning_rate": 1.9896471920794717e-06,
      "loss": 0.647,
      "step": 585
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7211371193610369,
      "learning_rate": 1.989587881987564e-06,
      "loss": 0.5243,
      "step": 586
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7083689444678922,
      "learning_rate": 1.9895284033791497e-06,
      "loss": 0.6235,
      "step": 587
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.881945082230812,
      "learning_rate": 1.9894687562643568e-06,
      "loss": 0.655,
      "step": 588
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6935289256962699,
      "learning_rate": 1.9894089406533434e-06,
      "loss": 0.5911,
      "step": 589
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0039966400809808,
      "learning_rate": 1.9893489565562947e-06,
      "loss": 0.6712,
      "step": 590
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.747033744133976,
      "learning_rate": 1.989288803983426e-06,
      "loss": 0.6162,
      "step": 591
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7436823348875847,
      "learning_rate": 1.9892284829449797e-06,
      "loss": 0.5801,
      "step": 592
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.033523517673034,
      "learning_rate": 1.989167993451229e-06,
      "loss": 0.7125,
      "step": 593
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9095181071253053,
      "learning_rate": 1.9891073355124733e-06,
      "loss": 0.6476,
      "step": 594
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7434436203530099,
      "learning_rate": 1.989046509139044e-06,
      "loss": 0.5981,
      "step": 595
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5802384952938089,
      "learning_rate": 1.9889855143412972e-06,
      "loss": 0.5591,
      "step": 596
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.085697433779284,
      "learning_rate": 1.9889243511296207e-06,
      "loss": 0.6402,
      "step": 597
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6303180573801902,
      "learning_rate": 1.9888630195144293e-06,
      "loss": 0.5474,
      "step": 598
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3413915865312702,
      "learning_rate": 1.9888015195061683e-06,
      "loss": 0.6136,
      "step": 599
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8511319653754635,
      "learning_rate": 1.9887398511153095e-06,
      "loss": 0.5975,
      "step": 600
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.715852057825207,
      "learning_rate": 1.9886780143523545e-06,
      "loss": 0.5588,
      "step": 601
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.659271833376575,
      "learning_rate": 1.988616009227834e-06,
      "loss": 0.5404,
      "step": 602
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9664161784962277,
      "learning_rate": 1.988553835752306e-06,
      "loss": 0.6136,
      "step": 603
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6026822005444106,
      "learning_rate": 1.9884914939363586e-06,
      "loss": 0.5593,
      "step": 604
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6705326196647514,
      "learning_rate": 1.988428983790608e-06,
      "loss": 0.5999,
      "step": 605
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6552312766630315,
      "learning_rate": 1.9883663053256985e-06,
      "loss": 0.5647,
      "step": 606
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6596561548714385,
      "learning_rate": 1.9883034585523046e-06,
      "loss": 0.5642,
      "step": 607
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7064415105841974,
      "learning_rate": 1.988240443481127e-06,
      "loss": 0.6005,
      "step": 608
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6904601283825006,
      "learning_rate": 1.9881772601228977e-06,
      "loss": 0.6041,
      "step": 609
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6404068538065457,
      "learning_rate": 1.9881139084883756e-06,
      "loss": 0.568,
      "step": 610
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8816219968018946,
      "learning_rate": 1.988050388588349e-06,
      "loss": 0.6788,
      "step": 611
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7543776958800256,
      "learning_rate": 1.987986700433635e-06,
      "loss": 0.5709,
      "step": 612
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8238846293414066,
      "learning_rate": 1.9879228440350785e-06,
      "loss": 0.6749,
      "step": 613
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8917840822332804,
      "learning_rate": 1.9878588194035535e-06,
      "loss": 0.6748,
      "step": 614
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7763996238794073,
      "learning_rate": 1.9877946265499633e-06,
      "loss": 0.654,
      "step": 615
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8261125263298917,
      "learning_rate": 1.987730265485239e-06,
      "loss": 0.6936,
      "step": 616
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.627811287734862,
      "learning_rate": 1.9876657362203402e-06,
      "loss": 0.5532,
      "step": 617
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0901250486987362,
      "learning_rate": 1.987601038766256e-06,
      "loss": 0.6665,
      "step": 618
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6094014173150708,
      "learning_rate": 1.987536173134004e-06,
      "loss": 0.5272,
      "step": 619
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6692087696129586,
      "learning_rate": 1.9874711393346297e-06,
      "loss": 0.5886,
      "step": 620
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.808027861904522,
      "learning_rate": 1.9874059373792074e-06,
      "loss": 0.6266,
      "step": 621
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6837399777962959,
      "learning_rate": 1.9873405672788414e-06,
      "loss": 0.5829,
      "step": 622
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8522117219788022,
      "learning_rate": 1.9872750290446623e-06,
      "loss": 0.6163,
      "step": 623
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5720629187471247,
      "learning_rate": 1.987209322687831e-06,
      "loss": 0.5127,
      "step": 624
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7999731719370359,
      "learning_rate": 1.987143448219537e-06,
      "loss": 0.6713,
      "step": 625
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8899122717311352,
      "learning_rate": 1.987077405650998e-06,
      "loss": 0.6459,
      "step": 626
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6563699550164905,
      "learning_rate": 1.9870111949934595e-06,
      "loss": 0.6049,
      "step": 627
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7830118377700895,
      "learning_rate": 1.9869448162581976e-06,
      "loss": 0.6171,
      "step": 628
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5708054261065402,
      "learning_rate": 1.986878269456515e-06,
      "loss": 0.5094,
      "step": 629
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9134654561678494,
      "learning_rate": 1.9868115545997446e-06,
      "loss": 0.6862,
      "step": 630
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8121496977579589,
      "learning_rate": 1.986744671699247e-06,
      "loss": 0.6473,
      "step": 631
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7623226263514385,
      "learning_rate": 1.9866776207664116e-06,
      "loss": 0.5886,
      "step": 632
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.93291751183117,
      "learning_rate": 1.9866104018126566e-06,
      "loss": 0.7166,
      "step": 633
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7678529839128109,
      "learning_rate": 1.9865430148494283e-06,
      "loss": 0.6146,
      "step": 634
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7059831230569459,
      "learning_rate": 1.9864754598882025e-06,
      "loss": 0.5803,
      "step": 635
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7476221086128951,
      "learning_rate": 1.986407736940483e-06,
      "loss": 0.5519,
      "step": 636
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7212173288812114,
      "learning_rate": 1.986339846017802e-06,
      "loss": 0.5859,
      "step": 637
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7176351810897198,
      "learning_rate": 1.986271787131721e-06,
      "loss": 0.571,
      "step": 638
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9956946442398525,
      "learning_rate": 1.9862035602938297e-06,
      "loss": 0.6726,
      "step": 639
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6564916395068399,
      "learning_rate": 1.986135165515746e-06,
      "loss": 0.5427,
      "step": 640
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6854838385962346,
      "learning_rate": 1.9860666028091173e-06,
      "loss": 0.5967,
      "step": 641
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7293395742362232,
      "learning_rate": 1.9859978721856184e-06,
      "loss": 0.578,
      "step": 642
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0096567650949269,
      "learning_rate": 1.9859289736569544e-06,
      "loss": 0.6577,
      "step": 643
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6861508890341291,
      "learning_rate": 1.9858599072348575e-06,
      "loss": 0.5239,
      "step": 644
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7025329605117591,
      "learning_rate": 1.9857906729310893e-06,
      "loss": 0.5244,
      "step": 645
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6780686207067961,
      "learning_rate": 1.985721270757439e-06,
      "loss": 0.5654,
      "step": 646
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7355168902551373,
      "learning_rate": 1.985651700725725e-06,
      "loss": 0.5985,
      "step": 647
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6782576859012697,
      "learning_rate": 1.985581962847796e-06,
      "loss": 0.567,
      "step": 648
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7748055135540624,
      "learning_rate": 1.9855120571355256e-06,
      "loss": 0.6489,
      "step": 649
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.797995101481714,
      "learning_rate": 1.9854419836008187e-06,
      "loss": 0.636,
      "step": 650
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.068919679452347,
      "learning_rate": 1.9853717422556086e-06,
      "loss": 0.6934,
      "step": 651
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.652277275370946,
      "learning_rate": 1.9853013331118566e-06,
      "loss": 0.5965,
      "step": 652
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6838579928501709,
      "learning_rate": 1.985230756181552e-06,
      "loss": 0.592,
      "step": 653
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8509369132939175,
      "learning_rate": 1.985160011476714e-06,
      "loss": 0.6302,
      "step": 654
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.638192128708149,
      "learning_rate": 1.9850890990093887e-06,
      "loss": 0.5739,
      "step": 655
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8472933976053362,
      "learning_rate": 1.9850180187916526e-06,
      "loss": 0.6324,
      "step": 656
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7302032593503742,
      "learning_rate": 1.98494677083561e-06,
      "loss": 0.5844,
      "step": 657
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6978153532388472,
      "learning_rate": 1.9848753551533938e-06,
      "loss": 0.5769,
      "step": 658
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8981173286577327,
      "learning_rate": 1.9848037717571643e-06,
      "loss": 0.6869,
      "step": 659
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3002925257464328,
      "learning_rate": 1.9847320206591127e-06,
      "loss": 0.6434,
      "step": 660
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6903143137091778,
      "learning_rate": 1.9846601018714565e-06,
      "loss": 0.6287,
      "step": 661
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6611786972979261,
      "learning_rate": 1.9845880154064427e-06,
      "loss": 0.6288,
      "step": 662
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8570696709079825,
      "learning_rate": 1.9845157612763476e-06,
      "loss": 0.7148,
      "step": 663
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8632101911499828,
      "learning_rate": 1.9844433394934752e-06,
      "loss": 0.6443,
      "step": 664
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7211289966916442,
      "learning_rate": 1.9843707500701576e-06,
      "loss": 0.5307,
      "step": 665
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.669404058601053,
      "learning_rate": 1.9842979930187567e-06,
      "loss": 0.5953,
      "step": 666
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9615249294392699,
      "learning_rate": 1.9842250683516613e-06,
      "loss": 0.7195,
      "step": 667
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8504393865612274,
      "learning_rate": 1.984151976081291e-06,
      "loss": 0.6468,
      "step": 668
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8216159208214725,
      "learning_rate": 1.984078716220092e-06,
      "loss": 0.6562,
      "step": 669
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6381573640965261,
      "learning_rate": 1.9840052887805393e-06,
      "loss": 0.5942,
      "step": 670
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1493429130505968,
      "learning_rate": 1.9839316937751375e-06,
      "loss": 0.7116,
      "step": 671
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7066314067389571,
      "learning_rate": 1.9838579312164183e-06,
      "loss": 0.5911,
      "step": 672
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9357411774356766,
      "learning_rate": 1.9837840011169438e-06,
      "loss": 0.6336,
      "step": 673
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6030118498947621,
      "learning_rate": 1.9837099034893023e-06,
      "loss": 0.5461,
      "step": 674
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6169748181955841,
      "learning_rate": 1.9836356383461127e-06,
      "loss": 0.5685,
      "step": 675
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6152438497554215,
      "learning_rate": 1.9835612057000216e-06,
      "loss": 0.5398,
      "step": 676
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9247181940605527,
      "learning_rate": 1.9834866055637036e-06,
      "loss": 0.5918,
      "step": 677
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9769797312158686,
      "learning_rate": 1.9834118379498627e-06,
      "loss": 0.6289,
      "step": 678
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5877982936371612,
      "learning_rate": 1.9833369028712305e-06,
      "loss": 0.5781,
      "step": 679
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6105643811295265,
      "learning_rate": 1.9832618003405687e-06,
      "loss": 0.5566,
      "step": 680
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8964336800230498,
      "learning_rate": 1.9831865303706654e-06,
      "loss": 0.6654,
      "step": 681
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.601525226866425,
      "learning_rate": 1.9831110929743387e-06,
      "loss": 0.5364,
      "step": 682
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7153807518569122,
      "learning_rate": 1.9830354881644352e-06,
      "loss": 0.6017,
      "step": 683
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6613277913097433,
      "learning_rate": 1.9829597159538285e-06,
      "loss": 0.6032,
      "step": 684
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9356859925227377,
      "learning_rate": 1.982883776355423e-06,
      "loss": 0.6625,
      "step": 685
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7458433874832644,
      "learning_rate": 1.9828076693821504e-06,
      "loss": 0.595,
      "step": 686
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6714002784813978,
      "learning_rate": 1.9827313950469703e-06,
      "loss": 0.5596,
      "step": 687
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5846272822520122,
      "learning_rate": 1.9826549533628715e-06,
      "loss": 0.5436,
      "step": 688
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6442604704176558,
      "learning_rate": 1.982578344342872e-06,
      "loss": 0.5682,
      "step": 689
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7033242533518118,
      "learning_rate": 1.982501568000016e-06,
      "loss": 0.6249,
      "step": 690
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8201422698235402,
      "learning_rate": 1.982424624347379e-06,
      "loss": 0.6259,
      "step": 691
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8269105215466174,
      "learning_rate": 1.982347513398063e-06,
      "loss": 0.6717,
      "step": 692
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6831797804868314,
      "learning_rate": 1.9822702351652e-06,
      "loss": 0.5644,
      "step": 693
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8786783283100819,
      "learning_rate": 1.9821927896619485e-06,
      "loss": 0.6106,
      "step": 694
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7369761423317939,
      "learning_rate": 1.982115176901498e-06,
      "loss": 0.5609,
      "step": 695
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6579635394648186,
      "learning_rate": 1.982037396897064e-06,
      "loss": 0.5416,
      "step": 696
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6304713145194579,
      "learning_rate": 1.9819594496618923e-06,
      "loss": 0.6018,
      "step": 697
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8358720668230432,
      "learning_rate": 1.9818813352092563e-06,
      "loss": 0.6196,
      "step": 698
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7491183598980263,
      "learning_rate": 1.981803053552458e-06,
      "loss": 0.637,
      "step": 699
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.721564288749998,
      "learning_rate": 1.9817246047048283e-06,
      "loss": 0.5666,
      "step": 700
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.850259146024705,
      "learning_rate": 1.9816459886797255e-06,
      "loss": 0.6978,
      "step": 701
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7261715688498724,
      "learning_rate": 1.981567205490538e-06,
      "loss": 0.5794,
      "step": 702
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8818505414130215,
      "learning_rate": 1.981488255150681e-06,
      "loss": 0.719,
      "step": 703
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.996778087177852,
      "learning_rate": 1.9814091376735994e-06,
      "loss": 0.6621,
      "step": 704
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7435369607876877,
      "learning_rate": 1.981329853072766e-06,
      "loss": 0.6197,
      "step": 705
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6614107989283773,
      "learning_rate": 1.9812504013616818e-06,
      "loss": 0.5634,
      "step": 706
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.822138660115289,
      "learning_rate": 1.981170782553877e-06,
      "loss": 0.5917,
      "step": 707
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.562372679884486,
      "learning_rate": 1.9810909966629095e-06,
      "loss": 0.5999,
      "step": 708
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6851523294104136,
      "learning_rate": 1.981011043702366e-06,
      "loss": 0.5803,
      "step": 709
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8202947131891288,
      "learning_rate": 1.980930923685862e-06,
      "loss": 0.6446,
      "step": 710
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8168578829032651,
      "learning_rate": 1.9808506366270416e-06,
      "loss": 0.6598,
      "step": 711
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7948613365837286,
      "learning_rate": 1.9807701825395754e-06,
      "loss": 0.6593,
      "step": 712
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0769684651848344,
      "learning_rate": 1.9806895614371652e-06,
      "loss": 0.6486,
      "step": 713
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7735145860441905,
      "learning_rate": 1.9806087733335392e-06,
      "loss": 0.6403,
      "step": 714
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6972305435984866,
      "learning_rate": 1.980527818242455e-06,
      "loss": 0.5982,
      "step": 715
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7035279575541616,
      "learning_rate": 1.9804466961776987e-06,
      "loss": 0.5373,
      "step": 716
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8513575120217315,
      "learning_rate": 1.980365407153084e-06,
      "loss": 0.6601,
      "step": 717
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6889164678998441,
      "learning_rate": 1.9802839511824537e-06,
      "loss": 0.5766,
      "step": 718
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6074745451393104,
      "learning_rate": 1.9802023282796794e-06,
      "loss": 0.5141,
      "step": 719
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7494585477055167,
      "learning_rate": 1.98012053845866e-06,
      "loss": 0.6424,
      "step": 720
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.744756094324546,
      "learning_rate": 1.980038581733324e-06,
      "loss": 0.5569,
      "step": 721
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6963258127690319,
      "learning_rate": 1.979956458117627e-06,
      "loss": 0.5898,
      "step": 722
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7443315653394612,
      "learning_rate": 1.979874167625555e-06,
      "loss": 0.6094,
      "step": 723
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6783430466692265,
      "learning_rate": 1.97979171027112e-06,
      "loss": 0.5341,
      "step": 724
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0547625462312304,
      "learning_rate": 1.979709086068364e-06,
      "loss": 0.6716,
      "step": 725
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9200871025087154,
      "learning_rate": 1.9796262950313574e-06,
      "loss": 0.7183,
      "step": 726
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7375319506273126,
      "learning_rate": 1.979543337174199e-06,
      "loss": 0.5304,
      "step": 727
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6500347364959113,
      "learning_rate": 1.979460212511014e-06,
      "loss": 0.5277,
      "step": 728
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8393473988890956,
      "learning_rate": 1.9793769210559593e-06,
      "loss": 0.6023,
      "step": 729
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7615589291819547,
      "learning_rate": 1.9792934628232182e-06,
      "loss": 0.6326,
      "step": 730
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.755037844078346,
      "learning_rate": 1.9792098378270024e-06,
      "loss": 0.6899,
      "step": 731
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7538385948811868,
      "learning_rate": 1.979126046081553e-06,
      "loss": 0.5357,
      "step": 732
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.957082455704836,
      "learning_rate": 1.9790420876011375e-06,
      "loss": 0.6522,
      "step": 733
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0596192482506819,
      "learning_rate": 1.9789579624000547e-06,
      "loss": 0.5922,
      "step": 734
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.68645085342628,
      "learning_rate": 1.9788736704926295e-06,
      "loss": 0.5492,
      "step": 735
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6237244323698963,
      "learning_rate": 1.9787892118932157e-06,
      "loss": 0.5538,
      "step": 736
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6661273263181292,
      "learning_rate": 1.9787045866161963e-06,
      "loss": 0.5469,
      "step": 737
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8873004904484652,
      "learning_rate": 1.9786197946759823e-06,
      "loss": 0.6573,
      "step": 738
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7708683347714198,
      "learning_rate": 1.978534836087012e-06,
      "loss": 0.6333,
      "step": 739
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6640724985353499,
      "learning_rate": 1.9784497108637535e-06,
      "loss": 0.5494,
      "step": 740
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7680990101790869,
      "learning_rate": 1.9783644190207025e-06,
      "loss": 0.5547,
      "step": 741
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8040354475152123,
      "learning_rate": 1.978278960572384e-06,
      "loss": 0.5584,
      "step": 742
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5995552064853391,
      "learning_rate": 1.9781933355333496e-06,
      "loss": 0.5583,
      "step": 743
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6078379065944832,
      "learning_rate": 1.9781075439181813e-06,
      "loss": 0.4987,
      "step": 744
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6836467921643848,
      "learning_rate": 1.9780215857414878e-06,
      "loss": 0.6178,
      "step": 745
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.614677285489946,
      "learning_rate": 1.977935461017907e-06,
      "loss": 0.4891,
      "step": 746
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9059187188466682,
      "learning_rate": 1.9778491697621055e-06,
      "loss": 0.6396,
      "step": 747
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6726997689046449,
      "learning_rate": 1.977762711988778e-06,
      "loss": 0.6019,
      "step": 748
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0082093747366472,
      "learning_rate": 1.977676087712646e-06,
      "loss": 0.6241,
      "step": 749
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8012353153348502,
      "learning_rate": 1.9775892969484622e-06,
      "loss": 0.6084,
      "step": 750
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7681270068283917,
      "learning_rate": 1.9775023397110056e-06,
      "loss": 0.6478,
      "step": 751
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6186195967705558,
      "learning_rate": 1.977415216015084e-06,
      "loss": 0.5715,
      "step": 752
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5727427056040725,
      "learning_rate": 1.9773279258755336e-06,
      "loss": 0.5509,
      "step": 753
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7936499913214147,
      "learning_rate": 1.9772404693072197e-06,
      "loss": 0.6106,
      "step": 754
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.050579912611296,
      "learning_rate": 1.9771528463250342e-06,
      "loss": 0.6438,
      "step": 755
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7738457636617446,
      "learning_rate": 1.9770650569438988e-06,
      "loss": 0.6052,
      "step": 756
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.617902845393335,
      "learning_rate": 1.9769771011787635e-06,
      "loss": 0.6029,
      "step": 757
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8106681826993078,
      "learning_rate": 1.976888979044606e-06,
      "loss": 0.5978,
      "step": 758
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6493905243301383,
      "learning_rate": 1.976800690556432e-06,
      "loss": 0.545,
      "step": 759
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8113768682418111,
      "learning_rate": 1.9767122357292772e-06,
      "loss": 0.618,
      "step": 760
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6743122755685925,
      "learning_rate": 1.976623614578204e-06,
      "loss": 0.552,
      "step": 761
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7090764247025748,
      "learning_rate": 1.9765348271183033e-06,
      "loss": 0.6299,
      "step": 762
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.611676256337836,
      "learning_rate": 1.976445873364695e-06,
      "loss": 0.5453,
      "step": 763
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7442564411507394,
      "learning_rate": 1.976356753332527e-06,
      "loss": 0.5545,
      "step": 764
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6769787023241824,
      "learning_rate": 1.9762674670369754e-06,
      "loss": 0.5113,
      "step": 765
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7499204560811521,
      "learning_rate": 1.976178014493245e-06,
      "loss": 0.6017,
      "step": 766
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8227860002787356,
      "learning_rate": 1.9760883957165687e-06,
      "loss": 0.6627,
      "step": 767
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6669506457369923,
      "learning_rate": 1.975998610722207e-06,
      "loss": 0.524,
      "step": 768
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6845905311063933,
      "learning_rate": 1.9759086595254503e-06,
      "loss": 0.6344,
      "step": 769
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7689295491697293,
      "learning_rate": 1.9758185421416154e-06,
      "loss": 0.5903,
      "step": 770
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8614482838117034,
      "learning_rate": 1.975728258586049e-06,
      "loss": 0.6386,
      "step": 771
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7993298795256107,
      "learning_rate": 1.9756378088741254e-06,
      "loss": 0.6282,
      "step": 772
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6164833275253789,
      "learning_rate": 1.9755471930212465e-06,
      "loss": 0.5859,
      "step": 773
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6283345723717823,
      "learning_rate": 1.975456411042844e-06,
      "loss": 0.5594,
      "step": 774
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9230906296433633,
      "learning_rate": 1.9753654629543775e-06,
      "loss": 0.6469,
      "step": 775
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8415304821767399,
      "learning_rate": 1.9752743487713333e-06,
      "loss": 0.6234,
      "step": 776
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7485345883437187,
      "learning_rate": 1.975183068509228e-06,
      "loss": 0.6548,
      "step": 777
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6999141079604583,
      "learning_rate": 1.9750916221836057e-06,
      "loss": 0.5655,
      "step": 778
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.720893133797929,
      "learning_rate": 1.9750000098100384e-06,
      "loss": 0.6021,
      "step": 779
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8770269361720169,
      "learning_rate": 1.974908231404127e-06,
      "loss": 0.7092,
      "step": 780
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7047161075071706,
      "learning_rate": 1.9748162869815e-06,
      "loss": 0.5996,
      "step": 781
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6967152116901362,
      "learning_rate": 1.974724176557815e-06,
      "loss": 0.5853,
      "step": 782
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6833733719656645,
      "learning_rate": 1.974631900148758e-06,
      "loss": 0.5669,
      "step": 783
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0753168782421432,
      "learning_rate": 1.9745394577700412e-06,
      "loss": 0.6255,
      "step": 784
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7924289733640827,
      "learning_rate": 1.974446849437408e-06,
      "loss": 0.535,
      "step": 785
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8641187993501612,
      "learning_rate": 1.9743540751666275e-06,
      "loss": 0.6754,
      "step": 786
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5993352150914808,
      "learning_rate": 1.9742611349734993e-06,
      "loss": 0.5303,
      "step": 787
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7612309741319502,
      "learning_rate": 1.9741680288738493e-06,
      "loss": 0.5318,
      "step": 788
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6563031969741099,
      "learning_rate": 1.9740747568835326e-06,
      "loss": 0.5954,
      "step": 789
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8653190011904542,
      "learning_rate": 1.973981319018433e-06,
      "loss": 0.6767,
      "step": 790
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8851278973729406,
      "learning_rate": 1.973887715294461e-06,
      "loss": 0.635,
      "step": 791
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8335661893994004,
      "learning_rate": 1.9737939457275577e-06,
      "loss": 0.6645,
      "step": 792
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7013083132095101,
      "learning_rate": 1.9737000103336904e-06,
      "loss": 0.5917,
      "step": 793
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8045268022193846,
      "learning_rate": 1.9736059091288547e-06,
      "loss": 0.6328,
      "step": 794
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8157123521063797,
      "learning_rate": 1.9735116421290763e-06,
      "loss": 0.7478,
      "step": 795
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8162946383338712,
      "learning_rate": 1.9734172093504066e-06,
      "loss": 0.6151,
      "step": 796
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0025998518349999,
      "learning_rate": 1.9733226108089274e-06,
      "loss": 0.6719,
      "step": 797
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7219412772478457,
      "learning_rate": 1.973227846520748e-06,
      "loss": 0.5632,
      "step": 798
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6539449767895833,
      "learning_rate": 1.9731329165020044e-06,
      "loss": 0.5341,
      "step": 799
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6308148494124889,
      "learning_rate": 1.9730378207688636e-06,
      "loss": 0.5423,
      "step": 800
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8244248254207861,
      "learning_rate": 1.9729425593375194e-06,
      "loss": 0.6382,
      "step": 801
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6446769236880466,
      "learning_rate": 1.972847132224193e-06,
      "loss": 0.4918,
      "step": 802
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9353138395325179,
      "learning_rate": 1.972751539445135e-06,
      "loss": 0.6713,
      "step": 803
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1516783449500636,
      "learning_rate": 1.9726557810166236e-06,
      "loss": 0.6724,
      "step": 804
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7267148545537543,
      "learning_rate": 1.9725598569549666e-06,
      "loss": 0.5869,
      "step": 805
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.152682738515493,
      "learning_rate": 1.972463767276497e-06,
      "loss": 0.6124,
      "step": 806
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7854841173684677,
      "learning_rate": 1.9723675119975794e-06,
      "loss": 0.6663,
      "step": 807
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6451516063426489,
      "learning_rate": 1.972271091134605e-06,
      "loss": 0.5704,
      "step": 808
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6692814395934845,
      "learning_rate": 1.972174504703992e-06,
      "loss": 0.5863,
      "step": 809
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7777329592678502,
      "learning_rate": 1.9720777527221895e-06,
      "loss": 0.678,
      "step": 810
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5661195228618701,
      "learning_rate": 1.9719808352056723e-06,
      "loss": 0.5203,
      "step": 811
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9646879244718761,
      "learning_rate": 1.971883752170945e-06,
      "loss": 0.667,
      "step": 812
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8074883107794468,
      "learning_rate": 1.97178650363454e-06,
      "loss": 0.6822,
      "step": 813
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8825110475294269,
      "learning_rate": 1.971689089613017e-06,
      "loss": 0.7125,
      "step": 814
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.629931632306053,
      "learning_rate": 1.9715915101229653e-06,
      "loss": 0.5206,
      "step": 815
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7516799941162974,
      "learning_rate": 1.9714937651810014e-06,
      "loss": 0.6123,
      "step": 816
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6652751750052173,
      "learning_rate": 1.97139585480377e-06,
      "loss": 0.5244,
      "step": 817
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6768018857299368,
      "learning_rate": 1.971297779007945e-06,
      "loss": 0.566,
      "step": 818
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6978941505933295,
      "learning_rate": 1.9711995378102266e-06,
      "loss": 0.547,
      "step": 819
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6861533062120846,
      "learning_rate": 1.971101131227345e-06,
      "loss": 0.5593,
      "step": 820
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9063479975989226,
      "learning_rate": 1.9710025592760577e-06,
      "loss": 0.655,
      "step": 821
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6790587749117465,
      "learning_rate": 1.97090382197315e-06,
      "loss": 0.5371,
      "step": 822
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7536524000166772,
      "learning_rate": 1.9708049193354365e-06,
      "loss": 0.5867,
      "step": 823
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8856482839691099,
      "learning_rate": 1.9707058513797594e-06,
      "loss": 0.6552,
      "step": 824
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.767530599739631,
      "learning_rate": 1.970606618122988e-06,
      "loss": 0.6333,
      "step": 825
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7464623585787192,
      "learning_rate": 1.970507219582021e-06,
      "loss": 0.5018,
      "step": 826
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7837843379036434,
      "learning_rate": 1.9704076557737855e-06,
      "loss": 0.6545,
      "step": 827
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7020161513102735,
      "learning_rate": 1.970307926715236e-06,
      "loss": 0.563,
      "step": 828
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8334557405686772,
      "learning_rate": 1.9702080324233552e-06,
      "loss": 0.5838,
      "step": 829
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8230648051282579,
      "learning_rate": 1.970107972915154e-06,
      "loss": 0.5788,
      "step": 830
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7091907961059148,
      "learning_rate": 1.9700077482076717e-06,
      "loss": 0.5616,
      "step": 831
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7750383593538468,
      "learning_rate": 1.9699073583179747e-06,
      "loss": 0.5997,
      "step": 832
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6740022634100381,
      "learning_rate": 1.9698068032631596e-06,
      "loss": 0.6132,
      "step": 833
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7882570790571856,
      "learning_rate": 1.969706083060349e-06,
      "loss": 0.5558,
      "step": 834
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6957575466412099,
      "learning_rate": 1.969605197726695e-06,
      "loss": 0.5424,
      "step": 835
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7208118349747086,
      "learning_rate": 1.9695041472793767e-06,
      "loss": 0.621,
      "step": 836
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.644365790890913,
      "learning_rate": 1.969402931735603e-06,
      "loss": 0.5271,
      "step": 837
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8177597935463959,
      "learning_rate": 1.9693015511126085e-06,
      "loss": 0.6274,
      "step": 838
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.072878796563849,
      "learning_rate": 1.969200005427658e-06,
      "loss": 0.6189,
      "step": 839
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8998226534320273,
      "learning_rate": 1.969098294698044e-06,
      "loss": 0.7163,
      "step": 840
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7604696326589294,
      "learning_rate": 1.9689964189410863e-06,
      "loss": 0.5678,
      "step": 841
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7911457679644602,
      "learning_rate": 1.9688943781741335e-06,
      "loss": 0.6205,
      "step": 842
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6680396094349825,
      "learning_rate": 1.9687921724145617e-06,
      "loss": 0.5554,
      "step": 843
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6661923195634976,
      "learning_rate": 1.968689801679776e-06,
      "loss": 0.5841,
      "step": 844
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8159686257413361,
      "learning_rate": 1.968587265987209e-06,
      "loss": 0.5559,
      "step": 845
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5730838723654572,
      "learning_rate": 1.9684845653543208e-06,
      "loss": 0.5074,
      "step": 846
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8333071881854135,
      "learning_rate": 1.9683816997986007e-06,
      "loss": 0.6619,
      "step": 847
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7575372228641677,
      "learning_rate": 1.968278669337566e-06,
      "loss": 0.6183,
      "step": 848
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.745386428295944,
      "learning_rate": 1.968175473988761e-06,
      "loss": 0.5712,
      "step": 849
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8320975073747173,
      "learning_rate": 1.9680721137697596e-06,
      "loss": 0.704,
      "step": 850
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7717622861848287,
      "learning_rate": 1.9679685886981623e-06,
      "loss": 0.5409,
      "step": 851
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9336652978512102,
      "learning_rate": 1.9678648987915986e-06,
      "loss": 0.662,
      "step": 852
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7288405466530615,
      "learning_rate": 1.9677610440677255e-06,
      "loss": 0.5965,
      "step": 853
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.620954768399733,
      "learning_rate": 1.9676570245442286e-06,
      "loss": 0.5472,
      "step": 854
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8738518908654507,
      "learning_rate": 1.967552840238822e-06,
      "loss": 0.6513,
      "step": 855
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9019528013021154,
      "learning_rate": 1.9674484911692463e-06,
      "loss": 0.6896,
      "step": 856
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6704085801939242,
      "learning_rate": 1.967343977353271e-06,
      "loss": 0.5332,
      "step": 857
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8028041816294855,
      "learning_rate": 1.9672392988086946e-06,
      "loss": 0.6112,
      "step": 858
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6130844073092024,
      "learning_rate": 1.9671344555533424e-06,
      "loss": 0.5117,
      "step": 859
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9980313890468107,
      "learning_rate": 1.967029447605067e-06,
      "loss": 0.5986,
      "step": 860
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7556821285252112,
      "learning_rate": 1.966924274981752e-06,
      "loss": 0.5601,
      "step": 861
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6003288460302754,
      "learning_rate": 1.9668189377013064e-06,
      "loss": 0.5582,
      "step": 862
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8872025885557276,
      "learning_rate": 1.9667134357816673e-06,
      "loss": 0.617,
      "step": 863
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.807200861570258,
      "learning_rate": 1.9666077692408014e-06,
      "loss": 0.6408,
      "step": 864
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6506359020241926,
      "learning_rate": 1.9665019380967026e-06,
      "loss": 0.5487,
      "step": 865
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1707914896724807,
      "learning_rate": 1.9663959423673927e-06,
      "loss": 0.6624,
      "step": 866
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8941151580606443,
      "learning_rate": 1.9662897820709218e-06,
      "loss": 0.6957,
      "step": 867
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7871389132599979,
      "learning_rate": 1.9661834572253675e-06,
      "loss": 0.6946,
      "step": 868
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6656862842487034,
      "learning_rate": 1.966076967848836e-06,
      "loss": 0.5343,
      "step": 869
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8925960749544167,
      "learning_rate": 1.9659703139594614e-06,
      "loss": 0.6423,
      "step": 870
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8469531925547247,
      "learning_rate": 1.965863495575406e-06,
      "loss": 0.5805,
      "step": 871
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1238085900777257,
      "learning_rate": 1.9657565127148595e-06,
      "loss": 0.679,
      "step": 872
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.992012246640505,
      "learning_rate": 1.96564936539604e-06,
      "loss": 0.6616,
      "step": 873
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8081121377049941,
      "learning_rate": 1.9655420536371937e-06,
      "loss": 0.6942,
      "step": 874
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8756769987784445,
      "learning_rate": 1.9654345774565946e-06,
      "loss": 0.6944,
      "step": 875
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9070341601901982,
      "learning_rate": 1.9653269368725447e-06,
      "loss": 0.6715,
      "step": 876
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6954680880537243,
      "learning_rate": 1.965219131903374e-06,
      "loss": 0.5463,
      "step": 877
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9178385644566754,
      "learning_rate": 1.965111162567441e-06,
      "loss": 0.67,
      "step": 878
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7087952952823579,
      "learning_rate": 1.9650030288831316e-06,
      "loss": 0.5693,
      "step": 879
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.899839627152507,
      "learning_rate": 1.9648947308688595e-06,
      "loss": 0.6994,
      "step": 880
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1177015279770126,
      "learning_rate": 1.9647862685430667e-06,
      "loss": 0.6434,
      "step": 881
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7648707249075081,
      "learning_rate": 1.9646776419242236e-06,
      "loss": 0.5736,
      "step": 882
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7836341961325934,
      "learning_rate": 1.964568851030828e-06,
      "loss": 0.6624,
      "step": 883
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7209668121937278,
      "learning_rate": 1.964459895881406e-06,
      "loss": 0.6178,
      "step": 884
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7395842368060255,
      "learning_rate": 1.9643507764945113e-06,
      "loss": 0.5613,
      "step": 885
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9523517694947433,
      "learning_rate": 1.964241492888726e-06,
      "loss": 0.712,
      "step": 886
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.759579988839089,
      "learning_rate": 1.96413204508266e-06,
      "loss": 0.6016,
      "step": 887
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6415294384338988,
      "learning_rate": 1.964022433094951e-06,
      "loss": 0.5848,
      "step": 888
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7033070413563287,
      "learning_rate": 1.9639126569442644e-06,
      "loss": 0.5537,
      "step": 889
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.862684507791676,
      "learning_rate": 1.9638027166492942e-06,
      "loss": 0.6993,
      "step": 890
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8488339878826154,
      "learning_rate": 1.9636926122287627e-06,
      "loss": 0.666,
      "step": 891
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7676733778553675,
      "learning_rate": 1.9635823437014184e-06,
      "loss": 0.6668,
      "step": 892
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6198485149103461,
      "learning_rate": 1.96347191108604e-06,
      "loss": 0.574,
      "step": 893
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7306946953524489,
      "learning_rate": 1.9633613144014324e-06,
      "loss": 0.5582,
      "step": 894
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1483857194936544,
      "learning_rate": 1.963250553666429e-06,
      "loss": 0.6477,
      "step": 895
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8361601279203786,
      "learning_rate": 1.9631396288998917e-06,
      "loss": 0.7013,
      "step": 896
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6508504075829842,
      "learning_rate": 1.9630285401207093e-06,
      "loss": 0.5468,
      "step": 897
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9719928320331945,
      "learning_rate": 1.9629172873477994e-06,
      "loss": 0.7296,
      "step": 898
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6445508578130046,
      "learning_rate": 1.9628058706001067e-06,
      "loss": 0.5512,
      "step": 899
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8914683069239387,
      "learning_rate": 1.962694289896605e-06,
      "loss": 0.6344,
      "step": 900
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8449164193588393,
      "learning_rate": 1.962582545256295e-06,
      "loss": 0.6636,
      "step": 901
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9756133621343571,
      "learning_rate": 1.9624706366982055e-06,
      "loss": 0.7175,
      "step": 902
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8818123628675876,
      "learning_rate": 1.9623585642413935e-06,
      "loss": 0.633,
      "step": 903
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6678019445350563,
      "learning_rate": 1.962246327904944e-06,
      "loss": 0.5766,
      "step": 904
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8317168953782076,
      "learning_rate": 1.96213392770797e-06,
      "loss": 0.6421,
      "step": 905
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6981180381964973,
      "learning_rate": 1.96202136366961e-06,
      "loss": 0.6194,
      "step": 906
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7622682944041159,
      "learning_rate": 1.9619086358090354e-06,
      "loss": 0.5531,
      "step": 907
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.75578595477043,
      "learning_rate": 1.961795744145441e-06,
      "loss": 0.6329,
      "step": 908
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8655359020541531,
      "learning_rate": 1.9616826886980505e-06,
      "loss": 0.6678,
      "step": 909
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.616476173207024,
      "learning_rate": 1.961569469486118e-06,
      "loss": 0.5736,
      "step": 910
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9039803077091871,
      "learning_rate": 1.9614560865289216e-06,
      "loss": 0.6836,
      "step": 911
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8515318698678036,
      "learning_rate": 1.9613425398457705e-06,
      "loss": 0.6925,
      "step": 912
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8825450267905519,
      "learning_rate": 1.9612288294559996e-06,
      "loss": 0.6572,
      "step": 913
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7856344574933853,
      "learning_rate": 1.961114955378973e-06,
      "loss": 0.6003,
      "step": 914
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8901537464858076,
      "learning_rate": 1.9610009176340827e-06,
      "loss": 0.7376,
      "step": 915
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6570273588128311,
      "learning_rate": 1.9608867162407478e-06,
      "loss": 0.5807,
      "step": 916
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.679573944086101,
      "learning_rate": 1.960772351218415e-06,
      "loss": 0.5499,
      "step": 917
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.44691379510409,
      "learning_rate": 1.960657822586561e-06,
      "loss": 0.6875,
      "step": 918
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8628824940057535,
      "learning_rate": 1.960543130364687e-06,
      "loss": 0.6168,
      "step": 919
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6644367545946832,
      "learning_rate": 1.960428274572325e-06,
      "loss": 0.5247,
      "step": 920
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1566375496791679,
      "learning_rate": 1.960313255229034e-06,
      "loss": 0.594,
      "step": 921
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8486626490370862,
      "learning_rate": 1.9601980723543995e-06,
      "loss": 0.6319,
      "step": 922
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.635568398261324,
      "learning_rate": 1.960082725968037e-06,
      "loss": 0.5464,
      "step": 923
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9806401726812076,
      "learning_rate": 1.959967216089588e-06,
      "loss": 0.6549,
      "step": 924
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7509605500822271,
      "learning_rate": 1.959851542738723e-06,
      "loss": 0.5991,
      "step": 925
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9731404587910542,
      "learning_rate": 1.95973570593514e-06,
      "loss": 0.678,
      "step": 926
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9326375492091475,
      "learning_rate": 1.9596197056985647e-06,
      "loss": 0.7253,
      "step": 927
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7179170579867882,
      "learning_rate": 1.959503542048751e-06,
      "loss": 0.5706,
      "step": 928
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6737422602401815,
      "learning_rate": 1.95938721500548e-06,
      "loss": 0.6426,
      "step": 929
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7240756204658625,
      "learning_rate": 1.959270724588561e-06,
      "loss": 0.574,
      "step": 930
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6409236687110115,
      "learning_rate": 1.9591540708178313e-06,
      "loss": 0.5373,
      "step": 931
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.825775295230597,
      "learning_rate": 1.9590372537131557e-06,
      "loss": 0.6122,
      "step": 932
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.01668741116574,
      "learning_rate": 1.9589202732944267e-06,
      "loss": 0.6588,
      "step": 933
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.71473525983015,
      "learning_rate": 1.9588031295815656e-06,
      "loss": 0.5933,
      "step": 934
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7219179053880909,
      "learning_rate": 1.95868582259452e-06,
      "loss": 0.5991,
      "step": 935
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6357498739505724,
      "learning_rate": 1.9585683523532663e-06,
      "loss": 0.5712,
      "step": 936
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.685819651357079,
      "learning_rate": 1.9584507188778088e-06,
      "loss": 0.5586,
      "step": 937
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6831178281641995,
      "learning_rate": 1.9583329221881787e-06,
      "loss": 0.6349,
      "step": 938
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6512072512955748,
      "learning_rate": 1.9582149623044355e-06,
      "loss": 0.5205,
      "step": 939
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6091723736912761,
      "learning_rate": 1.958096839246667e-06,
      "loss": 0.5683,
      "step": 940
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6268845064929525,
      "learning_rate": 1.9579785530349886e-06,
      "loss": 0.5942,
      "step": 941
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8077310486510025,
      "learning_rate": 1.9578601036895426e-06,
      "loss": 0.6525,
      "step": 942
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.652601740939098,
      "learning_rate": 1.9577414912304994e-06,
      "loss": 0.5633,
      "step": 943
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8628204471744981,
      "learning_rate": 1.9576227156780584e-06,
      "loss": 0.6376,
      "step": 944
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.664294270848172,
      "learning_rate": 1.957503777052445e-06,
      "loss": 0.5413,
      "step": 945
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7130336865682105,
      "learning_rate": 1.957384675373914e-06,
      "loss": 0.6602,
      "step": 946
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7017616579615986,
      "learning_rate": 1.9572654106627465e-06,
      "loss": 0.5861,
      "step": 947
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8793748076341631,
      "learning_rate": 1.957145982939252e-06,
      "loss": 0.6156,
      "step": 948
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8792784393219998,
      "learning_rate": 1.9570263922237684e-06,
      "loss": 0.6221,
      "step": 949
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8944236817388244,
      "learning_rate": 1.9569066385366606e-06,
      "loss": 0.6488,
      "step": 950
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6130274294674304,
      "learning_rate": 1.956786721898321e-06,
      "loss": 0.552,
      "step": 951
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9118972880536662,
      "learning_rate": 1.9566666423291703e-06,
      "loss": 0.6614,
      "step": 952
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.065551707146563,
      "learning_rate": 1.956546399849657e-06,
      "loss": 0.7326,
      "step": 953
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7582685110301709,
      "learning_rate": 1.956425994480257e-06,
      "loss": 0.531,
      "step": 954
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6653634723779718,
      "learning_rate": 1.956305426241474e-06,
      "loss": 0.5496,
      "step": 955
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7575504632353254,
      "learning_rate": 1.95618469515384e-06,
      "loss": 0.6576,
      "step": 956
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8653980295606406,
      "learning_rate": 1.9560638012379134e-06,
      "loss": 0.6313,
      "step": 957
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8547768248261564,
      "learning_rate": 1.9559427445142822e-06,
      "loss": 0.6521,
      "step": 958
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7431944461334282,
      "learning_rate": 1.9558215250035605e-06,
      "loss": 0.5849,
      "step": 959
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7654466003345309,
      "learning_rate": 1.9557001427263905e-06,
      "loss": 0.6681,
      "step": 960
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8455995576267222,
      "learning_rate": 1.9555785977034424e-06,
      "loss": 0.623,
      "step": 961
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.839411812822503,
      "learning_rate": 1.9554568899554146e-06,
      "loss": 0.6912,
      "step": 962
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7728414189959457,
      "learning_rate": 1.9553350195030323e-06,
      "loss": 0.595,
      "step": 963
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6091506068133549,
      "learning_rate": 1.955212986367049e-06,
      "loss": 0.5307,
      "step": 964
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.750434124018174,
      "learning_rate": 1.9550907905682447e-06,
      "loss": 0.5552,
      "step": 965
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7316859418842425,
      "learning_rate": 1.9549684321274295e-06,
      "loss": 0.6154,
      "step": 966
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8255528426582034,
      "learning_rate": 1.954845911065439e-06,
      "loss": 0.6492,
      "step": 967
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7060440370455678,
      "learning_rate": 1.954723227403137e-06,
      "loss": 0.5759,
      "step": 968
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6691216005609768,
      "learning_rate": 1.954600381161416e-06,
      "loss": 0.5781,
      "step": 969
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9218309446186043,
      "learning_rate": 1.954477372361195e-06,
      "loss": 0.6753,
      "step": 970
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8788392012012401,
      "learning_rate": 1.954354201023421e-06,
      "loss": 0.6261,
      "step": 971
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0672925516054856,
      "learning_rate": 1.9542308671690687e-06,
      "loss": 0.6717,
      "step": 972
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8246027701640113,
      "learning_rate": 1.9541073708191413e-06,
      "loss": 0.6718,
      "step": 973
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6586986780432992,
      "learning_rate": 1.9539837119946684e-06,
      "loss": 0.5095,
      "step": 974
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6308706666400716,
      "learning_rate": 1.953859890716708e-06,
      "loss": 0.5711,
      "step": 975
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8512473873273898,
      "learning_rate": 1.953735907006345e-06,
      "loss": 0.6957,
      "step": 976
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8637807721394538,
      "learning_rate": 1.9536117608846934e-06,
      "loss": 0.648,
      "step": 977
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.899870011434305,
      "learning_rate": 1.9534874523728934e-06,
      "loss": 0.6489,
      "step": 978
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6545434401739678,
      "learning_rate": 1.9533629814921135e-06,
      "loss": 0.5448,
      "step": 979
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7016465648287954,
      "learning_rate": 1.95323834826355e-06,
      "loss": 0.5948,
      "step": 980
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.885420103869679,
      "learning_rate": 1.9531135527084266e-06,
      "loss": 0.6262,
      "step": 981
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0647739163061707,
      "learning_rate": 1.952988594847995e-06,
      "loss": 0.6571,
      "step": 982
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6760371753851048,
      "learning_rate": 1.952863474703533e-06,
      "loss": 0.5715,
      "step": 983
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6315718168842409,
      "learning_rate": 1.9527381922963485e-06,
      "loss": 0.5408,
      "step": 984
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7097733938811038,
      "learning_rate": 1.952612747647776e-06,
      "loss": 0.5439,
      "step": 985
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8054186509689291,
      "learning_rate": 1.952487140779176e-06,
      "loss": 0.6633,
      "step": 986
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7057688267227478,
      "learning_rate": 1.9523613717119395e-06,
      "loss": 0.5785,
      "step": 987
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8099029387321683,
      "learning_rate": 1.9522354404674825e-06,
      "loss": 0.6959,
      "step": 988
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8145500774131236,
      "learning_rate": 1.9521093470672506e-06,
      "loss": 0.6414,
      "step": 989
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8786986949218748,
      "learning_rate": 1.951983091532716e-06,
      "loss": 0.6559,
      "step": 990
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8970142537100135,
      "learning_rate": 1.9518566738853783e-06,
      "loss": 0.6331,
      "step": 991
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8894416744581568,
      "learning_rate": 1.9517300941467654e-06,
      "loss": 0.6474,
      "step": 992
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9298782210216464,
      "learning_rate": 1.951603352338433e-06,
      "loss": 0.6992,
      "step": 993
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7446927148432636,
      "learning_rate": 1.951476448481963e-06,
      "loss": 0.5679,
      "step": 994
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.077894191402499,
      "learning_rate": 1.9513493825989663e-06,
      "loss": 0.7248,
      "step": 995
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9748796028914771,
      "learning_rate": 1.951222154711081e-06,
      "loss": 0.6901,
      "step": 996
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8218818474312494,
      "learning_rate": 1.951094764839972e-06,
      "loss": 0.6209,
      "step": 997
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7422075710612643,
      "learning_rate": 1.9509672130073336e-06,
      "loss": 0.5551,
      "step": 998
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8760039636783977,
      "learning_rate": 1.9508394992348855e-06,
      "loss": 0.6702,
      "step": 999
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6571775809869482,
      "learning_rate": 1.9507116235443768e-06,
      "loss": 0.5427,
      "step": 1000
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7412628423125,
      "learning_rate": 1.9505835859575827e-06,
      "loss": 0.5228,
      "step": 1001
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6826359699964017,
      "learning_rate": 1.9504553864963075e-06,
      "loss": 0.5629,
      "step": 1002
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8420154398581653,
      "learning_rate": 1.9503270251823813e-06,
      "loss": 0.6353,
      "step": 1003
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3194748711940778,
      "learning_rate": 1.950198502037663e-06,
      "loss": 0.6344,
      "step": 1004
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7165792860786188,
      "learning_rate": 1.9500698170840395e-06,
      "loss": 0.5669,
      "step": 1005
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8242020899197583,
      "learning_rate": 1.9499409703434235e-06,
      "loss": 0.6216,
      "step": 1006
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7774332047427653,
      "learning_rate": 1.9498119618377567e-06,
      "loss": 0.5632,
      "step": 1007
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8802185040286188,
      "learning_rate": 1.949682791589008e-06,
      "loss": 0.6679,
      "step": 1008
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8885269717467343,
      "learning_rate": 1.9495534596191737e-06,
      "loss": 0.6544,
      "step": 1009
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6753136727591234,
      "learning_rate": 1.9494239659502777e-06,
      "loss": 0.5372,
      "step": 1010
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.691133612907113,
      "learning_rate": 1.9492943106043707e-06,
      "loss": 0.5546,
      "step": 1011
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7857458950595669,
      "learning_rate": 1.9491644936035327e-06,
      "loss": 0.6296,
      "step": 1012
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7673595644084344,
      "learning_rate": 1.94903451496987e-06,
      "loss": 0.6087,
      "step": 1013
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9954746942129905,
      "learning_rate": 1.948904374725516e-06,
      "loss": 0.6648,
      "step": 1014
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0116626132420077,
      "learning_rate": 1.948774072892633e-06,
      "loss": 0.6337,
      "step": 1015
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8298368388961275,
      "learning_rate": 1.948643609493409e-06,
      "loss": 0.6495,
      "step": 1016
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7551172622682851,
      "learning_rate": 1.9485129845500617e-06,
      "loss": 0.6196,
      "step": 1017
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5858320444489276,
      "learning_rate": 1.9483821980848344e-06,
      "loss": 0.5332,
      "step": 1018
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8764961495990471,
      "learning_rate": 1.948251250119999e-06,
      "loss": 0.6675,
      "step": 1019
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8536165652066033,
      "learning_rate": 1.948120140677855e-06,
      "loss": 0.6727,
      "step": 1020
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0002376823235686,
      "learning_rate": 1.947988869780728e-06,
      "loss": 0.7281,
      "step": 1021
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5923346454463809,
      "learning_rate": 1.9478574374509723e-06,
      "loss": 0.5542,
      "step": 1022
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.889302258009896,
      "learning_rate": 1.94772584371097e-06,
      "loss": 0.6577,
      "step": 1023
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.704760735458918,
      "learning_rate": 1.94759408858313e-06,
      "loss": 0.6083,
      "step": 1024
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6406526632342663,
      "learning_rate": 1.9474621720898884e-06,
      "loss": 0.5366,
      "step": 1025
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8694413281708848,
      "learning_rate": 1.9473300942537097e-06,
      "loss": 0.5898,
      "step": 1026
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8963454623231697,
      "learning_rate": 1.947197855097085e-06,
      "loss": 0.6668,
      "step": 1027
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7813685713235143,
      "learning_rate": 1.9470654546425336e-06,
      "loss": 0.6389,
      "step": 1028
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7472000999867008,
      "learning_rate": 1.946932892912602e-06,
      "loss": 0.6222,
      "step": 1029
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7848193039094754,
      "learning_rate": 1.9468001699298633e-06,
      "loss": 0.541,
      "step": 1030
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7863990173191229,
      "learning_rate": 1.9466672857169195e-06,
      "loss": 0.6143,
      "step": 1031
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8157438486641113,
      "learning_rate": 1.9465342402963993e-06,
      "loss": 0.6673,
      "step": 1032
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6156427076836347,
      "learning_rate": 1.9464010336909586e-06,
      "loss": 0.5463,
      "step": 1033
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7647861765516883,
      "learning_rate": 1.946267665923282e-06,
      "loss": 0.6641,
      "step": 1034
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7117216345407701,
      "learning_rate": 1.9461341370160796e-06,
      "loss": 0.5857,
      "step": 1035
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9080814882338024,
      "learning_rate": 1.9460004469920905e-06,
      "loss": 0.6243,
      "step": 1036
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6305319940050684,
      "learning_rate": 1.9458665958740803e-06,
      "loss": 0.6011,
      "step": 1037
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7998340149246962,
      "learning_rate": 1.9457325836848435e-06,
      "loss": 0.653,
      "step": 1038
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7963617009389307,
      "learning_rate": 1.9455984104471995e-06,
      "loss": 0.6008,
      "step": 1039
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.692624102242373,
      "learning_rate": 1.945464076183998e-06,
      "loss": 0.5426,
      "step": 1040
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6760912764255483,
      "learning_rate": 1.9453295809181142e-06,
      "loss": 0.5765,
      "step": 1041
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7388277011341495,
      "learning_rate": 1.945194924672451e-06,
      "loss": 0.6318,
      "step": 1042
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7447928289285848,
      "learning_rate": 1.9450601074699386e-06,
      "loss": 0.5986,
      "step": 1043
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9642657349404752,
      "learning_rate": 1.944925129333536e-06,
      "loss": 0.6817,
      "step": 1044
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6877706625509188,
      "learning_rate": 1.9447899902862276e-06,
      "loss": 0.597,
      "step": 1045
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8698614090595488,
      "learning_rate": 1.944654690351027e-06,
      "loss": 0.6931,
      "step": 1046
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0981606813145115,
      "learning_rate": 1.9445192295509737e-06,
      "loss": 0.618,
      "step": 1047
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6518020122966293,
      "learning_rate": 1.9443836079091356e-06,
      "loss": 0.5644,
      "step": 1048
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8747205072849583,
      "learning_rate": 1.9442478254486074e-06,
      "loss": 0.6974,
      "step": 1049
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9746118674511695,
      "learning_rate": 1.944111882192512e-06,
      "loss": 0.5603,
      "step": 1050
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.653193248833744,
      "learning_rate": 1.9439757781639987e-06,
      "loss": 0.5498,
      "step": 1051
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.633627634503781,
      "learning_rate": 1.9438395133862445e-06,
      "loss": 0.5263,
      "step": 1052
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6637626784956492,
      "learning_rate": 1.9437030878824537e-06,
      "loss": 0.5283,
      "step": 1053
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0968658575531014,
      "learning_rate": 1.9435665016758593e-06,
      "loss": 0.6985,
      "step": 1054
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7148896291312722,
      "learning_rate": 1.9434297547897193e-06,
      "loss": 0.5965,
      "step": 1055
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6091746246190839,
      "learning_rate": 1.94329284724732e-06,
      "loss": 0.5453,
      "step": 1056
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8983334622505745,
      "learning_rate": 1.9431557790719767e-06,
      "loss": 0.6965,
      "step": 1057
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.826693885522798,
      "learning_rate": 1.94301855028703e-06,
      "loss": 0.6532,
      "step": 1058
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9314442890591906,
      "learning_rate": 1.942881160915848e-06,
      "loss": 0.681,
      "step": 1059
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7907594912255953,
      "learning_rate": 1.9427436109818273e-06,
      "loss": 0.5827,
      "step": 1060
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7616363607335165,
      "learning_rate": 1.9426059005083914e-06,
      "loss": 0.5528,
      "step": 1061
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8336706489849964,
      "learning_rate": 1.9424680295189902e-06,
      "loss": 0.6203,
      "step": 1062
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7967345028916949,
      "learning_rate": 1.942329998037103e-06,
      "loss": 0.6553,
      "step": 1063
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6806951122840847,
      "learning_rate": 1.942191806086233e-06,
      "loss": 0.5197,
      "step": 1064
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6617381784232839,
      "learning_rate": 1.9420534536899147e-06,
      "loss": 0.5088,
      "step": 1065
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7308480331089972,
      "learning_rate": 1.941914940871708e-06,
      "loss": 0.5363,
      "step": 1066
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6933216069577045,
      "learning_rate": 1.941776267655199e-06,
      "loss": 0.6042,
      "step": 1067
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9587349504708711,
      "learning_rate": 1.941637434064003e-06,
      "loss": 0.6103,
      "step": 1068
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8942256929096047,
      "learning_rate": 1.941498440121762e-06,
      "loss": 0.6581,
      "step": 1069
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6647307134916989,
      "learning_rate": 1.9413592858521454e-06,
      "loss": 0.5873,
      "step": 1070
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6955678592577135,
      "learning_rate": 1.941219971278849e-06,
      "loss": 0.5614,
      "step": 1071
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8754055677719764,
      "learning_rate": 1.941080496425597e-06,
      "loss": 0.5592,
      "step": 1072
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6329092371201205,
      "learning_rate": 1.940940861316141e-06,
      "loss": 0.4778,
      "step": 1073
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7674168003496451,
      "learning_rate": 1.940801065974259e-06,
      "loss": 0.6417,
      "step": 1074
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7114106878583818,
      "learning_rate": 1.9406611104237563e-06,
      "loss": 0.5752,
      "step": 1075
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.800945537690414,
      "learning_rate": 1.9405209946884663e-06,
      "loss": 0.6345,
      "step": 1076
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8979484550796473,
      "learning_rate": 1.9403807187922488e-06,
      "loss": 0.6309,
      "step": 1077
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6812372448121354,
      "learning_rate": 1.9402402827589922e-06,
      "loss": 0.575,
      "step": 1078
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7416745660763798,
      "learning_rate": 1.940099686612611e-06,
      "loss": 0.5853,
      "step": 1079
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8105997977534364,
      "learning_rate": 1.939958930377047e-06,
      "loss": 0.6405,
      "step": 1080
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7066066770629402,
      "learning_rate": 1.9398180140762693e-06,
      "loss": 0.571,
      "step": 1081
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8486341886656302,
      "learning_rate": 1.9396769377342746e-06,
      "loss": 0.6644,
      "step": 1082
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8071678200527111,
      "learning_rate": 1.939535701375087e-06,
      "loss": 0.5723,
      "step": 1083
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6930113320194837,
      "learning_rate": 1.9393943050227577e-06,
      "loss": 0.5229,
      "step": 1084
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9604468919910021,
      "learning_rate": 1.9392527487013644e-06,
      "loss": 0.5919,
      "step": 1085
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7988565519555352,
      "learning_rate": 1.9391110324350137e-06,
      "loss": 0.5853,
      "step": 1086
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6915793982238381,
      "learning_rate": 1.938969156247837e-06,
      "loss": 0.5946,
      "step": 1087
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8067418405855121,
      "learning_rate": 1.938827120163996e-06,
      "loss": 0.5705,
      "step": 1088
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1070727048737719,
      "learning_rate": 1.9386849242076765e-06,
      "loss": 0.7025,
      "step": 1089
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5964867751025502,
      "learning_rate": 1.9385425684030932e-06,
      "loss": 0.5059,
      "step": 1090
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6949607044823535,
      "learning_rate": 1.938400052774489e-06,
      "loss": 0.5495,
      "step": 1091
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9194822693613911,
      "learning_rate": 1.938257377346131e-06,
      "loss": 0.7148,
      "step": 1092
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9797311933107593,
      "learning_rate": 1.9381145421423174e-06,
      "loss": 0.687,
      "step": 1093
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9141821067600461,
      "learning_rate": 1.9379715471873698e-06,
      "loss": 0.6134,
      "step": 1094
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.668377195455847,
      "learning_rate": 1.9378283925056397e-06,
      "loss": 0.519,
      "step": 1095
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6993184531571733,
      "learning_rate": 1.9376850781215043e-06,
      "loss": 0.5371,
      "step": 1096
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7525847791352538,
      "learning_rate": 1.937541604059369e-06,
      "loss": 0.5674,
      "step": 1097
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.693140727198669,
      "learning_rate": 1.937397970343666e-06,
      "loss": 0.6131,
      "step": 1098
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7664767087052541,
      "learning_rate": 1.937254176998854e-06,
      "loss": 0.5932,
      "step": 1099
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9408770088213105,
      "learning_rate": 1.9371102240494204e-06,
      "loss": 0.6181,
      "step": 1100
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9873221004482454,
      "learning_rate": 1.9369661115198777e-06,
      "loss": 0.6332,
      "step": 1101
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8872834912327237,
      "learning_rate": 1.936821839434768e-06,
      "loss": 0.6027,
      "step": 1102
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8205679072103947,
      "learning_rate": 1.9366774078186585e-06,
      "loss": 0.6139,
      "step": 1103
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8048346361478271,
      "learning_rate": 1.936532816696145e-06,
      "loss": 0.6687,
      "step": 1104
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7004455698753029,
      "learning_rate": 1.936388066091849e-06,
      "loss": 0.5787,
      "step": 1105
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7495237974386355,
      "learning_rate": 1.9362431560304212e-06,
      "loss": 0.6173,
      "step": 1106
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5798545724881402,
      "learning_rate": 1.9360980865365374e-06,
      "loss": 0.5187,
      "step": 1107
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8217930525066944,
      "learning_rate": 1.9359528576349018e-06,
      "loss": 0.617,
      "step": 1108
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6276164211494447,
      "learning_rate": 1.9358074693502448e-06,
      "loss": 0.5076,
      "step": 1109
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7804785567807171,
      "learning_rate": 1.9356619217073252e-06,
      "loss": 0.5665,
      "step": 1110
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.666291863444723,
      "learning_rate": 1.9355162147309277e-06,
      "loss": 0.5794,
      "step": 1111
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8244689364191091,
      "learning_rate": 1.9353703484458656e-06,
      "loss": 0.6334,
      "step": 1112
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6449582618268463,
      "learning_rate": 1.9352243228769773e-06,
      "loss": 0.5536,
      "step": 1113
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8638339888611027,
      "learning_rate": 1.93507813804913e-06,
      "loss": 0.6589,
      "step": 1114
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6733519285145788,
      "learning_rate": 1.9349317939872172e-06,
      "loss": 0.5967,
      "step": 1115
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8664413783808915,
      "learning_rate": 1.9347852907161603e-06,
      "loss": 0.6614,
      "step": 1116
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6346462987485083,
      "learning_rate": 1.9346386282609065e-06,
      "loss": 0.5033,
      "step": 1117
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8142530139378948,
      "learning_rate": 1.934491806646432e-06,
      "loss": 0.6477,
      "step": 1118
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6169218432922565,
      "learning_rate": 1.9343448258977373e-06,
      "loss": 0.5653,
      "step": 1119
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.774852317712845,
      "learning_rate": 1.9341976860398537e-06,
      "loss": 0.6113,
      "step": 1120
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6720945225673164,
      "learning_rate": 1.9340503870978357e-06,
      "loss": 0.5514,
      "step": 1121
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6363619153800679,
      "learning_rate": 1.933902929096768e-06,
      "loss": 0.5394,
      "step": 1122
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.787582133429448,
      "learning_rate": 1.933755312061761e-06,
      "loss": 0.6362,
      "step": 1123
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5939045871277261,
      "learning_rate": 1.933607536017952e-06,
      "loss": 0.5286,
      "step": 1124
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6270578282888595,
      "learning_rate": 1.9334596009905057e-06,
      "loss": 0.5314,
      "step": 1125
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8158005581580233,
      "learning_rate": 1.9333115070046143e-06,
      "loss": 0.6076,
      "step": 1126
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.940805203778106,
      "learning_rate": 1.9331632540854964e-06,
      "loss": 0.6323,
      "step": 1127
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.938243551934598,
      "learning_rate": 1.9330148422583978e-06,
      "loss": 0.6594,
      "step": 1128
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6647476577798627,
      "learning_rate": 1.932866271548592e-06,
      "loss": 0.579,
      "step": 1129
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7710684211151441,
      "learning_rate": 1.932717541981378e-06,
      "loss": 0.5412,
      "step": 1130
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7015095693859049,
      "learning_rate": 1.932568653582084e-06,
      "loss": 0.6298,
      "step": 1131
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9974915380239072,
      "learning_rate": 1.9324196063760637e-06,
      "loss": 0.6829,
      "step": 1132
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7183317653473398,
      "learning_rate": 1.9322704003886987e-06,
      "loss": 0.5412,
      "step": 1133
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9903190718425698,
      "learning_rate": 1.932121035645396e-06,
      "loss": 0.68,
      "step": 1134
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6292104037721264,
      "learning_rate": 1.9319715121715927e-06,
      "loss": 0.5313,
      "step": 1135
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8115527558001782,
      "learning_rate": 1.9318218299927495e-06,
      "loss": 0.5862,
      "step": 1136
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.737012421342107,
      "learning_rate": 1.9316719891343564e-06,
      "loss": 0.5973,
      "step": 1137
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6210320997555163,
      "learning_rate": 1.9315219896219294e-06,
      "loss": 0.5288,
      "step": 1138
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8140718376612642,
      "learning_rate": 1.9313718314810125e-06,
      "loss": 0.6733,
      "step": 1139
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9391408687771978,
      "learning_rate": 1.931221514737175e-06,
      "loss": 0.6565,
      "step": 1140
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9767178375198257,
      "learning_rate": 1.9310710394160153e-06,
      "loss": 0.6191,
      "step": 1141
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8675583238319798,
      "learning_rate": 1.9309204055431576e-06,
      "loss": 0.6381,
      "step": 1142
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7599491843557507,
      "learning_rate": 1.930769613144253e-06,
      "loss": 0.5328,
      "step": 1143
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7414144123790753,
      "learning_rate": 1.93061866224498e-06,
      "loss": 0.5424,
      "step": 1144
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6698290718586358,
      "learning_rate": 1.930467552871044e-06,
      "loss": 0.565,
      "step": 1145
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7534501339864335,
      "learning_rate": 1.930316285048177e-06,
      "loss": 0.6427,
      "step": 1146
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8339853243125236,
      "learning_rate": 1.930164858802139e-06,
      "loss": 0.6028,
      "step": 1147
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9429979528669824,
      "learning_rate": 1.9300132741587156e-06,
      "loss": 0.5771,
      "step": 1148
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7354834951424498,
      "learning_rate": 1.9298615311437206e-06,
      "loss": 0.5634,
      "step": 1149
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.860742281353345,
      "learning_rate": 1.929709629782994e-06,
      "loss": 0.5994,
      "step": 1150
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6398553148802747,
      "learning_rate": 1.929557570102403e-06,
      "loss": 0.5468,
      "step": 1151
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4751903744678585,
      "learning_rate": 1.9294053521278423e-06,
      "loss": 0.6816,
      "step": 1152
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6097576223036705,
      "learning_rate": 1.929252975885232e-06,
      "loss": 0.4991,
      "step": 1153
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7700467787991755,
      "learning_rate": 1.9291004414005213e-06,
      "loss": 0.5818,
      "step": 1154
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8715224903533787,
      "learning_rate": 1.9289477486996842e-06,
      "loss": 0.6894,
      "step": 1155
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.786570041884722,
      "learning_rate": 1.928794897808724e-06,
      "loss": 0.5328,
      "step": 1156
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8336596147355181,
      "learning_rate": 1.9286418887536683e-06,
      "loss": 0.611,
      "step": 1157
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8712321263213032,
      "learning_rate": 1.9284887215605735e-06,
      "loss": 0.6658,
      "step": 1158
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7776757576728112,
      "learning_rate": 1.928335396255522e-06,
      "loss": 0.5644,
      "step": 1159
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7908444408008988,
      "learning_rate": 1.9281819128646246e-06,
      "loss": 0.5738,
      "step": 1160
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.670328077386675,
      "learning_rate": 1.9280282714140167e-06,
      "loss": 0.5772,
      "step": 1161
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7629518462574484,
      "learning_rate": 1.9278744719298622e-06,
      "loss": 0.6581,
      "step": 1162
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7432328372310353,
      "learning_rate": 1.927720514438352e-06,
      "loss": 0.5842,
      "step": 1163
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6532966563548461,
      "learning_rate": 1.9275663989657025e-06,
      "loss": 0.5392,
      "step": 1164
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6472521563349904,
      "learning_rate": 1.927412125538159e-06,
      "loss": 0.583,
      "step": 1165
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.724790562194451,
      "learning_rate": 1.927257694181993e-06,
      "loss": 0.5886,
      "step": 1166
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6812233207189767,
      "learning_rate": 1.9271031049235005e-06,
      "loss": 0.5393,
      "step": 1167
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7686983648087132,
      "learning_rate": 1.9269483577890083e-06,
      "loss": 0.6548,
      "step": 1168
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9370445233884142,
      "learning_rate": 1.9267934528048675e-06,
      "loss": 0.66,
      "step": 1169
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6936458927325159,
      "learning_rate": 1.9266383899974572e-06,
      "loss": 0.5285,
      "step": 1170
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.762879632520005,
      "learning_rate": 1.9264831693931825e-06,
      "loss": 0.5218,
      "step": 1171
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9400237105848452,
      "learning_rate": 1.926327791018476e-06,
      "loss": 0.6967,
      "step": 1172
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7673926913017607,
      "learning_rate": 1.9261722548997974e-06,
      "loss": 0.6223,
      "step": 1173
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8304891664403465,
      "learning_rate": 1.9260165610636326e-06,
      "loss": 0.6436,
      "step": 1174
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7012564208699866,
      "learning_rate": 1.9258607095364943e-06,
      "loss": 0.5651,
      "step": 1175
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6886271716997511,
      "learning_rate": 1.925704700344923e-06,
      "loss": 0.5646,
      "step": 1176
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7800014769126741,
      "learning_rate": 1.925548533515485e-06,
      "loss": 0.661,
      "step": 1177
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8427186840743779,
      "learning_rate": 1.925392209074774e-06,
      "loss": 0.6423,
      "step": 1178
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.698738182465362,
      "learning_rate": 1.9252357270494107e-06,
      "loss": 0.5776,
      "step": 1179
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7513397289871464,
      "learning_rate": 1.9250790874660422e-06,
      "loss": 0.5704,
      "step": 1180
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6303067424437214,
      "learning_rate": 1.924922290351342e-06,
      "loss": 0.5467,
      "step": 1181
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6395174639967565,
      "learning_rate": 1.9247653357320117e-06,
      "loss": 0.5346,
      "step": 1182
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7957693693675866,
      "learning_rate": 1.9246082236347792e-06,
      "loss": 0.6355,
      "step": 1183
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6398065719930426,
      "learning_rate": 1.9244509540863983e-06,
      "loss": 0.5209,
      "step": 1184
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.639712705423486,
      "learning_rate": 1.924293527113651e-06,
      "loss": 0.6506,
      "step": 1185
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6980691997113211,
      "learning_rate": 1.924135942743345e-06,
      "loss": 0.5528,
      "step": 1186
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.638585163911734,
      "learning_rate": 1.923978201002316e-06,
      "loss": 0.631,
      "step": 1187
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8966225408569751,
      "learning_rate": 1.9238203019174247e-06,
      "loss": 0.7119,
      "step": 1188
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9158143712687993,
      "learning_rate": 1.923662245515561e-06,
      "loss": 0.6357,
      "step": 1189
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9116909233445981,
      "learning_rate": 1.923504031823639e-06,
      "loss": 0.6825,
      "step": 1190
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9560421431791293,
      "learning_rate": 1.9233456608686013e-06,
      "loss": 0.7004,
      "step": 1191
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6541064378880194,
      "learning_rate": 1.9231871326774175e-06,
      "loss": 0.5368,
      "step": 1192
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6955362963247771,
      "learning_rate": 1.9230284472770823e-06,
      "loss": 0.5227,
      "step": 1193
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0697263745374144,
      "learning_rate": 1.9228696046946184e-06,
      "loss": 0.6743,
      "step": 1194
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7785250147133688,
      "learning_rate": 1.9227106049570754e-06,
      "loss": 0.6277,
      "step": 1195
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2999743228369813,
      "learning_rate": 1.922551448091529e-06,
      "loss": 0.6481,
      "step": 1196
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8350688508876185,
      "learning_rate": 1.9223921341250823e-06,
      "loss": 0.6411,
      "step": 1197
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7493061457870274,
      "learning_rate": 1.9222326630848642e-06,
      "loss": 0.5784,
      "step": 1198
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7494630702656563,
      "learning_rate": 1.922073034998032e-06,
      "loss": 0.6195,
      "step": 1199
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7440114217932107,
      "learning_rate": 1.921913249891767e-06,
      "loss": 0.5639,
      "step": 1200
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8520563469938737,
      "learning_rate": 1.9217533077932808e-06,
      "loss": 0.704,
      "step": 1201
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7426956896876387,
      "learning_rate": 1.921593208729809e-06,
      "loss": 0.5392,
      "step": 1202
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.634241890454269,
      "learning_rate": 1.9214329527286144e-06,
      "loss": 0.5168,
      "step": 1203
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5920910859766049,
      "learning_rate": 1.9212725398169876e-06,
      "loss": 0.5321,
      "step": 1204
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6291221979034487,
      "learning_rate": 1.921111970022245e-06,
      "loss": 0.5392,
      "step": 1205
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7099488534998689,
      "learning_rate": 1.92095124337173e-06,
      "loss": 0.5931,
      "step": 1206
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7095776035344237,
      "learning_rate": 1.9207903598928127e-06,
      "loss": 0.6143,
      "step": 1207
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9281897737044447,
      "learning_rate": 1.9206293196128898e-06,
      "loss": 0.6815,
      "step": 1208
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6131777117406543,
      "learning_rate": 1.9204681225593846e-06,
      "loss": 0.5465,
      "step": 1209
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6978483711135388,
      "learning_rate": 1.920306768759748e-06,
      "loss": 0.6274,
      "step": 1210
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7313055184658558,
      "learning_rate": 1.9201452582414557e-06,
      "loss": 0.5637,
      "step": 1211
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7262895437306012,
      "learning_rate": 1.9199835910320125e-06,
      "loss": 0.643,
      "step": 1212
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7738901016999956,
      "learning_rate": 1.919821767158948e-06,
      "loss": 0.6595,
      "step": 1213
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8209433126268498,
      "learning_rate": 1.919659786649819e-06,
      "loss": 0.5931,
      "step": 1214
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7406550822176572,
      "learning_rate": 1.9194976495322093e-06,
      "loss": 0.6336,
      "step": 1215
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7307479253798567,
      "learning_rate": 1.9193353558337295e-06,
      "loss": 0.6122,
      "step": 1216
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7776519926617271,
      "learning_rate": 1.9191729055820155e-06,
      "loss": 0.6052,
      "step": 1217
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.797352638810382,
      "learning_rate": 1.9190102988047317e-06,
      "loss": 0.7196,
      "step": 1218
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7274545846154133,
      "learning_rate": 1.9188475355295687e-06,
      "loss": 0.6051,
      "step": 1219
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9351566668143381,
      "learning_rate": 1.9186846157842425e-06,
      "loss": 0.6405,
      "step": 1220
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6407916109139107,
      "learning_rate": 1.918521539596497e-06,
      "loss": 0.519,
      "step": 1221
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6005573923046778,
      "learning_rate": 1.9183583069941023e-06,
      "loss": 0.577,
      "step": 1222
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7745414623542163,
      "learning_rate": 1.9181949180048557e-06,
      "loss": 0.6147,
      "step": 1223
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8834513708937011,
      "learning_rate": 1.91803137265658e-06,
      "loss": 0.663,
      "step": 1224
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5710287257601654,
      "learning_rate": 1.9178676709771255e-06,
      "loss": 0.5369,
      "step": 1225
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7079436663247748,
      "learning_rate": 1.917703812994369e-06,
      "loss": 0.5929,
      "step": 1226
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6460241075000596,
      "learning_rate": 1.9175397987362135e-06,
      "loss": 0.5378,
      "step": 1227
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7081389172815225,
      "learning_rate": 1.9173756282305896e-06,
      "loss": 0.5482,
      "step": 1228
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6461571994523665,
      "learning_rate": 1.9172113015054528e-06,
      "loss": 0.5734,
      "step": 1229
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7534929250182225,
      "learning_rate": 1.9170468185887872e-06,
      "loss": 0.6634,
      "step": 1230
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7345784124499537,
      "learning_rate": 1.916882179508602e-06,
      "loss": 0.5427,
      "step": 1231
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7672033332877664,
      "learning_rate": 1.916717384292934e-06,
      "loss": 0.5881,
      "step": 1232
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.829451785820827,
      "learning_rate": 1.916552432969845e-06,
      "loss": 0.6594,
      "step": 1233
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6975740148881469,
      "learning_rate": 1.9163873255674257e-06,
      "loss": 0.5478,
      "step": 1234
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6780230212112954,
      "learning_rate": 1.916222062113792e-06,
      "loss": 0.5676,
      "step": 1235
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7803842373117257,
      "learning_rate": 1.9160566426370855e-06,
      "loss": 0.6543,
      "step": 1236
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0437282719949326,
      "learning_rate": 1.9158910671654764e-06,
      "loss": 0.6369,
      "step": 1237
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6786858065995152,
      "learning_rate": 1.9157253357271602e-06,
      "loss": 0.5551,
      "step": 1238
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7318216384548416,
      "learning_rate": 1.9155594483503597e-06,
      "loss": 0.5494,
      "step": 1239
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7815951022116578,
      "learning_rate": 1.9153934050633234e-06,
      "loss": 0.6044,
      "step": 1240
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7414886556393671,
      "learning_rate": 1.915227205894326e-06,
      "loss": 0.6446,
      "step": 1241
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7273874048409005,
      "learning_rate": 1.9150608508716708e-06,
      "loss": 0.5764,
      "step": 1242
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7212465846983127,
      "learning_rate": 1.9148943400236855e-06,
      "loss": 0.5691,
      "step": 1243
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0413453606803564,
      "learning_rate": 1.9147276733787258e-06,
      "loss": 0.6958,
      "step": 1244
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8133200220645784,
      "learning_rate": 1.9145608509651724e-06,
      "loss": 0.6337,
      "step": 1245
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0306294731727854,
      "learning_rate": 1.9143938728114343e-06,
      "loss": 0.6847,
      "step": 1246
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6220869872372307,
      "learning_rate": 1.9142267389459453e-06,
      "loss": 0.5334,
      "step": 1247
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7080435666478909,
      "learning_rate": 1.914059449397167e-06,
      "loss": 0.5926,
      "step": 1248
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8293647595700109,
      "learning_rate": 1.9138920041935875e-06,
      "loss": 0.5953,
      "step": 1249
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9641881247719795,
      "learning_rate": 1.9137244033637203e-06,
      "loss": 0.6273,
      "step": 1250
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8277414240893562,
      "learning_rate": 1.9135566469361068e-06,
      "loss": 0.5815,
      "step": 1251
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8658589704596521,
      "learning_rate": 1.913388734939313e-06,
      "loss": 0.5862,
      "step": 1252
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6346157033126076,
      "learning_rate": 1.9132206674019335e-06,
      "loss": 0.5671,
      "step": 1253
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9127955686681459,
      "learning_rate": 1.9130524443525883e-06,
      "loss": 0.6379,
      "step": 1254
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6747840178417314,
      "learning_rate": 1.9128840658199237e-06,
      "loss": 0.5597,
      "step": 1255
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5813740513546736,
      "learning_rate": 1.9127155318326135e-06,
      "loss": 0.5278,
      "step": 1256
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0716128028851892,
      "learning_rate": 1.9125468424193565e-06,
      "loss": 0.7162,
      "step": 1257
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9188823255134607,
      "learning_rate": 1.912377997608879e-06,
      "loss": 0.6272,
      "step": 1258
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6186359647696887,
      "learning_rate": 1.9122089974299336e-06,
      "loss": 0.5559,
      "step": 1259
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.622892585064229,
      "learning_rate": 1.912039841911299e-06,
      "loss": 0.5312,
      "step": 1260
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.695871055324035,
      "learning_rate": 1.9118705310817817e-06,
      "loss": 0.5627,
      "step": 1261
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8405922177942429,
      "learning_rate": 1.911701064970212e-06,
      "loss": 0.6813,
      "step": 1262
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6578183848835084,
      "learning_rate": 1.91153144360545e-06,
      "loss": 0.6057,
      "step": 1263
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7935525933662255,
      "learning_rate": 1.9113616670163782e-06,
      "loss": 0.5378,
      "step": 1264
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6573479326253858,
      "learning_rate": 1.9111917352319098e-06,
      "loss": 0.5974,
      "step": 1265
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9396959459736868,
      "learning_rate": 1.911021648280981e-06,
      "loss": 0.5983,
      "step": 1266
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7973031799678535,
      "learning_rate": 1.9108514061925567e-06,
      "loss": 0.6374,
      "step": 1267
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9517347872214054,
      "learning_rate": 1.9106810089956274e-06,
      "loss": 0.5958,
      "step": 1268
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6765829249951395,
      "learning_rate": 1.9105104567192093e-06,
      "loss": 0.5347,
      "step": 1269
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5764316749898823,
      "learning_rate": 1.9103397493923463e-06,
      "loss": 0.5159,
      "step": 1270
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8133369206446016,
      "learning_rate": 1.910168887044108e-06,
      "loss": 0.6322,
      "step": 1271
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8914684733991114,
      "learning_rate": 1.9099978697035897e-06,
      "loss": 0.5212,
      "step": 1272
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6316762454061254,
      "learning_rate": 1.909826697399915e-06,
      "loss": 0.5137,
      "step": 1273
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6960176471699184,
      "learning_rate": 1.909655370162232e-06,
      "loss": 0.544,
      "step": 1274
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5947001165482533,
      "learning_rate": 1.909483888019716e-06,
      "loss": 0.5129,
      "step": 1275
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6213878790167677,
      "learning_rate": 1.9093122510015693e-06,
      "loss": 0.5589,
      "step": 1276
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9262035985184728,
      "learning_rate": 1.909140459137019e-06,
      "loss": 0.6347,
      "step": 1277
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9551474398199575,
      "learning_rate": 1.90896851245532e-06,
      "loss": 0.66,
      "step": 1278
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8494437024299266,
      "learning_rate": 1.908796410985753e-06,
      "loss": 0.6099,
      "step": 1279
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7206498411641772,
      "learning_rate": 1.908624154757625e-06,
      "loss": 0.6469,
      "step": 1280
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7388187024377662,
      "learning_rate": 1.908451743800269e-06,
      "loss": 0.5482,
      "step": 1281
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.884655458201063,
      "learning_rate": 1.9082791781430457e-06,
      "loss": 0.6916,
      "step": 1282
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8488393930851411,
      "learning_rate": 1.908106457815341e-06,
      "loss": 0.6106,
      "step": 1283
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8998056910210946,
      "learning_rate": 1.9079335828465663e-06,
      "loss": 0.6286,
      "step": 1284
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7517361194655062,
      "learning_rate": 1.9077605532661616e-06,
      "loss": 0.5916,
      "step": 1285
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8327857559207879,
      "learning_rate": 1.9075873691035917e-06,
      "loss": 0.6751,
      "step": 1286
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6916511469239959,
      "learning_rate": 1.907414030388348e-06,
      "loss": 0.5277,
      "step": 1287
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8240926328440982,
      "learning_rate": 1.9072405371499482e-06,
      "loss": 0.6406,
      "step": 1288
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6871242302185181,
      "learning_rate": 1.9070668894179366e-06,
      "loss": 0.573,
      "step": 1289
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9105669924162509,
      "learning_rate": 1.9068930872218836e-06,
      "loss": 0.5645,
      "step": 1290
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6602122580816693,
      "learning_rate": 1.9067191305913855e-06,
      "loss": 0.5355,
      "step": 1291
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8290096593865575,
      "learning_rate": 1.906545019556066e-06,
      "loss": 0.6341,
      "step": 1292
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8467532295273246,
      "learning_rate": 1.9063707541455735e-06,
      "loss": 0.5977,
      "step": 1293
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8897051722777256,
      "learning_rate": 1.9061963343895844e-06,
      "loss": 0.6406,
      "step": 1294
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7012413760275412,
      "learning_rate": 1.9060217603178002e-06,
      "loss": 0.6222,
      "step": 1295
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7577228171228438,
      "learning_rate": 1.9058470319599489e-06,
      "loss": 0.6474,
      "step": 1296
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9819909210794312,
      "learning_rate": 1.9056721493457852e-06,
      "loss": 0.685,
      "step": 1297
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7355255741767956,
      "learning_rate": 1.9054971125050897e-06,
      "loss": 0.6048,
      "step": 1298
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7972585266426047,
      "learning_rate": 1.9053219214676695e-06,
      "loss": 0.5389,
      "step": 1299
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1120679246374776,
      "learning_rate": 1.9051465762633573e-06,
      "loss": 0.687,
      "step": 1300
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7819321447005778,
      "learning_rate": 1.9049710769220132e-06,
      "loss": 0.6082,
      "step": 1301
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8263074917401989,
      "learning_rate": 1.9047954234735225e-06,
      "loss": 0.6115,
      "step": 1302
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6746510852270694,
      "learning_rate": 1.9046196159477973e-06,
      "loss": 0.4663,
      "step": 1303
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9490439852062861,
      "learning_rate": 1.904443654374776e-06,
      "loss": 0.6457,
      "step": 1304
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8460661359437417,
      "learning_rate": 1.9042675387844225e-06,
      "loss": 0.5976,
      "step": 1305
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6581950417367757,
      "learning_rate": 1.9040912692067276e-06,
      "loss": 0.551,
      "step": 1306
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6233308407007312,
      "learning_rate": 1.9039148456717086e-06,
      "loss": 0.5216,
      "step": 1307
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.869239516250715,
      "learning_rate": 1.9037382682094083e-06,
      "loss": 0.6254,
      "step": 1308
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8329661797218253,
      "learning_rate": 1.9035615368498958e-06,
      "loss": 0.6373,
      "step": 1309
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.513823221618247,
      "learning_rate": 1.903384651623267e-06,
      "loss": 0.6632,
      "step": 1310
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7684143636690005,
      "learning_rate": 1.9032076125596435e-06,
      "loss": 0.6959,
      "step": 1311
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9460109980724525,
      "learning_rate": 1.9030304196891728e-06,
      "loss": 0.6199,
      "step": 1312
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.828871077102763,
      "learning_rate": 1.9028530730420295e-06,
      "loss": 0.6186,
      "step": 1313
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.78262792739587,
      "learning_rate": 1.9026755726484138e-06,
      "loss": 0.6385,
      "step": 1314
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8712835251749558,
      "learning_rate": 1.9024979185385519e-06,
      "loss": 0.6376,
      "step": 1315
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7678760883656018,
      "learning_rate": 1.9023201107426968e-06,
      "loss": 0.5915,
      "step": 1316
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8310289211065707,
      "learning_rate": 1.902142149291127e-06,
      "loss": 0.6514,
      "step": 1317
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6408899267212803,
      "learning_rate": 1.9019640342141479e-06,
      "loss": 0.5114,
      "step": 1318
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6836346106253465,
      "learning_rate": 1.90178576554209e-06,
      "loss": 0.5579,
      "step": 1319
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9109105827040568,
      "learning_rate": 1.9016073433053111e-06,
      "loss": 0.6484,
      "step": 1320
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.889981745090896,
      "learning_rate": 1.9014287675341944e-06,
      "loss": 0.5997,
      "step": 1321
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6466073637549428,
      "learning_rate": 1.9012500382591499e-06,
      "loss": 0.5278,
      "step": 1322
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6473793041109502,
      "learning_rate": 1.9010711555106126e-06,
      "loss": 0.5243,
      "step": 1323
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.715345896876541,
      "learning_rate": 1.9008921193190455e-06,
      "loss": 0.596,
      "step": 1324
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8245356914229638,
      "learning_rate": 1.9007129297149353e-06,
      "loss": 0.6401,
      "step": 1325
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8044321719926135,
      "learning_rate": 1.9005335867287968e-06,
      "loss": 0.5835,
      "step": 1326
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8875345126597416,
      "learning_rate": 1.9003540903911706e-06,
      "loss": 0.6037,
      "step": 1327
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0666678654491368,
      "learning_rate": 1.9001744407326222e-06,
      "loss": 0.6784,
      "step": 1328
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7215700772546143,
      "learning_rate": 1.8999946377837445e-06,
      "loss": 0.5524,
      "step": 1329
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7933907643969997,
      "learning_rate": 1.8998146815751563e-06,
      "loss": 0.6494,
      "step": 1330
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6206367052227725,
      "learning_rate": 1.899634572137502e-06,
      "loss": 0.5694,
      "step": 1331
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7613027193343308,
      "learning_rate": 1.8994543095014523e-06,
      "loss": 0.5595,
      "step": 1332
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6734253750150848,
      "learning_rate": 1.8992738936977042e-06,
      "loss": 0.5371,
      "step": 1333
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8264198632205328,
      "learning_rate": 1.8990933247569805e-06,
      "loss": 0.6469,
      "step": 1334
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6348817567269129,
      "learning_rate": 1.8989126027100301e-06,
      "loss": 0.5296,
      "step": 1335
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8915956778382629,
      "learning_rate": 1.8987317275876284e-06,
      "loss": 0.6478,
      "step": 1336
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.748769976795235,
      "learning_rate": 1.8985506994205768e-06,
      "loss": 0.5881,
      "step": 1337
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.720236636621758,
      "learning_rate": 1.8983695182397015e-06,
      "loss": 0.5346,
      "step": 1338
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9649324795585792,
      "learning_rate": 1.8981881840758569e-06,
      "loss": 0.662,
      "step": 1339
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7755062849826476,
      "learning_rate": 1.8980066969599213e-06,
      "loss": 0.6363,
      "step": 1340
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7709400146488374,
      "learning_rate": 1.897825056922801e-06,
      "loss": 0.6234,
      "step": 1341
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9136686313914626,
      "learning_rate": 1.8976432639954267e-06,
      "loss": 0.6603,
      "step": 1342
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7425405742853465,
      "learning_rate": 1.8974613182087565e-06,
      "loss": 0.575,
      "step": 1343
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6157107016683719,
      "learning_rate": 1.897279219593773e-06,
      "loss": 0.5649,
      "step": 1344
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6213917904408933,
      "learning_rate": 1.8970969681814866e-06,
      "loss": 0.5413,
      "step": 1345
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7742290397722541,
      "learning_rate": 1.8969145640029323e-06,
      "loss": 0.6341,
      "step": 1346
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6185264232925686,
      "learning_rate": 1.8967320070891715e-06,
      "loss": 0.5416,
      "step": 1347
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7108215400423297,
      "learning_rate": 1.8965492974712923e-06,
      "loss": 0.579,
      "step": 1348
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9343243669509215,
      "learning_rate": 1.896366435180408e-06,
      "loss": 0.6123,
      "step": 1349
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.032110290756634,
      "learning_rate": 1.8961834202476576e-06,
      "loss": 0.6272,
      "step": 1350
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7663364315207016,
      "learning_rate": 1.8960002527042073e-06,
      "loss": 0.5984,
      "step": 1351
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8507335222219339,
      "learning_rate": 1.8958169325812485e-06,
      "loss": 0.6393,
      "step": 1352
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8420373119697528,
      "learning_rate": 1.8956334599099989e-06,
      "loss": 0.6453,
      "step": 1353
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8483904760776666,
      "learning_rate": 1.8954498347217011e-06,
      "loss": 0.591,
      "step": 1354
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7937243189604148,
      "learning_rate": 1.8952660570476256e-06,
      "loss": 0.5524,
      "step": 1355
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7695723245289143,
      "learning_rate": 1.8950821269190676e-06,
      "loss": 0.644,
      "step": 1356
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6052266500519187,
      "learning_rate": 1.894898044367348e-06,
      "loss": 0.598,
      "step": 1357
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7912854040229887,
      "learning_rate": 1.8947138094238143e-06,
      "loss": 0.5986,
      "step": 1358
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6425185916487481,
      "learning_rate": 1.89452942211984e-06,
      "loss": 0.5948,
      "step": 1359
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6513553896385886,
      "learning_rate": 1.8943448824868241e-06,
      "loss": 0.5715,
      "step": 1360
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6474812983139311,
      "learning_rate": 1.8941601905561923e-06,
      "loss": 0.5677,
      "step": 1361
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8622919181858628,
      "learning_rate": 1.893975346359395e-06,
      "loss": 0.6769,
      "step": 1362
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7392316950879241,
      "learning_rate": 1.89379034992791e-06,
      "loss": 0.5786,
      "step": 1363
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8828617522356548,
      "learning_rate": 1.8936052012932395e-06,
      "loss": 0.6977,
      "step": 1364
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9772946958451301,
      "learning_rate": 1.893419900486913e-06,
      "loss": 0.6513,
      "step": 1365
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8891138493186889,
      "learning_rate": 1.8932344475404846e-06,
      "loss": 0.673,
      "step": 1366
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8591831123617806,
      "learning_rate": 1.8930488424855357e-06,
      "loss": 0.6258,
      "step": 1367
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6307369962604498,
      "learning_rate": 1.8928630853536726e-06,
      "loss": 0.5535,
      "step": 1368
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7971084390737865,
      "learning_rate": 1.892677176176528e-06,
      "loss": 0.6752,
      "step": 1369
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8489528936099523,
      "learning_rate": 1.89249111498576e-06,
      "loss": 0.7058,
      "step": 1370
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7209278011399826,
      "learning_rate": 1.8923049018130531e-06,
      "loss": 0.6247,
      "step": 1371
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6756220714416059,
      "learning_rate": 1.892118536690117e-06,
      "loss": 0.5664,
      "step": 1372
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8532171815517061,
      "learning_rate": 1.8919320196486888e-06,
      "loss": 0.6183,
      "step": 1373
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7450629541248743,
      "learning_rate": 1.8917453507205292e-06,
      "loss": 0.6392,
      "step": 1374
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8815355095910226,
      "learning_rate": 1.8915585299374269e-06,
      "loss": 0.6638,
      "step": 1375
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7983113053103803,
      "learning_rate": 1.8913715573311948e-06,
      "loss": 0.6395,
      "step": 1376
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7705890612828493,
      "learning_rate": 1.8911844329336728e-06,
      "loss": 0.5366,
      "step": 1377
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.657974349547334,
      "learning_rate": 1.8909971567767263e-06,
      "loss": 0.5611,
      "step": 1378
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6246208838748795,
      "learning_rate": 1.8908097288922464e-06,
      "loss": 0.546,
      "step": 1379
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7700891831833455,
      "learning_rate": 1.8906221493121498e-06,
      "loss": 0.6498,
      "step": 1380
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7263520646987879,
      "learning_rate": 1.8904344180683798e-06,
      "loss": 0.5328,
      "step": 1381
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7330832675032019,
      "learning_rate": 1.8902465351929048e-06,
      "loss": 0.6025,
      "step": 1382
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5984790398867992,
      "learning_rate": 1.8900585007177195e-06,
      "loss": 0.5246,
      "step": 1383
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8701614869780259,
      "learning_rate": 1.889870314674844e-06,
      "loss": 0.6445,
      "step": 1384
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7907572025047873,
      "learning_rate": 1.8896819770963244e-06,
      "loss": 0.6354,
      "step": 1385
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6633698120613989,
      "learning_rate": 1.8894934880142329e-06,
      "loss": 0.5399,
      "step": 1386
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8213707931902501,
      "learning_rate": 1.8893048474606669e-06,
      "loss": 0.5547,
      "step": 1387
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.61504477406459,
      "learning_rate": 1.8891160554677504e-06,
      "loss": 0.493,
      "step": 1388
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9338211443552195,
      "learning_rate": 1.888927112067632e-06,
      "loss": 0.6933,
      "step": 1389
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7898151894960762,
      "learning_rate": 1.8887380172924874e-06,
      "loss": 0.6178,
      "step": 1390
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9130750236593735,
      "learning_rate": 1.8885487711745172e-06,
      "loss": 0.6901,
      "step": 1391
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7861914759513458,
      "learning_rate": 1.8883593737459481e-06,
      "loss": 0.6054,
      "step": 1392
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.93255765255696,
      "learning_rate": 1.8881698250390327e-06,
      "loss": 0.6066,
      "step": 1393
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7053209255910798,
      "learning_rate": 1.8879801250860488e-06,
      "loss": 0.5501,
      "step": 1394
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6746232384959575,
      "learning_rate": 1.8877902739193005e-06,
      "loss": 0.5506,
      "step": 1395
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8694554234808456,
      "learning_rate": 1.8876002715711174e-06,
      "loss": 0.6422,
      "step": 1396
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6488615831681229,
      "learning_rate": 1.8874101180738551e-06,
      "loss": 0.5564,
      "step": 1397
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7171262171924074,
      "learning_rate": 1.887219813459895e-06,
      "loss": 0.5002,
      "step": 1398
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8801699313849026,
      "learning_rate": 1.8870293577616431e-06,
      "loss": 0.7131,
      "step": 1399
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5774440785315017,
      "learning_rate": 1.886838751011533e-06,
      "loss": 0.5055,
      "step": 1400
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8679009795144756,
      "learning_rate": 1.8866479932420223e-06,
      "loss": 0.6527,
      "step": 1401
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7461308918842194,
      "learning_rate": 1.8864570844855955e-06,
      "loss": 0.5402,
      "step": 1402
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7330932003526601,
      "learning_rate": 1.8862660247747624e-06,
      "loss": 0.5584,
      "step": 1403
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.718373289951845,
      "learning_rate": 1.8860748141420584e-06,
      "loss": 0.6077,
      "step": 1404
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8260744124627369,
      "learning_rate": 1.8858834526200441e-06,
      "loss": 0.6898,
      "step": 1405
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7998337744758383,
      "learning_rate": 1.8856919402413074e-06,
      "loss": 0.5602,
      "step": 1406
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.689305456188954,
      "learning_rate": 1.88550027703846e-06,
      "loss": 0.4978,
      "step": 1407
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7148375569518989,
      "learning_rate": 1.8853084630441409e-06,
      "loss": 0.5797,
      "step": 1408
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8752247174372582,
      "learning_rate": 1.8851164982910133e-06,
      "loss": 0.6578,
      "step": 1409
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6594379961802735,
      "learning_rate": 1.8849243828117675e-06,
      "loss": 0.5284,
      "step": 1410
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6523415768785581,
      "learning_rate": 1.8847321166391179e-06,
      "loss": 0.531,
      "step": 1411
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9246346703975636,
      "learning_rate": 1.884539699805806e-06,
      "loss": 0.6255,
      "step": 1412
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9958921693800686,
      "learning_rate": 1.8843471323445984e-06,
      "loss": 0.6974,
      "step": 1413
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9048531078986073,
      "learning_rate": 1.8841544142882872e-06,
      "loss": 0.6526,
      "step": 1414
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6733251265732655,
      "learning_rate": 1.8839615456696905e-06,
      "loss": 0.5144,
      "step": 1415
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.864988634447487,
      "learning_rate": 1.8837685265216514e-06,
      "loss": 0.688,
      "step": 1416
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0865077309639943,
      "learning_rate": 1.8835753568770391e-06,
      "loss": 0.582,
      "step": 1417
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.758120683673968,
      "learning_rate": 1.883382036768749e-06,
      "loss": 0.5942,
      "step": 1418
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7365233432680053,
      "learning_rate": 1.8831885662297004e-06,
      "loss": 0.5309,
      "step": 1419
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8534301458209403,
      "learning_rate": 1.8829949452928405e-06,
      "loss": 0.6677,
      "step": 1420
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7406340910491646,
      "learning_rate": 1.8828011739911403e-06,
      "loss": 0.5937,
      "step": 1421
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9042631175936158,
      "learning_rate": 1.882607252357597e-06,
      "loss": 0.609,
      "step": 1422
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.700002435330286,
      "learning_rate": 1.8824131804252338e-06,
      "loss": 0.5579,
      "step": 1423
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.90234925065053,
      "learning_rate": 1.8822189582270985e-06,
      "loss": 0.6723,
      "step": 1424
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6288808774549141,
      "learning_rate": 1.8820245857962659e-06,
      "loss": 0.5218,
      "step": 1425
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8957382750808092,
      "learning_rate": 1.881830063165835e-06,
      "loss": 0.6804,
      "step": 1426
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6544487602944442,
      "learning_rate": 1.881635390368931e-06,
      "loss": 0.5396,
      "step": 1427
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8346212058866364,
      "learning_rate": 1.8814405674387052e-06,
      "loss": 0.6149,
      "step": 1428
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.016528017418266,
      "learning_rate": 1.8812455944083335e-06,
      "loss": 0.6898,
      "step": 1429
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8592181226518223,
      "learning_rate": 1.8810504713110177e-06,
      "loss": 0.6622,
      "step": 1430
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8736610215251372,
      "learning_rate": 1.8808551981799852e-06,
      "loss": 0.6772,
      "step": 1431
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9904027370266641,
      "learning_rate": 1.8806597750484894e-06,
      "loss": 0.6989,
      "step": 1432
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.949093262548183,
      "learning_rate": 1.8804642019498084e-06,
      "loss": 0.6716,
      "step": 1433
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9175499727798004,
      "learning_rate": 1.8802684789172465e-06,
      "loss": 0.6272,
      "step": 1434
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7995299428628458,
      "learning_rate": 1.8800726059841331e-06,
      "loss": 0.5741,
      "step": 1435
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7744835209488642,
      "learning_rate": 1.8798765831838238e-06,
      "loss": 0.592,
      "step": 1436
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7061534253852233,
      "learning_rate": 1.8796804105496984e-06,
      "loss": 0.5686,
      "step": 1437
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9463607835406957,
      "learning_rate": 1.8794840881151637e-06,
      "loss": 0.6328,
      "step": 1438
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7546901513434331,
      "learning_rate": 1.8792876159136513e-06,
      "loss": 0.5685,
      "step": 1439
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6764873211596125,
      "learning_rate": 1.8790909939786182e-06,
      "loss": 0.5176,
      "step": 1440
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6149637875791749,
      "learning_rate": 1.878894222343547e-06,
      "loss": 0.5695,
      "step": 1441
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9719458895275768,
      "learning_rate": 1.878697301041946e-06,
      "loss": 0.679,
      "step": 1442
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7790371186504014,
      "learning_rate": 1.878500230107349e-06,
      "loss": 0.5944,
      "step": 1443
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7781016825361322,
      "learning_rate": 1.8783030095733148e-06,
      "loss": 0.6313,
      "step": 1444
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6454660905715187,
      "learning_rate": 1.8781056394734282e-06,
      "loss": 0.5388,
      "step": 1445
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7437658858289923,
      "learning_rate": 1.8779081198412991e-06,
      "loss": 0.5244,
      "step": 1446
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9140722313688361,
      "learning_rate": 1.8777104507105635e-06,
      "loss": 0.6512,
      "step": 1447
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9498360509598618,
      "learning_rate": 1.877512632114882e-06,
      "loss": 0.6699,
      "step": 1448
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8165953212973491,
      "learning_rate": 1.8773146640879411e-06,
      "loss": 0.6272,
      "step": 1449
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8967076765131066,
      "learning_rate": 1.8771165466634526e-06,
      "loss": 0.6881,
      "step": 1450
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9111754379537457,
      "learning_rate": 1.876918279875154e-06,
      "loss": 0.6934,
      "step": 1451
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.670068748256077,
      "learning_rate": 1.8767198637568078e-06,
      "loss": 0.5638,
      "step": 1452
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6573602871606139,
      "learning_rate": 1.8765212983422027e-06,
      "loss": 0.5716,
      "step": 1453
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7923046841795471,
      "learning_rate": 1.8763225836651517e-06,
      "loss": 0.6405,
      "step": 1454
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8059996935637745,
      "learning_rate": 1.8761237197594943e-06,
      "loss": 0.5589,
      "step": 1455
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6789931413183536,
      "learning_rate": 1.8759247066590951e-06,
      "loss": 0.5066,
      "step": 1456
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6072192764639288,
      "learning_rate": 1.8757255443978436e-06,
      "loss": 0.5229,
      "step": 1457
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8538674155270647,
      "learning_rate": 1.8755262330096548e-06,
      "loss": 0.6169,
      "step": 1458
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7914518981982325,
      "learning_rate": 1.8753267725284698e-06,
      "loss": 0.5698,
      "step": 1459
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6599571130296518,
      "learning_rate": 1.8751271629882548e-06,
      "loss": 0.5431,
      "step": 1460
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6823738125902976,
      "learning_rate": 1.8749274044230008e-06,
      "loss": 0.5907,
      "step": 1461
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7232856278055456,
      "learning_rate": 1.8747274968667246e-06,
      "loss": 0.6202,
      "step": 1462
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0068455656891342,
      "learning_rate": 1.8745274403534688e-06,
      "loss": 0.6749,
      "step": 1463
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6057762565423743,
      "learning_rate": 1.8743272349173004e-06,
      "loss": 0.5567,
      "step": 1464
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6185598676221526,
      "learning_rate": 1.8741268805923126e-06,
      "loss": 0.5663,
      "step": 1465
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8030462414919578,
      "learning_rate": 1.8739263774126237e-06,
      "loss": 0.595,
      "step": 1466
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8888718951191156,
      "learning_rate": 1.8737257254123771e-06,
      "loss": 0.648,
      "step": 1467
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4006107934925882,
      "learning_rate": 1.8735249246257421e-06,
      "loss": 0.6308,
      "step": 1468
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7135154310043672,
      "learning_rate": 1.8733239750869125e-06,
      "loss": 0.5917,
      "step": 1469
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.782394831700157,
      "learning_rate": 1.8731228768301085e-06,
      "loss": 0.5976,
      "step": 1470
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6689222476452088,
      "learning_rate": 1.8729216298895744e-06,
      "loss": 0.5752,
      "step": 1471
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8826201579171571,
      "learning_rate": 1.872720234299581e-06,
      "loss": 0.6374,
      "step": 1472
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9010196246609141,
      "learning_rate": 1.8725186900944235e-06,
      "loss": 0.7016,
      "step": 1473
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9678219103782075,
      "learning_rate": 1.8723169973084229e-06,
      "loss": 0.691,
      "step": 1474
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.822075764884198,
      "learning_rate": 1.8721151559759254e-06,
      "loss": 0.5876,
      "step": 1475
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8389367359319551,
      "learning_rate": 1.8719131661313025e-06,
      "loss": 0.6117,
      "step": 1476
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.614925444113953,
      "learning_rate": 1.8717110278089509e-06,
      "loss": 0.484,
      "step": 1477
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8665390833131461,
      "learning_rate": 1.8715087410432927e-06,
      "loss": 0.5645,
      "step": 1478
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8261241258026107,
      "learning_rate": 1.8713063058687753e-06,
      "loss": 0.6678,
      "step": 1479
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7401957025562489,
      "learning_rate": 1.8711037223198711e-06,
      "loss": 0.6047,
      "step": 1480
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6207232410826065,
      "learning_rate": 1.8709009904310779e-06,
      "loss": 0.4907,
      "step": 1481
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7086837531272505,
      "learning_rate": 1.8706981102369192e-06,
      "loss": 0.5852,
      "step": 1482
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8126501151940565,
      "learning_rate": 1.8704950817719434e-06,
      "loss": 0.6094,
      "step": 1483
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.92484605431923,
      "learning_rate": 1.8702919050707235e-06,
      "loss": 0.5841,
      "step": 1484
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.761858247086483,
      "learning_rate": 1.8700885801678592e-06,
      "loss": 0.5544,
      "step": 1485
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.91800126943183,
      "learning_rate": 1.869885107097974e-06,
      "loss": 0.648,
      "step": 1486
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7985310063862452,
      "learning_rate": 1.8696814858957176e-06,
      "loss": 0.5989,
      "step": 1487
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7129341482784679,
      "learning_rate": 1.8694777165957643e-06,
      "loss": 0.5124,
      "step": 1488
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7871653487663614,
      "learning_rate": 1.869273799232814e-06,
      "loss": 0.6334,
      "step": 1489
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8335509787552419,
      "learning_rate": 1.8690697338415919e-06,
      "loss": 0.6284,
      "step": 1490
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8424956207943637,
      "learning_rate": 1.8688655204568477e-06,
      "loss": 0.6148,
      "step": 1491
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9133961533385199,
      "learning_rate": 1.8686611591133574e-06,
      "loss": 0.6915,
      "step": 1492
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7936779004477725,
      "learning_rate": 1.8684566498459214e-06,
      "loss": 0.5274,
      "step": 1493
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6657406363979287,
      "learning_rate": 1.8682519926893653e-06,
      "loss": 0.6057,
      "step": 1494
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6808686452955117,
      "learning_rate": 1.8680471876785402e-06,
      "loss": 0.6002,
      "step": 1495
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6818550759296733,
      "learning_rate": 1.8678422348483226e-06,
      "loss": 0.599,
      "step": 1496
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8437556983559646,
      "learning_rate": 1.8676371342336133e-06,
      "loss": 0.6966,
      "step": 1497
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8137366309667806,
      "learning_rate": 1.867431885869339e-06,
      "loss": 0.6524,
      "step": 1498
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.616784488141124,
      "learning_rate": 1.8672264897904513e-06,
      "loss": 0.572,
      "step": 1499
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9246619262601333,
      "learning_rate": 1.867020946031927e-06,
      "loss": 0.6364,
      "step": 1500
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.704886930088419,
      "learning_rate": 1.8668152546287686e-06,
      "loss": 0.5452,
      "step": 1501
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9156452026597057,
      "learning_rate": 1.8666094156160022e-06,
      "loss": 0.6582,
      "step": 1502
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7206072647687064,
      "learning_rate": 1.866403429028681e-06,
      "loss": 0.5776,
      "step": 1503
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8498808736027476,
      "learning_rate": 1.8661972949018817e-06,
      "loss": 0.6199,
      "step": 1504
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7714233311502476,
      "learning_rate": 1.8659910132707072e-06,
      "loss": 0.5968,
      "step": 1505
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7529971436371257,
      "learning_rate": 1.8657845841702851e-06,
      "loss": 0.6058,
      "step": 1506
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9634732061273164,
      "learning_rate": 1.8655780076357676e-06,
      "loss": 0.6427,
      "step": 1507
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6611391815528458,
      "learning_rate": 1.8653712837023331e-06,
      "loss": 0.6008,
      "step": 1508
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6026054575525503,
      "learning_rate": 1.8651644124051846e-06,
      "loss": 0.5112,
      "step": 1509
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.759700431303414,
      "learning_rate": 1.8649573937795497e-06,
      "loss": 0.5964,
      "step": 1510
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2853988612736746,
      "learning_rate": 1.8647502278606817e-06,
      "loss": 0.719,
      "step": 1511
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6856213266623999,
      "learning_rate": 1.864542914683859e-06,
      "loss": 0.6191,
      "step": 1512
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7615831952053184,
      "learning_rate": 1.8643354542843843e-06,
      "loss": 0.584,
      "step": 1513
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.846341851913998,
      "learning_rate": 1.8641278466975866e-06,
      "loss": 0.6433,
      "step": 1514
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8215122282671824,
      "learning_rate": 1.8639200919588188e-06,
      "loss": 0.6283,
      "step": 1515
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.795272875064585,
      "learning_rate": 1.8637121901034596e-06,
      "loss": 0.6254,
      "step": 1516
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7240585297503832,
      "learning_rate": 1.8635041411669128e-06,
      "loss": 0.5854,
      "step": 1517
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.714945504217413,
      "learning_rate": 1.8632959451846064e-06,
      "loss": 0.5721,
      "step": 1518
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.743579998599788,
      "learning_rate": 1.8630876021919945e-06,
      "loss": 0.5831,
      "step": 1519
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9240985924157026,
      "learning_rate": 1.8628791122245552e-06,
      "loss": 0.6851,
      "step": 1520
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8025818524376312,
      "learning_rate": 1.862670475317793e-06,
      "loss": 0.7181,
      "step": 1521
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8822210286432239,
      "learning_rate": 1.8624616915072357e-06,
      "loss": 0.7238,
      "step": 1522
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6700355783987793,
      "learning_rate": 1.8622527608284375e-06,
      "loss": 0.5768,
      "step": 1523
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7030711299139792,
      "learning_rate": 1.862043683316977e-06,
      "loss": 0.6145,
      "step": 1524
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7115454925875814,
      "learning_rate": 1.8618344590084581e-06,
      "loss": 0.6172,
      "step": 1525
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.704657357650042,
      "learning_rate": 1.861625087938509e-06,
      "loss": 0.5725,
      "step": 1526
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6957507793091998,
      "learning_rate": 1.8614155701427842e-06,
      "loss": 0.5127,
      "step": 1527
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6130606203074686,
      "learning_rate": 1.8612059056569617e-06,
      "loss": 0.5121,
      "step": 1528
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9082709271088465,
      "learning_rate": 1.8609960945167457e-06,
      "loss": 0.5822,
      "step": 1529
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8234701937282133,
      "learning_rate": 1.860786136757864e-06,
      "loss": 0.6577,
      "step": 1530
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8457913960111877,
      "learning_rate": 1.8605760324160716e-06,
      "loss": 0.5995,
      "step": 1531
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.842631693621379,
      "learning_rate": 1.860365781527146e-06,
      "loss": 0.6386,
      "step": 1532
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8178491529706742,
      "learning_rate": 1.8601553841268908e-06,
      "loss": 0.5644,
      "step": 1533
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7358100734343288,
      "learning_rate": 1.8599448402511349e-06,
      "loss": 0.5729,
      "step": 1534
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7461473102745995,
      "learning_rate": 1.8597341499357313e-06,
      "loss": 0.5649,
      "step": 1535
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.724348912661096,
      "learning_rate": 1.859523313216559e-06,
      "loss": 0.5984,
      "step": 1536
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7011046106966219,
      "learning_rate": 1.8593123301295206e-06,
      "loss": 0.6444,
      "step": 1537
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7196170194623498,
      "learning_rate": 1.8591012007105446e-06,
      "loss": 0.5995,
      "step": 1538
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6903779304166274,
      "learning_rate": 1.8588899249955845e-06,
      "loss": 0.5809,
      "step": 1539
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7052761456163396,
      "learning_rate": 1.8586785030206178e-06,
      "loss": 0.5761,
      "step": 1540
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9335321038714185,
      "learning_rate": 1.8584669348216478e-06,
      "loss": 0.6777,
      "step": 1541
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8711748668844923,
      "learning_rate": 1.8582552204347022e-06,
      "loss": 0.6091,
      "step": 1542
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6871931266865535,
      "learning_rate": 1.8580433598958336e-06,
      "loss": 0.5659,
      "step": 1543
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8363310541937206,
      "learning_rate": 1.8578313532411201e-06,
      "loss": 0.6582,
      "step": 1544
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9181072818304548,
      "learning_rate": 1.8576192005066638e-06,
      "loss": 0.6524,
      "step": 1545
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7588096734473051,
      "learning_rate": 1.8574069017285923e-06,
      "loss": 0.6215,
      "step": 1546
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7000373063035533,
      "learning_rate": 1.8571944569430579e-06,
      "loss": 0.5948,
      "step": 1547
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.639826043496221,
      "learning_rate": 1.8569818661862378e-06,
      "loss": 0.5206,
      "step": 1548
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6643999658948555,
      "learning_rate": 1.8567691294943338e-06,
      "loss": 0.5573,
      "step": 1549
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9411235801371197,
      "learning_rate": 1.8565562469035724e-06,
      "loss": 0.6485,
      "step": 1550
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.890023205918796,
      "learning_rate": 1.856343218450206e-06,
      "loss": 0.6709,
      "step": 1551
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8699130101007695,
      "learning_rate": 1.856130044170511e-06,
      "loss": 0.5302,
      "step": 1552
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7681743423180455,
      "learning_rate": 1.8559167241007882e-06,
      "loss": 0.5833,
      "step": 1553
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7158332723218168,
      "learning_rate": 1.8557032582773644e-06,
      "loss": 0.6029,
      "step": 1554
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8991008720968421,
      "learning_rate": 1.8554896467365904e-06,
      "loss": 0.6147,
      "step": 1555
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6981369116680725,
      "learning_rate": 1.855275889514842e-06,
      "loss": 0.5879,
      "step": 1556
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7575680684905965,
      "learning_rate": 1.8550619866485198e-06,
      "loss": 0.6812,
      "step": 1557
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6210549294180444,
      "learning_rate": 1.854847938174049e-06,
      "loss": 0.6182,
      "step": 1558
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6113870208978385,
      "learning_rate": 1.8546337441278804e-06,
      "loss": 0.545,
      "step": 1559
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6891737176087768,
      "learning_rate": 1.8544194045464886e-06,
      "loss": 0.5571,
      "step": 1560
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7808280946484666,
      "learning_rate": 1.8542049194663733e-06,
      "loss": 0.629,
      "step": 1561
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8041643510081159,
      "learning_rate": 1.8539902889240592e-06,
      "loss": 0.6545,
      "step": 1562
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7271054027740419,
      "learning_rate": 1.853775512956096e-06,
      "loss": 0.5696,
      "step": 1563
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.724937895189766,
      "learning_rate": 1.8535605915990575e-06,
      "loss": 0.5593,
      "step": 1564
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7125528779112926,
      "learning_rate": 1.8533455248895424e-06,
      "loss": 0.5653,
      "step": 1565
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8766515601495122,
      "learning_rate": 1.8531303128641747e-06,
      "loss": 0.5966,
      "step": 1566
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8201049937340776,
      "learning_rate": 1.8529149555596023e-06,
      "loss": 0.6521,
      "step": 1567
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6927285733566425,
      "learning_rate": 1.852699453012499e-06,
      "loss": 0.6279,
      "step": 1568
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8261624818313235,
      "learning_rate": 1.8524838052595615e-06,
      "loss": 0.5984,
      "step": 1569
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7491854715114691,
      "learning_rate": 1.8522680123375137e-06,
      "loss": 0.6153,
      "step": 1570
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8175076377427374,
      "learning_rate": 1.852052074283102e-06,
      "loss": 0.6289,
      "step": 1571
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8888378084094476,
      "learning_rate": 1.8518359911330985e-06,
      "loss": 0.5817,
      "step": 1572
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8494571473596824,
      "learning_rate": 1.8516197629243004e-06,
      "loss": 0.6519,
      "step": 1573
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8541460569904148,
      "learning_rate": 1.8514033896935289e-06,
      "loss": 0.5678,
      "step": 1574
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9843343628358772,
      "learning_rate": 1.8511868714776297e-06,
      "loss": 0.6382,
      "step": 1575
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6755790039685406,
      "learning_rate": 1.8509702083134736e-06,
      "loss": 0.5584,
      "step": 1576
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7788887557572672,
      "learning_rate": 1.850753400237957e-06,
      "loss": 0.5836,
      "step": 1577
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8598998733318207,
      "learning_rate": 1.8505364472879991e-06,
      "loss": 0.5706,
      "step": 1578
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8425294085246251,
      "learning_rate": 1.850319349500545e-06,
      "loss": 0.5908,
      "step": 1579
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7908951978455421,
      "learning_rate": 1.8501021069125646e-06,
      "loss": 0.7094,
      "step": 1580
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.899689484122659,
      "learning_rate": 1.8498847195610513e-06,
      "loss": 0.6277,
      "step": 1581
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7636934706126499,
      "learning_rate": 1.8496671874830245e-06,
      "loss": 0.606,
      "step": 1582
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8536968776789445,
      "learning_rate": 1.8494495107155275e-06,
      "loss": 0.6591,
      "step": 1583
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9416638224501291,
      "learning_rate": 1.8492316892956281e-06,
      "loss": 0.6614,
      "step": 1584
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6546502213605624,
      "learning_rate": 1.8490137232604193e-06,
      "loss": 0.5608,
      "step": 1585
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7190647756823703,
      "learning_rate": 1.8487956126470183e-06,
      "loss": 0.5661,
      "step": 1586
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6849726006403514,
      "learning_rate": 1.8485773574925673e-06,
      "loss": 0.5645,
      "step": 1587
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6625476578229635,
      "learning_rate": 1.8483589578342328e-06,
      "loss": 0.5734,
      "step": 1588
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8173091134547499,
      "learning_rate": 1.8481404137092058e-06,
      "loss": 0.6444,
      "step": 1589
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7714196727777476,
      "learning_rate": 1.8479217251547024e-06,
      "loss": 0.5746,
      "step": 1590
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8704460499297837,
      "learning_rate": 1.8477028922079625e-06,
      "loss": 0.6517,
      "step": 1591
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8852174635292996,
      "learning_rate": 1.8474839149062515e-06,
      "loss": 0.647,
      "step": 1592
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.71944605210508,
      "learning_rate": 1.8472647932868589e-06,
      "loss": 0.5802,
      "step": 1593
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6792289397359085,
      "learning_rate": 1.8470455273870984e-06,
      "loss": 0.5714,
      "step": 1594
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8619672738229428,
      "learning_rate": 1.8468261172443093e-06,
      "loss": 0.6432,
      "step": 1595
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6812147663616255,
      "learning_rate": 1.8466065628958548e-06,
      "loss": 0.5761,
      "step": 1596
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7189688282279665,
      "learning_rate": 1.8463868643791222e-06,
      "loss": 0.602,
      "step": 1597
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8268830126333812,
      "learning_rate": 1.8461670217315245e-06,
      "loss": 0.618,
      "step": 1598
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8163055792325702,
      "learning_rate": 1.8459470349904982e-06,
      "loss": 0.6359,
      "step": 1599
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8167488815846617,
      "learning_rate": 1.8457269041935046e-06,
      "loss": 0.6057,
      "step": 1600
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6130862347226241,
      "learning_rate": 1.8455066293780301e-06,
      "loss": 0.492,
      "step": 1601
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8797532647556527,
      "learning_rate": 1.8452862105815852e-06,
      "loss": 0.6898,
      "step": 1602
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8575619584769479,
      "learning_rate": 1.8450656478417046e-06,
      "loss": 0.6626,
      "step": 1603
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8842413009105622,
      "learning_rate": 1.844844941195948e-06,
      "loss": 0.6487,
      "step": 1604
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8671488568461713,
      "learning_rate": 1.8446240906818992e-06,
      "loss": 0.6336,
      "step": 1605
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8136743670512881,
      "learning_rate": 1.8444030963371673e-06,
      "loss": 0.6218,
      "step": 1606
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8026436637794968,
      "learning_rate": 1.8441819581993853e-06,
      "loss": 0.5952,
      "step": 1607
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6048654511049967,
      "learning_rate": 1.8439606763062102e-06,
      "loss": 0.5721,
      "step": 1608
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8577062998504724,
      "learning_rate": 1.843739250695324e-06,
      "loss": 0.6058,
      "step": 1609
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9576287252393064,
      "learning_rate": 1.8435176814044335e-06,
      "loss": 0.6244,
      "step": 1610
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8530261475714226,
      "learning_rate": 1.8432959684712696e-06,
      "loss": 0.625,
      "step": 1611
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6204601050428569,
      "learning_rate": 1.8430741119335878e-06,
      "loss": 0.552,
      "step": 1612
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6750004670512109,
      "learning_rate": 1.8428521118291677e-06,
      "loss": 0.5606,
      "step": 1613
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7177576521021638,
      "learning_rate": 1.8426299681958137e-06,
      "loss": 0.529,
      "step": 1614
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.721978236216542,
      "learning_rate": 1.8424076810713546e-06,
      "loss": 0.6059,
      "step": 1615
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.734533118829263,
      "learning_rate": 1.8421852504936438e-06,
      "loss": 0.5548,
      "step": 1616
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8390804804927139,
      "learning_rate": 1.8419626765005582e-06,
      "loss": 0.661,
      "step": 1617
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6626508871287348,
      "learning_rate": 1.8417399591300006e-06,
      "loss": 0.5709,
      "step": 1618
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8192595083432876,
      "learning_rate": 1.8415170984198972e-06,
      "loss": 0.6367,
      "step": 1619
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6564226787446984,
      "learning_rate": 1.8412940944081985e-06,
      "loss": 0.5851,
      "step": 1620
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7411693365720023,
      "learning_rate": 1.8410709471328802e-06,
      "loss": 0.5881,
      "step": 1621
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6385309303484951,
      "learning_rate": 1.8408476566319417e-06,
      "loss": 0.568,
      "step": 1622
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9445621766163685,
      "learning_rate": 1.8406242229434077e-06,
      "loss": 0.6242,
      "step": 1623
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.69876762095836,
      "learning_rate": 1.8404006461053254e-06,
      "loss": 0.5812,
      "step": 1624
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8081198671482689,
      "learning_rate": 1.8401769261557689e-06,
      "loss": 0.5827,
      "step": 1625
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0521242807327829,
      "learning_rate": 1.8399530631328343e-06,
      "loss": 0.6856,
      "step": 1626
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8597917404694845,
      "learning_rate": 1.8397290570746437e-06,
      "loss": 0.686,
      "step": 1627
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6759553242308229,
      "learning_rate": 1.839504908019343e-06,
      "loss": 0.535,
      "step": 1628
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6898034013754697,
      "learning_rate": 1.8392806160051024e-06,
      "loss": 0.5055,
      "step": 1629
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7590441117950109,
      "learning_rate": 1.8390561810701166e-06,
      "loss": 0.5686,
      "step": 1630
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8738872756606386,
      "learning_rate": 1.8388316032526041e-06,
      "loss": 0.5802,
      "step": 1631
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7929798309793638,
      "learning_rate": 1.8386068825908087e-06,
      "loss": 0.6955,
      "step": 1632
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1056654293640331,
      "learning_rate": 1.8383820191229976e-06,
      "loss": 0.626,
      "step": 1633
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7657061595178817,
      "learning_rate": 1.8381570128874628e-06,
      "loss": 0.6203,
      "step": 1634
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6328234068891088,
      "learning_rate": 1.8379318639225208e-06,
      "loss": 0.5859,
      "step": 1635
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7256429380307872,
      "learning_rate": 1.8377065722665116e-06,
      "loss": 0.5432,
      "step": 1636
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7145570773435055,
      "learning_rate": 1.8374811379578004e-06,
      "loss": 0.5923,
      "step": 1637
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.873654850708191,
      "learning_rate": 1.8372555610347762e-06,
      "loss": 0.6396,
      "step": 1638
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.83740763688326,
      "learning_rate": 1.8370298415358524e-06,
      "loss": 0.6106,
      "step": 1639
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9507709414012546,
      "learning_rate": 1.836803979499467e-06,
      "loss": 0.5879,
      "step": 1640
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7162572026325102,
      "learning_rate": 1.836577974964081e-06,
      "loss": 0.5486,
      "step": 1641
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8452709874405113,
      "learning_rate": 1.836351827968182e-06,
      "loss": 0.6542,
      "step": 1642
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6893632087623576,
      "learning_rate": 1.8361255385502794e-06,
      "loss": 0.6106,
      "step": 1643
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6506495065444791,
      "learning_rate": 1.8358991067489082e-06,
      "loss": 0.5207,
      "step": 1644
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8825450209886359,
      "learning_rate": 1.8356725326026275e-06,
      "loss": 0.6257,
      "step": 1645
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7559691618728528,
      "learning_rate": 1.8354458161500206e-06,
      "loss": 0.6567,
      "step": 1646
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6704168487885038,
      "learning_rate": 1.8352189574296948e-06,
      "loss": 0.5112,
      "step": 1647
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6894047726274033,
      "learning_rate": 1.8349919564802815e-06,
      "loss": 0.6116,
      "step": 1648
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6948196917908928,
      "learning_rate": 1.8347648133404375e-06,
      "loss": 0.5751,
      "step": 1649
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.692581434834775,
      "learning_rate": 1.8345375280488418e-06,
      "loss": 0.5461,
      "step": 1650
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8194038517587576,
      "learning_rate": 1.8343101006441994e-06,
      "loss": 0.6431,
      "step": 1651
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6407235700782669,
      "learning_rate": 1.8340825311652383e-06,
      "loss": 0.5589,
      "step": 1652
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8468378631389253,
      "learning_rate": 1.8338548196507121e-06,
      "loss": 0.6439,
      "step": 1653
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8734794123389028,
      "learning_rate": 1.833626966139397e-06,
      "loss": 0.6749,
      "step": 1654
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8013533866339752,
      "learning_rate": 1.8333989706700939e-06,
      "loss": 0.6677,
      "step": 1655
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5928042213781552,
      "learning_rate": 1.8331708332816288e-06,
      "loss": 0.5278,
      "step": 1656
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8144997164678949,
      "learning_rate": 1.8329425540128504e-06,
      "loss": 0.6359,
      "step": 1657
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.977124050802364,
      "learning_rate": 1.8327141329026327e-06,
      "loss": 0.6846,
      "step": 1658
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0087841667824307,
      "learning_rate": 1.8324855699898733e-06,
      "loss": 0.6619,
      "step": 1659
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8986226558281747,
      "learning_rate": 1.8322568653134942e-06,
      "loss": 0.6628,
      "step": 1660
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7921554826764078,
      "learning_rate": 1.8320280189124412e-06,
      "loss": 0.5922,
      "step": 1661
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5876378741089907,
      "learning_rate": 1.8317990308256848e-06,
      "loss": 0.5087,
      "step": 1662
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7370933222467385,
      "learning_rate": 1.8315699010922187e-06,
      "loss": 0.5453,
      "step": 1663
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6018550117834901,
      "learning_rate": 1.8313406297510618e-06,
      "loss": 0.5572,
      "step": 1664
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7644657865460001,
      "learning_rate": 1.8311112168412568e-06,
      "loss": 0.6237,
      "step": 1665
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5795405957249056,
      "learning_rate": 1.83088166240187e-06,
      "loss": 0.5211,
      "step": 1666
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.963861691677456,
      "learning_rate": 1.8306519664719917e-06,
      "loss": 0.6206,
      "step": 1667
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.215607269140423,
      "learning_rate": 1.8304221290907376e-06,
      "loss": 0.624,
      "step": 1668
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6081156974487874,
      "learning_rate": 1.8301921502972459e-06,
      "loss": 0.5061,
      "step": 1669
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8615416937711721,
      "learning_rate": 1.82996203013068e-06,
      "loss": 0.6891,
      "step": 1670
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6322853460124794,
      "learning_rate": 1.8297317686302269e-06,
      "loss": 0.5688,
      "step": 1671
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8375491865368652,
      "learning_rate": 1.8295013658350978e-06,
      "loss": 0.6053,
      "step": 1672
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6625210088498944,
      "learning_rate": 1.8292708217845276e-06,
      "loss": 0.5334,
      "step": 1673
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9419629383492433,
      "learning_rate": 1.8290401365177757e-06,
      "loss": 0.5954,
      "step": 1674
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8105340052486486,
      "learning_rate": 1.8288093100741256e-06,
      "loss": 0.5891,
      "step": 1675
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6243193170773744,
      "learning_rate": 1.8285783424928842e-06,
      "loss": 0.5527,
      "step": 1676
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6761909696830369,
      "learning_rate": 1.8283472338133831e-06,
      "loss": 0.5493,
      "step": 1677
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6400055123021755,
      "learning_rate": 1.8281159840749782e-06,
      "loss": 0.5792,
      "step": 1678
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9514989781518169,
      "learning_rate": 1.827884593317048e-06,
      "loss": 0.6216,
      "step": 1679
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.704368678874323,
      "learning_rate": 1.8276530615789968e-06,
      "loss": 0.5719,
      "step": 1680
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6376935677805395,
      "learning_rate": 1.8274213889002513e-06,
      "loss": 0.5404,
      "step": 1681
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7016259978977398,
      "learning_rate": 1.8271895753202636e-06,
      "loss": 0.585,
      "step": 1682
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.667379014621767,
      "learning_rate": 1.8269576208785087e-06,
      "loss": 0.5768,
      "step": 1683
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7169820387837189,
      "learning_rate": 1.826725525614486e-06,
      "loss": 0.5238,
      "step": 1684
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8535744437583227,
      "learning_rate": 1.8264932895677193e-06,
      "loss": 0.6208,
      "step": 1685
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.759325381467501,
      "learning_rate": 1.8262609127777555e-06,
      "loss": 0.5592,
      "step": 1686
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6867251456947394,
      "learning_rate": 1.8260283952841662e-06,
      "loss": 0.5821,
      "step": 1687
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6844502400655375,
      "learning_rate": 1.8257957371265466e-06,
      "loss": 0.5078,
      "step": 1688
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0501574860131644,
      "learning_rate": 1.8255629383445164e-06,
      "loss": 0.6296,
      "step": 1689
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7359696825495957,
      "learning_rate": 1.8253299989777182e-06,
      "loss": 0.6298,
      "step": 1690
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5944791727806555,
      "learning_rate": 1.8250969190658194e-06,
      "loss": 0.5334,
      "step": 1691
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7086113176730976,
      "learning_rate": 1.824863698648511e-06,
      "loss": 0.5977,
      "step": 1692
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6682657789934663,
      "learning_rate": 1.8246303377655083e-06,
      "loss": 0.5774,
      "step": 1693
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7311237871921148,
      "learning_rate": 1.82439683645655e-06,
      "loss": 0.5557,
      "step": 1694
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9351573993722992,
      "learning_rate": 1.824163194761399e-06,
      "loss": 0.6239,
      "step": 1695
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0573278920230318,
      "learning_rate": 1.8239294127198416e-06,
      "loss": 0.6899,
      "step": 1696
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8344512612404636,
      "learning_rate": 1.8236954903716892e-06,
      "loss": 0.5651,
      "step": 1697
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6888017534953439,
      "learning_rate": 1.823461427756776e-06,
      "loss": 0.533,
      "step": 1698
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7732419714499459,
      "learning_rate": 1.8232272249149604e-06,
      "loss": 0.616,
      "step": 1699
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8433822631234789,
      "learning_rate": 1.8229928818861244e-06,
      "loss": 0.6228,
      "step": 1700
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9359366082002993,
      "learning_rate": 1.8227583987101749e-06,
      "loss": 0.6523,
      "step": 1701
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2195359312216238,
      "learning_rate": 1.8225237754270413e-06,
      "loss": 0.6511,
      "step": 1702
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.717747879583717,
      "learning_rate": 1.822289012076678e-06,
      "loss": 0.5978,
      "step": 1703
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5640604308698282,
      "learning_rate": 1.822054108699062e-06,
      "loss": 0.4917,
      "step": 1704
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8031217560664929,
      "learning_rate": 1.8218190653341959e-06,
      "loss": 0.6298,
      "step": 1705
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5787144426985771,
      "learning_rate": 1.8215838820221044e-06,
      "loss": 0.5319,
      "step": 1706
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8937066735404819,
      "learning_rate": 1.8213485588028367e-06,
      "loss": 0.6945,
      "step": 1707
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6863783673075151,
      "learning_rate": 1.8211130957164666e-06,
      "loss": 0.5567,
      "step": 1708
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7265154892039355,
      "learning_rate": 1.8208774928030905e-06,
      "loss": 0.5754,
      "step": 1709
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.772943659476795,
      "learning_rate": 1.8206417501028292e-06,
      "loss": 0.6054,
      "step": 1710
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9176846667592462,
      "learning_rate": 1.820405867655827e-06,
      "loss": 0.6722,
      "step": 1711
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5169606561675282,
      "learning_rate": 1.820169845502253e-06,
      "loss": 0.5989,
      "step": 1712
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9321053736372229,
      "learning_rate": 1.8199336836822984e-06,
      "loss": 0.6108,
      "step": 1713
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6613246859591668,
      "learning_rate": 1.8196973822361796e-06,
      "loss": 0.5643,
      "step": 1714
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8621861741978805,
      "learning_rate": 1.8194609412041362e-06,
      "loss": 0.6714,
      "step": 1715
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7322492213774321,
      "learning_rate": 1.8192243606264314e-06,
      "loss": 0.5575,
      "step": 1716
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6400037426835998,
      "learning_rate": 1.8189876405433527e-06,
      "loss": 0.5424,
      "step": 1717
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7130718263249135,
      "learning_rate": 1.818750780995211e-06,
      "loss": 0.5323,
      "step": 1718
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7786314679117163,
      "learning_rate": 1.8185137820223408e-06,
      "loss": 0.6134,
      "step": 1719
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.863903543277782,
      "learning_rate": 1.818276643665101e-06,
      "loss": 0.6036,
      "step": 1720
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7695356755769173,
      "learning_rate": 1.8180393659638733e-06,
      "loss": 0.5403,
      "step": 1721
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7961285574841794,
      "learning_rate": 1.817801948959064e-06,
      "loss": 0.6411,
      "step": 1722
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6933802464700348,
      "learning_rate": 1.8175643926911026e-06,
      "loss": 0.5944,
      "step": 1723
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7414849206408384,
      "learning_rate": 1.8173266972004424e-06,
      "loss": 0.5961,
      "step": 1724
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6705604842210674,
      "learning_rate": 1.8170888625275605e-06,
      "loss": 0.5409,
      "step": 1725
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6495933271204001,
      "learning_rate": 1.8168508887129576e-06,
      "loss": 0.5894,
      "step": 1726
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9815922397408414,
      "learning_rate": 1.8166127757971583e-06,
      "loss": 0.6456,
      "step": 1727
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8000133973683352,
      "learning_rate": 1.8163745238207109e-06,
      "loss": 0.6337,
      "step": 1728
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8675632344121847,
      "learning_rate": 1.8161361328241868e-06,
      "loss": 0.5925,
      "step": 1729
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6320894598546554,
      "learning_rate": 1.8158976028481816e-06,
      "loss": 0.5524,
      "step": 1730
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7189990178330681,
      "learning_rate": 1.815658933933315e-06,
      "loss": 0.585,
      "step": 1731
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7963348748625861,
      "learning_rate": 1.815420126120229e-06,
      "loss": 0.5724,
      "step": 1732
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7083215336159754,
      "learning_rate": 1.815181179449591e-06,
      "loss": 0.5717,
      "step": 1733
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8559419765520545,
      "learning_rate": 1.8149420939620904e-06,
      "loss": 0.5772,
      "step": 1734
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7659617275875781,
      "learning_rate": 1.8147028696984408e-06,
      "loss": 0.5821,
      "step": 1735
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6793268043214676,
      "learning_rate": 1.8144635066993806e-06,
      "loss": 0.5852,
      "step": 1736
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8470579173275415,
      "learning_rate": 1.81422400500567e-06,
      "loss": 0.6695,
      "step": 1737
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7237771145517838,
      "learning_rate": 1.8139843646580937e-06,
      "loss": 0.5823,
      "step": 1738
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7518205025159073,
      "learning_rate": 1.8137445856974602e-06,
      "loss": 0.5894,
      "step": 1739
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.609674305048156,
      "learning_rate": 1.8135046681646014e-06,
      "loss": 0.5869,
      "step": 1740
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8015002984003656,
      "learning_rate": 1.8132646121003728e-06,
      "loss": 0.6024,
      "step": 1741
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9621602535567867,
      "learning_rate": 1.8130244175456531e-06,
      "loss": 0.7251,
      "step": 1742
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6922865833202008,
      "learning_rate": 1.812784084541345e-06,
      "loss": 0.5429,
      "step": 1743
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7716825465299049,
      "learning_rate": 1.812543613128375e-06,
      "loss": 0.6176,
      "step": 1744
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7450223219492245,
      "learning_rate": 1.8123030033476924e-06,
      "loss": 0.6246,
      "step": 1745
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.719050302690064,
      "learning_rate": 1.812062255240271e-06,
      "loss": 0.5622,
      "step": 1746
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7811080873572938,
      "learning_rate": 1.8118213688471077e-06,
      "loss": 0.6673,
      "step": 1747
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7601059733631147,
      "learning_rate": 1.8115803442092226e-06,
      "loss": 0.5587,
      "step": 1748
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0545204832270898,
      "learning_rate": 1.8113391813676602e-06,
      "loss": 0.648,
      "step": 1749
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4159443131826095,
      "learning_rate": 1.8110978803634873e-06,
      "loss": 0.6413,
      "step": 1750
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7753009876390837,
      "learning_rate": 1.8108564412377954e-06,
      "loss": 0.6135,
      "step": 1751
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6402930474897482,
      "learning_rate": 1.8106148640316991e-06,
      "loss": 0.5219,
      "step": 1752
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7184571126578188,
      "learning_rate": 1.8103731487863363e-06,
      "loss": 0.5541,
      "step": 1753
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8164535530610884,
      "learning_rate": 1.8101312955428689e-06,
      "loss": 0.7062,
      "step": 1754
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7036791649722098,
      "learning_rate": 1.8098893043424814e-06,
      "loss": 0.5044,
      "step": 1755
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8196196271565832,
      "learning_rate": 1.809647175226383e-06,
      "loss": 0.5597,
      "step": 1756
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9076930818021154,
      "learning_rate": 1.8094049082358053e-06,
      "loss": 0.6464,
      "step": 1757
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3252900544878141,
      "learning_rate": 1.8091625034120042e-06,
      "loss": 0.6324,
      "step": 1758
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8820197688188727,
      "learning_rate": 1.8089199607962586e-06,
      "loss": 0.6482,
      "step": 1759
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9436784053090133,
      "learning_rate": 1.8086772804298706e-06,
      "loss": 0.6262,
      "step": 1760
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8388288289573467,
      "learning_rate": 1.808434462354167e-06,
      "loss": 0.6119,
      "step": 1761
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7436074536410334,
      "learning_rate": 1.8081915066104965e-06,
      "loss": 0.6424,
      "step": 1762
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6391458944845819,
      "learning_rate": 1.807948413240232e-06,
      "loss": 0.5543,
      "step": 1763
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6616441871034829,
      "learning_rate": 1.8077051822847697e-06,
      "loss": 0.5654,
      "step": 1764
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7153939696029903,
      "learning_rate": 1.8074618137855298e-06,
      "loss": 0.5288,
      "step": 1765
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7166501300969018,
      "learning_rate": 1.8072183077839549e-06,
      "loss": 0.5613,
      "step": 1766
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9375966353132854,
      "learning_rate": 1.8069746643215117e-06,
      "loss": 0.6596,
      "step": 1767
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6168753495415233,
      "learning_rate": 1.8067308834396903e-06,
      "loss": 0.5634,
      "step": 1768
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.793592030923942,
      "learning_rate": 1.8064869651800038e-06,
      "loss": 0.5713,
      "step": 1769
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7221843069016274,
      "learning_rate": 1.806242909583989e-06,
      "loss": 0.5747,
      "step": 1770
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.754349000250631,
      "learning_rate": 1.8059987166932062e-06,
      "loss": 0.5353,
      "step": 1771
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7377007516511078,
      "learning_rate": 1.8057543865492389e-06,
      "loss": 0.5858,
      "step": 1772
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8914179174487533,
      "learning_rate": 1.8055099191936938e-06,
      "loss": 0.6039,
      "step": 1773
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8358315477250097,
      "learning_rate": 1.805265314668201e-06,
      "loss": 0.612,
      "step": 1774
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6752937680619854,
      "learning_rate": 1.8050205730144144e-06,
      "loss": 0.5876,
      "step": 1775
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9635903635095093,
      "learning_rate": 1.804775694274011e-06,
      "loss": 0.6358,
      "step": 1776
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1013710186720425,
      "learning_rate": 1.8045306784886907e-06,
      "loss": 0.6133,
      "step": 1777
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8217142853232547,
      "learning_rate": 1.8042855257001777e-06,
      "loss": 0.6286,
      "step": 1778
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7735564763486008,
      "learning_rate": 1.8040402359502185e-06,
      "loss": 0.6093,
      "step": 1779
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6718518428706335,
      "learning_rate": 1.8037948092805834e-06,
      "loss": 0.546,
      "step": 1780
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6967613842264047,
      "learning_rate": 1.8035492457330663e-06,
      "loss": 0.5527,
      "step": 1781
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5857458394611412,
      "learning_rate": 1.8033035453494839e-06,
      "loss": 0.5344,
      "step": 1782
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7777387049701806,
      "learning_rate": 1.8030577081716767e-06,
      "loss": 0.6168,
      "step": 1783
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7292091758705487,
      "learning_rate": 1.8028117342415076e-06,
      "loss": 0.5525,
      "step": 1784
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8399172280979899,
      "learning_rate": 1.8025656236008642e-06,
      "loss": 0.6659,
      "step": 1785
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7873550511642023,
      "learning_rate": 1.802319376291656e-06,
      "loss": 0.5645,
      "step": 1786
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7011668376376946,
      "learning_rate": 1.8020729923558165e-06,
      "loss": 0.6104,
      "step": 1787
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6575708396640998,
      "learning_rate": 1.8018264718353024e-06,
      "loss": 0.5711,
      "step": 1788
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.075276911909434,
      "learning_rate": 1.8015798147720935e-06,
      "loss": 0.6481,
      "step": 1789
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6511075333869601,
      "learning_rate": 1.801333021208193e-06,
      "loss": 0.5462,
      "step": 1790
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7116117749272244,
      "learning_rate": 1.8010860911856271e-06,
      "loss": 0.5712,
      "step": 1791
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7749589141582869,
      "learning_rate": 1.8008390247464455e-06,
      "loss": 0.6493,
      "step": 1792
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5759491996597453,
      "learning_rate": 1.8005918219327212e-06,
      "loss": 0.4923,
      "step": 1793
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.709414032631128,
      "learning_rate": 1.8003444827865497e-06,
      "loss": 0.5401,
      "step": 1794
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7270460076709832,
      "learning_rate": 1.8000970073500512e-06,
      "loss": 0.5872,
      "step": 1795
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7303684683271232,
      "learning_rate": 1.7998493956653675e-06,
      "loss": 0.5505,
      "step": 1796
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6632609688806054,
      "learning_rate": 1.7996016477746644e-06,
      "loss": 0.5627,
      "step": 1797
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6716472636542259,
      "learning_rate": 1.7993537637201313e-06,
      "loss": 0.5347,
      "step": 1798
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8940057563993772,
      "learning_rate": 1.7991057435439796e-06,
      "loss": 0.6268,
      "step": 1799
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9901882066021813,
      "learning_rate": 1.7988575872884448e-06,
      "loss": 0.668,
      "step": 1800
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7736175987787409,
      "learning_rate": 1.7986092949957851e-06,
      "loss": 0.6248,
      "step": 1801
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6297204859594099,
      "learning_rate": 1.7983608667082825e-06,
      "loss": 0.5351,
      "step": 1802
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0389765068198484,
      "learning_rate": 1.7981123024682417e-06,
      "loss": 0.6717,
      "step": 1803
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6310245126822918,
      "learning_rate": 1.7978636023179902e-06,
      "loss": 0.5346,
      "step": 1804
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8822144262115715,
      "learning_rate": 1.7976147662998795e-06,
      "loss": 0.6763,
      "step": 1805
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8367915734638388,
      "learning_rate": 1.7973657944562835e-06,
      "loss": 0.5568,
      "step": 1806
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6700320410292668,
      "learning_rate": 1.7971166868295996e-06,
      "loss": 0.5901,
      "step": 1807
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8039083210573608,
      "learning_rate": 1.7968674434622484e-06,
      "loss": 0.6085,
      "step": 1808
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7332711376306712,
      "learning_rate": 1.7966180643966731e-06,
      "loss": 0.5897,
      "step": 1809
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8017395546653407,
      "learning_rate": 1.7963685496753404e-06,
      "loss": 0.623,
      "step": 1810
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9731423940169147,
      "learning_rate": 1.7961188993407405e-06,
      "loss": 0.7114,
      "step": 1811
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9136533455331627,
      "learning_rate": 1.7958691134353859e-06,
      "loss": 0.6612,
      "step": 1812
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6824813024463036,
      "learning_rate": 1.7956191920018124e-06,
      "loss": 0.5442,
      "step": 1813
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6547783313914429,
      "learning_rate": 1.7953691350825793e-06,
      "loss": 0.5869,
      "step": 1814
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9107461437175577,
      "learning_rate": 1.7951189427202686e-06,
      "loss": 0.6044,
      "step": 1815
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9049523623551637,
      "learning_rate": 1.7948686149574854e-06,
      "loss": 0.6302,
      "step": 1816
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6533488673986121,
      "learning_rate": 1.7946181518368582e-06,
      "loss": 0.5485,
      "step": 1817
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7732942418058559,
      "learning_rate": 1.7943675534010376e-06,
      "loss": 0.6102,
      "step": 1818
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7161350974579639,
      "learning_rate": 1.7941168196926987e-06,
      "loss": 0.582,
      "step": 1819
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7637448549180914,
      "learning_rate": 1.7938659507545385e-06,
      "loss": 0.5291,
      "step": 1820
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8450502837249684,
      "learning_rate": 1.793614946629277e-06,
      "loss": 0.6706,
      "step": 1821
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.874345592622125,
      "learning_rate": 1.793363807359658e-06,
      "loss": 0.6393,
      "step": 1822
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8434333773652687,
      "learning_rate": 1.7931125329884478e-06,
      "loss": 0.6015,
      "step": 1823
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6753504198295516,
      "learning_rate": 1.7928611235584356e-06,
      "loss": 0.5709,
      "step": 1824
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.840266876293863,
      "learning_rate": 1.7926095791124344e-06,
      "loss": 0.66,
      "step": 1825
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.784821383743405,
      "learning_rate": 1.7923578996932794e-06,
      "loss": 0.6212,
      "step": 1826
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6884032469451853,
      "learning_rate": 1.7921060853438284e-06,
      "loss": 0.542,
      "step": 1827
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8075422832975773,
      "learning_rate": 1.7918541361069633e-06,
      "loss": 0.6222,
      "step": 1828
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7399777559833074,
      "learning_rate": 1.7916020520255884e-06,
      "loss": 0.5651,
      "step": 1829
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6104505732127692,
      "learning_rate": 1.7913498331426308e-06,
      "loss": 0.5023,
      "step": 1830
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6845071756461392,
      "learning_rate": 1.7910974795010405e-06,
      "loss": 0.5669,
      "step": 1831
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6983308791905123,
      "learning_rate": 1.7908449911437914e-06,
      "loss": 0.5664,
      "step": 1832
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5641879746773053,
      "learning_rate": 1.7905923681138792e-06,
      "loss": 0.6193,
      "step": 1833
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9189165504242603,
      "learning_rate": 1.7903396104543232e-06,
      "loss": 0.586,
      "step": 1834
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6768418230283442,
      "learning_rate": 1.7900867182081648e-06,
      "loss": 0.5129,
      "step": 1835
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7459100265506579,
      "learning_rate": 1.7898336914184698e-06,
      "loss": 0.5502,
      "step": 1836
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.894648220461959,
      "learning_rate": 1.7895805301283256e-06,
      "loss": 0.6167,
      "step": 1837
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7826766630910629,
      "learning_rate": 1.7893272343808426e-06,
      "loss": 0.6205,
      "step": 1838
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6585168820675301,
      "learning_rate": 1.789073804219155e-06,
      "loss": 0.5373,
      "step": 1839
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6929156481527876,
      "learning_rate": 1.788820239686419e-06,
      "loss": 0.5531,
      "step": 1840
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8118485495405864,
      "learning_rate": 1.7885665408258138e-06,
      "loss": 0.6576,
      "step": 1841
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.633392084229221,
      "learning_rate": 1.788312707680542e-06,
      "loss": 0.5176,
      "step": 1842
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5939779440611529,
      "learning_rate": 1.7880587402938287e-06,
      "loss": 0.458,
      "step": 1843
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7921128822794303,
      "learning_rate": 1.787804638708922e-06,
      "loss": 0.5956,
      "step": 1844
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7262286736980104,
      "learning_rate": 1.7875504029690926e-06,
      "loss": 0.5532,
      "step": 1845
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8430102284907283,
      "learning_rate": 1.7872960331176344e-06,
      "loss": 0.6652,
      "step": 1846
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8025138987104528,
      "learning_rate": 1.7870415291978633e-06,
      "loss": 0.6081,
      "step": 1847
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7528508808124932,
      "learning_rate": 1.7867868912531195e-06,
      "loss": 0.5858,
      "step": 1848
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.82065606932345,
      "learning_rate": 1.7865321193267645e-06,
      "loss": 0.6887,
      "step": 1849
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6852720152451314,
      "learning_rate": 1.7862772134621835e-06,
      "loss": 0.513,
      "step": 1850
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8240005478968955,
      "learning_rate": 1.786022173702785e-06,
      "loss": 0.6599,
      "step": 1851
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6526852813261855,
      "learning_rate": 1.7857670000919986e-06,
      "loss": 0.5462,
      "step": 1852
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8147762917629007,
      "learning_rate": 1.7855116926732783e-06,
      "loss": 0.6072,
      "step": 1853
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6855452714477639,
      "learning_rate": 1.7852562514901e-06,
      "loss": 0.5743,
      "step": 1854
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7763801136740046,
      "learning_rate": 1.785000676585963e-06,
      "loss": 0.6082,
      "step": 1855
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.69595133939554,
      "learning_rate": 1.7847449680043886e-06,
      "loss": 0.4729,
      "step": 1856
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8918771608434011,
      "learning_rate": 1.784489125788922e-06,
      "loss": 0.6664,
      "step": 1857
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9657679285298948,
      "learning_rate": 1.7842331499831294e-06,
      "loss": 0.7128,
      "step": 1858
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8663473904549109,
      "learning_rate": 1.783977040630602e-06,
      "loss": 0.7076,
      "step": 1859
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6453726744615014,
      "learning_rate": 1.7837207977749516e-06,
      "loss": 0.5352,
      "step": 1860
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6530170249749625,
      "learning_rate": 1.7834644214598145e-06,
      "loss": 0.5621,
      "step": 1861
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.95251570310003,
      "learning_rate": 1.7832079117288483e-06,
      "loss": 0.7001,
      "step": 1862
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6722346808365075,
      "learning_rate": 1.782951268625734e-06,
      "loss": 0.558,
      "step": 1863
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9274179274129589,
      "learning_rate": 1.7826944921941753e-06,
      "loss": 0.6076,
      "step": 1864
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6853838023398372,
      "learning_rate": 1.782437582477899e-06,
      "loss": 0.5541,
      "step": 1865
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7599640349084955,
      "learning_rate": 1.7821805395206535e-06,
      "loss": 0.6366,
      "step": 1866
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7769829132601709,
      "learning_rate": 1.7819233633662109e-06,
      "loss": 0.624,
      "step": 1867
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8693199033196352,
      "learning_rate": 1.7816660540583656e-06,
      "loss": 0.6348,
      "step": 1868
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7822921014276808,
      "learning_rate": 1.7814086116409346e-06,
      "loss": 0.5689,
      "step": 1869
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6611646884305945,
      "learning_rate": 1.7811510361577576e-06,
      "loss": 0.5696,
      "step": 1870
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.800443096287394,
      "learning_rate": 1.7808933276526971e-06,
      "loss": 0.6115,
      "step": 1871
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8331911398045193,
      "learning_rate": 1.7806354861696384e-06,
      "loss": 0.6356,
      "step": 1872
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6844827744325065,
      "learning_rate": 1.7803775117524888e-06,
      "loss": 0.5514,
      "step": 1873
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8547481653654526,
      "learning_rate": 1.7801194044451791e-06,
      "loss": 0.5512,
      "step": 1874
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7221053148217083,
      "learning_rate": 1.7798611642916619e-06,
      "loss": 0.5735,
      "step": 1875
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7739495685772367,
      "learning_rate": 1.7796027913359129e-06,
      "loss": 0.5793,
      "step": 1876
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7297646355104602,
      "learning_rate": 1.7793442856219304e-06,
      "loss": 0.5421,
      "step": 1877
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6461841130311093,
      "learning_rate": 1.7790856471937352e-06,
      "loss": 0.5127,
      "step": 1878
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9236566265321192,
      "learning_rate": 1.778826876095371e-06,
      "loss": 0.66,
      "step": 1879
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7539636166850503,
      "learning_rate": 1.778567972370903e-06,
      "loss": 0.6375,
      "step": 1880
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7286288032686404,
      "learning_rate": 1.7783089360644209e-06,
      "loss": 0.579,
      "step": 1881
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8678458234518728,
      "learning_rate": 1.7780497672200352e-06,
      "loss": 0.6182,
      "step": 1882
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8712181169348084,
      "learning_rate": 1.7777904658818797e-06,
      "loss": 0.5817,
      "step": 1883
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9606644706029104,
      "learning_rate": 1.777531032094111e-06,
      "loss": 0.6993,
      "step": 1884
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7675116112509338,
      "learning_rate": 1.7772714659009075e-06,
      "loss": 0.5638,
      "step": 1885
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.804147269510392,
      "learning_rate": 1.777011767346471e-06,
      "loss": 0.6418,
      "step": 1886
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8462590077494584,
      "learning_rate": 1.7767519364750254e-06,
      "loss": 0.6466,
      "step": 1887
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7649575442190123,
      "learning_rate": 1.7764919733308172e-06,
      "loss": 0.6224,
      "step": 1888
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6641919143395882,
      "learning_rate": 1.7762318779581155e-06,
      "loss": 0.5599,
      "step": 1889
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6191015615607506,
      "learning_rate": 1.7759716504012117e-06,
      "loss": 0.5471,
      "step": 1890
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.826367261722918,
      "learning_rate": 1.7757112907044198e-06,
      "loss": 0.6312,
      "step": 1891
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7329093125374098,
      "learning_rate": 1.7754507989120762e-06,
      "loss": 0.5807,
      "step": 1892
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0880791361518078,
      "learning_rate": 1.7751901750685402e-06,
      "loss": 0.6877,
      "step": 1893
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7794289140878874,
      "learning_rate": 1.7749294192181936e-06,
      "loss": 0.6075,
      "step": 1894
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9488668021981287,
      "learning_rate": 1.7746685314054397e-06,
      "loss": 0.6564,
      "step": 1895
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7419709195127533,
      "learning_rate": 1.7744075116747055e-06,
      "loss": 0.5439,
      "step": 1896
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.609804270726911,
      "learning_rate": 1.7741463600704396e-06,
      "loss": 0.5878,
      "step": 1897
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9678376361798278,
      "learning_rate": 1.7738850766371138e-06,
      "loss": 0.7068,
      "step": 1898
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7030072723511392,
      "learning_rate": 1.7736236614192216e-06,
      "loss": 0.5142,
      "step": 1899
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7356532445456269,
      "learning_rate": 1.7733621144612796e-06,
      "loss": 0.5814,
      "step": 1900
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6125566865459748,
      "learning_rate": 1.7731004358078261e-06,
      "loss": 0.4986,
      "step": 1901
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6870762972652394,
      "learning_rate": 1.7728386255034224e-06,
      "loss": 0.5201,
      "step": 1902
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6047305834695749,
      "learning_rate": 1.7725766835926526e-06,
      "loss": 0.5208,
      "step": 1903
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6297307208200389,
      "learning_rate": 1.7723146101201218e-06,
      "loss": 0.517,
      "step": 1904
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7478567894659794,
      "learning_rate": 1.772052405130459e-06,
      "loss": 0.6344,
      "step": 1905
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8826083647122114,
      "learning_rate": 1.7717900686683148e-06,
      "loss": 0.5932,
      "step": 1906
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.836202440522907,
      "learning_rate": 1.7715276007783621e-06,
      "loss": 0.6345,
      "step": 1907
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7268851100974629,
      "learning_rate": 1.771265001505297e-06,
      "loss": 0.5825,
      "step": 1908
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6382083365342568,
      "learning_rate": 1.771002270893837e-06,
      "loss": 0.5736,
      "step": 1909
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8316729488570054,
      "learning_rate": 1.7707394089887226e-06,
      "loss": 0.6759,
      "step": 1910
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.602748423526947,
      "learning_rate": 1.7704764158347164e-06,
      "loss": 0.4984,
      "step": 1911
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1711787851824031,
      "learning_rate": 1.7702132914766033e-06,
      "loss": 0.6554,
      "step": 1912
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8060070938716503,
      "learning_rate": 1.7699500359591904e-06,
      "loss": 0.6326,
      "step": 1913
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7796377904710955,
      "learning_rate": 1.7696866493273079e-06,
      "loss": 0.5641,
      "step": 1914
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9707287094226319,
      "learning_rate": 1.7694231316258077e-06,
      "loss": 0.65,
      "step": 1915
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9936923167622201,
      "learning_rate": 1.7691594828995637e-06,
      "loss": 0.6512,
      "step": 1916
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8277719101354267,
      "learning_rate": 1.768895703193473e-06,
      "loss": 0.6593,
      "step": 1917
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7802760148694542,
      "learning_rate": 1.7686317925524544e-06,
      "loss": 0.659,
      "step": 1918
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.876995905368891,
      "learning_rate": 1.7683677510214489e-06,
      "loss": 0.6658,
      "step": 1919
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6567710691631918,
      "learning_rate": 1.7681035786454202e-06,
      "loss": 0.5623,
      "step": 1920
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7822242399840562,
      "learning_rate": 1.767839275469354e-06,
      "loss": 0.5837,
      "step": 1921
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9110085841157554,
      "learning_rate": 1.7675748415382587e-06,
      "loss": 0.6497,
      "step": 1922
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7521348787254326,
      "learning_rate": 1.7673102768971644e-06,
      "loss": 0.5934,
      "step": 1923
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8471070197928361,
      "learning_rate": 1.7670455815911236e-06,
      "loss": 0.6324,
      "step": 1924
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8958541573756849,
      "learning_rate": 1.7667807556652115e-06,
      "loss": 0.6207,
      "step": 1925
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7457863893156891,
      "learning_rate": 1.7665157991645247e-06,
      "loss": 0.5553,
      "step": 1926
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7302968150403704,
      "learning_rate": 1.766250712134183e-06,
      "loss": 0.6165,
      "step": 1927
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6122864161430793,
      "learning_rate": 1.7659854946193278e-06,
      "loss": 0.5737,
      "step": 1928
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7549936233355308,
      "learning_rate": 1.7657201466651226e-06,
      "loss": 0.5881,
      "step": 1929
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.655692417023371,
      "learning_rate": 1.7654546683167539e-06,
      "loss": 0.5723,
      "step": 1930
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6368001873576175,
      "learning_rate": 1.7651890596194293e-06,
      "loss": 0.5573,
      "step": 1931
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.702391316569007,
      "learning_rate": 1.7649233206183797e-06,
      "loss": 0.4837,
      "step": 1932
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7221252873810466,
      "learning_rate": 1.764657451358858e-06,
      "loss": 0.5834,
      "step": 1933
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8900272411860402,
      "learning_rate": 1.7643914518861377e-06,
      "loss": 0.6337,
      "step": 1934
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.717110735990919,
      "learning_rate": 1.7641253222455169e-06,
      "loss": 0.5198,
      "step": 1935
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8049182601354071,
      "learning_rate": 1.7638590624823144e-06,
      "loss": 0.6588,
      "step": 1936
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.835161513069722,
      "learning_rate": 1.763592672641871e-06,
      "loss": 0.5984,
      "step": 1937
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6904690097870843,
      "learning_rate": 1.763326152769551e-06,
      "loss": 0.591,
      "step": 1938
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8632441537915783,
      "learning_rate": 1.7630595029107393e-06,
      "loss": 0.648,
      "step": 1939
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7826437403444958,
      "learning_rate": 1.7627927231108436e-06,
      "loss": 0.5874,
      "step": 1940
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6287215463480835,
      "learning_rate": 1.762525813415294e-06,
      "loss": 0.5535,
      "step": 1941
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7659824403455862,
      "learning_rate": 1.7622587738695423e-06,
      "loss": 0.5302,
      "step": 1942
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.843256894496395,
      "learning_rate": 1.7619916045190623e-06,
      "loss": 0.6414,
      "step": 1943
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7774421131888085,
      "learning_rate": 1.7617243054093507e-06,
      "loss": 0.623,
      "step": 1944
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6506932904808445,
      "learning_rate": 1.7614568765859256e-06,
      "loss": 0.5609,
      "step": 1945
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7815221517381211,
      "learning_rate": 1.7611893180943268e-06,
      "loss": 0.5653,
      "step": 1946
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6522070242029065,
      "learning_rate": 1.7609216299801174e-06,
      "loss": 0.5913,
      "step": 1947
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7507391850490923,
      "learning_rate": 1.7606538122888817e-06,
      "loss": 0.5346,
      "step": 1948
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5720325102953173,
      "learning_rate": 1.7603858650662259e-06,
      "loss": 0.5217,
      "step": 1949
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6441855593318525,
      "learning_rate": 1.7601177883577791e-06,
      "loss": 0.5129,
      "step": 1950
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7232599441643058,
      "learning_rate": 1.7598495822091919e-06,
      "loss": 0.5437,
      "step": 1951
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6517641379863649,
      "learning_rate": 1.759581246666137e-06,
      "loss": 0.6005,
      "step": 1952
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7861779904728772,
      "learning_rate": 1.7593127817743087e-06,
      "loss": 0.5863,
      "step": 1953
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.698028513976124,
      "learning_rate": 1.7590441875794244e-06,
      "loss": 0.5555,
      "step": 1954
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6573264891159366,
      "learning_rate": 1.7587754641272226e-06,
      "loss": 0.5982,
      "step": 1955
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7823142755271408,
      "learning_rate": 1.758506611463464e-06,
      "loss": 0.6726,
      "step": 1956
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9503604097595598,
      "learning_rate": 1.7582376296339317e-06,
      "loss": 0.6091,
      "step": 1957
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8649748996681877,
      "learning_rate": 1.7579685186844302e-06,
      "loss": 0.6053,
      "step": 1958
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6890272099073813,
      "learning_rate": 1.7576992786607865e-06,
      "loss": 0.5852,
      "step": 1959
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6892385383971602,
      "learning_rate": 1.7574299096088493e-06,
      "loss": 0.5764,
      "step": 1960
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6998493343120642,
      "learning_rate": 1.757160411574489e-06,
      "loss": 0.5651,
      "step": 1961
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1327296386247903,
      "learning_rate": 1.7568907846035992e-06,
      "loss": 0.6668,
      "step": 1962
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9262251267845324,
      "learning_rate": 1.7566210287420935e-06,
      "loss": 0.6537,
      "step": 1963
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7848158578849249,
      "learning_rate": 1.7563511440359094e-06,
      "loss": 0.5477,
      "step": 1964
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.730317574898393,
      "learning_rate": 1.7560811305310051e-06,
      "loss": 0.5811,
      "step": 1965
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8278870905522837,
      "learning_rate": 1.7558109882733606e-06,
      "loss": 0.6411,
      "step": 1966
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8054967152969693,
      "learning_rate": 1.7555407173089794e-06,
      "loss": 0.6234,
      "step": 1967
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8773575157836968,
      "learning_rate": 1.7552703176838848e-06,
      "loss": 0.6383,
      "step": 1968
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6508041652847512,
      "learning_rate": 1.7549997894441236e-06,
      "loss": 0.5225,
      "step": 1969
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9175595680321516,
      "learning_rate": 1.7547291326357638e-06,
      "loss": 0.6402,
      "step": 1970
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6227471094894387,
      "learning_rate": 1.7544583473048955e-06,
      "loss": 0.5223,
      "step": 1971
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8155018910323264,
      "learning_rate": 1.7541874334976303e-06,
      "loss": 0.5812,
      "step": 1972
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6350887075095119,
      "learning_rate": 1.7539163912601023e-06,
      "loss": 0.4831,
      "step": 1973
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8050582414820544,
      "learning_rate": 1.7536452206384671e-06,
      "loss": 0.6321,
      "step": 1974
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7538737987100852,
      "learning_rate": 1.7533739216789024e-06,
      "loss": 0.5709,
      "step": 1975
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9600290402927429,
      "learning_rate": 1.7531024944276074e-06,
      "loss": 0.6369,
      "step": 1976
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8174533281317452,
      "learning_rate": 1.7528309389308032e-06,
      "loss": 0.6263,
      "step": 1977
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6971373595206908,
      "learning_rate": 1.752559255234733e-06,
      "loss": 0.5605,
      "step": 1978
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9526745824675443,
      "learning_rate": 1.752287443385662e-06,
      "loss": 0.6108,
      "step": 1979
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.689269069695771,
      "learning_rate": 1.7520155034298768e-06,
      "loss": 0.5268,
      "step": 1980
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8708647272556966,
      "learning_rate": 1.7517434354136854e-06,
      "loss": 0.6028,
      "step": 1981
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6361585543414858,
      "learning_rate": 1.7514712393834188e-06,
      "loss": 0.5453,
      "step": 1982
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8392434249678136,
      "learning_rate": 1.7511989153854288e-06,
      "loss": 0.6148,
      "step": 1983
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7075448213622311,
      "learning_rate": 1.7509264634660894e-06,
      "loss": 0.5685,
      "step": 1984
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8528814133011654,
      "learning_rate": 1.7506538836717964e-06,
      "loss": 0.6456,
      "step": 1985
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8265923423981631,
      "learning_rate": 1.7503811760489673e-06,
      "loss": 0.5389,
      "step": 1986
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8008823004752446,
      "learning_rate": 1.7501083406440412e-06,
      "loss": 0.5195,
      "step": 1987
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8033964058898991,
      "learning_rate": 1.7498353775034796e-06,
      "loss": 0.6242,
      "step": 1988
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0737669195935626,
      "learning_rate": 1.7495622866737647e-06,
      "loss": 0.6399,
      "step": 1989
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8321079576771585,
      "learning_rate": 1.7492890682014014e-06,
      "loss": 0.5917,
      "step": 1990
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9312648408941783,
      "learning_rate": 1.7490157221329155e-06,
      "loss": 0.699,
      "step": 1991
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7327398403417205,
      "learning_rate": 1.7487422485148557e-06,
      "loss": 0.5712,
      "step": 1992
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6849808079150778,
      "learning_rate": 1.748468647393791e-06,
      "loss": 0.5024,
      "step": 1993
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9348076194541277,
      "learning_rate": 1.7481949188163134e-06,
      "loss": 0.6158,
      "step": 1994
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6982390069361648,
      "learning_rate": 1.7479210628290356e-06,
      "loss": 0.5888,
      "step": 1995
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6678979061957547,
      "learning_rate": 1.7476470794785927e-06,
      "loss": 0.5531,
      "step": 1996
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6122456060231373,
      "learning_rate": 1.747372968811641e-06,
      "loss": 0.5081,
      "step": 1997
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.847510734132804,
      "learning_rate": 1.7470987308748588e-06,
      "loss": 0.6683,
      "step": 1998
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8597071874715856,
      "learning_rate": 1.746824365714946e-06,
      "loss": 0.5579,
      "step": 1999
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0049671843803445,
      "learning_rate": 1.7465498733786238e-06,
      "loss": 0.6882,
      "step": 2000
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9632850134203702,
      "learning_rate": 1.7462752539126359e-06,
      "loss": 0.6602,
      "step": 2001
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6204375705688481,
      "learning_rate": 1.746000507363747e-06,
      "loss": 0.4932,
      "step": 2002
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7255976172712437,
      "learning_rate": 1.7457256337787434e-06,
      "loss": 0.5694,
      "step": 2003
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0890434377212321,
      "learning_rate": 1.7454506332044331e-06,
      "loss": 0.6209,
      "step": 2004
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8036445124670614,
      "learning_rate": 1.7451755056876466e-06,
      "loss": 0.6673,
      "step": 2005
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8681376660133959,
      "learning_rate": 1.744900251275234e-06,
      "loss": 0.6258,
      "step": 2006
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7520154982144631,
      "learning_rate": 1.744624870014069e-06,
      "loss": 0.5474,
      "step": 2007
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7584632840173634,
      "learning_rate": 1.7443493619510465e-06,
      "loss": 0.5703,
      "step": 2008
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6902469840265371,
      "learning_rate": 1.7440737271330818e-06,
      "loss": 0.5149,
      "step": 2009
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6299731996901248,
      "learning_rate": 1.7437979656071135e-06,
      "loss": 0.5525,
      "step": 2010
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7647183068795858,
      "learning_rate": 1.7435220774201e-06,
      "loss": 0.5799,
      "step": 2011
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9352919537488072,
      "learning_rate": 1.7432460626190228e-06,
      "loss": 0.6229,
      "step": 2012
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0063283423286586,
      "learning_rate": 1.7429699212508845e-06,
      "loss": 0.6647,
      "step": 2013
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5984889037514833,
      "learning_rate": 1.7426936533627088e-06,
      "loss": 0.4911,
      "step": 2014
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.92772233450929,
      "learning_rate": 1.742417259001541e-06,
      "loss": 0.6753,
      "step": 2015
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6863813135188469,
      "learning_rate": 1.7421407382144486e-06,
      "loss": 0.563,
      "step": 2016
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.793328552887762,
      "learning_rate": 1.7418640910485198e-06,
      "loss": 0.607,
      "step": 2017
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7328694300580574,
      "learning_rate": 1.7415873175508654e-06,
      "loss": 0.6,
      "step": 2018
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9601012962013009,
      "learning_rate": 1.7413104177686165e-06,
      "loss": 0.6537,
      "step": 2019
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7216811978034054,
      "learning_rate": 1.741033391748926e-06,
      "loss": 0.572,
      "step": 2020
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7626175276009577,
      "learning_rate": 1.7407562395389693e-06,
      "loss": 0.5778,
      "step": 2021
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8036142869090829,
      "learning_rate": 1.7404789611859419e-06,
      "loss": 0.6058,
      "step": 2022
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8210347776463577,
      "learning_rate": 1.740201556737062e-06,
      "loss": 0.6407,
      "step": 2023
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4226676888999619,
      "learning_rate": 1.739924026239568e-06,
      "loss": 0.7019,
      "step": 2024
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8600482768017905,
      "learning_rate": 1.739646369740721e-06,
      "loss": 0.6404,
      "step": 2025
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6722306567439322,
      "learning_rate": 1.7393685872878026e-06,
      "loss": 0.5103,
      "step": 2026
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8070650001578221,
      "learning_rate": 1.7390906789281167e-06,
      "loss": 0.5428,
      "step": 2027
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8182578516738027,
      "learning_rate": 1.7388126447089879e-06,
      "loss": 0.5756,
      "step": 2028
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6760540250660055,
      "learning_rate": 1.7385344846777628e-06,
      "loss": 0.5388,
      "step": 2029
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8547478698231441,
      "learning_rate": 1.7382561988818086e-06,
      "loss": 0.6966,
      "step": 2030
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6028461219783835,
      "learning_rate": 1.7379777873685152e-06,
      "loss": 0.5249,
      "step": 2031
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7751331584886236,
      "learning_rate": 1.7376992501852922e-06,
      "loss": 0.5576,
      "step": 2032
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8529732318524347,
      "learning_rate": 1.7374205873795727e-06,
      "loss": 0.6323,
      "step": 2033
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.784708571185131,
      "learning_rate": 1.7371417989988095e-06,
      "loss": 0.6677,
      "step": 2034
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8445203811228591,
      "learning_rate": 1.736862885090477e-06,
      "loss": 0.5922,
      "step": 2035
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.626953814265976,
      "learning_rate": 1.7365838457020723e-06,
      "loss": 0.5533,
      "step": 2036
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9373658271261294,
      "learning_rate": 1.7363046808811123e-06,
      "loss": 0.5383,
      "step": 2037
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8206727013516608,
      "learning_rate": 1.7360253906751358e-06,
      "loss": 0.6051,
      "step": 2038
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7839039452778948,
      "learning_rate": 1.735745975131703e-06,
      "loss": 0.6538,
      "step": 2039
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7428453455709109,
      "learning_rate": 1.7354664342983959e-06,
      "loss": 0.5713,
      "step": 2040
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6882113452087494,
      "learning_rate": 1.7351867682228167e-06,
      "loss": 0.5198,
      "step": 2041
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7405770388955921,
      "learning_rate": 1.7349069769525902e-06,
      "loss": 0.5891,
      "step": 2042
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6889721181823543,
      "learning_rate": 1.7346270605353617e-06,
      "loss": 0.5596,
      "step": 2043
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7022138186757454,
      "learning_rate": 1.734347019018798e-06,
      "loss": 0.5185,
      "step": 2044
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6419877180474133,
      "learning_rate": 1.7340668524505873e-06,
      "loss": 0.5331,
      "step": 2045
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8225161794894839,
      "learning_rate": 1.7337865608784392e-06,
      "loss": 0.5974,
      "step": 2046
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9045042253625015,
      "learning_rate": 1.7335061443500845e-06,
      "loss": 0.6199,
      "step": 2047
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8477206326338558,
      "learning_rate": 1.7332256029132748e-06,
      "loss": 0.6107,
      "step": 2048
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6724255637797462,
      "learning_rate": 1.7329449366157833e-06,
      "loss": 0.5241,
      "step": 2049
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8198267896718139,
      "learning_rate": 1.7326641455054055e-06,
      "loss": 0.6457,
      "step": 2050
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7062066621521382,
      "learning_rate": 1.7323832296299562e-06,
      "loss": 0.5616,
      "step": 2051
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6419552377439518,
      "learning_rate": 1.7321021890372727e-06,
      "loss": 0.5069,
      "step": 2052
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8959814502871365,
      "learning_rate": 1.7318210237752135e-06,
      "loss": 0.675,
      "step": 2053
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0490021598480384,
      "learning_rate": 1.7315397338916581e-06,
      "loss": 0.6811,
      "step": 2054
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8716286483232623,
      "learning_rate": 1.731258319434507e-06,
      "loss": 0.6459,
      "step": 2055
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.698011633756415,
      "learning_rate": 1.7309767804516823e-06,
      "loss": 0.5216,
      "step": 2056
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7508879306314754,
      "learning_rate": 1.7306951169911275e-06,
      "loss": 0.6254,
      "step": 2057
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9881541263076405,
      "learning_rate": 1.7304133291008062e-06,
      "loss": 0.6911,
      "step": 2058
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7116438983719325,
      "learning_rate": 1.7301314168287046e-06,
      "loss": 0.5544,
      "step": 2059
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6403762726257071,
      "learning_rate": 1.7298493802228289e-06,
      "loss": 0.5406,
      "step": 2060
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8413679826564827,
      "learning_rate": 1.7295672193312076e-06,
      "loss": 0.5872,
      "step": 2061
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7017827772766783,
      "learning_rate": 1.7292849342018893e-06,
      "loss": 0.5858,
      "step": 2062
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5594918324840721,
      "learning_rate": 1.7290025248829443e-06,
      "loss": 0.4846,
      "step": 2063
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8874858545529425,
      "learning_rate": 1.7287199914224639e-06,
      "loss": 0.6424,
      "step": 2064
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.641578037508466,
      "learning_rate": 1.7284373338685608e-06,
      "loss": 0.5471,
      "step": 2065
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8688531300229159,
      "learning_rate": 1.7281545522693687e-06,
      "loss": 0.6333,
      "step": 2066
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.629678099551423,
      "learning_rate": 1.727871646673042e-06,
      "loss": 0.5726,
      "step": 2067
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7194016810169793,
      "learning_rate": 1.7275886171277568e-06,
      "loss": 0.6058,
      "step": 2068
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7756462750011853,
      "learning_rate": 1.72730546368171e-06,
      "loss": 0.6118,
      "step": 2069
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6106209097449659,
      "learning_rate": 1.72702218638312e-06,
      "loss": 0.4811,
      "step": 2070
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6379225934836319,
      "learning_rate": 1.7267387852802255e-06,
      "loss": 0.5492,
      "step": 2071
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.73557929057043,
      "learning_rate": 1.7264552604212871e-06,
      "loss": 0.5174,
      "step": 2072
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6568481489244096,
      "learning_rate": 1.7261716118545859e-06,
      "loss": 0.5299,
      "step": 2073
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7522097401125556,
      "learning_rate": 1.7258878396284246e-06,
      "loss": 0.5839,
      "step": 2074
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8063774593821758,
      "learning_rate": 1.7256039437911263e-06,
      "loss": 0.5549,
      "step": 2075
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6184881228464356,
      "learning_rate": 1.7253199243910354e-06,
      "loss": 0.5388,
      "step": 2076
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6959889378114918,
      "learning_rate": 1.7250357814765182e-06,
      "loss": 0.5626,
      "step": 2077
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8693089443641364,
      "learning_rate": 1.7247515150959603e-06,
      "loss": 0.6737,
      "step": 2078
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6741355987290173,
      "learning_rate": 1.7244671252977702e-06,
      "loss": 0.564,
      "step": 2079
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6584876047266681,
      "learning_rate": 1.724182612130376e-06,
      "loss": 0.5259,
      "step": 2080
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5970392328563369,
      "learning_rate": 1.7238979756422275e-06,
      "loss": 0.5478,
      "step": 2081
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7172137035039096,
      "learning_rate": 1.723613215881795e-06,
      "loss": 0.5559,
      "step": 2082
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9840110340669319,
      "learning_rate": 1.7233283328975707e-06,
      "loss": 0.6053,
      "step": 2083
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6411436631160083,
      "learning_rate": 1.723043326738067e-06,
      "loss": 0.5191,
      "step": 2084
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6482826657422097,
      "learning_rate": 1.7227581974518168e-06,
      "loss": 0.5563,
      "step": 2085
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0901317704013287,
      "learning_rate": 1.7224729450873757e-06,
      "loss": 0.6605,
      "step": 2086
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8055626110352178,
      "learning_rate": 1.7221875696933184e-06,
      "loss": 0.5819,
      "step": 2087
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7366317969484671,
      "learning_rate": 1.7219020713182417e-06,
      "loss": 0.6884,
      "step": 2088
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8276012714795393,
      "learning_rate": 1.721616450010763e-06,
      "loss": 0.5683,
      "step": 2089
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6588587281175609,
      "learning_rate": 1.7213307058195207e-06,
      "loss": 0.5875,
      "step": 2090
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8109735591974573,
      "learning_rate": 1.7210448387931739e-06,
      "loss": 0.652,
      "step": 2091
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7802042736312507,
      "learning_rate": 1.7207588489804033e-06,
      "loss": 0.5691,
      "step": 2092
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.697161140030107,
      "learning_rate": 1.720472736429909e-06,
      "loss": 0.5457,
      "step": 2093
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7997910661808749,
      "learning_rate": 1.7201865011904135e-06,
      "loss": 0.5979,
      "step": 2094
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.945852131385185,
      "learning_rate": 1.71990014331066e-06,
      "loss": 0.651,
      "step": 2095
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6451086111413631,
      "learning_rate": 1.7196136628394116e-06,
      "loss": 0.5439,
      "step": 2096
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6548722712150665,
      "learning_rate": 1.7193270598254537e-06,
      "loss": 0.5457,
      "step": 2097
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5997866798412184,
      "learning_rate": 1.719040334317591e-06,
      "loss": 0.4917,
      "step": 2098
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8882770414282587,
      "learning_rate": 1.7187534863646506e-06,
      "loss": 0.6783,
      "step": 2099
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0264187187429725,
      "learning_rate": 1.7184665160154793e-06,
      "loss": 0.6868,
      "step": 2100
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7550911790740567,
      "learning_rate": 1.7181794233189453e-06,
      "loss": 0.6052,
      "step": 2101
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8415172948757969,
      "learning_rate": 1.7178922083239372e-06,
      "loss": 0.5525,
      "step": 2102
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8350341661703067,
      "learning_rate": 1.7176048710793651e-06,
      "loss": 0.5958,
      "step": 2103
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8350758247524337,
      "learning_rate": 1.7173174116341596e-06,
      "loss": 0.6172,
      "step": 2104
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9668490416955452,
      "learning_rate": 1.7170298300372716e-06,
      "loss": 0.6661,
      "step": 2105
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6326829767314572,
      "learning_rate": 1.7167421263376736e-06,
      "loss": 0.4648,
      "step": 2106
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6216500110967164,
      "learning_rate": 1.7164543005843585e-06,
      "loss": 0.5154,
      "step": 2107
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9075321451175721,
      "learning_rate": 1.7161663528263398e-06,
      "loss": 0.6852,
      "step": 2108
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5629755789408005,
      "learning_rate": 1.7158782831126525e-06,
      "loss": 0.4656,
      "step": 2109
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6178415172064557,
      "learning_rate": 1.7155900914923515e-06,
      "loss": 0.5403,
      "step": 2110
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8607887958474726,
      "learning_rate": 1.7153017780145126e-06,
      "loss": 0.6229,
      "step": 2111
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7276456747468057,
      "learning_rate": 1.7150133427282332e-06,
      "loss": 0.5528,
      "step": 2112
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9511762620559335,
      "learning_rate": 1.7147247856826303e-06,
      "loss": 0.6678,
      "step": 2113
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7305804001135221,
      "learning_rate": 1.7144361069268425e-06,
      "loss": 0.5621,
      "step": 2114
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7952042001228986,
      "learning_rate": 1.7141473065100283e-06,
      "loss": 0.5323,
      "step": 2115
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8422396313309024,
      "learning_rate": 1.7138583844813678e-06,
      "loss": 0.671,
      "step": 2116
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8628427171623998,
      "learning_rate": 1.7135693408900615e-06,
      "loss": 0.6349,
      "step": 2117
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5829056475092363,
      "learning_rate": 1.7132801757853304e-06,
      "loss": 0.5336,
      "step": 2118
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7426735022617582,
      "learning_rate": 1.7129908892164161e-06,
      "loss": 0.5609,
      "step": 2119
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7630862592830553,
      "learning_rate": 1.7127014812325816e-06,
      "loss": 0.6093,
      "step": 2120
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8410352422783963,
      "learning_rate": 1.7124119518831095e-06,
      "loss": 0.6607,
      "step": 2121
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6977230046435009,
      "learning_rate": 1.7121223012173035e-06,
      "loss": 0.5645,
      "step": 2122
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9439437599152608,
      "learning_rate": 1.711832529284489e-06,
      "loss": 0.623,
      "step": 2123
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8083589046331039,
      "learning_rate": 1.7115426361340103e-06,
      "loss": 0.5622,
      "step": 2124
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6963257663638225,
      "learning_rate": 1.7112526218152334e-06,
      "loss": 0.5976,
      "step": 2125
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.07970299910784,
      "learning_rate": 1.710962486377545e-06,
      "loss": 0.649,
      "step": 2126
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8541522078462299,
      "learning_rate": 1.7106722298703517e-06,
      "loss": 0.5869,
      "step": 2127
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7878792812618934,
      "learning_rate": 1.7103818523430814e-06,
      "loss": 0.589,
      "step": 2128
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8478693557248129,
      "learning_rate": 1.7100913538451826e-06,
      "loss": 0.5821,
      "step": 2129
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9474649894452436,
      "learning_rate": 1.7098007344261237e-06,
      "loss": 0.7052,
      "step": 2130
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8630035329237031,
      "learning_rate": 1.7095099941353943e-06,
      "loss": 0.6952,
      "step": 2131
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6534126745300457,
      "learning_rate": 1.7092191330225045e-06,
      "loss": 0.5817,
      "step": 2132
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.963705659323309,
      "learning_rate": 1.7089281511369854e-06,
      "loss": 0.6528,
      "step": 2133
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7096005633145203,
      "learning_rate": 1.7086370485283874e-06,
      "loss": 0.5686,
      "step": 2134
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8818279876651732,
      "learning_rate": 1.7083458252462824e-06,
      "loss": 0.6102,
      "step": 2135
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.028091000548899,
      "learning_rate": 1.708054481340263e-06,
      "loss": 0.672,
      "step": 2136
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.796414800709641,
      "learning_rate": 1.707763016859942e-06,
      "loss": 0.5378,
      "step": 2137
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7351851921067623,
      "learning_rate": 1.7074714318549525e-06,
      "loss": 0.572,
      "step": 2138
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9407748319376167,
      "learning_rate": 1.7071797263749485e-06,
      "loss": 0.6571,
      "step": 2139
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7340355185471747,
      "learning_rate": 1.7068879004696047e-06,
      "loss": 0.5459,
      "step": 2140
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8183131026333355,
      "learning_rate": 1.7065959541886156e-06,
      "loss": 0.6309,
      "step": 2141
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6822109820329837,
      "learning_rate": 1.7063038875816966e-06,
      "loss": 0.54,
      "step": 2142
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6513381943331386,
      "learning_rate": 1.7060117006985842e-06,
      "loss": 0.5288,
      "step": 2143
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7558075510653973,
      "learning_rate": 1.7057193935890339e-06,
      "loss": 0.6757,
      "step": 2144
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9018917721839882,
      "learning_rate": 1.7054269663028232e-06,
      "loss": 0.674,
      "step": 2145
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9186939348802294,
      "learning_rate": 1.7051344188897491e-06,
      "loss": 0.5707,
      "step": 2146
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8249037723100083,
      "learning_rate": 1.7048417513996296e-06,
      "loss": 0.5581,
      "step": 2147
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9342802725875994,
      "learning_rate": 1.704548963882303e-06,
      "loss": 0.6334,
      "step": 2148
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8866595308803102,
      "learning_rate": 1.7042560563876277e-06,
      "loss": 0.6711,
      "step": 2149
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7065738011302267,
      "learning_rate": 1.7039630289654827e-06,
      "loss": 0.5646,
      "step": 2150
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7656869103137477,
      "learning_rate": 1.7036698816657677e-06,
      "loss": 0.6056,
      "step": 2151
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7387149381313486,
      "learning_rate": 1.7033766145384027e-06,
      "loss": 0.5454,
      "step": 2152
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0575021119036667,
      "learning_rate": 1.703083227633328e-06,
      "loss": 0.6513,
      "step": 2153
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9676258533584826,
      "learning_rate": 1.7027897210005045e-06,
      "loss": 0.6428,
      "step": 2154
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7409357932862124,
      "learning_rate": 1.7024960946899128e-06,
      "loss": 0.5933,
      "step": 2155
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6695974485071609,
      "learning_rate": 1.7022023487515548e-06,
      "loss": 0.5797,
      "step": 2156
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0754988509306311,
      "learning_rate": 1.7019084832354524e-06,
      "loss": 0.615,
      "step": 2157
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7124817761920776,
      "learning_rate": 1.7016144981916476e-06,
      "loss": 0.5159,
      "step": 2158
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7691015840482227,
      "learning_rate": 1.7013203936702035e-06,
      "loss": 0.5417,
      "step": 2159
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.839744772127702,
      "learning_rate": 1.7010261697212022e-06,
      "loss": 0.6657,
      "step": 2160
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.990325147278458,
      "learning_rate": 1.7007318263947478e-06,
      "loss": 0.6372,
      "step": 2161
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7801825037429803,
      "learning_rate": 1.7004373637409632e-06,
      "loss": 0.6368,
      "step": 2162
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8370609305765528,
      "learning_rate": 1.7001427818099927e-06,
      "loss": 0.6605,
      "step": 2163
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2888462949319373,
      "learning_rate": 1.6998480806520008e-06,
      "loss": 0.6666,
      "step": 2164
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7252536869720716,
      "learning_rate": 1.6995532603171716e-06,
      "loss": 0.5664,
      "step": 2165
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8545147508933769,
      "learning_rate": 1.69925832085571e-06,
      "loss": 0.6451,
      "step": 2166
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8812759949943069,
      "learning_rate": 1.6989632623178412e-06,
      "loss": 0.6652,
      "step": 2167
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6882869846149489,
      "learning_rate": 1.6986680847538104e-06,
      "loss": 0.5652,
      "step": 2168
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8762816365596262,
      "learning_rate": 1.6983727882138833e-06,
      "loss": 0.65,
      "step": 2169
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.658497060245278,
      "learning_rate": 1.6980773727483464e-06,
      "loss": 0.5033,
      "step": 2170
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8674016520601762,
      "learning_rate": 1.6977818384075053e-06,
      "loss": 0.5227,
      "step": 2171
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7927083404164387,
      "learning_rate": 1.6974861852416867e-06,
      "loss": 0.6301,
      "step": 2172
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6360577575131724,
      "learning_rate": 1.6971904133012368e-06,
      "loss": 0.5262,
      "step": 2173
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.820140594018224,
      "learning_rate": 1.696894522636523e-06,
      "loss": 0.6283,
      "step": 2174
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8620870491879562,
      "learning_rate": 1.6965985132979322e-06,
      "loss": 0.6135,
      "step": 2175
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.652693623692442,
      "learning_rate": 1.6963023853358715e-06,
      "loss": 0.5596,
      "step": 2176
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7554056127491318,
      "learning_rate": 1.6960061388007686e-06,
      "loss": 0.6226,
      "step": 2177
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7479137007145724,
      "learning_rate": 1.6957097737430714e-06,
      "loss": 0.6069,
      "step": 2178
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8956987391539133,
      "learning_rate": 1.6954132902132476e-06,
      "loss": 0.6982,
      "step": 2179
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8309677659853003,
      "learning_rate": 1.6951166882617853e-06,
      "loss": 0.6087,
      "step": 2180
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6053209532532329,
      "learning_rate": 1.6948199679391926e-06,
      "loss": 0.5292,
      "step": 2181
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7523583404994801,
      "learning_rate": 1.6945231292959975e-06,
      "loss": 0.5881,
      "step": 2182
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7684195868276319,
      "learning_rate": 1.6942261723827495e-06,
      "loss": 0.5566,
      "step": 2183
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8987275600204758,
      "learning_rate": 1.6939290972500166e-06,
      "loss": 0.6278,
      "step": 2184
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6671804008480465,
      "learning_rate": 1.6936319039483876e-06,
      "loss": 0.5623,
      "step": 2185
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8694903380476243,
      "learning_rate": 1.6933345925284718e-06,
      "loss": 0.6667,
      "step": 2186
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9155380878028245,
      "learning_rate": 1.6930371630408979e-06,
      "loss": 0.6425,
      "step": 2187
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7419916258652419,
      "learning_rate": 1.6927396155363151e-06,
      "loss": 0.5502,
      "step": 2188
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.732808851770632,
      "learning_rate": 1.6924419500653927e-06,
      "loss": 0.6347,
      "step": 2189
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7653711880219417,
      "learning_rate": 1.6921441666788198e-06,
      "loss": 0.64,
      "step": 2190
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6934293603697798,
      "learning_rate": 1.6918462654273062e-06,
      "loss": 0.5029,
      "step": 2191
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0112695200560367,
      "learning_rate": 1.6915482463615813e-06,
      "loss": 0.6946,
      "step": 2192
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.738589792174627,
      "learning_rate": 1.6912501095323943e-06,
      "loss": 0.5959,
      "step": 2193
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8591417693415365,
      "learning_rate": 1.6909518549905153e-06,
      "loss": 0.5827,
      "step": 2194
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6855854364045008,
      "learning_rate": 1.6906534827867335e-06,
      "loss": 0.517,
      "step": 2195
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6330346039478522,
      "learning_rate": 1.690354992971859e-06,
      "loss": 0.5681,
      "step": 2196
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7019132553112578,
      "learning_rate": 1.6900563855967212e-06,
      "loss": 0.5604,
      "step": 2197
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6539614271170227,
      "learning_rate": 1.6897576607121697e-06,
      "loss": 0.5317,
      "step": 2198
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8485886921396372,
      "learning_rate": 1.6894588183690745e-06,
      "loss": 0.636,
      "step": 2199
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6416593480921033,
      "learning_rate": 1.6891598586183253e-06,
      "loss": 0.507,
      "step": 2200
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7501176460034265,
      "learning_rate": 1.6888607815108318e-06,
      "loss": 0.5204,
      "step": 2201
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7224924742172312,
      "learning_rate": 1.6885615870975237e-06,
      "loss": 0.5474,
      "step": 2202
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6771364641228486,
      "learning_rate": 1.6882622754293507e-06,
      "loss": 0.5473,
      "step": 2203
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7022359759512,
      "learning_rate": 1.6879628465572823e-06,
      "loss": 0.5482,
      "step": 2204
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7317940359312319,
      "learning_rate": 1.6876633005323083e-06,
      "loss": 0.5889,
      "step": 2205
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7363327691843138,
      "learning_rate": 1.6873636374054385e-06,
      "loss": 0.558,
      "step": 2206
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8099894789713618,
      "learning_rate": 1.6870638572277021e-06,
      "loss": 0.6176,
      "step": 2207
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8634233915749285,
      "learning_rate": 1.6867639600501486e-06,
      "loss": 0.6528,
      "step": 2208
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8468848127915406,
      "learning_rate": 1.6864639459238472e-06,
      "loss": 0.6448,
      "step": 2209
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6383206096613858,
      "learning_rate": 1.6861638148998872e-06,
      "loss": 0.5535,
      "step": 2210
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6960432536249203,
      "learning_rate": 1.6858635670293784e-06,
      "loss": 0.5437,
      "step": 2211
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6220077639527575,
      "learning_rate": 1.6855632023634493e-06,
      "loss": 0.5992,
      "step": 2212
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9292095849828487,
      "learning_rate": 1.6852627209532484e-06,
      "loss": 0.6438,
      "step": 2213
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8107418055082478,
      "learning_rate": 1.6849621228499457e-06,
      "loss": 0.6005,
      "step": 2214
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6527167311307057,
      "learning_rate": 1.6846614081047292e-06,
      "loss": 0.4994,
      "step": 2215
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7091420183396735,
      "learning_rate": 1.6843605767688076e-06,
      "loss": 0.5786,
      "step": 2216
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6839856739860061,
      "learning_rate": 1.6840596288934097e-06,
      "loss": 0.5489,
      "step": 2217
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.025710684817184,
      "learning_rate": 1.6837585645297832e-06,
      "loss": 0.6287,
      "step": 2218
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6923564110321747,
      "learning_rate": 1.6834573837291966e-06,
      "loss": 0.5646,
      "step": 2219
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.787324540757855,
      "learning_rate": 1.6831560865429375e-06,
      "loss": 0.7215,
      "step": 2220
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8933669440639492,
      "learning_rate": 1.6828546730223143e-06,
      "loss": 0.6658,
      "step": 2221
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6783579195856242,
      "learning_rate": 1.682553143218654e-06,
      "loss": 0.5428,
      "step": 2222
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8570040110131358,
      "learning_rate": 1.6822514971833042e-06,
      "loss": 0.6425,
      "step": 2223
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6936390810707274,
      "learning_rate": 1.6819497349676324e-06,
      "loss": 0.5189,
      "step": 2224
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8854234399893237,
      "learning_rate": 1.6816478566230248e-06,
      "loss": 0.6357,
      "step": 2225
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7097785662217849,
      "learning_rate": 1.6813458622008887e-06,
      "loss": 0.5515,
      "step": 2226
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7523741797678828,
      "learning_rate": 1.6810437517526505e-06,
      "loss": 0.572,
      "step": 2227
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9036820917288725,
      "learning_rate": 1.6807415253297563e-06,
      "loss": 0.6496,
      "step": 2228
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8331417571658979,
      "learning_rate": 1.6804391829836726e-06,
      "loss": 0.6016,
      "step": 2229
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9526594615281615,
      "learning_rate": 1.6801367247658847e-06,
      "loss": 0.692,
      "step": 2230
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8370655857106266,
      "learning_rate": 1.6798341507278982e-06,
      "loss": 0.6133,
      "step": 2231
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8143533186848736,
      "learning_rate": 1.6795314609212379e-06,
      "loss": 0.6026,
      "step": 2232
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7100219613158313,
      "learning_rate": 1.6792286553974493e-06,
      "loss": 0.542,
      "step": 2233
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6208700582065056,
      "learning_rate": 1.678925734208097e-06,
      "loss": 0.514,
      "step": 2234
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6913565943684202,
      "learning_rate": 1.6786226974047653e-06,
      "loss": 0.5515,
      "step": 2235
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7052182927874939,
      "learning_rate": 1.678319545039058e-06,
      "loss": 0.5403,
      "step": 2236
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9051046305580936,
      "learning_rate": 1.6780162771625985e-06,
      "loss": 0.6663,
      "step": 2237
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7315610295571439,
      "learning_rate": 1.6777128938270308e-06,
      "loss": 0.5223,
      "step": 2238
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8231513082580345,
      "learning_rate": 1.6774093950840175e-06,
      "loss": 0.6256,
      "step": 2239
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8845994669674028,
      "learning_rate": 1.6771057809852416e-06,
      "loss": 0.5548,
      "step": 2240
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6945980589547308,
      "learning_rate": 1.6768020515824051e-06,
      "loss": 0.553,
      "step": 2241
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6577207950136928,
      "learning_rate": 1.67649820692723e-06,
      "loss": 0.559,
      "step": 2242
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6085064589394716,
      "learning_rate": 1.6761942470714582e-06,
      "loss": 0.524,
      "step": 2243
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8926260411514289,
      "learning_rate": 1.6758901720668504e-06,
      "loss": 0.6265,
      "step": 2244
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8207606742559838,
      "learning_rate": 1.675585981965188e-06,
      "loss": 0.529,
      "step": 2245
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8049246284630345,
      "learning_rate": 1.6752816768182708e-06,
      "loss": 0.6373,
      "step": 2246
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8578391069480552,
      "learning_rate": 1.6749772566779191e-06,
      "loss": 0.6124,
      "step": 2247
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7511569100479492,
      "learning_rate": 1.6746727215959729e-06,
      "loss": 0.5504,
      "step": 2248
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7736810626021798,
      "learning_rate": 1.6743680716242903e-06,
      "loss": 0.5563,
      "step": 2249
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6583082914792537,
      "learning_rate": 1.6740633068147507e-06,
      "loss": 0.5874,
      "step": 2250
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.72117599210976,
      "learning_rate": 1.6737584272192526e-06,
      "loss": 0.5713,
      "step": 2251
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9200312396943667,
      "learning_rate": 1.6734534328897135e-06,
      "loss": 0.6364,
      "step": 2252
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6207388513138897,
      "learning_rate": 1.6731483238780703e-06,
      "loss": 0.5673,
      "step": 2253
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6309643338798087,
      "learning_rate": 1.6728431002362809e-06,
      "loss": 0.5518,
      "step": 2254
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7768519097348239,
      "learning_rate": 1.672537762016321e-06,
      "loss": 0.5521,
      "step": 2255
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6849134614583908,
      "learning_rate": 1.6722323092701864e-06,
      "loss": 0.4893,
      "step": 2256
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6725543501430364,
      "learning_rate": 1.671926742049893e-06,
      "loss": 0.5008,
      "step": 2257
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6851289843771293,
      "learning_rate": 1.6716210604074755e-06,
      "loss": 0.6357,
      "step": 2258
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8675747215822092,
      "learning_rate": 1.6713152643949883e-06,
      "loss": 0.5649,
      "step": 2259
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9673955956468508,
      "learning_rate": 1.6710093540645054e-06,
      "loss": 0.5983,
      "step": 2260
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.964842357696227,
      "learning_rate": 1.6707033294681198e-06,
      "loss": 0.6021,
      "step": 2261
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6928061079172201,
      "learning_rate": 1.6703971906579448e-06,
      "loss": 0.5774,
      "step": 2262
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9382228022663371,
      "learning_rate": 1.6700909376861123e-06,
      "loss": 0.6263,
      "step": 2263
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8421502593237556,
      "learning_rate": 1.669784570604774e-06,
      "loss": 0.5932,
      "step": 2264
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6425987452711236,
      "learning_rate": 1.6694780894661012e-06,
      "loss": 0.5011,
      "step": 2265
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7526716561326535,
      "learning_rate": 1.6691714943222844e-06,
      "loss": 0.5227,
      "step": 2266
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7221312555383687,
      "learning_rate": 1.6688647852255332e-06,
      "loss": 0.5323,
      "step": 2267
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8119916887480131,
      "learning_rate": 1.6685579622280774e-06,
      "loss": 0.5432,
      "step": 2268
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7171680935434755,
      "learning_rate": 1.6682510253821653e-06,
      "loss": 0.5566,
      "step": 2269
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7719739126137437,
      "learning_rate": 1.6679439747400658e-06,
      "loss": 0.639,
      "step": 2270
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8923696052523371,
      "learning_rate": 1.667636810354066e-06,
      "loss": 0.6324,
      "step": 2271
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7345210794312194,
      "learning_rate": 1.6673295322764727e-06,
      "loss": 0.6282,
      "step": 2272
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7436414393996811,
      "learning_rate": 1.6670221405596122e-06,
      "loss": 0.5206,
      "step": 2273
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8816672301127467,
      "learning_rate": 1.6667146352558301e-06,
      "loss": 0.5875,
      "step": 2274
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7447957577804639,
      "learning_rate": 1.6664070164174912e-06,
      "loss": 0.5688,
      "step": 2275
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8616723679693312,
      "learning_rate": 1.6660992840969806e-06,
      "loss": 0.5899,
      "step": 2276
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7858703729609936,
      "learning_rate": 1.6657914383467006e-06,
      "loss": 0.6384,
      "step": 2277
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8024283384801922,
      "learning_rate": 1.6654834792190752e-06,
      "loss": 0.6317,
      "step": 2278
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8712190824907251,
      "learning_rate": 1.6651754067665463e-06,
      "loss": 0.689,
      "step": 2279
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8997270818758489,
      "learning_rate": 1.6648672210415753e-06,
      "loss": 0.6101,
      "step": 2280
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8029948043294698,
      "learning_rate": 1.664558922096643e-06,
      "loss": 0.6127,
      "step": 2281
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0726666760859815,
      "learning_rate": 1.6642505099842494e-06,
      "loss": 0.6659,
      "step": 2282
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7398779796528825,
      "learning_rate": 1.6639419847569145e-06,
      "loss": 0.5747,
      "step": 2283
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8316765691468839,
      "learning_rate": 1.6636333464671762e-06,
      "loss": 0.6217,
      "step": 2284
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8852604912307752,
      "learning_rate": 1.6633245951675929e-06,
      "loss": 0.6752,
      "step": 2285
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.803011091492394,
      "learning_rate": 1.6630157309107414e-06,
      "loss": 0.6076,
      "step": 2286
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6817290443298669,
      "learning_rate": 1.662706753749218e-06,
      "loss": 0.5343,
      "step": 2287
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.917512307434402,
      "learning_rate": 1.6623976637356386e-06,
      "loss": 0.5907,
      "step": 2288
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7541601285437791,
      "learning_rate": 1.662088460922638e-06,
      "loss": 0.5925,
      "step": 2289
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8691287099435786,
      "learning_rate": 1.66177914536287e-06,
      "loss": 0.5871,
      "step": 2290
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6689808027758937,
      "learning_rate": 1.6614697171090079e-06,
      "loss": 0.5408,
      "step": 2291
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7006442914430474,
      "learning_rate": 1.6611601762137441e-06,
      "loss": 0.6893,
      "step": 2292
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9187981664622983,
      "learning_rate": 1.66085052272979e-06,
      "loss": 0.6508,
      "step": 2293
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8936349090826531,
      "learning_rate": 1.660540756709877e-06,
      "loss": 0.6993,
      "step": 2294
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.809402071303986,
      "learning_rate": 1.6602308782067542e-06,
      "loss": 0.6045,
      "step": 2295
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0191718764605853,
      "learning_rate": 1.6599208872731913e-06,
      "loss": 0.5613,
      "step": 2296
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6462000918388331,
      "learning_rate": 1.6596107839619764e-06,
      "loss": 0.5177,
      "step": 2297
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7818637801795143,
      "learning_rate": 1.6593005683259166e-06,
      "loss": 0.5711,
      "step": 2298
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7832499060108764,
      "learning_rate": 1.6589902404178387e-06,
      "loss": 0.6449,
      "step": 2299
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7167426861659866,
      "learning_rate": 1.6586798002905879e-06,
      "loss": 0.4956,
      "step": 2300
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7981649902841359,
      "learning_rate": 1.6583692479970296e-06,
      "loss": 0.5745,
      "step": 2301
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0394544041561944,
      "learning_rate": 1.6580585835900468e-06,
      "loss": 0.6887,
      "step": 2302
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7675248187917575,
      "learning_rate": 1.6577478071225431e-06,
      "loss": 0.6302,
      "step": 2303
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6444274444867941,
      "learning_rate": 1.6574369186474399e-06,
      "loss": 0.5546,
      "step": 2304
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6841058641790826,
      "learning_rate": 1.657125918217679e-06,
      "loss": 0.5303,
      "step": 2305
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8565966713599275,
      "learning_rate": 1.6568148058862196e-06,
      "loss": 0.5888,
      "step": 2306
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6317157520018506,
      "learning_rate": 1.6565035817060417e-06,
      "loss": 0.4773,
      "step": 2307
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.750669986215123,
      "learning_rate": 1.6561922457301433e-06,
      "loss": 0.5648,
      "step": 2308
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6503446444656306,
      "learning_rate": 1.6558807980115414e-06,
      "loss": 0.5903,
      "step": 2309
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7782256286129265,
      "learning_rate": 1.6555692386032728e-06,
      "loss": 0.6047,
      "step": 2310
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8272628479115911,
      "learning_rate": 1.6552575675583925e-06,
      "loss": 0.5708,
      "step": 2311
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7166399877044437,
      "learning_rate": 1.654945784929975e-06,
      "loss": 0.521,
      "step": 2312
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6774154561178422,
      "learning_rate": 1.654633890771113e-06,
      "loss": 0.5682,
      "step": 2313
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6533579453885231,
      "learning_rate": 1.6543218851349197e-06,
      "loss": 0.5088,
      "step": 2314
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6312431647770271,
      "learning_rate": 1.654009768074526e-06,
      "loss": 0.5896,
      "step": 2315
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7387827647138251,
      "learning_rate": 1.6536975396430818e-06,
      "loss": 0.5641,
      "step": 2316
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7239294104014358,
      "learning_rate": 1.6533851998937573e-06,
      "loss": 0.5516,
      "step": 2317
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6570674184915435,
      "learning_rate": 1.65307274887974e-06,
      "loss": 0.5285,
      "step": 2318
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6624897315764817,
      "learning_rate": 1.652760186654237e-06,
      "loss": 0.5046,
      "step": 2319
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6657928223274541,
      "learning_rate": 1.6524475132704751e-06,
      "loss": 0.4946,
      "step": 2320
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8645205927978623,
      "learning_rate": 1.6521347287816988e-06,
      "loss": 0.6241,
      "step": 2321
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.589597953878077,
      "learning_rate": 1.6518218332411718e-06,
      "loss": 0.5053,
      "step": 2322
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7594268110390386,
      "learning_rate": 1.651508826702177e-06,
      "loss": 0.5863,
      "step": 2323
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2129293039840563,
      "learning_rate": 1.651195709218017e-06,
      "loss": 0.6448,
      "step": 2324
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7119140943137071,
      "learning_rate": 1.6508824808420112e-06,
      "loss": 0.4943,
      "step": 2325
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6363155883333226,
      "learning_rate": 1.6505691416275003e-06,
      "loss": 0.5251,
      "step": 2326
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1308335713379853,
      "learning_rate": 1.6502556916278418e-06,
      "loss": 0.6279,
      "step": 2327
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8068985149723921,
      "learning_rate": 1.6499421308964135e-06,
      "loss": 0.6217,
      "step": 2328
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8901190320921132,
      "learning_rate": 1.6496284594866112e-06,
      "loss": 0.6038,
      "step": 2329
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7626283782749874,
      "learning_rate": 1.6493146774518506e-06,
      "loss": 0.5759,
      "step": 2330
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7799676810791333,
      "learning_rate": 1.649000784845564e-06,
      "loss": 0.5832,
      "step": 2331
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6360549269754372,
      "learning_rate": 1.6486867817212056e-06,
      "loss": 0.5435,
      "step": 2332
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7779230395910909,
      "learning_rate": 1.6483726681322461e-06,
      "loss": 0.6128,
      "step": 2333
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6112762446758263,
      "learning_rate": 1.648058444132176e-06,
      "loss": 0.5072,
      "step": 2334
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9881742657538033,
      "learning_rate": 1.6477441097745042e-06,
      "loss": 0.6259,
      "step": 2335
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7780897752507852,
      "learning_rate": 1.6474296651127588e-06,
      "loss": 0.6123,
      "step": 2336
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6749877450734497,
      "learning_rate": 1.647115110200486e-06,
      "loss": 0.5516,
      "step": 2337
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6851667217340366,
      "learning_rate": 1.6468004450912516e-06,
      "loss": 0.5948,
      "step": 2338
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7570637587155784,
      "learning_rate": 1.64648566983864e-06,
      "loss": 0.506,
      "step": 2339
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7578852869872799,
      "learning_rate": 1.6461707844962539e-06,
      "loss": 0.6284,
      "step": 2340
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9736562576339315,
      "learning_rate": 1.6458557891177145e-06,
      "loss": 0.672,
      "step": 2341
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7911551009825062,
      "learning_rate": 1.645540683756663e-06,
      "loss": 0.565,
      "step": 2342
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7337107681992371,
      "learning_rate": 1.6452254684667581e-06,
      "loss": 0.5699,
      "step": 2343
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0215494421752733,
      "learning_rate": 1.644910143301678e-06,
      "loss": 0.696,
      "step": 2344
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0468356409979203,
      "learning_rate": 1.644594708315119e-06,
      "loss": 0.6238,
      "step": 2345
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7596438252754523,
      "learning_rate": 1.6442791635607964e-06,
      "loss": 0.5861,
      "step": 2346
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7256526369079759,
      "learning_rate": 1.6439635090924445e-06,
      "loss": 0.5942,
      "step": 2347
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0619954182828173,
      "learning_rate": 1.6436477449638155e-06,
      "loss": 0.6431,
      "step": 2348
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7237890976675088,
      "learning_rate": 1.6433318712286814e-06,
      "loss": 0.6125,
      "step": 2349
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9616126805869909,
      "learning_rate": 1.6430158879408316e-06,
      "loss": 0.6363,
      "step": 2350
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8294412405010342,
      "learning_rate": 1.642699795154075e-06,
      "loss": 0.5791,
      "step": 2351
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1519112918432959,
      "learning_rate": 1.642383592922239e-06,
      "loss": 0.7158,
      "step": 2352
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7875231071807859,
      "learning_rate": 1.6420672812991692e-06,
      "loss": 0.5756,
      "step": 2353
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6571544104245998,
      "learning_rate": 1.6417508603387305e-06,
      "loss": 0.5562,
      "step": 2354
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8412274194169784,
      "learning_rate": 1.641434330094806e-06,
      "loss": 0.6497,
      "step": 2355
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8363808880091436,
      "learning_rate": 1.6411176906212977e-06,
      "loss": 0.5551,
      "step": 2356
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8563361692298546,
      "learning_rate": 1.6408009419721257e-06,
      "loss": 0.637,
      "step": 2357
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.057885992611512,
      "learning_rate": 1.6404840842012289e-06,
      "loss": 0.6173,
      "step": 2358
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6113859591085109,
      "learning_rate": 1.6401671173625653e-06,
      "loss": 0.5186,
      "step": 2359
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8486987977833653,
      "learning_rate": 1.6398500415101105e-06,
      "loss": 0.6464,
      "step": 2360
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9936103341651286,
      "learning_rate": 1.6395328566978598e-06,
      "loss": 0.6316,
      "step": 2361
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.758076046012566,
      "learning_rate": 1.6392155629798264e-06,
      "loss": 0.6283,
      "step": 2362
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9906072994713594,
      "learning_rate": 1.6388981604100414e-06,
      "loss": 0.6951,
      "step": 2363
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7185617787091048,
      "learning_rate": 1.6385806490425563e-06,
      "loss": 0.5307,
      "step": 2364
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6620030528498446,
      "learning_rate": 1.6382630289314387e-06,
      "loss": 0.5502,
      "step": 2365
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8647116780674396,
      "learning_rate": 1.637945300130777e-06,
      "loss": 0.6361,
      "step": 2366
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9200256011638845,
      "learning_rate": 1.6376274626946768e-06,
      "loss": 0.6973,
      "step": 2367
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9230542867983506,
      "learning_rate": 1.6373095166772623e-06,
      "loss": 0.642,
      "step": 2368
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.824400848646835,
      "learning_rate": 1.6369914621326766e-06,
      "loss": 0.6024,
      "step": 2369
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8806979314894896,
      "learning_rate": 1.636673299115081e-06,
      "loss": 0.6113,
      "step": 2370
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.706173534729791,
      "learning_rate": 1.6363550276786556e-06,
      "loss": 0.6201,
      "step": 2371
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8001607992274602,
      "learning_rate": 1.6360366478775985e-06,
      "loss": 0.6365,
      "step": 2372
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8543371330877693,
      "learning_rate": 1.6357181597661262e-06,
      "loss": 0.5995,
      "step": 2373
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6659332907908532,
      "learning_rate": 1.635399563398474e-06,
      "loss": 0.5875,
      "step": 2374
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8040486016647955,
      "learning_rate": 1.6350808588288963e-06,
      "loss": 0.6191,
      "step": 2375
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7173793405162839,
      "learning_rate": 1.6347620461116641e-06,
      "loss": 0.5852,
      "step": 2376
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6196421944512187,
      "learning_rate": 1.6344431253010681e-06,
      "loss": 0.5434,
      "step": 2377
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7323373030450488,
      "learning_rate": 1.634124096451418e-06,
      "loss": 0.5758,
      "step": 2378
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6751335987404704,
      "learning_rate": 1.6338049596170404e-06,
      "loss": 0.5678,
      "step": 2379
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7790034151519865,
      "learning_rate": 1.633485714852281e-06,
      "loss": 0.6175,
      "step": 2380
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9696325653204507,
      "learning_rate": 1.633166362211504e-06,
      "loss": 0.6251,
      "step": 2381
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6632506370100224,
      "learning_rate": 1.6328469017490916e-06,
      "loss": 0.5802,
      "step": 2382
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5983570118186374,
      "learning_rate": 1.6325273335194453e-06,
      "loss": 0.4923,
      "step": 2383
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8677759100084825,
      "learning_rate": 1.6322076575769831e-06,
      "loss": 0.6506,
      "step": 2384
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7335365848007822,
      "learning_rate": 1.6318878739761438e-06,
      "loss": 0.5485,
      "step": 2385
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9250802287361223,
      "learning_rate": 1.6315679827713818e-06,
      "loss": 0.638,
      "step": 2386
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8205803849399236,
      "learning_rate": 1.6312479840171723e-06,
      "loss": 0.5749,
      "step": 2387
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7085837658599741,
      "learning_rate": 1.6309278777680072e-06,
      "loss": 0.5229,
      "step": 2388
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8861137858527499,
      "learning_rate": 1.630607664078398e-06,
      "loss": 0.6483,
      "step": 2389
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7156129198086779,
      "learning_rate": 1.6302873430028727e-06,
      "loss": 0.5732,
      "step": 2390
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8344517049046373,
      "learning_rate": 1.6299669145959793e-06,
      "loss": 0.5696,
      "step": 2391
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7321636055022906,
      "learning_rate": 1.6296463789122836e-06,
      "loss": 0.5457,
      "step": 2392
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6911067974935775,
      "learning_rate": 1.6293257360063688e-06,
      "loss": 0.5685,
      "step": 2393
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6658619910325687,
      "learning_rate": 1.6290049859328376e-06,
      "loss": 0.5165,
      "step": 2394
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6493384821902125,
      "learning_rate": 1.6286841287463101e-06,
      "loss": 0.6049,
      "step": 2395
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7008670596026451,
      "learning_rate": 1.628363164501425e-06,
      "loss": 0.5421,
      "step": 2396
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8431782441593939,
      "learning_rate": 1.6280420932528396e-06,
      "loss": 0.6518,
      "step": 2397
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9823599729694824,
      "learning_rate": 1.6277209150552285e-06,
      "loss": 0.6794,
      "step": 2398
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6943816521833125,
      "learning_rate": 1.6273996299632852e-06,
      "loss": 0.5739,
      "step": 2399
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6467356624099669,
      "learning_rate": 1.6270782380317204e-06,
      "loss": 0.5248,
      "step": 2400
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.938370308704877,
      "learning_rate": 1.6267567393152652e-06,
      "loss": 0.7083,
      "step": 2401
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9920508713784483,
      "learning_rate": 1.6264351338686667e-06,
      "loss": 0.6965,
      "step": 2402
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6590284702696324,
      "learning_rate": 1.626113421746691e-06,
      "loss": 0.5191,
      "step": 2403
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7747546202901052,
      "learning_rate": 1.6257916030041226e-06,
      "loss": 0.5541,
      "step": 2404
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6075229475271532,
      "learning_rate": 1.6254696776957632e-06,
      "loss": 0.4826,
      "step": 2405
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5841065106206605,
      "learning_rate": 1.6251476458764341e-06,
      "loss": 0.5255,
      "step": 2406
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6223450101011191,
      "learning_rate": 1.6248255076009737e-06,
      "loss": 0.5612,
      "step": 2407
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.765984493381814,
      "learning_rate": 1.6245032629242386e-06,
      "loss": 0.6309,
      "step": 2408
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6204132870654709,
      "learning_rate": 1.6241809119011039e-06,
      "loss": 0.5515,
      "step": 2409
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6264563112729826,
      "learning_rate": 1.6238584545864624e-06,
      "loss": 0.5584,
      "step": 2410
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7593805124146197,
      "learning_rate": 1.6235358910352254e-06,
      "loss": 0.5881,
      "step": 2411
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6655583567106697,
      "learning_rate": 1.6232132213023223e-06,
      "loss": 0.56,
      "step": 2412
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8740795942896238,
      "learning_rate": 1.6228904454427004e-06,
      "loss": 0.5877,
      "step": 2413
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7012817919546442,
      "learning_rate": 1.6225675635113243e-06,
      "loss": 0.5318,
      "step": 2414
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9690617995791702,
      "learning_rate": 1.6222445755631784e-06,
      "loss": 0.6033,
      "step": 2415
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6427159761259156,
      "learning_rate": 1.6219214816532638e-06,
      "loss": 0.5024,
      "step": 2416
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7242575030527468,
      "learning_rate": 1.6215982818366e-06,
      "loss": 0.536,
      "step": 2417
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.764664173744657,
      "learning_rate": 1.6212749761682242e-06,
      "loss": 0.5973,
      "step": 2418
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.951898856320164,
      "learning_rate": 1.6209515647031926e-06,
      "loss": 0.6705,
      "step": 2419
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8360996482643221,
      "learning_rate": 1.6206280474965786e-06,
      "loss": 0.5914,
      "step": 2420
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7676795201261866,
      "learning_rate": 1.6203044246034738e-06,
      "loss": 0.5706,
      "step": 2421
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7999125974543403,
      "learning_rate": 1.6199806960789877e-06,
      "loss": 0.6025,
      "step": 2422
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6383580183100955,
      "learning_rate": 1.6196568619782477e-06,
      "loss": 0.55,
      "step": 2423
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8731628886419787,
      "learning_rate": 1.6193329223564e-06,
      "loss": 0.6439,
      "step": 2424
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6323919993838435,
      "learning_rate": 1.619008877268608e-06,
      "loss": 0.5126,
      "step": 2425
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8206851311123455,
      "learning_rate": 1.6186847267700526e-06,
      "loss": 0.6454,
      "step": 2426
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.762313230538518,
      "learning_rate": 1.6183604709159338e-06,
      "loss": 0.5798,
      "step": 2427
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8283840566294349,
      "learning_rate": 1.6180361097614687e-06,
      "loss": 0.6523,
      "step": 2428
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8172610893630339,
      "learning_rate": 1.617711643361893e-06,
      "loss": 0.6468,
      "step": 2429
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8843385755315869,
      "learning_rate": 1.6173870717724593e-06,
      "loss": 0.7126,
      "step": 2430
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6610234427777778,
      "learning_rate": 1.6170623950484395e-06,
      "loss": 0.574,
      "step": 2431
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1011538083067407,
      "learning_rate": 1.6167376132451223e-06,
      "loss": 0.6353,
      "step": 2432
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8249121693532128,
      "learning_rate": 1.6164127264178148e-06,
      "loss": 0.6096,
      "step": 2433
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7202904134932954,
      "learning_rate": 1.6160877346218412e-06,
      "loss": 0.5653,
      "step": 2434
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7015551908514589,
      "learning_rate": 1.6157626379125452e-06,
      "loss": 0.543,
      "step": 2435
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8594224000593357,
      "learning_rate": 1.615437436345287e-06,
      "loss": 0.597,
      "step": 2436
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6934066909545274,
      "learning_rate": 1.615112129975445e-06,
      "loss": 0.5328,
      "step": 2437
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9620232933009184,
      "learning_rate": 1.614786718858415e-06,
      "loss": 0.6071,
      "step": 2438
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8831929393948162,
      "learning_rate": 1.614461203049612e-06,
      "loss": 0.6358,
      "step": 2439
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4336418674581353,
      "learning_rate": 1.6141355826044673e-06,
      "loss": 0.6582,
      "step": 2440
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6975080236078062,
      "learning_rate": 1.6138098575784312e-06,
      "loss": 0.5566,
      "step": 2441
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6507088099613131,
      "learning_rate": 1.6134840280269706e-06,
      "loss": 0.5283,
      "step": 2442
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9363721258016622,
      "learning_rate": 1.6131580940055712e-06,
      "loss": 0.6425,
      "step": 2443
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6122468905708789,
      "learning_rate": 1.6128320555697363e-06,
      "loss": 0.5576,
      "step": 2444
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9294130186039169,
      "learning_rate": 1.6125059127749872e-06,
      "loss": 0.672,
      "step": 2445
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7875480254101632,
      "learning_rate": 1.6121796656768617e-06,
      "loss": 0.6259,
      "step": 2446
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8145221918196556,
      "learning_rate": 1.6118533143309172e-06,
      "loss": 0.5752,
      "step": 2447
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6496847006190792,
      "learning_rate": 1.6115268587927273e-06,
      "loss": 0.5428,
      "step": 2448
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6883355407075407,
      "learning_rate": 1.6112002991178844e-06,
      "loss": 0.5449,
      "step": 2449
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7052931724647741,
      "learning_rate": 1.6108736353619976e-06,
      "loss": 0.5357,
      "step": 2450
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5941846904261885,
      "learning_rate": 1.610546867580695e-06,
      "loss": 0.532,
      "step": 2451
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8982521189621095,
      "learning_rate": 1.6102199958296212e-06,
      "loss": 0.6702,
      "step": 2452
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0101531571169227,
      "learning_rate": 1.6098930201644397e-06,
      "loss": 0.6758,
      "step": 2453
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6646981900247438,
      "learning_rate": 1.6095659406408303e-06,
      "loss": 0.5389,
      "step": 2454
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7416050659828799,
      "learning_rate": 1.6092387573144922e-06,
      "loss": 0.6317,
      "step": 2455
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7900815299170257,
      "learning_rate": 1.6089114702411402e-06,
      "loss": 0.5627,
      "step": 2456
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8010280613785924,
      "learning_rate": 1.6085840794765086e-06,
      "loss": 0.5433,
      "step": 2457
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8742315414879183,
      "learning_rate": 1.6082565850763485e-06,
      "loss": 0.6507,
      "step": 2458
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.67646055167846,
      "learning_rate": 1.6079289870964287e-06,
      "loss": 0.5436,
      "step": 2459
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7843736049200927,
      "learning_rate": 1.6076012855925355e-06,
      "loss": 0.5358,
      "step": 2460
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7003149444836645,
      "learning_rate": 1.6072734806204736e-06,
      "loss": 0.5866,
      "step": 2461
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7792713533690936,
      "learning_rate": 1.6069455722360646e-06,
      "loss": 0.6319,
      "step": 2462
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6356591196424144,
      "learning_rate": 1.6066175604951475e-06,
      "loss": 0.4954,
      "step": 2463
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6809362063775646,
      "learning_rate": 1.6062894454535795e-06,
      "loss": 0.5494,
      "step": 2464
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7203944359582988,
      "learning_rate": 1.6059612271672357e-06,
      "loss": 0.5601,
      "step": 2465
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7245022804153479,
      "learning_rate": 1.6056329056920072e-06,
      "loss": 0.5512,
      "step": 2466
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8686957250749607,
      "learning_rate": 1.6053044810838046e-06,
      "loss": 0.6285,
      "step": 2467
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8110071934068561,
      "learning_rate": 1.6049759533985548e-06,
      "loss": 0.572,
      "step": 2468
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9144998841884497,
      "learning_rate": 1.6046473226922027e-06,
      "loss": 0.6431,
      "step": 2469
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6434320653532378,
      "learning_rate": 1.6043185890207108e-06,
      "loss": 0.5543,
      "step": 2470
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6661684012779162,
      "learning_rate": 1.603989752440059e-06,
      "loss": 0.6089,
      "step": 2471
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7193466316599335,
      "learning_rate": 1.6036608130062446e-06,
      "loss": 0.5608,
      "step": 2472
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7573869097575038,
      "learning_rate": 1.6033317707752825e-06,
      "loss": 0.6222,
      "step": 2473
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6001335926397549,
      "learning_rate": 1.6030026258032052e-06,
      "loss": 0.5133,
      "step": 2474
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0595959896185636,
      "learning_rate": 1.6026733781460631e-06,
      "loss": 0.632,
      "step": 2475
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9067400249753934,
      "learning_rate": 1.6023440278599231e-06,
      "loss": 0.7087,
      "step": 2476
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8598448498347431,
      "learning_rate": 1.6020145750008704e-06,
      "loss": 0.6318,
      "step": 2477
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8204190221856763,
      "learning_rate": 1.6016850196250076e-06,
      "loss": 0.5861,
      "step": 2478
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6125485697791082,
      "learning_rate": 1.6013553617884538e-06,
      "loss": 0.4971,
      "step": 2479
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7783806425118769,
      "learning_rate": 1.601025601547347e-06,
      "loss": 0.6048,
      "step": 2480
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.980387288644324,
      "learning_rate": 1.6006957389578418e-06,
      "loss": 0.6171,
      "step": 2481
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9886067736108141,
      "learning_rate": 1.6003657740761098e-06,
      "loss": 0.7051,
      "step": 2482
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8058527532276617,
      "learning_rate": 1.6000357069583416e-06,
      "loss": 0.5703,
      "step": 2483
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.86366129071733,
      "learning_rate": 1.599705537660743e-06,
      "loss": 0.6252,
      "step": 2484
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6802258492773239,
      "learning_rate": 1.5993752662395396e-06,
      "loss": 0.5407,
      "step": 2485
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7175130793056197,
      "learning_rate": 1.5990448927509725e-06,
      "loss": 0.5794,
      "step": 2486
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8156238219024209,
      "learning_rate": 1.5987144172513008e-06,
      "loss": 0.5665,
      "step": 2487
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0249659995464693,
      "learning_rate": 1.5983838397968015e-06,
      "loss": 0.6434,
      "step": 2488
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6435962513942913,
      "learning_rate": 1.5980531604437679e-06,
      "loss": 0.5541,
      "step": 2489
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7087593495119783,
      "learning_rate": 1.5977223792485118e-06,
      "loss": 0.5924,
      "step": 2490
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9495528512309235,
      "learning_rate": 1.597391496267361e-06,
      "loss": 0.538,
      "step": 2491
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8381530399046914,
      "learning_rate": 1.5970605115566626e-06,
      "loss": 0.5916,
      "step": 2492
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8604637729409753,
      "learning_rate": 1.596729425172779e-06,
      "loss": 0.6059,
      "step": 2493
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7723417897315928,
      "learning_rate": 1.5963982371720908e-06,
      "loss": 0.6123,
      "step": 2494
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.670312886232341,
      "learning_rate": 1.5960669476109961e-06,
      "loss": 0.5862,
      "step": 2495
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7716198652033283,
      "learning_rate": 1.59573555654591e-06,
      "loss": 0.5354,
      "step": 2496
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8622364136133446,
      "learning_rate": 1.595404064033265e-06,
      "loss": 0.6311,
      "step": 2497
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8987069583454859,
      "learning_rate": 1.5950724701295108e-06,
      "loss": 0.6282,
      "step": 2498
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6151092477882707,
      "learning_rate": 1.5947407748911143e-06,
      "loss": 0.5578,
      "step": 2499
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.664115475728601,
      "learning_rate": 1.59440897837456e-06,
      "loss": 0.4931,
      "step": 2500
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7232250085254985,
      "learning_rate": 1.594077080636349e-06,
      "loss": 0.5689,
      "step": 2501
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9236167824439739,
      "learning_rate": 1.5937450817330001e-06,
      "loss": 0.5629,
      "step": 2502
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.699353727957285,
      "learning_rate": 1.5934129817210495e-06,
      "loss": 0.5517,
      "step": 2503
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8514409644870283,
      "learning_rate": 1.5930807806570504e-06,
      "loss": 0.576,
      "step": 2504
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6053863337802154,
      "learning_rate": 1.592748478597573e-06,
      "loss": 0.5335,
      "step": 2505
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8120303052027267,
      "learning_rate": 1.5924160755992046e-06,
      "loss": 0.5797,
      "step": 2506
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8915178960787578,
      "learning_rate": 1.5920835717185505e-06,
      "loss": 0.652,
      "step": 2507
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9805617462708859,
      "learning_rate": 1.5917509670122325e-06,
      "loss": 0.6406,
      "step": 2508
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8741302904383479,
      "learning_rate": 1.5914182615368895e-06,
      "loss": 0.6164,
      "step": 2509
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7965169578604819,
      "learning_rate": 1.5910854553491779e-06,
      "loss": 0.639,
      "step": 2510
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6230515659073382,
      "learning_rate": 1.5907525485057711e-06,
      "loss": 0.5203,
      "step": 2511
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6826528847712071,
      "learning_rate": 1.59041954106336e-06,
      "loss": 0.6116,
      "step": 2512
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7552508212742363,
      "learning_rate": 1.5900864330786517e-06,
      "loss": 0.5262,
      "step": 2513
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6961164460872856,
      "learning_rate": 1.5897532246083714e-06,
      "loss": 0.497,
      "step": 2514
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9377466296813155,
      "learning_rate": 1.5894199157092608e-06,
      "loss": 0.6382,
      "step": 2515
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.848146698078804,
      "learning_rate": 1.5890865064380794e-06,
      "loss": 0.6622,
      "step": 2516
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8903305119855955,
      "learning_rate": 1.588752996851603e-06,
      "loss": 0.6549,
      "step": 2517
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9407229285183042,
      "learning_rate": 1.5884193870066245e-06,
      "loss": 0.6432,
      "step": 2518
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7741778203903938,
      "learning_rate": 1.5880856769599547e-06,
      "loss": 0.6066,
      "step": 2519
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7359609943961934,
      "learning_rate": 1.5877518667684206e-06,
      "loss": 0.6127,
      "step": 2520
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6953691945317098,
      "learning_rate": 1.587417956488867e-06,
      "loss": 0.5536,
      "step": 2521
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8787307548852056,
      "learning_rate": 1.5870839461781552e-06,
      "loss": 0.6405,
      "step": 2522
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7184219862136817,
      "learning_rate": 1.5867498358931632e-06,
      "loss": 0.6011,
      "step": 2523
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.996028043678537,
      "learning_rate": 1.5864156256907867e-06,
      "loss": 0.6764,
      "step": 2524
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.829706731799562,
      "learning_rate": 1.5860813156279388e-06,
      "loss": 0.6686,
      "step": 2525
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8589054483102626,
      "learning_rate": 1.5857469057615484e-06,
      "loss": 0.6433,
      "step": 2526
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8026838624368913,
      "learning_rate": 1.5854123961485623e-06,
      "loss": 0.5964,
      "step": 2527
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6534581656985162,
      "learning_rate": 1.585077786845944e-06,
      "loss": 0.5147,
      "step": 2528
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6788244109516006,
      "learning_rate": 1.5847430779106738e-06,
      "loss": 0.5607,
      "step": 2529
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.78794488184345,
      "learning_rate": 1.5844082693997492e-06,
      "loss": 0.545,
      "step": 2530
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5785985190393186,
      "learning_rate": 1.5840733613701846e-06,
      "loss": 0.517,
      "step": 2531
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.765495744139225,
      "learning_rate": 1.5837383538790117e-06,
      "loss": 0.5553,
      "step": 2532
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4622075492059616,
      "learning_rate": 1.5834032469832779e-06,
      "loss": 0.6404,
      "step": 2533
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7118768854225761,
      "learning_rate": 1.5830680407400493e-06,
      "loss": 0.5494,
      "step": 2534
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7239251765369579,
      "learning_rate": 1.582732735206408e-06,
      "loss": 0.5996,
      "step": 2535
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6336755093853782,
      "learning_rate": 1.5823973304394525e-06,
      "loss": 0.5032,
      "step": 2536
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8569447892908659,
      "learning_rate": 1.5820618264962991e-06,
      "loss": 0.6205,
      "step": 2537
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9217174191761583,
      "learning_rate": 1.5817262234340802e-06,
      "loss": 0.5595,
      "step": 2538
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6321285215224897,
      "learning_rate": 1.5813905213099461e-06,
      "loss": 0.5294,
      "step": 2539
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9349849489719363,
      "learning_rate": 1.581054720181063e-06,
      "loss": 0.5887,
      "step": 2540
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6290311246925899,
      "learning_rate": 1.5807188201046148e-06,
      "loss": 0.4475,
      "step": 2541
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6644933028940619,
      "learning_rate": 1.580382821137801e-06,
      "loss": 0.5329,
      "step": 2542
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7187526847658339,
      "learning_rate": 1.5800467233378392e-06,
      "loss": 0.5623,
      "step": 2543
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6989345234753084,
      "learning_rate": 1.5797105267619635e-06,
      "loss": 0.5808,
      "step": 2544
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6969569742660636,
      "learning_rate": 1.5793742314674244e-06,
      "loss": 0.5406,
      "step": 2545
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7094646639337747,
      "learning_rate": 1.5790378375114893e-06,
      "loss": 0.5616,
      "step": 2546
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6416630524813071,
      "learning_rate": 1.5787013449514431e-06,
      "loss": 0.526,
      "step": 2547
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6417409229303236,
      "learning_rate": 1.5783647538445867e-06,
      "loss": 0.4926,
      "step": 2548
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6827726362657659,
      "learning_rate": 1.578028064248238e-06,
      "loss": 0.5623,
      "step": 2549
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5857222717958704,
      "learning_rate": 1.5776912762197316e-06,
      "loss": 0.4899,
      "step": 2550
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8168226975915017,
      "learning_rate": 1.5773543898164197e-06,
      "loss": 0.5597,
      "step": 2551
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8172095171983793,
      "learning_rate": 1.5770174050956697e-06,
      "loss": 0.6461,
      "step": 2552
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6520127149793095,
      "learning_rate": 1.5766803221148673e-06,
      "loss": 0.555,
      "step": 2553
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9341753512799217,
      "learning_rate": 1.5763431409314134e-06,
      "loss": 0.6497,
      "step": 2554
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6074898402501409,
      "learning_rate": 1.576005861602727e-06,
      "loss": 0.4805,
      "step": 2555
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9411713466066226,
      "learning_rate": 1.5756684841862433e-06,
      "loss": 0.6332,
      "step": 2556
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6541898654687741,
      "learning_rate": 1.5753310087394138e-06,
      "loss": 0.5227,
      "step": 2557
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6946972794125692,
      "learning_rate": 1.5749934353197077e-06,
      "loss": 0.4887,
      "step": 2558
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7959812327613129,
      "learning_rate": 1.5746557639846094e-06,
      "loss": 0.5427,
      "step": 2559
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.804697109798102,
      "learning_rate": 1.5743179947916215e-06,
      "loss": 0.6305,
      "step": 2560
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9152540668240107,
      "learning_rate": 1.573980127798262e-06,
      "loss": 0.6052,
      "step": 2561
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7504039316691035,
      "learning_rate": 1.5736421630620664e-06,
      "loss": 0.621,
      "step": 2562
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7499569737288814,
      "learning_rate": 1.573304100640587e-06,
      "loss": 0.5588,
      "step": 2563
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9488640031365498,
      "learning_rate": 1.5729659405913918e-06,
      "loss": 0.6668,
      "step": 2564
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8631889542941432,
      "learning_rate": 1.5726276829720656e-06,
      "loss": 0.6285,
      "step": 2565
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.665238955751868,
      "learning_rate": 1.572289327840211e-06,
      "loss": 0.5662,
      "step": 2566
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8616516322435755,
      "learning_rate": 1.5719508752534457e-06,
      "loss": 0.6374,
      "step": 2567
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6970467853670573,
      "learning_rate": 1.5716123252694052e-06,
      "loss": 0.5167,
      "step": 2568
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7295224205098898,
      "learning_rate": 1.5712736779457405e-06,
      "loss": 0.5925,
      "step": 2569
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0316168982120097,
      "learning_rate": 1.5709349333401198e-06,
      "loss": 0.6559,
      "step": 2570
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8984282965748758,
      "learning_rate": 1.570596091510228e-06,
      "loss": 0.5505,
      "step": 2571
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8315430479178283,
      "learning_rate": 1.5702571525137662e-06,
      "loss": 0.5757,
      "step": 2572
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6238297765901686,
      "learning_rate": 1.5699181164084525e-06,
      "loss": 0.513,
      "step": 2573
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6806460929370697,
      "learning_rate": 1.5695789832520208e-06,
      "loss": 0.538,
      "step": 2574
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7542735790422822,
      "learning_rate": 1.5692397531022219e-06,
      "loss": 0.6493,
      "step": 2575
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6332481844776309,
      "learning_rate": 1.5689004260168235e-06,
      "loss": 0.4859,
      "step": 2576
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6591643176035831,
      "learning_rate": 1.5685610020536092e-06,
      "loss": 0.506,
      "step": 2577
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0120765311465691,
      "learning_rate": 1.5682214812703799e-06,
      "loss": 0.6555,
      "step": 2578
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0649137168994636,
      "learning_rate": 1.5678818637249516e-06,
      "loss": 0.6878,
      "step": 2579
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6035584493561864,
      "learning_rate": 1.5675421494751579e-06,
      "loss": 0.5196,
      "step": 2580
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9110625809264739,
      "learning_rate": 1.5672023385788494e-06,
      "loss": 0.5959,
      "step": 2581
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8972276439204921,
      "learning_rate": 1.5668624310938912e-06,
      "loss": 0.6396,
      "step": 2582
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7180481987408107,
      "learning_rate": 1.5665224270781664e-06,
      "loss": 0.5579,
      "step": 2583
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6053509147967748,
      "learning_rate": 1.5661823265895745e-06,
      "loss": 0.5417,
      "step": 2584
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9241062013651185,
      "learning_rate": 1.565842129686031e-06,
      "loss": 0.5897,
      "step": 2585
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5915035287666525,
      "learning_rate": 1.5655018364254676e-06,
      "loss": 0.5161,
      "step": 2586
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6803520701884275,
      "learning_rate": 1.5651614468658326e-06,
      "loss": 0.5278,
      "step": 2587
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.714474250923458,
      "learning_rate": 1.564820961065091e-06,
      "loss": 0.4996,
      "step": 2588
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7686287571519791,
      "learning_rate": 1.5644803790812242e-06,
      "loss": 0.4935,
      "step": 2589
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8511682260474583,
      "learning_rate": 1.5641397009722295e-06,
      "loss": 0.6221,
      "step": 2590
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5845829630963592,
      "learning_rate": 1.5637989267961208e-06,
      "loss": 0.484,
      "step": 2591
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7938446202665874,
      "learning_rate": 1.5634580566109287e-06,
      "loss": 0.5211,
      "step": 2592
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6402248829404993,
      "learning_rate": 1.5631170904746998e-06,
      "loss": 0.5522,
      "step": 2593
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8252877052754484,
      "learning_rate": 1.5627760284454965e-06,
      "loss": 0.5961,
      "step": 2594
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7571600186175518,
      "learning_rate": 1.5624348705813985e-06,
      "loss": 0.6209,
      "step": 2595
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8450428323230402,
      "learning_rate": 1.5620936169405017e-06,
      "loss": 0.5823,
      "step": 2596
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7534341431685809,
      "learning_rate": 1.5617522675809176e-06,
      "loss": 0.5713,
      "step": 2597
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8955971057312625,
      "learning_rate": 1.5614108225607744e-06,
      "loss": 0.6297,
      "step": 2598
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6934151626076508,
      "learning_rate": 1.5610692819382171e-06,
      "loss": 0.5837,
      "step": 2599
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8515024880142362,
      "learning_rate": 1.5607276457714064e-06,
      "loss": 0.6559,
      "step": 2600
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6892410379736257,
      "learning_rate": 1.5603859141185192e-06,
      "loss": 0.5376,
      "step": 2601
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6390983788952416,
      "learning_rate": 1.5600440870377487e-06,
      "loss": 0.5057,
      "step": 2602
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8613673266229193,
      "learning_rate": 1.5597021645873048e-06,
      "loss": 0.5624,
      "step": 2603
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8647907981066635,
      "learning_rate": 1.5593601468254132e-06,
      "loss": 0.6273,
      "step": 2604
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6267038146529416,
      "learning_rate": 1.5590180338103159e-06,
      "loss": 0.5009,
      "step": 2605
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6934131044956916,
      "learning_rate": 1.5586758256002712e-06,
      "loss": 0.5126,
      "step": 2606
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0414196184194449,
      "learning_rate": 1.5583335222535538e-06,
      "loss": 0.652,
      "step": 2607
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7398529670522938,
      "learning_rate": 1.5579911238284544e-06,
      "loss": 0.5883,
      "step": 2608
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.965627394189997,
      "learning_rate": 1.5576486303832794e-06,
      "loss": 0.6914,
      "step": 2609
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7185502354647558,
      "learning_rate": 1.5573060419763523e-06,
      "loss": 0.5233,
      "step": 2610
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6331843843806698,
      "learning_rate": 1.556963358666012e-06,
      "loss": 0.577,
      "step": 2611
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7284391326409974,
      "learning_rate": 1.5566205805106145e-06,
      "loss": 0.5263,
      "step": 2612
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6932440436696805,
      "learning_rate": 1.556277707568531e-06,
      "loss": 0.5964,
      "step": 2613
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5746316977070192,
      "learning_rate": 1.5559347398981488e-06,
      "loss": 0.5233,
      "step": 2614
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0539568152592735,
      "learning_rate": 1.5555916775578724e-06,
      "loss": 0.6742,
      "step": 2615
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.718179580752316,
      "learning_rate": 1.5552485206061213e-06,
      "loss": 0.5824,
      "step": 2616
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7661280116288639,
      "learning_rate": 1.5549052691013315e-06,
      "loss": 0.5629,
      "step": 2617
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0025493822600118,
      "learning_rate": 1.5545619231019556e-06,
      "loss": 0.6283,
      "step": 2618
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9264561063452965,
      "learning_rate": 1.5542184826664614e-06,
      "loss": 0.5819,
      "step": 2619
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6946940380093737,
      "learning_rate": 1.5538749478533335e-06,
      "loss": 0.5804,
      "step": 2620
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7984629008591758,
      "learning_rate": 1.553531318721072e-06,
      "loss": 0.5786,
      "step": 2621
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8243519179312014,
      "learning_rate": 1.5531875953281941e-06,
      "loss": 0.6575,
      "step": 2622
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6586957314476799,
      "learning_rate": 1.5528437777332312e-06,
      "loss": 0.5644,
      "step": 2623
    },
    {
      "epoch": 0.33,
      "grad_norm": 13.25863321152729,
      "learning_rate": 1.5524998659947328e-06,
      "loss": 0.5221,
      "step": 2624
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6014881760340378,
      "learning_rate": 1.552155860171263e-06,
      "loss": 0.5394,
      "step": 2625
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7419662625122913,
      "learning_rate": 1.5518117603214024e-06,
      "loss": 0.5465,
      "step": 2626
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6409063539651462,
      "learning_rate": 1.551467566503748e-06,
      "loss": 0.5719,
      "step": 2627
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7021138527211542,
      "learning_rate": 1.5511232787769122e-06,
      "loss": 0.5568,
      "step": 2628
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7025644408801217,
      "learning_rate": 1.5507788971995233e-06,
      "loss": 0.5563,
      "step": 2629
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7527568901142943,
      "learning_rate": 1.5504344218302264e-06,
      "loss": 0.5389,
      "step": 2630
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8043522246569156,
      "learning_rate": 1.5500898527276816e-06,
      "loss": 0.6632,
      "step": 2631
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8994663693422271,
      "learning_rate": 1.5497451899505659e-06,
      "loss": 0.6169,
      "step": 2632
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8580824636401024,
      "learning_rate": 1.5494004335575714e-06,
      "loss": 0.5901,
      "step": 2633
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6848044735678862,
      "learning_rate": 1.549055583607407e-06,
      "loss": 0.5377,
      "step": 2634
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5721144793992392,
      "learning_rate": 1.5487106401587964e-06,
      "loss": 0.5033,
      "step": 2635
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9792911523867096,
      "learning_rate": 1.5483656032704802e-06,
      "loss": 0.6435,
      "step": 2636
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7590737426095436,
      "learning_rate": 1.5480204730012147e-06,
      "loss": 0.5843,
      "step": 2637
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8757857293005646,
      "learning_rate": 1.5476752494097718e-06,
      "loss": 0.6589,
      "step": 2638
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8201512994882567,
      "learning_rate": 1.54732993255494e-06,
      "loss": 0.6211,
      "step": 2639
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8233198522442915,
      "learning_rate": 1.546984522495522e-06,
      "loss": 0.5813,
      "step": 2640
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7545402015931093,
      "learning_rate": 1.5466390192903389e-06,
      "loss": 0.5498,
      "step": 2641
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8911395834575572,
      "learning_rate": 1.5462934229982253e-06,
      "loss": 0.6497,
      "step": 2642
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8700750056893659,
      "learning_rate": 1.5459477336780333e-06,
      "loss": 0.6964,
      "step": 2643
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8467776322339813,
      "learning_rate": 1.5456019513886296e-06,
      "loss": 0.5749,
      "step": 2644
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6413629750641375,
      "learning_rate": 1.5452560761888976e-06,
      "loss": 0.5533,
      "step": 2645
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6376908601672958,
      "learning_rate": 1.5449101081377365e-06,
      "loss": 0.5282,
      "step": 2646
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.672180386303089,
      "learning_rate": 1.5445640472940603e-06,
      "loss": 0.5538,
      "step": 2647
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.721379258380877,
      "learning_rate": 1.5442178937168008e-06,
      "loss": 0.5749,
      "step": 2648
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7966303071124852,
      "learning_rate": 1.5438716474649033e-06,
      "loss": 0.5702,
      "step": 2649
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9026711106169523,
      "learning_rate": 1.54352530859733e-06,
      "loss": 0.6773,
      "step": 2650
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7636672857345796,
      "learning_rate": 1.5431788771730597e-06,
      "loss": 0.5948,
      "step": 2651
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7065413830477413,
      "learning_rate": 1.5428323532510847e-06,
      "loss": 0.5152,
      "step": 2652
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7222540732276216,
      "learning_rate": 1.5424857368904154e-06,
      "loss": 0.6065,
      "step": 2653
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7580740319422532,
      "learning_rate": 1.5421390281500768e-06,
      "loss": 0.5466,
      "step": 2654
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7121332525199152,
      "learning_rate": 1.5417922270891096e-06,
      "loss": 0.5579,
      "step": 2655
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7773608159610655,
      "learning_rate": 1.5414453337665705e-06,
      "loss": 0.6659,
      "step": 2656
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7004766212385289,
      "learning_rate": 1.5410983482415318e-06,
      "loss": 0.5615,
      "step": 2657
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7280071581017894,
      "learning_rate": 1.5407512705730815e-06,
      "loss": 0.5269,
      "step": 2658
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6751841557706953,
      "learning_rate": 1.5404041008203232e-06,
      "loss": 0.5954,
      "step": 2659
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7021591715651088,
      "learning_rate": 1.5400568390423765e-06,
      "loss": 0.5752,
      "step": 2660
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7591204428905055,
      "learning_rate": 1.5397094852983765e-06,
      "loss": 0.5546,
      "step": 2661
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7181087615662662,
      "learning_rate": 1.5393620396474738e-06,
      "loss": 0.6051,
      "step": 2662
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7828763797756219,
      "learning_rate": 1.5390145021488348e-06,
      "loss": 0.6174,
      "step": 2663
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6934627441020993,
      "learning_rate": 1.5386668728616412e-06,
      "loss": 0.5559,
      "step": 2664
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.5735505747640333,
      "learning_rate": 1.5383191518450912e-06,
      "loss": 0.6212,
      "step": 2665
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.401230211660289,
      "learning_rate": 1.5379713391583977e-06,
      "loss": 0.5869,
      "step": 2666
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6693739799842298,
      "learning_rate": 1.53762343486079e-06,
      "loss": 0.5573,
      "step": 2667
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7949351107821199,
      "learning_rate": 1.5372754390115116e-06,
      "loss": 0.6494,
      "step": 2668
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7574293124796494,
      "learning_rate": 1.5369273516698238e-06,
      "loss": 0.5419,
      "step": 2669
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7608196273776373,
      "learning_rate": 1.5365791728950016e-06,
      "loss": 0.5567,
      "step": 2670
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6949936688001659,
      "learning_rate": 1.536230902746336e-06,
      "loss": 0.5462,
      "step": 2671
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7792803931478585,
      "learning_rate": 1.5358825412831342e-06,
      "loss": 0.594,
      "step": 2672
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.319311447755787,
      "learning_rate": 1.5355340885647185e-06,
      "loss": 0.5978,
      "step": 2673
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6711776727009926,
      "learning_rate": 1.5351855446504267e-06,
      "loss": 0.5762,
      "step": 2674
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9042131609964323,
      "learning_rate": 1.5348369095996124e-06,
      "loss": 0.6052,
      "step": 2675
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8977802272894979,
      "learning_rate": 1.534488183471644e-06,
      "loss": 0.6486,
      "step": 2676
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8882656406496038,
      "learning_rate": 1.5341393663259062e-06,
      "loss": 0.6168,
      "step": 2677
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6326980024734311,
      "learning_rate": 1.533790458221799e-06,
      "loss": 0.4965,
      "step": 2678
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6801749988222119,
      "learning_rate": 1.5334414592187375e-06,
      "loss": 0.5782,
      "step": 2679
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9635715184246961,
      "learning_rate": 1.533092369376153e-06,
      "loss": 0.6547,
      "step": 2680
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6873721846576595,
      "learning_rate": 1.5327431887534918e-06,
      "loss": 0.5819,
      "step": 2681
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6532313658259804,
      "learning_rate": 1.5323939174102157e-06,
      "loss": 0.5122,
      "step": 2682
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8968013897358722,
      "learning_rate": 1.5320445554058017e-06,
      "loss": 0.716,
      "step": 2683
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7974510128954213,
      "learning_rate": 1.531695102799743e-06,
      "loss": 0.5615,
      "step": 2684
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8318503927719663,
      "learning_rate": 1.5313455596515471e-06,
      "loss": 0.647,
      "step": 2685
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6467112436761154,
      "learning_rate": 1.5309959260207382e-06,
      "loss": 0.5461,
      "step": 2686
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8362236594797428,
      "learning_rate": 1.5306462019668549e-06,
      "loss": 0.5586,
      "step": 2687
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7132130331418559,
      "learning_rate": 1.5302963875494515e-06,
      "loss": 0.5165,
      "step": 2688
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7382787231691627,
      "learning_rate": 1.5299464828280982e-06,
      "loss": 0.5823,
      "step": 2689
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7616743477518817,
      "learning_rate": 1.5295964878623795e-06,
      "loss": 0.565,
      "step": 2690
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8228494077737853,
      "learning_rate": 1.5292464027118963e-06,
      "loss": 0.6518,
      "step": 2691
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8944313482742176,
      "learning_rate": 1.5288962274362644e-06,
      "loss": 0.6315,
      "step": 2692
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6619007072477948,
      "learning_rate": 1.528545962095115e-06,
      "loss": 0.5152,
      "step": 2693
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8333396114998701,
      "learning_rate": 1.5281956067480946e-06,
      "loss": 0.5385,
      "step": 2694
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9393108186707092,
      "learning_rate": 1.527845161454865e-06,
      "loss": 0.6351,
      "step": 2695
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9081170153823361,
      "learning_rate": 1.5274946262751031e-06,
      "loss": 0.6442,
      "step": 2696
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7271146582802742,
      "learning_rate": 1.5271440012685025e-06,
      "loss": 0.6141,
      "step": 2697
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.871481643528823,
      "learning_rate": 1.5267932864947697e-06,
      "loss": 0.6061,
      "step": 2698
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.820711219643142,
      "learning_rate": 1.5264424820136284e-06,
      "loss": 0.604,
      "step": 2699
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6934625101454092,
      "learning_rate": 1.526091587884817e-06,
      "loss": 0.5738,
      "step": 2700
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9303136764778933,
      "learning_rate": 1.5257406041680888e-06,
      "loss": 0.6371,
      "step": 2701
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8238363027177719,
      "learning_rate": 1.525389530923213e-06,
      "loss": 0.5884,
      "step": 2702
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6846651392697702,
      "learning_rate": 1.5250383682099734e-06,
      "loss": 0.6141,
      "step": 2703
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8096055578771191,
      "learning_rate": 1.5246871160881696e-06,
      "loss": 0.5942,
      "step": 2704
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8198797547309176,
      "learning_rate": 1.524335774617616e-06,
      "loss": 0.6364,
      "step": 2705
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8683203268013842,
      "learning_rate": 1.5239843438581428e-06,
      "loss": 0.6124,
      "step": 2706
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7076189648847233,
      "learning_rate": 1.5236328238695944e-06,
      "loss": 0.5785,
      "step": 2707
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8542546187791273,
      "learning_rate": 1.5232812147118312e-06,
      "loss": 0.5986,
      "step": 2708
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7117857628580446,
      "learning_rate": 1.5229295164447287e-06,
      "loss": 0.541,
      "step": 2709
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7678354301752931,
      "learning_rate": 1.5225777291281773e-06,
      "loss": 0.6051,
      "step": 2710
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7191800430412006,
      "learning_rate": 1.5222258528220827e-06,
      "loss": 0.5587,
      "step": 2711
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.836783627728259,
      "learning_rate": 1.5218738875863662e-06,
      "loss": 0.6526,
      "step": 2712
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.933315406061919,
      "learning_rate": 1.5215218334809632e-06,
      "loss": 0.6604,
      "step": 2713
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7998197801255282,
      "learning_rate": 1.5211696905658253e-06,
      "loss": 0.6083,
      "step": 2714
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7234124947493389,
      "learning_rate": 1.5208174589009186e-06,
      "loss": 0.5658,
      "step": 2715
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9049726544674671,
      "learning_rate": 1.5204651385462245e-06,
      "loss": 0.6345,
      "step": 2716
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7478737383979305,
      "learning_rate": 1.5201127295617395e-06,
      "loss": 0.5742,
      "step": 2717
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6849346212885078,
      "learning_rate": 1.519760232007475e-06,
      "loss": 0.5422,
      "step": 2718
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7069993737189381,
      "learning_rate": 1.5194076459434582e-06,
      "loss": 0.5409,
      "step": 2719
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1248963155834466,
      "learning_rate": 1.5190549714297301e-06,
      "loss": 0.6115,
      "step": 2720
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6455352371672562,
      "learning_rate": 1.5187022085263479e-06,
      "loss": 0.5708,
      "step": 2721
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5891843057979788,
      "learning_rate": 1.5183493572933833e-06,
      "loss": 0.5254,
      "step": 2722
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8090551137943908,
      "learning_rate": 1.5179964177909237e-06,
      "loss": 0.5554,
      "step": 2723
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7338631039466889,
      "learning_rate": 1.5176433900790703e-06,
      "loss": 0.5495,
      "step": 2724
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8385053041376656,
      "learning_rate": 1.5172902742179398e-06,
      "loss": 0.5966,
      "step": 2725
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6479743866352093,
      "learning_rate": 1.5169370702676654e-06,
      "loss": 0.5331,
      "step": 2726
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6640071581864379,
      "learning_rate": 1.516583778288393e-06,
      "loss": 0.5401,
      "step": 2727
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7844153558121632,
      "learning_rate": 1.5162303983402852e-06,
      "loss": 0.5855,
      "step": 2728
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5987192666997421,
      "learning_rate": 1.5158769304835183e-06,
      "loss": 0.5359,
      "step": 2729
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7039900578000646,
      "learning_rate": 1.5155233747782848e-06,
      "loss": 0.5257,
      "step": 2730
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7553581939022301,
      "learning_rate": 1.515169731284791e-06,
      "loss": 0.6061,
      "step": 2731
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.728643723089328,
      "learning_rate": 1.5148160000632593e-06,
      "loss": 0.5546,
      "step": 2732
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6774852140911092,
      "learning_rate": 1.5144621811739257e-06,
      "loss": 0.5009,
      "step": 2733
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5982599252176639,
      "learning_rate": 1.5141082746770425e-06,
      "loss": 0.5066,
      "step": 2734
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7794687556181809,
      "learning_rate": 1.513754280632876e-06,
      "loss": 0.5856,
      "step": 2735
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.762073971344144,
      "learning_rate": 1.5134001991017079e-06,
      "loss": 0.6338,
      "step": 2736
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7895890497786219,
      "learning_rate": 1.5130460301438343e-06,
      "loss": 0.7224,
      "step": 2737
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.603641608550316,
      "learning_rate": 1.5126917738195663e-06,
      "loss": 0.5036,
      "step": 2738
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7622576154880916,
      "learning_rate": 1.512337430189231e-06,
      "loss": 0.5549,
      "step": 2739
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8449130382612617,
      "learning_rate": 1.5119829993131688e-06,
      "loss": 0.6073,
      "step": 2740
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6942946227993302,
      "learning_rate": 1.5116284812517352e-06,
      "loss": 0.5307,
      "step": 2741
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6841852898616607,
      "learning_rate": 1.5112738760653017e-06,
      "loss": 0.5277,
      "step": 2742
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6514421506194327,
      "learning_rate": 1.5109191838142534e-06,
      "loss": 0.5553,
      "step": 2743
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7595529152611123,
      "learning_rate": 1.5105644045589907e-06,
      "loss": 0.6087,
      "step": 2744
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6229450328247342,
      "learning_rate": 1.5102095383599291e-06,
      "loss": 0.5246,
      "step": 2745
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6892993834567736,
      "learning_rate": 1.5098545852774984e-06,
      "loss": 0.5769,
      "step": 2746
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8436699065958566,
      "learning_rate": 1.5094995453721433e-06,
      "loss": 0.6344,
      "step": 2747
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8568324571966234,
      "learning_rate": 1.5091444187043238e-06,
      "loss": 0.6211,
      "step": 2748
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7779622307365157,
      "learning_rate": 1.5087892053345138e-06,
      "loss": 0.554,
      "step": 2749
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7576944072476369,
      "learning_rate": 1.5084339053232027e-06,
      "loss": 0.5932,
      "step": 2750
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6756978614860609,
      "learning_rate": 1.5080785187308944e-06,
      "loss": 0.6188,
      "step": 2751
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5988848029112629,
      "learning_rate": 1.5077230456181075e-06,
      "loss": 0.5242,
      "step": 2752
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7492957318785779,
      "learning_rate": 1.5073674860453753e-06,
      "loss": 0.5597,
      "step": 2753
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7166376408933957,
      "learning_rate": 1.5070118400732458e-06,
      "loss": 0.5071,
      "step": 2754
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8986433350250779,
      "learning_rate": 1.506656107762282e-06,
      "loss": 0.5994,
      "step": 2755
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8321511006444927,
      "learning_rate": 1.506300289173062e-06,
      "loss": 0.6704,
      "step": 2756
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9131508275823171,
      "learning_rate": 1.5059443843661765e-06,
      "loss": 0.6587,
      "step": 2757
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.70888542543407,
      "learning_rate": 1.5055883934022337e-06,
      "loss": 0.5876,
      "step": 2758
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8627024208851585,
      "learning_rate": 1.5052323163418548e-06,
      "loss": 0.5649,
      "step": 2759
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8349774162336141,
      "learning_rate": 1.5048761532456755e-06,
      "loss": 0.607,
      "step": 2760
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7957402233973435,
      "learning_rate": 1.5045199041743473e-06,
      "loss": 0.6034,
      "step": 2761
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7084440018877063,
      "learning_rate": 1.5041635691885356e-06,
      "loss": 0.5183,
      "step": 2762
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9475343341965512,
      "learning_rate": 1.5038071483489198e-06,
      "loss": 0.6376,
      "step": 2763
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.866160550320788,
      "learning_rate": 1.5034506417161959e-06,
      "loss": 0.6165,
      "step": 2764
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7353045429084609,
      "learning_rate": 1.5030940493510725e-06,
      "loss": 0.5819,
      "step": 2765
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6121072871793188,
      "learning_rate": 1.5027373713142733e-06,
      "loss": 0.5233,
      "step": 2766
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6701214150061907,
      "learning_rate": 1.5023806076665375e-06,
      "loss": 0.5397,
      "step": 2767
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6611134483316184,
      "learning_rate": 1.502023758468618e-06,
      "loss": 0.6005,
      "step": 2768
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8047607500676393,
      "learning_rate": 1.5016668237812824e-06,
      "loss": 0.6107,
      "step": 2769
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8756824725920811,
      "learning_rate": 1.501309803665313e-06,
      "loss": 0.6251,
      "step": 2770
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6971907337980469,
      "learning_rate": 1.5009526981815071e-06,
      "loss": 0.5642,
      "step": 2771
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9330454267014099,
      "learning_rate": 1.500595507390675e-06,
      "loss": 0.6399,
      "step": 2772
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7790275898812877,
      "learning_rate": 1.5002382313536434e-06,
      "loss": 0.5169,
      "step": 2773
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8339705375493703,
      "learning_rate": 1.4998808701312527e-06,
      "loss": 0.6005,
      "step": 2774
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8264765201551842,
      "learning_rate": 1.4995234237843573e-06,
      "loss": 0.715,
      "step": 2775
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8523558755788513,
      "learning_rate": 1.4991658923738267e-06,
      "loss": 0.6297,
      "step": 2776
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7363873528091556,
      "learning_rate": 1.4988082759605449e-06,
      "loss": 0.5881,
      "step": 2777
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6480111023873047,
      "learning_rate": 1.4984505746054107e-06,
      "loss": 0.548,
      "step": 2778
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6320121842214815,
      "learning_rate": 1.498092788369336e-06,
      "loss": 0.5256,
      "step": 2779
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8398572977393959,
      "learning_rate": 1.4977349173132487e-06,
      "loss": 0.6366,
      "step": 2780
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8723385410414921,
      "learning_rate": 1.49737696149809e-06,
      "loss": 0.6517,
      "step": 2781
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6248675120915862,
      "learning_rate": 1.4970189209848167e-06,
      "loss": 0.5592,
      "step": 2782
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3276181095905166,
      "learning_rate": 1.496660795834399e-06,
      "loss": 0.6757,
      "step": 2783
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6417394041741807,
      "learning_rate": 1.4963025861078222e-06,
      "loss": 0.5412,
      "step": 2784
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8788874658431991,
      "learning_rate": 1.4959442918660852e-06,
      "loss": 0.665,
      "step": 2785
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7497994614758708,
      "learning_rate": 1.495585913170202e-06,
      "loss": 0.5713,
      "step": 2786
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6804242412535482,
      "learning_rate": 1.495227450081201e-06,
      "loss": 0.5469,
      "step": 2787
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.991025611563954,
      "learning_rate": 1.4948689026601243e-06,
      "loss": 0.6434,
      "step": 2788
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8070624838200096,
      "learning_rate": 1.4945102709680289e-06,
      "loss": 0.5259,
      "step": 2789
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6973771458339837,
      "learning_rate": 1.4941515550659861e-06,
      "loss": 0.5371,
      "step": 2790
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6130608976219806,
      "learning_rate": 1.4937927550150817e-06,
      "loss": 0.5215,
      "step": 2791
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.752072773044262,
      "learning_rate": 1.4934338708764155e-06,
      "loss": 0.5728,
      "step": 2792
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7596111844720105,
      "learning_rate": 1.4930749027111013e-06,
      "loss": 0.6334,
      "step": 2793
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7547208706911472,
      "learning_rate": 1.492715850580268e-06,
      "loss": 0.5898,
      "step": 2794
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8256548376558105,
      "learning_rate": 1.4923567145450587e-06,
      "loss": 0.5682,
      "step": 2795
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.878382667385202,
      "learning_rate": 1.49199749466663e-06,
      "loss": 0.6889,
      "step": 2796
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6844584629431314,
      "learning_rate": 1.4916381910061537e-06,
      "loss": 0.5099,
      "step": 2797
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9417801837383127,
      "learning_rate": 1.4912788036248154e-06,
      "loss": 0.6883,
      "step": 2798
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9993115709276478,
      "learning_rate": 1.490919332583815e-06,
      "loss": 0.6205,
      "step": 2799
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.712431838687457,
      "learning_rate": 1.4905597779443662e-06,
      "loss": 0.5624,
      "step": 2800
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9954101807394483,
      "learning_rate": 1.490200139767698e-06,
      "loss": 0.6487,
      "step": 2801
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7703680125427543,
      "learning_rate": 1.4898404181150533e-06,
      "loss": 0.5292,
      "step": 2802
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6801230389210384,
      "learning_rate": 1.489480613047688e-06,
      "loss": 0.5354,
      "step": 2803
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6869755998172585,
      "learning_rate": 1.4891207246268737e-06,
      "loss": 0.5433,
      "step": 2804
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0097618511519242,
      "learning_rate": 1.488760752913896e-06,
      "loss": 0.5963,
      "step": 2805
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9611198456023415,
      "learning_rate": 1.4884006979700536e-06,
      "loss": 0.63,
      "step": 2806
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8568639259941129,
      "learning_rate": 1.4880405598566608e-06,
      "loss": 0.6519,
      "step": 2807
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7807086486932002,
      "learning_rate": 1.4876803386350447e-06,
      "loss": 0.5361,
      "step": 2808
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9452024241556469,
      "learning_rate": 1.4873200343665478e-06,
      "loss": 0.5776,
      "step": 2809
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7422888166236278,
      "learning_rate": 1.4869596471125253e-06,
      "loss": 0.5894,
      "step": 2810
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8395395847453998,
      "learning_rate": 1.4865991769343488e-06,
      "loss": 0.5987,
      "step": 2811
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8145206628244746,
      "learning_rate": 1.4862386238934015e-06,
      "loss": 0.6312,
      "step": 2812
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6211170414080287,
      "learning_rate": 1.485877988051082e-06,
      "loss": 0.5155,
      "step": 2813
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9451681959565611,
      "learning_rate": 1.4855172694688032e-06,
      "loss": 0.618,
      "step": 2814
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7405673746521159,
      "learning_rate": 1.4851564682079913e-06,
      "loss": 0.5899,
      "step": 2815
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6604511842601581,
      "learning_rate": 1.4847955843300875e-06,
      "loss": 0.55,
      "step": 2816
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7123543532700324,
      "learning_rate": 1.484434617896546e-06,
      "loss": 0.5873,
      "step": 2817
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8855042096725597,
      "learning_rate": 1.484073568968836e-06,
      "loss": 0.6245,
      "step": 2818
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8806194203178593,
      "learning_rate": 1.4837124376084403e-06,
      "loss": 0.5766,
      "step": 2819
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8791602683071743,
      "learning_rate": 1.4833512238768555e-06,
      "loss": 0.6288,
      "step": 2820
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7665832568695121,
      "learning_rate": 1.482989927835593e-06,
      "loss": 0.6016,
      "step": 2821
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8047272116294628,
      "learning_rate": 1.4826285495461773e-06,
      "loss": 0.5624,
      "step": 2822
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9131174662422704,
      "learning_rate": 1.4822670890701476e-06,
      "loss": 0.6033,
      "step": 2823
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8821364233152592,
      "learning_rate": 1.4819055464690571e-06,
      "loss": 0.6315,
      "step": 2824
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8429396770684198,
      "learning_rate": 1.4815439218044726e-06,
      "loss": 0.61,
      "step": 2825
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9525227212937452,
      "learning_rate": 1.4811822151379745e-06,
      "loss": 0.6842,
      "step": 2826
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7585383691230887,
      "learning_rate": 1.4808204265311583e-06,
      "loss": 0.5962,
      "step": 2827
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7895312482008328,
      "learning_rate": 1.4804585560456327e-06,
      "loss": 0.6291,
      "step": 2828
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7991188232234464,
      "learning_rate": 1.4800966037430204e-06,
      "loss": 0.592,
      "step": 2829
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8834088301011445,
      "learning_rate": 1.479734569684958e-06,
      "loss": 0.6057,
      "step": 2830
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6791035197970052,
      "learning_rate": 1.479372453933096e-06,
      "loss": 0.5297,
      "step": 2831
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8498852482456768,
      "learning_rate": 1.4790102565490994e-06,
      "loss": 0.621,
      "step": 2832
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0275595307419705,
      "learning_rate": 1.4786479775946466e-06,
      "loss": 0.6571,
      "step": 2833
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6908995911305638,
      "learning_rate": 1.4782856171314296e-06,
      "loss": 0.5729,
      "step": 2834
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.367175357037824,
      "learning_rate": 1.4779231752211546e-06,
      "loss": 0.6134,
      "step": 2835
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6338766342409193,
      "learning_rate": 1.4775606519255414e-06,
      "loss": 0.5281,
      "step": 2836
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8166535003548172,
      "learning_rate": 1.477198047306325e-06,
      "loss": 0.6466,
      "step": 2837
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8271732737616625,
      "learning_rate": 1.4768353614252525e-06,
      "loss": 0.5821,
      "step": 2838
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8602234663684609,
      "learning_rate": 1.4764725943440854e-06,
      "loss": 0.5774,
      "step": 2839
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8149732392319375,
      "learning_rate": 1.4761097461245995e-06,
      "loss": 0.5995,
      "step": 2840
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0375300098110611,
      "learning_rate": 1.4757468168285838e-06,
      "loss": 0.6654,
      "step": 2841
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8720895143222749,
      "learning_rate": 1.4753838065178416e-06,
      "loss": 0.6445,
      "step": 2842
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8401376207135789,
      "learning_rate": 1.4750207152541896e-06,
      "loss": 0.5922,
      "step": 2843
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7137166118185889,
      "learning_rate": 1.4746575430994585e-06,
      "loss": 0.5949,
      "step": 2844
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.904275787449756,
      "learning_rate": 1.474294290115493e-06,
      "loss": 0.6895,
      "step": 2845
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6278408433970168,
      "learning_rate": 1.4739309563641507e-06,
      "loss": 0.533,
      "step": 2846
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9317759418471361,
      "learning_rate": 1.4735675419073038e-06,
      "loss": 0.6687,
      "step": 2847
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8541067919028597,
      "learning_rate": 1.4732040468068385e-06,
      "loss": 0.6245,
      "step": 2848
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6409720847446649,
      "learning_rate": 1.4728404711246537e-06,
      "loss": 0.5416,
      "step": 2849
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9758490467491027,
      "learning_rate": 1.4724768149226625e-06,
      "loss": 0.6967,
      "step": 2850
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9595496507858726,
      "learning_rate": 1.472113078262792e-06,
      "loss": 0.6379,
      "step": 2851
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7398188097774925,
      "learning_rate": 1.4717492612069826e-06,
      "loss": 0.5374,
      "step": 2852
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6662992799903374,
      "learning_rate": 1.471385363817189e-06,
      "loss": 0.5476,
      "step": 2853
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7834642834856776,
      "learning_rate": 1.4710213861553787e-06,
      "loss": 0.6131,
      "step": 2854
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7295562995003764,
      "learning_rate": 1.4706573282835332e-06,
      "loss": 0.6309,
      "step": 2855
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9048690458751024,
      "learning_rate": 1.4702931902636483e-06,
      "loss": 0.6745,
      "step": 2856
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6431363382022175,
      "learning_rate": 1.4699289721577323e-06,
      "loss": 0.5243,
      "step": 2857
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8723334360771513,
      "learning_rate": 1.4695646740278084e-06,
      "loss": 0.6523,
      "step": 2858
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6340711863349675,
      "learning_rate": 1.469200295935912e-06,
      "loss": 0.5409,
      "step": 2859
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6390113345264566,
      "learning_rate": 1.4688358379440935e-06,
      "loss": 0.5456,
      "step": 2860
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6242937988933909,
      "learning_rate": 1.468471300114416e-06,
      "loss": 0.5319,
      "step": 2861
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.049302200516771,
      "learning_rate": 1.4681066825089568e-06,
      "loss": 0.6488,
      "step": 2862
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0021388610490307,
      "learning_rate": 1.467741985189806e-06,
      "loss": 0.6531,
      "step": 2863
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.688816073543492,
      "learning_rate": 1.4673772082190682e-06,
      "loss": 0.529,
      "step": 2864
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7954243258998963,
      "learning_rate": 1.4670123516588607e-06,
      "loss": 0.6383,
      "step": 2865
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6231693026395826,
      "learning_rate": 1.466647415571315e-06,
      "loss": 0.5526,
      "step": 2866
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6612116878820933,
      "learning_rate": 1.4662824000185762e-06,
      "loss": 0.507,
      "step": 2867
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7228177788745414,
      "learning_rate": 1.4659173050628024e-06,
      "loss": 0.5903,
      "step": 2868
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7297713297166086,
      "learning_rate": 1.4655521307661654e-06,
      "loss": 0.5533,
      "step": 2869
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6761672205441746,
      "learning_rate": 1.4651868771908504e-06,
      "loss": 0.6158,
      "step": 2870
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5985113491739767,
      "learning_rate": 1.4648215443990566e-06,
      "loss": 0.4987,
      "step": 2871
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9215464434007808,
      "learning_rate": 1.4644561324529962e-06,
      "loss": 0.6571,
      "step": 2872
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9393647895589455,
      "learning_rate": 1.4640906414148955e-06,
      "loss": 0.5502,
      "step": 2873
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7430131579773921,
      "learning_rate": 1.463725071346993e-06,
      "loss": 0.5574,
      "step": 2874
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8321939537438792,
      "learning_rate": 1.463359422311542e-06,
      "loss": 0.5985,
      "step": 2875
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6758182032167424,
      "learning_rate": 1.4629936943708088e-06,
      "loss": 0.5883,
      "step": 2876
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8417081927849046,
      "learning_rate": 1.4626278875870726e-06,
      "loss": 0.7009,
      "step": 2877
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6306337002851999,
      "learning_rate": 1.462262002022627e-06,
      "loss": 0.5223,
      "step": 2878
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.059711686595755,
      "learning_rate": 1.4618960377397779e-06,
      "loss": 0.6694,
      "step": 2879
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8022784255350268,
      "learning_rate": 1.461529994800846e-06,
      "loss": 0.5821,
      "step": 2880
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6730733279404993,
      "learning_rate": 1.4611638732681639e-06,
      "loss": 0.5142,
      "step": 2881
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7054927846232008,
      "learning_rate": 1.4607976732040788e-06,
      "loss": 0.5754,
      "step": 2882
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8192668455977995,
      "learning_rate": 1.4604313946709502e-06,
      "loss": 0.5604,
      "step": 2883
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6230196729695052,
      "learning_rate": 1.460065037731152e-06,
      "loss": 0.5403,
      "step": 2884
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6641421122507739,
      "learning_rate": 1.459698602447071e-06,
      "loss": 0.5808,
      "step": 2885
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8781730793043107,
      "learning_rate": 1.4593320888811066e-06,
      "loss": 0.6291,
      "step": 2886
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7726082555341107,
      "learning_rate": 1.4589654970956732e-06,
      "loss": 0.5692,
      "step": 2887
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9369017412148846,
      "learning_rate": 1.4585988271531971e-06,
      "loss": 0.6818,
      "step": 2888
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.643393742497125,
      "learning_rate": 1.4582320791161181e-06,
      "loss": 0.5383,
      "step": 2889
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7461200581658407,
      "learning_rate": 1.4578652530468902e-06,
      "loss": 0.5833,
      "step": 2890
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7767761854201827,
      "learning_rate": 1.4574983490079795e-06,
      "loss": 0.6174,
      "step": 2891
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7910762618030133,
      "learning_rate": 1.4571313670618662e-06,
      "loss": 0.6166,
      "step": 2892
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9296065308190369,
      "learning_rate": 1.4567643072710431e-06,
      "loss": 0.6027,
      "step": 2893
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1739797110213461,
      "learning_rate": 1.4563971696980176e-06,
      "loss": 0.6332,
      "step": 2894
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6382087052426432,
      "learning_rate": 1.4560299544053086e-06,
      "loss": 0.5518,
      "step": 2895
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6421374223954901,
      "learning_rate": 1.455662661455449e-06,
      "loss": 0.599,
      "step": 2896
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7957639045037196,
      "learning_rate": 1.4552952909109856e-06,
      "loss": 0.5561,
      "step": 2897
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7121533918037124,
      "learning_rate": 1.454927842834477e-06,
      "loss": 0.5764,
      "step": 2898
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8269561110956469,
      "learning_rate": 1.4545603172884964e-06,
      "loss": 0.5928,
      "step": 2899
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6188116101653932,
      "learning_rate": 1.4541927143356292e-06,
      "loss": 0.5533,
      "step": 2900
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.783880788082337,
      "learning_rate": 1.4538250340384745e-06,
      "loss": 0.5221,
      "step": 2901
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.912251620763405,
      "learning_rate": 1.4534572764596445e-06,
      "loss": 0.6544,
      "step": 2902
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7014016807515112,
      "learning_rate": 1.4530894416617644e-06,
      "loss": 0.5702,
      "step": 2903
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8340215462439355,
      "learning_rate": 1.4527215297074726e-06,
      "loss": 0.6467,
      "step": 2904
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8029867615115054,
      "learning_rate": 1.452353540659421e-06,
      "loss": 0.5442,
      "step": 2905
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7897695705767267,
      "learning_rate": 1.4519854745802738e-06,
      "loss": 0.5633,
      "step": 2906
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7537989229950761,
      "learning_rate": 1.4516173315327088e-06,
      "loss": 0.5711,
      "step": 2907
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7630143942739561,
      "learning_rate": 1.4512491115794173e-06,
      "loss": 0.6208,
      "step": 2908
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6708950374021295,
      "learning_rate": 1.4508808147831034e-06,
      "loss": 0.5783,
      "step": 2909
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6723952230171419,
      "learning_rate": 1.4505124412064836e-06,
      "loss": 0.5084,
      "step": 2910
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8366969820925751,
      "learning_rate": 1.4501439909122888e-06,
      "loss": 0.6213,
      "step": 2911
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8287191932939644,
      "learning_rate": 1.4497754639632619e-06,
      "loss": 0.6112,
      "step": 2912
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7110835605944674,
      "learning_rate": 1.449406860422159e-06,
      "loss": 0.5916,
      "step": 2913
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9402308625779777,
      "learning_rate": 1.4490381803517497e-06,
      "loss": 0.6397,
      "step": 2914
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8443431419548582,
      "learning_rate": 1.4486694238148164e-06,
      "loss": 0.6066,
      "step": 2915
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6579200597845478,
      "learning_rate": 1.4483005908741545e-06,
      "loss": 0.5455,
      "step": 2916
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6535907733263981,
      "learning_rate": 1.4479316815925722e-06,
      "loss": 0.555,
      "step": 2917
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9902580977837102,
      "learning_rate": 1.4475626960328913e-06,
      "loss": 0.6462,
      "step": 2918
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7236014878111918,
      "learning_rate": 1.4471936342579455e-06,
      "loss": 0.5563,
      "step": 2919
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7341781045361712,
      "learning_rate": 1.4468244963305832e-06,
      "loss": 0.5364,
      "step": 2920
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.183641005048583,
      "learning_rate": 1.4464552823136635e-06,
      "loss": 0.6825,
      "step": 2921
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6909245131386688,
      "learning_rate": 1.4460859922700608e-06,
      "loss": 0.5861,
      "step": 2922
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7223689064219212,
      "learning_rate": 1.4457166262626608e-06,
      "loss": 0.5159,
      "step": 2923
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.665803479727205,
      "learning_rate": 1.4453471843543628e-06,
      "loss": 0.5532,
      "step": 2924
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7120534362115426,
      "learning_rate": 1.4449776666080786e-06,
      "loss": 0.6129,
      "step": 2925
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6686969939002663,
      "learning_rate": 1.4446080730867341e-06,
      "loss": 0.5209,
      "step": 2926
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7118750910800716,
      "learning_rate": 1.4442384038532662e-06,
      "loss": 0.5839,
      "step": 2927
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6910578137628784,
      "learning_rate": 1.4438686589706264e-06,
      "loss": 0.5311,
      "step": 2928
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.638091606586386,
      "learning_rate": 1.4434988385017781e-06,
      "loss": 0.5577,
      "step": 2929
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9138323103990887,
      "learning_rate": 1.443128942509698e-06,
      "loss": 0.6284,
      "step": 2930
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.847941011196858,
      "learning_rate": 1.4427589710573754e-06,
      "loss": 0.658,
      "step": 2931
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7998018825646488,
      "learning_rate": 1.4423889242078128e-06,
      "loss": 0.6142,
      "step": 2932
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7124106139576353,
      "learning_rate": 1.4420188020240248e-06,
      "loss": 0.5786,
      "step": 2933
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6466545513567014,
      "learning_rate": 1.4416486045690397e-06,
      "loss": 0.556,
      "step": 2934
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8484743945606571,
      "learning_rate": 1.4412783319058987e-06,
      "loss": 0.6077,
      "step": 2935
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7720705726445568,
      "learning_rate": 1.4409079840976548e-06,
      "loss": 0.5729,
      "step": 2936
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0143677895973267,
      "learning_rate": 1.4405375612073746e-06,
      "loss": 0.6857,
      "step": 2937
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6898835444306005,
      "learning_rate": 1.4401670632981372e-06,
      "loss": 0.5898,
      "step": 2938
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8562239409554988,
      "learning_rate": 1.4397964904330347e-06,
      "loss": 0.6312,
      "step": 2939
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6275942633881774,
      "learning_rate": 1.4394258426751715e-06,
      "loss": 0.5333,
      "step": 2940
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8306793606486005,
      "learning_rate": 1.4390551200876653e-06,
      "loss": 0.639,
      "step": 2941
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8698843479431698,
      "learning_rate": 1.4386843227336461e-06,
      "loss": 0.707,
      "step": 2942
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0234986116008342,
      "learning_rate": 1.4383134506762569e-06,
      "loss": 0.6562,
      "step": 2943
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7935783697085956,
      "learning_rate": 1.4379425039786533e-06,
      "loss": 0.5731,
      "step": 2944
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7231751743066511,
      "learning_rate": 1.437571482704004e-06,
      "loss": 0.5423,
      "step": 2945
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6201009366394794,
      "learning_rate": 1.4372003869154894e-06,
      "loss": 0.5388,
      "step": 2946
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6704760156784091,
      "learning_rate": 1.4368292166763036e-06,
      "loss": 0.6096,
      "step": 2947
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7873334139932725,
      "learning_rate": 1.4364579720496534e-06,
      "loss": 0.5379,
      "step": 2948
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.076230516332461,
      "learning_rate": 1.4360866530987573e-06,
      "loss": 0.629,
      "step": 2949
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8936582269689555,
      "learning_rate": 1.4357152598868476e-06,
      "loss": 0.6616,
      "step": 2950
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6852108334721988,
      "learning_rate": 1.4353437924771683e-06,
      "loss": 0.582,
      "step": 2951
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8820267243875591,
      "learning_rate": 1.4349722509329762e-06,
      "loss": 0.6588,
      "step": 2952
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6501467731105091,
      "learning_rate": 1.4346006353175415e-06,
      "loss": 0.5353,
      "step": 2953
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6364286013092798,
      "learning_rate": 1.434228945694146e-06,
      "loss": 0.5451,
      "step": 2954
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8567306597962844,
      "learning_rate": 1.4338571821260851e-06,
      "loss": 0.5639,
      "step": 2955
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7656357571916407,
      "learning_rate": 1.4334853446766661e-06,
      "loss": 0.5687,
      "step": 2956
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8219383776743547,
      "learning_rate": 1.4331134334092088e-06,
      "loss": 0.5937,
      "step": 2957
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6597001514861968,
      "learning_rate": 1.4327414483870458e-06,
      "loss": 0.5559,
      "step": 2958
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7711896010893342,
      "learning_rate": 1.4323693896735226e-06,
      "loss": 0.5518,
      "step": 2959
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9238898201112902,
      "learning_rate": 1.4319972573319966e-06,
      "loss": 0.7341,
      "step": 2960
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2172468232222828,
      "learning_rate": 1.431625051425838e-06,
      "loss": 0.6553,
      "step": 2961
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8476898689698178,
      "learning_rate": 1.4312527720184301e-06,
      "loss": 0.6205,
      "step": 2962
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8852796017840445,
      "learning_rate": 1.4308804191731678e-06,
      "loss": 0.6729,
      "step": 2963
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9393060547855648,
      "learning_rate": 1.430507992953459e-06,
      "loss": 0.6203,
      "step": 2964
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7592432396521749,
      "learning_rate": 1.4301354934227238e-06,
      "loss": 0.5786,
      "step": 2965
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6694951255797104,
      "learning_rate": 1.4297629206443956e-06,
      "loss": 0.5381,
      "step": 2966
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7917638356329877,
      "learning_rate": 1.4293902746819188e-06,
      "loss": 0.6327,
      "step": 2967
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7838022234930193,
      "learning_rate": 1.4290175555987518e-06,
      "loss": 0.5735,
      "step": 2968
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.685394067805382,
      "learning_rate": 1.4286447634583647e-06,
      "loss": 0.558,
      "step": 2969
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9383921313502919,
      "learning_rate": 1.4282718983242398e-06,
      "loss": 0.6604,
      "step": 2970
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9219988580776781,
      "learning_rate": 1.4278989602598724e-06,
      "loss": 0.6025,
      "step": 2971
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2579449453242686,
      "learning_rate": 1.42752594932877e-06,
      "loss": 0.6656,
      "step": 2972
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8465667951234505,
      "learning_rate": 1.427152865594452e-06,
      "loss": 0.6122,
      "step": 2973
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6760139087277035,
      "learning_rate": 1.4267797091204513e-06,
      "loss": 0.5723,
      "step": 2974
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7926683835961463,
      "learning_rate": 1.4264064799703121e-06,
      "loss": 0.5981,
      "step": 2975
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9138611273730386,
      "learning_rate": 1.4260331782075915e-06,
      "loss": 0.6187,
      "step": 2976
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6638668854685227,
      "learning_rate": 1.4256598038958595e-06,
      "loss": 0.4685,
      "step": 2977
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7188822482728612,
      "learning_rate": 1.425286357098697e-06,
      "loss": 0.5485,
      "step": 2978
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7546331998820988,
      "learning_rate": 1.4249128378796988e-06,
      "loss": 0.5901,
      "step": 2979
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7591776088816059,
      "learning_rate": 1.4245392463024707e-06,
      "loss": 0.5778,
      "step": 2980
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8272853667040697,
      "learning_rate": 1.424165582430632e-06,
      "loss": 0.5813,
      "step": 2981
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2093875956963143,
      "learning_rate": 1.4237918463278135e-06,
      "loss": 0.6989,
      "step": 2982
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8927824448074608,
      "learning_rate": 1.4234180380576586e-06,
      "loss": 0.6271,
      "step": 2983
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.650836045109679,
      "learning_rate": 1.423044157683823e-06,
      "loss": 0.59,
      "step": 2984
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.853130869444366,
      "learning_rate": 1.4226702052699746e-06,
      "loss": 0.6401,
      "step": 2985
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8142910491659119,
      "learning_rate": 1.4222961808797935e-06,
      "loss": 0.6347,
      "step": 2986
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.664138194848671,
      "learning_rate": 1.4219220845769723e-06,
      "loss": 0.5771,
      "step": 2987
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.035956849405899,
      "learning_rate": 1.4215479164252154e-06,
      "loss": 0.6414,
      "step": 2988
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6543999571962653,
      "learning_rate": 1.4211736764882404e-06,
      "loss": 0.4885,
      "step": 2989
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7516219631136275,
      "learning_rate": 1.4207993648297757e-06,
      "loss": 0.5815,
      "step": 2990
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7298867242030105,
      "learning_rate": 1.4204249815135631e-06,
      "loss": 0.5296,
      "step": 2991
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6563684290607786,
      "learning_rate": 1.4200505266033562e-06,
      "loss": 0.5572,
      "step": 2992
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8828322877649945,
      "learning_rate": 1.4196760001629208e-06,
      "loss": 0.6054,
      "step": 2993
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9215149141934957,
      "learning_rate": 1.4193014022560345e-06,
      "loss": 0.6476,
      "step": 2994
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.670080022202974,
      "learning_rate": 1.4189267329464878e-06,
      "loss": 0.5583,
      "step": 2995
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8516403528710943,
      "learning_rate": 1.418551992298083e-06,
      "loss": 0.6149,
      "step": 2996
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2123042609569503,
      "learning_rate": 1.4181771803746338e-06,
      "loss": 0.6236,
      "step": 2997
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8213229933310824,
      "learning_rate": 1.4178022972399678e-06,
      "loss": 0.5624,
      "step": 2998
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8121597080178251,
      "learning_rate": 1.417427342957923e-06,
      "loss": 0.554,
      "step": 2999
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7638328428765224,
      "learning_rate": 1.4170523175923506e-06,
      "loss": 0.5722,
      "step": 3000
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7944807682202171,
      "learning_rate": 1.4166772212071129e-06,
      "loss": 0.6668,
      "step": 3001
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7069733834003264,
      "learning_rate": 1.416302053866086e-06,
      "loss": 0.5926,
      "step": 3002
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9058787848456458,
      "learning_rate": 1.4159268156331557e-06,
      "loss": 0.6613,
      "step": 3003
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6519594140127106,
      "learning_rate": 1.4155515065722221e-06,
      "loss": 0.5665,
      "step": 3004
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8104227746843757,
      "learning_rate": 1.415176126747196e-06,
      "loss": 0.594,
      "step": 3005
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6700789093562772,
      "learning_rate": 1.4148006762220007e-06,
      "loss": 0.526,
      "step": 3006
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.824616350017847,
      "learning_rate": 1.4144251550605718e-06,
      "loss": 0.5188,
      "step": 3007
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.750720628544232,
      "learning_rate": 1.4140495633268563e-06,
      "loss": 0.6419,
      "step": 3008
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6071372547875775,
      "learning_rate": 1.413673901084814e-06,
      "loss": 0.5189,
      "step": 3009
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8659769801520558,
      "learning_rate": 1.4132981683984157e-06,
      "loss": 0.6489,
      "step": 3010
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7356404101400091,
      "learning_rate": 1.4129223653316451e-06,
      "loss": 0.5602,
      "step": 3011
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1566432139450225,
      "learning_rate": 1.4125464919484976e-06,
      "loss": 0.7056,
      "step": 3012
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.743582772039973,
      "learning_rate": 1.4121705483129803e-06,
      "loss": 0.5227,
      "step": 3013
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7186620456629237,
      "learning_rate": 1.4117945344891125e-06,
      "loss": 0.5518,
      "step": 3014
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9094206463847527,
      "learning_rate": 1.4114184505409256e-06,
      "loss": 0.6286,
      "step": 3015
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8690500851435035,
      "learning_rate": 1.4110422965324627e-06,
      "loss": 0.6014,
      "step": 3016
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7398355711169552,
      "learning_rate": 1.4106660725277789e-06,
      "loss": 0.5439,
      "step": 3017
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7255804476456763,
      "learning_rate": 1.410289778590941e-06,
      "loss": 0.5492,
      "step": 3018
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7063142306127926,
      "learning_rate": 1.4099134147860286e-06,
      "loss": 0.5334,
      "step": 3019
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6079952451320196,
      "learning_rate": 1.4095369811771318e-06,
      "loss": 0.5111,
      "step": 3020
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7804555742539693,
      "learning_rate": 1.4091604778283536e-06,
      "loss": 0.5797,
      "step": 3021
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8194947326422135,
      "learning_rate": 1.4087839048038089e-06,
      "loss": 0.6465,
      "step": 3022
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7537315301762143,
      "learning_rate": 1.408407262167624e-06,
      "loss": 0.6193,
      "step": 3023
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6700254409911592,
      "learning_rate": 1.4080305499839366e-06,
      "loss": 0.5404,
      "step": 3024
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9170093021613263,
      "learning_rate": 1.4076537683168978e-06,
      "loss": 0.6207,
      "step": 3025
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7117474086503168,
      "learning_rate": 1.4072769172306692e-06,
      "loss": 0.5453,
      "step": 3026
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.964764291098878,
      "learning_rate": 1.4068999967894241e-06,
      "loss": 0.6272,
      "step": 3027
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6562264011715662,
      "learning_rate": 1.4065230070573491e-06,
      "loss": 0.5573,
      "step": 3028
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6618160073636681,
      "learning_rate": 1.4061459480986408e-06,
      "loss": 0.5492,
      "step": 3029
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8266839236692214,
      "learning_rate": 1.4057688199775089e-06,
      "loss": 0.6083,
      "step": 3030
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8762895024041505,
      "learning_rate": 1.4053916227581741e-06,
      "loss": 0.5438,
      "step": 3031
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.808153798077984,
      "learning_rate": 1.4050143565048694e-06,
      "loss": 0.6486,
      "step": 3032
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.639288981719959,
      "learning_rate": 1.404637021281839e-06,
      "loss": 0.5381,
      "step": 3033
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7924822386103259,
      "learning_rate": 1.4042596171533396e-06,
      "loss": 0.5984,
      "step": 3034
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7499464198484329,
      "learning_rate": 1.403882144183639e-06,
      "loss": 0.5027,
      "step": 3035
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8992025054839136,
      "learning_rate": 1.4035046024370164e-06,
      "loss": 0.6477,
      "step": 3036
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.051836967969425,
      "learning_rate": 1.4031269919777642e-06,
      "loss": 0.6676,
      "step": 3037
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7277688500631552,
      "learning_rate": 1.4027493128701848e-06,
      "loss": 0.5815,
      "step": 3038
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6729576222209144,
      "learning_rate": 1.4023715651785933e-06,
      "loss": 0.5789,
      "step": 3039
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7468178601292582,
      "learning_rate": 1.4019937489673161e-06,
      "loss": 0.5135,
      "step": 3040
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7244252516724392,
      "learning_rate": 1.4016158643006912e-06,
      "loss": 0.5282,
      "step": 3041
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8936983683746277,
      "learning_rate": 1.401237911243069e-06,
      "loss": 0.5878,
      "step": 3042
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.987620945908303,
      "learning_rate": 1.4008598898588102e-06,
      "loss": 0.6095,
      "step": 3043
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6558393297049334,
      "learning_rate": 1.4004818002122885e-06,
      "loss": 0.4651,
      "step": 3044
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8562602435726057,
      "learning_rate": 1.4001036423678882e-06,
      "loss": 0.6527,
      "step": 3045
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6621995669639869,
      "learning_rate": 1.3997254163900062e-06,
      "loss": 0.5093,
      "step": 3046
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7160092328534885,
      "learning_rate": 1.3993471223430498e-06,
      "loss": 0.6176,
      "step": 3047
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7467335392432723,
      "learning_rate": 1.3989687602914389e-06,
      "loss": 0.5445,
      "step": 3048
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7874308348344078,
      "learning_rate": 1.3985903302996046e-06,
      "loss": 0.5782,
      "step": 3049
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7158873000819497,
      "learning_rate": 1.3982118324319893e-06,
      "loss": 0.578,
      "step": 3050
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8480559730948038,
      "learning_rate": 1.3978332667530478e-06,
      "loss": 0.6524,
      "step": 3051
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.660844913526248,
      "learning_rate": 1.3974546333272455e-06,
      "loss": 0.4931,
      "step": 3052
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0188867200790888,
      "learning_rate": 1.3970759322190597e-06,
      "loss": 0.6447,
      "step": 3053
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.596417535339059,
      "learning_rate": 1.3966971634929794e-06,
      "loss": 0.5162,
      "step": 3054
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6187896284640257,
      "learning_rate": 1.3963183272135048e-06,
      "loss": 0.5228,
      "step": 3055
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9067817932211217,
      "learning_rate": 1.395939423445148e-06,
      "loss": 0.6231,
      "step": 3056
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7661487121994174,
      "learning_rate": 1.3955604522524325e-06,
      "loss": 0.5615,
      "step": 3057
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7956239030597039,
      "learning_rate": 1.3951814136998925e-06,
      "loss": 0.5917,
      "step": 3058
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7205426572844849,
      "learning_rate": 1.3948023078520749e-06,
      "loss": 0.5969,
      "step": 3059
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7345846234321981,
      "learning_rate": 1.3944231347735374e-06,
      "loss": 0.6286,
      "step": 3060
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7401107859636606,
      "learning_rate": 1.394043894528849e-06,
      "loss": 0.5552,
      "step": 3061
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8503840356783579,
      "learning_rate": 1.3936645871825903e-06,
      "loss": 0.6239,
      "step": 3062
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9026937311097736,
      "learning_rate": 1.3932852127993537e-06,
      "loss": 0.6779,
      "step": 3063
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8313075353277407,
      "learning_rate": 1.3929057714437429e-06,
      "loss": 0.6999,
      "step": 3064
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7472509686513,
      "learning_rate": 1.392526263180372e-06,
      "loss": 0.5924,
      "step": 3065
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7998443157356848,
      "learning_rate": 1.3921466880738682e-06,
      "loss": 0.6034,
      "step": 3066
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7669568935498641,
      "learning_rate": 1.3917670461888685e-06,
      "loss": 0.5476,
      "step": 3067
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8210074144624843,
      "learning_rate": 1.3913873375900224e-06,
      "loss": 0.5335,
      "step": 3068
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5805165683117435,
      "learning_rate": 1.3910075623419902e-06,
      "loss": 0.4953,
      "step": 3069
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7357027164797455,
      "learning_rate": 1.3906277205094434e-06,
      "loss": 0.5573,
      "step": 3070
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6370652266256153,
      "learning_rate": 1.3902478121570654e-06,
      "loss": 0.541,
      "step": 3071
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.640638932286323,
      "learning_rate": 1.3898678373495507e-06,
      "loss": 0.5923,
      "step": 3072
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8434632039172507,
      "learning_rate": 1.3894877961516047e-06,
      "loss": 0.6602,
      "step": 3073
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7189335733407953,
      "learning_rate": 1.3891076886279444e-06,
      "loss": 0.549,
      "step": 3074
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.870419102859188,
      "learning_rate": 1.3887275148432988e-06,
      "loss": 0.5516,
      "step": 3075
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6543415411451026,
      "learning_rate": 1.3883472748624075e-06,
      "loss": 0.5195,
      "step": 3076
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.198898936309915,
      "learning_rate": 1.3879669687500207e-06,
      "loss": 0.6356,
      "step": 3077
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8617258263879347,
      "learning_rate": 1.3875865965709008e-06,
      "loss": 0.635,
      "step": 3078
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7976794770841682,
      "learning_rate": 1.3872061583898216e-06,
      "loss": 0.5621,
      "step": 3079
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.911971024750434,
      "learning_rate": 1.3868256542715673e-06,
      "loss": 0.6053,
      "step": 3080
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5894705140986786,
      "learning_rate": 1.3864450842809342e-06,
      "loss": 0.5087,
      "step": 3081
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7114305446740168,
      "learning_rate": 1.3860644484827291e-06,
      "loss": 0.5772,
      "step": 3082
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8065196633682881,
      "learning_rate": 1.3856837469417704e-06,
      "loss": 0.5648,
      "step": 3083
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5785131775924585,
      "learning_rate": 1.3853029797228878e-06,
      "loss": 0.5044,
      "step": 3084
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8108596744535257,
      "learning_rate": 1.3849221468909215e-06,
      "loss": 0.5644,
      "step": 3085
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7784592870930646,
      "learning_rate": 1.3845412485107238e-06,
      "loss": 0.5499,
      "step": 3086
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7725760553097077,
      "learning_rate": 1.3841602846471575e-06,
      "loss": 0.5861,
      "step": 3087
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6924838413344677,
      "learning_rate": 1.3837792553650968e-06,
      "loss": 0.5775,
      "step": 3088
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9237084490598435,
      "learning_rate": 1.3833981607294274e-06,
      "loss": 0.6616,
      "step": 3089
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7075983734160326,
      "learning_rate": 1.3830170008050455e-06,
      "loss": 0.5589,
      "step": 3090
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7489193476917966,
      "learning_rate": 1.3826357756568584e-06,
      "loss": 0.565,
      "step": 3091
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.763601664487913,
      "learning_rate": 1.3822544853497848e-06,
      "loss": 0.5747,
      "step": 3092
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8524374001901855,
      "learning_rate": 1.381873129948755e-06,
      "loss": 0.5659,
      "step": 3093
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8185151052335679,
      "learning_rate": 1.3814917095187096e-06,
      "loss": 0.6405,
      "step": 3094
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6260629653074958,
      "learning_rate": 1.3811102241246e-06,
      "loss": 0.5432,
      "step": 3095
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9674082594299532,
      "learning_rate": 1.3807286738313896e-06,
      "loss": 0.6274,
      "step": 3096
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9438089849834327,
      "learning_rate": 1.380347058704053e-06,
      "loss": 0.5551,
      "step": 3097
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6236022789990685,
      "learning_rate": 1.3799653788075742e-06,
      "loss": 0.5221,
      "step": 3098
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.838193510354797,
      "learning_rate": 1.37958363420695e-06,
      "loss": 0.6315,
      "step": 3099
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.899187492836811,
      "learning_rate": 1.3792018249671874e-06,
      "loss": 0.6108,
      "step": 3100
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7708992022961609,
      "learning_rate": 1.3788199511533044e-06,
      "loss": 0.5763,
      "step": 3101
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.848960857513316,
      "learning_rate": 1.3784380128303301e-06,
      "loss": 0.6074,
      "step": 3102
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6297093655135994,
      "learning_rate": 1.3780560100633052e-06,
      "loss": 0.583,
      "step": 3103
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6218245145001554,
      "learning_rate": 1.3776739429172798e-06,
      "loss": 0.526,
      "step": 3104
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7730874329693123,
      "learning_rate": 1.3772918114573169e-06,
      "loss": 0.6303,
      "step": 3105
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7149510079393127,
      "learning_rate": 1.3769096157484889e-06,
      "loss": 0.5415,
      "step": 3106
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0364738409190069,
      "learning_rate": 1.37652735585588e-06,
      "loss": 0.6031,
      "step": 3107
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9678099774230313,
      "learning_rate": 1.3761450318445853e-06,
      "loss": 0.6382,
      "step": 3108
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8108791160273627,
      "learning_rate": 1.37576264377971e-06,
      "loss": 0.6645,
      "step": 3109
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9724002909238344,
      "learning_rate": 1.3753801917263714e-06,
      "loss": 0.649,
      "step": 3110
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6211448285366288,
      "learning_rate": 1.3749976757496967e-06,
      "loss": 0.5451,
      "step": 3111
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6568689205316481,
      "learning_rate": 1.3746150959148243e-06,
      "loss": 0.5728,
      "step": 3112
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8233192986327745,
      "learning_rate": 1.3742324522869041e-06,
      "loss": 0.56,
      "step": 3113
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6809746166867059,
      "learning_rate": 1.373849744931096e-06,
      "loss": 0.5019,
      "step": 3114
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.943166931435314,
      "learning_rate": 1.373466973912571e-06,
      "loss": 0.6573,
      "step": 3115
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5882815260604138,
      "learning_rate": 1.373084139296511e-06,
      "loss": 0.4802,
      "step": 3116
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8068045842217108,
      "learning_rate": 1.3727012411481087e-06,
      "loss": 0.6063,
      "step": 3117
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.86223879857831,
      "learning_rate": 1.372318279532568e-06,
      "loss": 0.6278,
      "step": 3118
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.664244863660214,
      "learning_rate": 1.3719352545151033e-06,
      "loss": 0.4939,
      "step": 3119
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7072949414515083,
      "learning_rate": 1.3715521661609393e-06,
      "loss": 0.5278,
      "step": 3120
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7384809252601513,
      "learning_rate": 1.371169014535312e-06,
      "loss": 0.5434,
      "step": 3121
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.846938334630315,
      "learning_rate": 1.3707857997034688e-06,
      "loss": 0.5968,
      "step": 3122
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8544331089889083,
      "learning_rate": 1.3704025217306665e-06,
      "loss": 0.6514,
      "step": 3123
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8370021669868832,
      "learning_rate": 1.3700191806821736e-06,
      "loss": 0.6372,
      "step": 3124
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7986911461133066,
      "learning_rate": 1.3696357766232688e-06,
      "loss": 0.6058,
      "step": 3125
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6689093839745374,
      "learning_rate": 1.3692523096192426e-06,
      "loss": 0.5592,
      "step": 3126
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0717142386647223,
      "learning_rate": 1.3688687797353947e-06,
      "loss": 0.6302,
      "step": 3127
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.747197193819771,
      "learning_rate": 1.3684851870370361e-06,
      "loss": 0.6016,
      "step": 3128
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.877188497479501,
      "learning_rate": 1.3681015315894894e-06,
      "loss": 0.6387,
      "step": 3129
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9247276902703885,
      "learning_rate": 1.3677178134580864e-06,
      "loss": 0.6822,
      "step": 3130
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6887574677773939,
      "learning_rate": 1.3673340327081708e-06,
      "loss": 0.5489,
      "step": 3131
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8526943693413179,
      "learning_rate": 1.3669501894050965e-06,
      "loss": 0.6922,
      "step": 3132
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8304518035434634,
      "learning_rate": 1.3665662836142276e-06,
      "loss": 0.5929,
      "step": 3133
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6331114649829831,
      "learning_rate": 1.3661823154009395e-06,
      "loss": 0.5417,
      "step": 3134
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8502887746345837,
      "learning_rate": 1.365798284830618e-06,
      "loss": 0.6374,
      "step": 3135
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8642413734434841,
      "learning_rate": 1.3654141919686595e-06,
      "loss": 0.6276,
      "step": 3136
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7091491060224494,
      "learning_rate": 1.3650300368804708e-06,
      "loss": 0.5453,
      "step": 3137
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9307605324887979,
      "learning_rate": 1.3646458196314695e-06,
      "loss": 0.623,
      "step": 3138
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6415044754138837,
      "learning_rate": 1.3642615402870843e-06,
      "loss": 0.5221,
      "step": 3139
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.83617105723574,
      "learning_rate": 1.3638771989127539e-06,
      "loss": 0.5845,
      "step": 3140
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7822973256829734,
      "learning_rate": 1.3634927955739268e-06,
      "loss": 0.5476,
      "step": 3141
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6476854661631996,
      "learning_rate": 1.3631083303360638e-06,
      "loss": 0.5269,
      "step": 3142
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8666166236841678,
      "learning_rate": 1.3627238032646344e-06,
      "loss": 0.6059,
      "step": 3143
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6463605818779308,
      "learning_rate": 1.3623392144251208e-06,
      "loss": 0.4798,
      "step": 3144
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7624725078840393,
      "learning_rate": 1.3619545638830135e-06,
      "loss": 0.5876,
      "step": 3145
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6723072603347222,
      "learning_rate": 1.3615698517038147e-06,
      "loss": 0.5645,
      "step": 3146
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.727938924759253,
      "learning_rate": 1.3611850779530372e-06,
      "loss": 0.5521,
      "step": 3147
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6754738633814731,
      "learning_rate": 1.3608002426962037e-06,
      "loss": 0.5658,
      "step": 3148
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.978854654804893,
      "learning_rate": 1.3604153459988477e-06,
      "loss": 0.6582,
      "step": 3149
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6935635550638868,
      "learning_rate": 1.360030387926513e-06,
      "loss": 0.5103,
      "step": 3150
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.807264348455229,
      "learning_rate": 1.359645368544754e-06,
      "loss": 0.6146,
      "step": 3151
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8422202752377788,
      "learning_rate": 1.3592602879191355e-06,
      "loss": 0.6492,
      "step": 3152
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6273807932752843,
      "learning_rate": 1.3588751461152326e-06,
      "loss": 0.5396,
      "step": 3153
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8055941314827889,
      "learning_rate": 1.3584899431986312e-06,
      "loss": 0.5858,
      "step": 3154
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7446448759696299,
      "learning_rate": 1.3581046792349268e-06,
      "loss": 0.5673,
      "step": 3155
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7601649806980579,
      "learning_rate": 1.3577193542897268e-06,
      "loss": 0.6,
      "step": 3156
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7409455094714247,
      "learning_rate": 1.357333968428647e-06,
      "loss": 0.5341,
      "step": 3157
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.81195555360734,
      "learning_rate": 1.3569485217173154e-06,
      "loss": 0.5782,
      "step": 3158
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4578402339710572,
      "learning_rate": 1.3565630142213688e-06,
      "loss": 0.5474,
      "step": 3159
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7311758558170522,
      "learning_rate": 1.3561774460064558e-06,
      "loss": 0.5755,
      "step": 3160
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6113590002475466,
      "learning_rate": 1.3557918171382346e-06,
      "loss": 0.5995,
      "step": 3161
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7257949969455444,
      "learning_rate": 1.3554061276823734e-06,
      "loss": 0.5561,
      "step": 3162
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7246678457553412,
      "learning_rate": 1.355020377704551e-06,
      "loss": 0.5298,
      "step": 3163
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7895414637291118,
      "learning_rate": 1.354634567270457e-06,
      "loss": 0.5584,
      "step": 3164
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8923554945155004,
      "learning_rate": 1.3542486964457908e-06,
      "loss": 0.6145,
      "step": 3165
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6591103543333665,
      "learning_rate": 1.3538627652962623e-06,
      "loss": 0.4964,
      "step": 3166
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6018531932359873,
      "learning_rate": 1.3534767738875912e-06,
      "loss": 0.5199,
      "step": 3167
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8488573210629706,
      "learning_rate": 1.3530907222855082e-06,
      "loss": 0.6438,
      "step": 3168
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7989660437156402,
      "learning_rate": 1.3527046105557533e-06,
      "loss": 0.5447,
      "step": 3169
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8978216842932966,
      "learning_rate": 1.3523184387640778e-06,
      "loss": 0.5685,
      "step": 3170
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.809185518203035,
      "learning_rate": 1.3519322069762426e-06,
      "loss": 0.6423,
      "step": 3171
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7597523496026483,
      "learning_rate": 1.3515459152580188e-06,
      "loss": 0.5692,
      "step": 3172
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.770231835490272,
      "learning_rate": 1.3511595636751884e-06,
      "loss": 0.6442,
      "step": 3173
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8101266705127789,
      "learning_rate": 1.3507731522935423e-06,
      "loss": 0.5725,
      "step": 3174
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6753879915997955,
      "learning_rate": 1.350386681178883e-06,
      "loss": 0.5731,
      "step": 3175
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8328720995548307,
      "learning_rate": 1.3500001503970222e-06,
      "loss": 0.6846,
      "step": 3176
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9374435250492774,
      "learning_rate": 1.3496135600137819e-06,
      "loss": 0.6713,
      "step": 3177
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7788667464014785,
      "learning_rate": 1.3492269100949947e-06,
      "loss": 0.5946,
      "step": 3178
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8083630862370003,
      "learning_rate": 1.3488402007065028e-06,
      "loss": 0.5938,
      "step": 3179
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7373585914822439,
      "learning_rate": 1.3484534319141589e-06,
      "loss": 0.603,
      "step": 3180
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6743197913779339,
      "learning_rate": 1.3480666037838258e-06,
      "loss": 0.5753,
      "step": 3181
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7392879429291906,
      "learning_rate": 1.3476797163813763e-06,
      "loss": 0.5936,
      "step": 3182
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6553076300650169,
      "learning_rate": 1.347292769772693e-06,
      "loss": 0.5536,
      "step": 3183
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6875834155421443,
      "learning_rate": 1.3469057640236692e-06,
      "loss": 0.4904,
      "step": 3184
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9341915234218361,
      "learning_rate": 1.3465186992002076e-06,
      "loss": 0.6325,
      "step": 3185
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7687823625064975,
      "learning_rate": 1.3461315753682217e-06,
      "loss": 0.5817,
      "step": 3186
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6790283646174142,
      "learning_rate": 1.3457443925936345e-06,
      "loss": 0.5369,
      "step": 3187
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8637083633029924,
      "learning_rate": 1.345357150942379e-06,
      "loss": 0.6311,
      "step": 3188
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.04125083334269,
      "learning_rate": 1.3449698504803987e-06,
      "loss": 0.662,
      "step": 3189
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7817874265299143,
      "learning_rate": 1.344582491273647e-06,
      "loss": 0.5921,
      "step": 3190
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9366924946113052,
      "learning_rate": 1.3441950733880868e-06,
      "loss": 0.6464,
      "step": 3191
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9837820954024494,
      "learning_rate": 1.3438075968896912e-06,
      "loss": 0.6376,
      "step": 3192
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8581081309300206,
      "learning_rate": 1.3434200618444437e-06,
      "loss": 0.606,
      "step": 3193
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7747160596833594,
      "learning_rate": 1.3430324683183374e-06,
      "loss": 0.6469,
      "step": 3194
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9041638445708565,
      "learning_rate": 1.3426448163773754e-06,
      "loss": 0.5664,
      "step": 3195
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9632457733948682,
      "learning_rate": 1.342257106087571e-06,
      "loss": 0.6173,
      "step": 3196
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5577196273315523,
      "learning_rate": 1.3418693375149468e-06,
      "loss": 0.4592,
      "step": 3197
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0150767730592254,
      "learning_rate": 1.3414815107255356e-06,
      "loss": 0.6254,
      "step": 3198
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7086105708653562,
      "learning_rate": 1.3410936257853813e-06,
      "loss": 0.5075,
      "step": 3199
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9861404915567523,
      "learning_rate": 1.3407056827605358e-06,
      "loss": 0.6755,
      "step": 3200
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8312200664055616,
      "learning_rate": 1.340317681717062e-06,
      "loss": 0.5709,
      "step": 3201
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8509505972842122,
      "learning_rate": 1.3399296227210325e-06,
      "loss": 0.6523,
      "step": 3202
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8269904426955954,
      "learning_rate": 1.3395415058385295e-06,
      "loss": 0.5561,
      "step": 3203
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6520606831182475,
      "learning_rate": 1.3391533311356456e-06,
      "loss": 0.5512,
      "step": 3204
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6815837590776441,
      "learning_rate": 1.338765098678482e-06,
      "loss": 0.5671,
      "step": 3205
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7026734045243225,
      "learning_rate": 1.3383768085331519e-06,
      "loss": 0.5241,
      "step": 3206
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5779496527062108,
      "learning_rate": 1.3379884607657762e-06,
      "loss": 0.5244,
      "step": 3207
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7166224513347719,
      "learning_rate": 1.337600055442487e-06,
      "loss": 0.5523,
      "step": 3208
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6868285517960843,
      "learning_rate": 1.3372115926294255e-06,
      "loss": 0.5573,
      "step": 3209
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.828572894529487,
      "learning_rate": 1.336823072392743e-06,
      "loss": 0.6514,
      "step": 3210
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7674812262136249,
      "learning_rate": 1.3364344947986e-06,
      "loss": 0.5723,
      "step": 3211
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7343969418610498,
      "learning_rate": 1.3360458599131674e-06,
      "loss": 0.6015,
      "step": 3212
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8302521786110378,
      "learning_rate": 1.335657167802626e-06,
      "loss": 0.6106,
      "step": 3213
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8567005059301759,
      "learning_rate": 1.3352684185331653e-06,
      "loss": 0.6102,
      "step": 3214
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8547885192745827,
      "learning_rate": 1.334879612170986e-06,
      "loss": 0.6209,
      "step": 3215
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8263934551928097,
      "learning_rate": 1.3344907487822976e-06,
      "loss": 0.6231,
      "step": 3216
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8144031028232864,
      "learning_rate": 1.3341018284333191e-06,
      "loss": 0.5717,
      "step": 3217
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1071137090028118,
      "learning_rate": 1.3337128511902802e-06,
      "loss": 0.6306,
      "step": 3218
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9926357057765708,
      "learning_rate": 1.3333238171194193e-06,
      "loss": 0.6937,
      "step": 3219
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7061901927485962,
      "learning_rate": 1.3329347262869849e-06,
      "loss": 0.5681,
      "step": 3220
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7621318450675911,
      "learning_rate": 1.332545578759235e-06,
      "loss": 0.5232,
      "step": 3221
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0593868469669958,
      "learning_rate": 1.3321563746024374e-06,
      "loss": 0.6153,
      "step": 3222
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6190417366021056,
      "learning_rate": 1.33176711388287e-06,
      "loss": 0.5182,
      "step": 3223
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.682193214251899,
      "learning_rate": 1.3313777966668189e-06,
      "loss": 0.5336,
      "step": 3224
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8293320959855693,
      "learning_rate": 1.3309884230205815e-06,
      "loss": 0.6275,
      "step": 3225
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7229915027538876,
      "learning_rate": 1.3305989930104637e-06,
      "loss": 0.5411,
      "step": 3226
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.825661076950281,
      "learning_rate": 1.330209506702782e-06,
      "loss": 0.5892,
      "step": 3227
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7284466195676561,
      "learning_rate": 1.3298199641638608e-06,
      "loss": 0.5836,
      "step": 3228
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9874980091651275,
      "learning_rate": 1.3294303654600362e-06,
      "loss": 0.6325,
      "step": 3229
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6350749365439009,
      "learning_rate": 1.3290407106576522e-06,
      "loss": 0.5867,
      "step": 3230
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7394017616640288,
      "learning_rate": 1.328650999823063e-06,
      "loss": 0.5954,
      "step": 3231
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7490331417970126,
      "learning_rate": 1.3282612330226327e-06,
      "loss": 0.5265,
      "step": 3232
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8923344869383985,
      "learning_rate": 1.327871410322734e-06,
      "loss": 0.6457,
      "step": 3233
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6437171311499403,
      "learning_rate": 1.3274815317897496e-06,
      "loss": 0.5205,
      "step": 3234
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6408675055178553,
      "learning_rate": 1.3270915974900726e-06,
      "loss": 0.5797,
      "step": 3235
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8419146248207814,
      "learning_rate": 1.3267016074901036e-06,
      "loss": 0.6205,
      "step": 3236
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9393184627688516,
      "learning_rate": 1.3263115618562548e-06,
      "loss": 0.6865,
      "step": 3237
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8540253360010526,
      "learning_rate": 1.3259214606549462e-06,
      "loss": 0.6216,
      "step": 3238
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6238602301338959,
      "learning_rate": 1.3255313039526086e-06,
      "loss": 0.5279,
      "step": 3239
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8825220536807993,
      "learning_rate": 1.3251410918156812e-06,
      "loss": 0.6451,
      "step": 3240
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6972580561275145,
      "learning_rate": 1.3247508243106128e-06,
      "loss": 0.5618,
      "step": 3241
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8725191295321476,
      "learning_rate": 1.3243605015038626e-06,
      "loss": 0.5624,
      "step": 3242
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9364130124977239,
      "learning_rate": 1.3239701234618982e-06,
      "loss": 0.6032,
      "step": 3243
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7939521178433963,
      "learning_rate": 1.3235796902511967e-06,
      "loss": 0.6004,
      "step": 3244
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.84684225317031,
      "learning_rate": 1.3231892019382448e-06,
      "loss": 0.6399,
      "step": 3245
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6953278979413027,
      "learning_rate": 1.3227986585895392e-06,
      "loss": 0.5439,
      "step": 3246
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6257395945902634,
      "learning_rate": 1.3224080602715849e-06,
      "loss": 0.5885,
      "step": 3247
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.794537493401352,
      "learning_rate": 1.3220174070508965e-06,
      "loss": 0.5203,
      "step": 3248
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.875391361082337,
      "learning_rate": 1.3216266989939985e-06,
      "loss": 0.6744,
      "step": 3249
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.707289779302033,
      "learning_rate": 1.3212359361674244e-06,
      "loss": 0.5614,
      "step": 3250
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5675252365456321,
      "learning_rate": 1.320845118637717e-06,
      "loss": 0.5006,
      "step": 3251
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7992583593614434,
      "learning_rate": 1.3204542464714284e-06,
      "loss": 0.632,
      "step": 3252
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9768740470911037,
      "learning_rate": 1.32006331973512e-06,
      "loss": 0.5968,
      "step": 3253
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.835988949470314,
      "learning_rate": 1.3196723384953625e-06,
      "loss": 0.5635,
      "step": 3254
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6600769389206114,
      "learning_rate": 1.3192813028187365e-06,
      "loss": 0.5329,
      "step": 3255
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7223176993339757,
      "learning_rate": 1.3188902127718307e-06,
      "loss": 0.5909,
      "step": 3256
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.0284075119527447,
      "learning_rate": 1.3184990684212436e-06,
      "loss": 0.68,
      "step": 3257
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8717203464863248,
      "learning_rate": 1.3181078698335835e-06,
      "loss": 0.6254,
      "step": 3258
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6887856399857266,
      "learning_rate": 1.3177166170754672e-06,
      "loss": 0.5874,
      "step": 3259
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7372800398653568,
      "learning_rate": 1.317325310213521e-06,
      "loss": 0.5696,
      "step": 3260
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9353425292776814,
      "learning_rate": 1.3169339493143804e-06,
      "loss": 0.6404,
      "step": 3261
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6599754107419443,
      "learning_rate": 1.3165425344446897e-06,
      "loss": 0.5374,
      "step": 3262
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8273748784814803,
      "learning_rate": 1.3161510656711032e-06,
      "loss": 0.5818,
      "step": 3263
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.639812927435432,
      "learning_rate": 1.3157595430602842e-06,
      "loss": 0.5337,
      "step": 3264
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9335668778257221,
      "learning_rate": 1.315367966678904e-06,
      "loss": 0.6793,
      "step": 3265
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7864941229991357,
      "learning_rate": 1.314976336593645e-06,
      "loss": 0.5976,
      "step": 3266
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7610356316307099,
      "learning_rate": 1.3145846528711972e-06,
      "loss": 0.5805,
      "step": 3267
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9453438087375742,
      "learning_rate": 1.3141929155782606e-06,
      "loss": 0.6748,
      "step": 3268
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6789994110028671,
      "learning_rate": 1.3138011247815432e-06,
      "loss": 0.5562,
      "step": 3269
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7401736692214915,
      "learning_rate": 1.3134092805477635e-06,
      "loss": 0.5712,
      "step": 3270
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6694640474481076,
      "learning_rate": 1.3130173829436486e-06,
      "loss": 0.5625,
      "step": 3271
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8352922375837759,
      "learning_rate": 1.3126254320359342e-06,
      "loss": 0.6711,
      "step": 3272
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7444006361870095,
      "learning_rate": 1.3122334278913657e-06,
      "loss": 0.5561,
      "step": 3273
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6154834963246166,
      "learning_rate": 1.3118413705766973e-06,
      "loss": 0.5421,
      "step": 3274
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8050774339312807,
      "learning_rate": 1.311449260158692e-06,
      "loss": 0.6038,
      "step": 3275
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6343657576946908,
      "learning_rate": 1.3110570967041229e-06,
      "loss": 0.5213,
      "step": 3276
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6937406759934448,
      "learning_rate": 1.3106648802797704e-06,
      "loss": 0.5782,
      "step": 3277
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0076431919405162,
      "learning_rate": 1.3102726109524254e-06,
      "loss": 0.5861,
      "step": 3278
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6563030106577998,
      "learning_rate": 1.3098802887888871e-06,
      "loss": 0.4955,
      "step": 3279
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6538920608749379,
      "learning_rate": 1.3094879138559639e-06,
      "loss": 0.5663,
      "step": 3280
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.652825857247094,
      "learning_rate": 1.3090954862204732e-06,
      "loss": 0.5542,
      "step": 3281
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8587935110954399,
      "learning_rate": 1.3087030059492415e-06,
      "loss": 0.5759,
      "step": 3282
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8457222209373402,
      "learning_rate": 1.3083104731091036e-06,
      "loss": 0.6263,
      "step": 3283
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8517683680205455,
      "learning_rate": 1.3079178877669043e-06,
      "loss": 0.6748,
      "step": 3284
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8418843932122059,
      "learning_rate": 1.3075252499894962e-06,
      "loss": 0.574,
      "step": 3285
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6956242145074383,
      "learning_rate": 1.3071325598437422e-06,
      "loss": 0.5483,
      "step": 3286
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8422642122805161,
      "learning_rate": 1.3067398173965129e-06,
      "loss": 0.6222,
      "step": 3287
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7163594974871501,
      "learning_rate": 1.3063470227146883e-06,
      "loss": 0.5916,
      "step": 3288
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8523255636621831,
      "learning_rate": 1.3059541758651573e-06,
      "loss": 0.6061,
      "step": 3289
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8168275933198325,
      "learning_rate": 1.3055612769148174e-06,
      "loss": 0.6589,
      "step": 3290
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6877761232641633,
      "learning_rate": 1.3051683259305754e-06,
      "loss": 0.5294,
      "step": 3291
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6688557810007784,
      "learning_rate": 1.3047753229793467e-06,
      "loss": 0.5574,
      "step": 3292
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6395487869905357,
      "learning_rate": 1.3043822681280557e-06,
      "loss": 0.525,
      "step": 3293
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7653283145590872,
      "learning_rate": 1.3039891614436353e-06,
      "loss": 0.5012,
      "step": 3294
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.878619747254257,
      "learning_rate": 1.3035960029930278e-06,
      "loss": 0.6576,
      "step": 3295
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6534335364800078,
      "learning_rate": 1.3032027928431836e-06,
      "loss": 0.5299,
      "step": 3296
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8724891374183859,
      "learning_rate": 1.3028095310610628e-06,
      "loss": 0.6419,
      "step": 3297
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8839956409009281,
      "learning_rate": 1.3024162177136334e-06,
      "loss": 0.6162,
      "step": 3298
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7734297232894441,
      "learning_rate": 1.3020228528678728e-06,
      "loss": 0.5555,
      "step": 3299
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7711563165219623,
      "learning_rate": 1.3016294365907667e-06,
      "loss": 0.5824,
      "step": 3300
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7771673801541849,
      "learning_rate": 1.3012359689493098e-06,
      "loss": 0.6379,
      "step": 3301
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7285088780928793,
      "learning_rate": 1.3008424500105057e-06,
      "loss": 0.5607,
      "step": 3302
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6257770971449789,
      "learning_rate": 1.3004488798413665e-06,
      "loss": 0.4782,
      "step": 3303
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8799810489433684,
      "learning_rate": 1.3000552585089132e-06,
      "loss": 0.7265,
      "step": 3304
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6606651364879231,
      "learning_rate": 1.2996615860801755e-06,
      "loss": 0.5383,
      "step": 3305
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6590870983392961,
      "learning_rate": 1.2992678626221913e-06,
      "loss": 0.5474,
      "step": 3306
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9381068943661567,
      "learning_rate": 1.2988740882020078e-06,
      "loss": 0.6289,
      "step": 3307
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8984664136537935,
      "learning_rate": 1.2984802628866806e-06,
      "loss": 0.6477,
      "step": 3308
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6860120465549272,
      "learning_rate": 1.2980863867432744e-06,
      "loss": 0.5146,
      "step": 3309
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6714889889524687,
      "learning_rate": 1.2976924598388622e-06,
      "loss": 0.4868,
      "step": 3310
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7798243901241998,
      "learning_rate": 1.297298482240525e-06,
      "loss": 0.6203,
      "step": 3311
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6594567957805334,
      "learning_rate": 1.2969044540153538e-06,
      "loss": 0.5382,
      "step": 3312
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6834619348970734,
      "learning_rate": 1.296510375230447e-06,
      "loss": 0.5571,
      "step": 3313
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7238979531337046,
      "learning_rate": 1.2961162459529124e-06,
      "loss": 0.5606,
      "step": 3314
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5727852685586161,
      "learning_rate": 1.295722066249866e-06,
      "loss": 0.4845,
      "step": 3315
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5646015288128052,
      "learning_rate": 1.2953278361884327e-06,
      "loss": 0.5245,
      "step": 3316
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1400377531401094,
      "learning_rate": 1.2949335558357455e-06,
      "loss": 0.6418,
      "step": 3317
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9525625221539132,
      "learning_rate": 1.2945392252589464e-06,
      "loss": 0.6791,
      "step": 3318
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0096711081335639,
      "learning_rate": 1.2941448445251857e-06,
      "loss": 0.638,
      "step": 3319
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7258459002494552,
      "learning_rate": 1.2937504137016222e-06,
      "loss": 0.5738,
      "step": 3320
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7556854520973495,
      "learning_rate": 1.2933559328554239e-06,
      "loss": 0.6384,
      "step": 3321
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8801951243075732,
      "learning_rate": 1.292961402053766e-06,
      "loss": 0.6369,
      "step": 3322
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7045401739039692,
      "learning_rate": 1.2925668213638334e-06,
      "loss": 0.519,
      "step": 3323
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7538596239881666,
      "learning_rate": 1.292172190852819e-06,
      "loss": 0.5844,
      "step": 3324
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8846163323567655,
      "learning_rate": 1.2917775105879245e-06,
      "loss": 0.6177,
      "step": 3325
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7303755344174485,
      "learning_rate": 1.2913827806363598e-06,
      "loss": 0.586,
      "step": 3326
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8934896893571518,
      "learning_rate": 1.2909880010653429e-06,
      "loss": 0.6802,
      "step": 3327
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7102038836386658,
      "learning_rate": 1.290593171942101e-06,
      "loss": 0.537,
      "step": 3328
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8312001312753735,
      "learning_rate": 1.2901982933338693e-06,
      "loss": 0.6384,
      "step": 3329
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7529674224506352,
      "learning_rate": 1.2898033653078918e-06,
      "loss": 0.5845,
      "step": 3330
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1005330143752516,
      "learning_rate": 1.28940838793142e-06,
      "loss": 0.6755,
      "step": 3331
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7747228312033063,
      "learning_rate": 1.289013361271715e-06,
      "loss": 0.525,
      "step": 3332
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8844564195390425,
      "learning_rate": 1.2886182853960453e-06,
      "loss": 0.5565,
      "step": 3333
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7812192245778813,
      "learning_rate": 1.2882231603716882e-06,
      "loss": 0.5603,
      "step": 3334
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8699741417611265,
      "learning_rate": 1.2878279862659299e-06,
      "loss": 0.6544,
      "step": 3335
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6618041237046565,
      "learning_rate": 1.287432763146064e-06,
      "loss": 0.5897,
      "step": 3336
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9823142683133963,
      "learning_rate": 1.2870374910793931e-06,
      "loss": 0.6819,
      "step": 3337
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6236889047936274,
      "learning_rate": 1.2866421701332277e-06,
      "loss": 0.5178,
      "step": 3338
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6613926178290512,
      "learning_rate": 1.2862468003748868e-06,
      "loss": 0.5339,
      "step": 3339
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.655019620224501,
      "learning_rate": 1.285851381871698e-06,
      "loss": 0.5688,
      "step": 3340
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.64133204393391,
      "learning_rate": 1.285455914690997e-06,
      "loss": 0.5617,
      "step": 3341
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6712138773433193,
      "learning_rate": 1.2850603989001274e-06,
      "loss": 0.5488,
      "step": 3342
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8116479914918476,
      "learning_rate": 1.2846648345664417e-06,
      "loss": 0.5971,
      "step": 3343
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6145426345684483,
      "learning_rate": 1.2842692217573004e-06,
      "loss": 0.5668,
      "step": 3344
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8499326715066393,
      "learning_rate": 1.283873560540072e-06,
      "loss": 0.6602,
      "step": 3345
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6574847498141632,
      "learning_rate": 1.2834778509821336e-06,
      "loss": 0.5525,
      "step": 3346
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.813409694800284,
      "learning_rate": 1.2830820931508703e-06,
      "loss": 0.6433,
      "step": 3347
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6303618460433006,
      "learning_rate": 1.2826862871136758e-06,
      "loss": 0.5202,
      "step": 3348
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8732138105022681,
      "learning_rate": 1.2822904329379518e-06,
      "loss": 0.6628,
      "step": 3349
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6681841802266762,
      "learning_rate": 1.2818945306911078e-06,
      "loss": 0.5268,
      "step": 3350
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8421769768090649,
      "learning_rate": 1.2814985804405618e-06,
      "loss": 0.6236,
      "step": 3351
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.7755706909416378,
      "learning_rate": 1.2811025822537403e-06,
      "loss": 0.6961,
      "step": 3352
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8632243224795411,
      "learning_rate": 1.2807065361980775e-06,
      "loss": 0.6615,
      "step": 3353
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8390964227970069,
      "learning_rate": 1.2803104423410163e-06,
      "loss": 0.6404,
      "step": 3354
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6815920601011772,
      "learning_rate": 1.2799143007500066e-06,
      "loss": 0.5491,
      "step": 3355
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6282068161711633,
      "learning_rate": 1.2795181114925077e-06,
      "loss": 0.5493,
      "step": 3356
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6853154986537563,
      "learning_rate": 1.2791218746359865e-06,
      "loss": 0.5722,
      "step": 3357
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9280089992684355,
      "learning_rate": 1.2787255902479177e-06,
      "loss": 0.6027,
      "step": 3358
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6067628356687095,
      "learning_rate": 1.278329258395785e-06,
      "loss": 0.5114,
      "step": 3359
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9800808081891872,
      "learning_rate": 1.2779328791470788e-06,
      "loss": 0.6647,
      "step": 3360
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.710059558208818,
      "learning_rate": 1.2775364525692992e-06,
      "loss": 0.5244,
      "step": 3361
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7042081052339655,
      "learning_rate": 1.277139978729953e-06,
      "loss": 0.5481,
      "step": 3362
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7912707458068898,
      "learning_rate": 1.2767434576965554e-06,
      "loss": 0.5789,
      "step": 3363
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5971891433469877,
      "learning_rate": 1.2763468895366302e-06,
      "loss": 0.5297,
      "step": 3364
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7322991169431246,
      "learning_rate": 1.275950274317709e-06,
      "loss": 0.5311,
      "step": 3365
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6092525470296591,
      "learning_rate": 1.2755536121073303e-06,
      "loss": 0.4881,
      "step": 3366
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9742116412972592,
      "learning_rate": 1.2751569029730425e-06,
      "loss": 0.6631,
      "step": 3367
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7695403692536669,
      "learning_rate": 1.2747601469824004e-06,
      "loss": 0.5214,
      "step": 3368
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9341757611024345,
      "learning_rate": 1.274363344202968e-06,
      "loss": 0.5945,
      "step": 3369
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7867373247523365,
      "learning_rate": 1.273966494702316e-06,
      "loss": 0.5663,
      "step": 3370
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8870320601230012,
      "learning_rate": 1.2735695985480246e-06,
      "loss": 0.6485,
      "step": 3371
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8316187401911879,
      "learning_rate": 1.2731726558076804e-06,
      "loss": 0.5677,
      "step": 3372
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9058338470846122,
      "learning_rate": 1.272775666548879e-06,
      "loss": 0.641,
      "step": 3373
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8170332040092044,
      "learning_rate": 1.272378630839223e-06,
      "loss": 0.6372,
      "step": 3374
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.693806391769718,
      "learning_rate": 1.271981548746324e-06,
      "loss": 0.6364,
      "step": 3375
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6577175872382137,
      "learning_rate": 1.2715844203378006e-06,
      "loss": 0.5645,
      "step": 3376
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7509998844162311,
      "learning_rate": 1.2711872456812798e-06,
      "loss": 0.5516,
      "step": 3377
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6461950249872955,
      "learning_rate": 1.2707900248443963e-06,
      "loss": 0.528,
      "step": 3378
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7271464198795982,
      "learning_rate": 1.270392757894793e-06,
      "loss": 0.546,
      "step": 3379
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8273230341103045,
      "learning_rate": 1.2699954449001194e-06,
      "loss": 0.666,
      "step": 3380
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6995459867274946,
      "learning_rate": 1.269598085928034e-06,
      "loss": 0.5265,
      "step": 3381
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8456678224394687,
      "learning_rate": 1.2692006810462038e-06,
      "loss": 0.6585,
      "step": 3382
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7565589148455836,
      "learning_rate": 1.2688032303223023e-06,
      "loss": 0.6431,
      "step": 3383
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1385869516841591,
      "learning_rate": 1.2684057338240107e-06,
      "loss": 0.6291,
      "step": 3384
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5987028065304266,
      "learning_rate": 1.268008191619019e-06,
      "loss": 0.5064,
      "step": 3385
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7164745984306417,
      "learning_rate": 1.2676106037750245e-06,
      "loss": 0.5608,
      "step": 3386
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6261083118948386,
      "learning_rate": 1.2672129703597319e-06,
      "loss": 0.5081,
      "step": 3387
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8388443636536251,
      "learning_rate": 1.2668152914408544e-06,
      "loss": 0.5922,
      "step": 3388
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0239046358596777,
      "learning_rate": 1.2664175670861123e-06,
      "loss": 0.5991,
      "step": 3389
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6922970028050641,
      "learning_rate": 1.2660197973632342e-06,
      "loss": 0.5644,
      "step": 3390
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9427557757947126,
      "learning_rate": 1.265621982339956e-06,
      "loss": 0.5483,
      "step": 3391
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9088472495794239,
      "learning_rate": 1.2652241220840212e-06,
      "loss": 0.623,
      "step": 3392
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8145879286954059,
      "learning_rate": 1.2648262166631817e-06,
      "loss": 0.5934,
      "step": 3393
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8177304800390499,
      "learning_rate": 1.264428266145196e-06,
      "loss": 0.5675,
      "step": 3394
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9238719271570184,
      "learning_rate": 1.2640302705978316e-06,
      "loss": 0.6824,
      "step": 3395
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1621413569030212,
      "learning_rate": 1.263632230088863e-06,
      "loss": 0.5993,
      "step": 3396
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8158202483390974,
      "learning_rate": 1.263234144686072e-06,
      "loss": 0.6005,
      "step": 3397
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9803404650474931,
      "learning_rate": 1.2628360144572485e-06,
      "loss": 0.6591,
      "step": 3398
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5899113826155633,
      "learning_rate": 1.26243783947019e-06,
      "loss": 0.502,
      "step": 3399
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1623375560181854,
      "learning_rate": 1.2620396197927015e-06,
      "loss": 0.6717,
      "step": 3400
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7003538770088509,
      "learning_rate": 1.2616413554925955e-06,
      "loss": 0.5325,
      "step": 3401
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8885872131233911,
      "learning_rate": 1.2612430466376927e-06,
      "loss": 0.6773,
      "step": 3402
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7284611836028604,
      "learning_rate": 1.2608446932958204e-06,
      "loss": 0.5384,
      "step": 3403
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.448395725539661,
      "learning_rate": 1.2604462955348142e-06,
      "loss": 0.6121,
      "step": 3404
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9604914172554712,
      "learning_rate": 1.2600478534225177e-06,
      "loss": 0.5951,
      "step": 3405
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.006396651048926,
      "learning_rate": 1.2596493670267804e-06,
      "loss": 0.6359,
      "step": 3406
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7283109912120818,
      "learning_rate": 1.259250836415461e-06,
      "loss": 0.5391,
      "step": 3407
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8738993678837599,
      "learning_rate": 1.258852261656425e-06,
      "loss": 0.6537,
      "step": 3408
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8292802022782253,
      "learning_rate": 1.2584536428175457e-06,
      "loss": 0.6267,
      "step": 3409
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8139289064334948,
      "learning_rate": 1.2580549799667034e-06,
      "loss": 0.5526,
      "step": 3410
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3019620613581468,
      "learning_rate": 1.2576562731717867e-06,
      "loss": 0.6566,
      "step": 3411
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8960845023587747,
      "learning_rate": 1.2572575225006908e-06,
      "loss": 0.6441,
      "step": 3412
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6769975101071087,
      "learning_rate": 1.2568587280213185e-06,
      "loss": 0.5693,
      "step": 3413
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7355051910269776,
      "learning_rate": 1.2564598898015812e-06,
      "loss": 0.5069,
      "step": 3414
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8015446960604529,
      "learning_rate": 1.2560610079093964e-06,
      "loss": 0.5668,
      "step": 3415
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8208171487654028,
      "learning_rate": 1.2556620824126895e-06,
      "loss": 0.5877,
      "step": 3416
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5899352498327027,
      "learning_rate": 1.2552631133793937e-06,
      "loss": 0.5108,
      "step": 3417
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0333316429996067,
      "learning_rate": 1.2548641008774487e-06,
      "loss": 0.5889,
      "step": 3418
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.923566005586143,
      "learning_rate": 1.2544650449748026e-06,
      "loss": 0.6344,
      "step": 3419
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7676206718343277,
      "learning_rate": 1.2540659457394102e-06,
      "loss": 0.5556,
      "step": 3420
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8219237885679513,
      "learning_rate": 1.2536668032392344e-06,
      "loss": 0.6028,
      "step": 3421
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8959544582212658,
      "learning_rate": 1.2532676175422443e-06,
      "loss": 0.5876,
      "step": 3422
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8011427328712676,
      "learning_rate": 1.2528683887164175e-06,
      "loss": 0.6148,
      "step": 3423
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6699374892247882,
      "learning_rate": 1.2524691168297385e-06,
      "loss": 0.5409,
      "step": 3424
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8616310503183756,
      "learning_rate": 1.252069801950199e-06,
      "loss": 0.5899,
      "step": 3425
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6242208496552898,
      "learning_rate": 1.2516704441457989e-06,
      "loss": 0.5264,
      "step": 3426
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1189441273570602,
      "learning_rate": 1.2512710434845436e-06,
      "loss": 0.5859,
      "step": 3427
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6141265533412329,
      "learning_rate": 1.2508716000344472e-06,
      "loss": 0.5402,
      "step": 3428
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6698615109814212,
      "learning_rate": 1.250472113863531e-06,
      "loss": 0.5331,
      "step": 3429
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7540188602215551,
      "learning_rate": 1.2500725850398233e-06,
      "loss": 0.5849,
      "step": 3430
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7458633727259069,
      "learning_rate": 1.2496730136313596e-06,
      "loss": 0.604,
      "step": 3431
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.757186897604443,
      "learning_rate": 1.2492733997061826e-06,
      "loss": 0.5496,
      "step": 3432
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.605860575374263,
      "learning_rate": 1.2488737433323427e-06,
      "loss": 0.4774,
      "step": 3433
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8706353454028379,
      "learning_rate": 1.2484740445778968e-06,
      "loss": 0.6354,
      "step": 3434
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0812780453949584,
      "learning_rate": 1.24807430351091e-06,
      "loss": 0.6232,
      "step": 3435
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6570292955181182,
      "learning_rate": 1.2476745201994534e-06,
      "loss": 0.4746,
      "step": 3436
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6425051683417388,
      "learning_rate": 1.2472746947116064e-06,
      "loss": 0.6216,
      "step": 3437
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9994540936014995,
      "learning_rate": 1.246874827115455e-06,
      "loss": 0.6239,
      "step": 3438
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6760532823680406,
      "learning_rate": 1.2464749174790924e-06,
      "loss": 0.5912,
      "step": 3439
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8049445617718202,
      "learning_rate": 1.2460749658706194e-06,
      "loss": 0.6635,
      "step": 3440
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8227554407782018,
      "learning_rate": 1.245674972358143e-06,
      "loss": 0.603,
      "step": 3441
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6314904416614506,
      "learning_rate": 1.2452749370097784e-06,
      "loss": 0.5372,
      "step": 3442
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8620222120396919,
      "learning_rate": 1.2448748598936474e-06,
      "loss": 0.6668,
      "step": 3443
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6588789954602593,
      "learning_rate": 1.2444747410778786e-06,
      "loss": 0.5692,
      "step": 3444
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.626607037013441,
      "learning_rate": 1.2440745806306084e-06,
      "loss": 0.5008,
      "step": 3445
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7248826159058039,
      "learning_rate": 1.24367437861998e-06,
      "loss": 0.5571,
      "step": 3446
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8476510343427638,
      "learning_rate": 1.2432741351141433e-06,
      "loss": 0.637,
      "step": 3447
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8191762717670863,
      "learning_rate": 1.242873850181256e-06,
      "loss": 0.5802,
      "step": 3448
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7661319521650517,
      "learning_rate": 1.2424735238894817e-06,
      "loss": 0.623,
      "step": 3449
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7413009685244539,
      "learning_rate": 1.2420731563069929e-06,
      "loss": 0.5773,
      "step": 3450
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7389292123156826,
      "learning_rate": 1.2416727475019672e-06,
      "loss": 0.5446,
      "step": 3451
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8642685681826094,
      "learning_rate": 1.2412722975425907e-06,
      "loss": 0.6654,
      "step": 3452
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8070867740593191,
      "learning_rate": 1.240871806497055e-06,
      "loss": 0.55,
      "step": 3453
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6704846689688515,
      "learning_rate": 1.2404712744335606e-06,
      "loss": 0.5459,
      "step": 3454
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6647625967910236,
      "learning_rate": 1.2400707014203131e-06,
      "loss": 0.565,
      "step": 3455
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6628246977028461,
      "learning_rate": 1.2396700875255261e-06,
      "loss": 0.5507,
      "step": 3456
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6763865887843218,
      "learning_rate": 1.23926943281742e-06,
      "loss": 0.532,
      "step": 3457
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7450907036108734,
      "learning_rate": 1.2388687373642224e-06,
      "loss": 0.573,
      "step": 3458
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6748485167281576,
      "learning_rate": 1.238468001234167e-06,
      "loss": 0.4873,
      "step": 3459
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.841741974100605,
      "learning_rate": 1.2380672244954958e-06,
      "loss": 0.6526,
      "step": 3460
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7609501005344621,
      "learning_rate": 1.237666407216456e-06,
      "loss": 0.5615,
      "step": 3461
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.786562005395296,
      "learning_rate": 1.2372655494653034e-06,
      "loss": 0.5958,
      "step": 3462
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7064669514094032,
      "learning_rate": 1.2368646513102995e-06,
      "loss": 0.595,
      "step": 3463
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7480501674597492,
      "learning_rate": 1.2364637128197128e-06,
      "loss": 0.5844,
      "step": 3464
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7073759640954551,
      "learning_rate": 1.2360627340618195e-06,
      "loss": 0.5604,
      "step": 3465
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7983536201032089,
      "learning_rate": 1.2356617151049019e-06,
      "loss": 0.6597,
      "step": 3466
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.010896235137408,
      "learning_rate": 1.2352606560172493e-06,
      "loss": 0.6371,
      "step": 3467
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6820091353286809,
      "learning_rate": 1.2348595568671579e-06,
      "loss": 0.5269,
      "step": 3468
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8201409636545012,
      "learning_rate": 1.2344584177229308e-06,
      "loss": 0.6261,
      "step": 3469
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9449405762468769,
      "learning_rate": 1.2340572386528774e-06,
      "loss": 0.6991,
      "step": 3470
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.956736777566163,
      "learning_rate": 1.233656019725315e-06,
      "loss": 0.6395,
      "step": 3471
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6905920414599401,
      "learning_rate": 1.2332547610085667e-06,
      "loss": 0.5717,
      "step": 3472
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6399633310230656,
      "learning_rate": 1.2328534625709623e-06,
      "loss": 0.5305,
      "step": 3473
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7312166848146624,
      "learning_rate": 1.2324521244808392e-06,
      "loss": 0.5237,
      "step": 3474
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.74368641764576,
      "learning_rate": 1.232050746806541e-06,
      "loss": 0.5498,
      "step": 3475
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.868664332311871,
      "learning_rate": 1.2316493296164184e-06,
      "loss": 0.5361,
      "step": 3476
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8351169527651191,
      "learning_rate": 1.2312478729788279e-06,
      "loss": 0.6267,
      "step": 3477
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8820336230346343,
      "learning_rate": 1.2308463769621337e-06,
      "loss": 0.6849,
      "step": 3478
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6226316920184087,
      "learning_rate": 1.2304448416347064e-06,
      "loss": 0.5273,
      "step": 3479
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6916069414038876,
      "learning_rate": 1.2300432670649236e-06,
      "loss": 0.5444,
      "step": 3480
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8228558755845252,
      "learning_rate": 1.229641653321169e-06,
      "loss": 0.5367,
      "step": 3481
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6164059343360774,
      "learning_rate": 1.229240000471833e-06,
      "loss": 0.5208,
      "step": 3482
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6586283733931749,
      "learning_rate": 1.2288383085853134e-06,
      "loss": 0.5523,
      "step": 3483
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8085229269288404,
      "learning_rate": 1.2284365777300136e-06,
      "loss": 0.5216,
      "step": 3484
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.699436947228876,
      "learning_rate": 1.2280348079743448e-06,
      "loss": 0.5485,
      "step": 3485
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6590446660419724,
      "learning_rate": 1.2276329993867235e-06,
      "loss": 0.5542,
      "step": 3486
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7500688209600497,
      "learning_rate": 1.2272311520355741e-06,
      "loss": 0.5671,
      "step": 3487
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7564204262637355,
      "learning_rate": 1.2268292659893267e-06,
      "loss": 0.6206,
      "step": 3488
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0072123234702361,
      "learning_rate": 1.2264273413164183e-06,
      "loss": 0.6375,
      "step": 3489
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8200357213764622,
      "learning_rate": 1.2260253780852928e-06,
      "loss": 0.6346,
      "step": 3490
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.65039404510652,
      "learning_rate": 1.2256233763644e-06,
      "loss": 0.5703,
      "step": 3491
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7245000953603701,
      "learning_rate": 1.2252213362221963e-06,
      "loss": 0.633,
      "step": 3492
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6982241036020059,
      "learning_rate": 1.2248192577271458e-06,
      "loss": 0.5615,
      "step": 3493
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7150806297184724,
      "learning_rate": 1.224417140947718e-06,
      "loss": 0.5227,
      "step": 3494
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9414136688035221,
      "learning_rate": 1.224014985952389e-06,
      "loss": 0.6703,
      "step": 3495
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8782191815859718,
      "learning_rate": 1.2236127928096416e-06,
      "loss": 0.6325,
      "step": 3496
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7626170396020309,
      "learning_rate": 1.223210561587965e-06,
      "loss": 0.558,
      "step": 3497
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7838715416650078,
      "learning_rate": 1.2228082923558555e-06,
      "loss": 0.6677,
      "step": 3498
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7176376461562528,
      "learning_rate": 1.2224059851818147e-06,
      "loss": 0.5711,
      "step": 3499
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8268322531912228,
      "learning_rate": 1.2220036401343517e-06,
      "loss": 0.6459,
      "step": 3500
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9829849582498553,
      "learning_rate": 1.2216012572819816e-06,
      "loss": 0.6522,
      "step": 3501
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8431369153822154,
      "learning_rate": 1.221198836693226e-06,
      "loss": 0.5731,
      "step": 3502
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9031175798980924,
      "learning_rate": 1.220796378436613e-06,
      "loss": 0.5263,
      "step": 3503
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7422289770319375,
      "learning_rate": 1.220393882580677e-06,
      "loss": 0.536,
      "step": 3504
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7354224347397338,
      "learning_rate": 1.2199913491939587e-06,
      "loss": 0.6256,
      "step": 3505
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2435135239274218,
      "learning_rate": 1.2195887783450057e-06,
      "loss": 0.6702,
      "step": 3506
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6357125744487298,
      "learning_rate": 1.2191861701023713e-06,
      "loss": 0.5134,
      "step": 3507
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.80221880768022,
      "learning_rate": 1.2187835245346157e-06,
      "loss": 0.6338,
      "step": 3508
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8558452782699452,
      "learning_rate": 1.218380841710305e-06,
      "loss": 0.6288,
      "step": 3509
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6784191302069856,
      "learning_rate": 1.2179781216980119e-06,
      "loss": 0.5725,
      "step": 3510
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8606559269711934,
      "learning_rate": 1.2175753645663158e-06,
      "loss": 0.6282,
      "step": 3511
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0389767753729784,
      "learning_rate": 1.2171725703838016e-06,
      "loss": 0.6481,
      "step": 3512
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8516509288230759,
      "learning_rate": 1.2167697392190615e-06,
      "loss": 0.5788,
      "step": 3513
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9621465340911541,
      "learning_rate": 1.2163668711406927e-06,
      "loss": 0.6841,
      "step": 3514
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6604704361449661,
      "learning_rate": 1.2159639662173002e-06,
      "loss": 0.5037,
      "step": 3515
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9442474852325309,
      "learning_rate": 1.2155610245174938e-06,
      "loss": 0.6403,
      "step": 3516
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8288217901532556,
      "learning_rate": 1.215158046109891e-06,
      "loss": 0.592,
      "step": 3517
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6995881457148487,
      "learning_rate": 1.2147550310631144e-06,
      "loss": 0.5935,
      "step": 3518
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.829631410530379,
      "learning_rate": 1.214351979445793e-06,
      "loss": 0.6048,
      "step": 3519
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8898685045932799,
      "learning_rate": 1.2139488913265627e-06,
      "loss": 0.6588,
      "step": 3520
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8126242333817216,
      "learning_rate": 1.213545766774065e-06,
      "loss": 0.4989,
      "step": 3521
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6797205780512917,
      "learning_rate": 1.213142605856948e-06,
      "loss": 0.5708,
      "step": 3522
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6475613630795397,
      "learning_rate": 1.212739408643866e-06,
      "loss": 0.5133,
      "step": 3523
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9009108059159847,
      "learning_rate": 1.2123361752034788e-06,
      "loss": 0.701,
      "step": 3524
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7861965965835049,
      "learning_rate": 1.2119329056044531e-06,
      "loss": 0.6197,
      "step": 3525
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7545364981942796,
      "learning_rate": 1.2115295999154616e-06,
      "loss": 0.595,
      "step": 3526
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8987236680853078,
      "learning_rate": 1.2111262582051828e-06,
      "loss": 0.5923,
      "step": 3527
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.961434028963848,
      "learning_rate": 1.2107228805423017e-06,
      "loss": 0.5643,
      "step": 3528
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9019295295037031,
      "learning_rate": 1.2103194669955095e-06,
      "loss": 0.6859,
      "step": 3529
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5989102138813082,
      "learning_rate": 1.209916017633503e-06,
      "loss": 0.5494,
      "step": 3530
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9381521708968847,
      "learning_rate": 1.2095125325249855e-06,
      "loss": 0.6185,
      "step": 3531
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9430053574763092,
      "learning_rate": 1.2091090117386667e-06,
      "loss": 0.7124,
      "step": 3532
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8340145910181755,
      "learning_rate": 1.2087054553432613e-06,
      "loss": 0.6287,
      "step": 3533
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.816956118263287,
      "learning_rate": 1.2083018634074913e-06,
      "loss": 0.6121,
      "step": 3534
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8054961930566695,
      "learning_rate": 1.207898236000084e-06,
      "loss": 0.6078,
      "step": 3535
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9359150661914656,
      "learning_rate": 1.2074945731897733e-06,
      "loss": 0.6107,
      "step": 3536
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.656594809588484,
      "learning_rate": 1.2070908750452981e-06,
      "loss": 0.5402,
      "step": 3537
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6922193443069765,
      "learning_rate": 1.2066871416354046e-06,
      "loss": 0.5859,
      "step": 3538
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8318259615102429,
      "learning_rate": 1.2062833730288437e-06,
      "loss": 0.6157,
      "step": 3539
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7613914655063581,
      "learning_rate": 1.2058795692943737e-06,
      "loss": 0.5618,
      "step": 3540
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6902467577418846,
      "learning_rate": 1.2054757305007578e-06,
      "loss": 0.5449,
      "step": 3541
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8834482235574148,
      "learning_rate": 1.205071856716766e-06,
      "loss": 0.5409,
      "step": 3542
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7400526361213282,
      "learning_rate": 1.204667948011173e-06,
      "loss": 0.5472,
      "step": 3543
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7140997336484658,
      "learning_rate": 1.2042640044527608e-06,
      "loss": 0.5131,
      "step": 3544
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7843670136104182,
      "learning_rate": 1.2038600261103166e-06,
      "loss": 0.6106,
      "step": 3545
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7658831103915218,
      "learning_rate": 1.203456013052634e-06,
      "loss": 0.5923,
      "step": 3546
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7938317778517147,
      "learning_rate": 1.2030519653485114e-06,
      "loss": 0.6578,
      "step": 3547
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8036531896012116,
      "learning_rate": 1.202647883066755e-06,
      "loss": 0.6071,
      "step": 3548
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6535604999248441,
      "learning_rate": 1.2022437662761751e-06,
      "loss": 0.5462,
      "step": 3549
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7322384545035954,
      "learning_rate": 1.2018396150455887e-06,
      "loss": 0.5963,
      "step": 3550
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6845579907614295,
      "learning_rate": 1.2014354294438187e-06,
      "loss": 0.5342,
      "step": 3551
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9334066464255726,
      "learning_rate": 1.2010312095396937e-06,
      "loss": 0.6211,
      "step": 3552
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5922909587093564,
      "learning_rate": 1.2006269554020478e-06,
      "loss": 0.5212,
      "step": 3553
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7731553485473244,
      "learning_rate": 1.2002226670997217e-06,
      "loss": 0.5956,
      "step": 3554
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8253981664558916,
      "learning_rate": 1.1998183447015614e-06,
      "loss": 0.5833,
      "step": 3555
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.851786114519458,
      "learning_rate": 1.1994139882764184e-06,
      "loss": 0.6396,
      "step": 3556
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7888010578851617,
      "learning_rate": 1.199009597893151e-06,
      "loss": 0.5633,
      "step": 3557
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6215559814448506,
      "learning_rate": 1.1986051736206222e-06,
      "loss": 0.518,
      "step": 3558
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7999177239958768,
      "learning_rate": 1.1982007155277015e-06,
      "loss": 0.5845,
      "step": 3559
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8665466274285062,
      "learning_rate": 1.1977962236832633e-06,
      "loss": 0.619,
      "step": 3560
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7670270411518367,
      "learning_rate": 1.1973916981561892e-06,
      "loss": 0.567,
      "step": 3561
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.887300188600661,
      "learning_rate": 1.1969871390153651e-06,
      "loss": 0.6309,
      "step": 3562
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9260395919546728,
      "learning_rate": 1.1965825463296835e-06,
      "loss": 0.634,
      "step": 3563
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8479487607795843,
      "learning_rate": 1.1961779201680425e-06,
      "loss": 0.57,
      "step": 3564
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9141854212178816,
      "learning_rate": 1.195773260599345e-06,
      "loss": 0.5999,
      "step": 3565
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8684873873171398,
      "learning_rate": 1.1953685676925013e-06,
      "loss": 0.646,
      "step": 3566
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5810265591366123,
      "learning_rate": 1.1949638415164256e-06,
      "loss": 0.5137,
      "step": 3567
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6225818658740272,
      "learning_rate": 1.1945590821400389e-06,
      "loss": 0.5562,
      "step": 3568
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8324657505468169,
      "learning_rate": 1.1941542896322673e-06,
      "loss": 0.6534,
      "step": 3569
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7548587874902005,
      "learning_rate": 1.1937494640620432e-06,
      "loss": 0.5012,
      "step": 3570
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.712866040846828,
      "learning_rate": 1.1933446054983033e-06,
      "loss": 0.6043,
      "step": 3571
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9162511901355743,
      "learning_rate": 1.1929397140099917e-06,
      "loss": 0.6013,
      "step": 3572
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8577123638222974,
      "learning_rate": 1.1925347896660566e-06,
      "loss": 0.6185,
      "step": 3573
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7767311538423525,
      "learning_rate": 1.192129832535453e-06,
      "loss": 0.5669,
      "step": 3574
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7024101133439205,
      "learning_rate": 1.1917248426871403e-06,
      "loss": 0.5057,
      "step": 3575
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8294044919234587,
      "learning_rate": 1.191319820190084e-06,
      "loss": 0.6465,
      "step": 3576
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.238667729040125,
      "learning_rate": 1.1909147651132557e-06,
      "loss": 0.5808,
      "step": 3577
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9885318384501535,
      "learning_rate": 1.1905096775256317e-06,
      "loss": 0.6625,
      "step": 3578
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7157083943182658,
      "learning_rate": 1.1901045574961948e-06,
      "loss": 0.5752,
      "step": 3579
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6792223639703365,
      "learning_rate": 1.1896994050939318e-06,
      "loss": 0.5072,
      "step": 3580
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7970826960305226,
      "learning_rate": 1.1892942203878367e-06,
      "loss": 0.5703,
      "step": 3581
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7502523933950223,
      "learning_rate": 1.1888890034469077e-06,
      "loss": 0.5724,
      "step": 3582
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.824172217398551,
      "learning_rate": 1.1884837543401492e-06,
      "loss": 0.6502,
      "step": 3583
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9276335087977379,
      "learning_rate": 1.1880784731365712e-06,
      "loss": 0.6335,
      "step": 3584
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6427226445088194,
      "learning_rate": 1.1876731599051883e-06,
      "loss": 0.5143,
      "step": 3585
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.720098945112339,
      "learning_rate": 1.1872678147150218e-06,
      "loss": 0.5373,
      "step": 3586
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9961433033558384,
      "learning_rate": 1.186862437635097e-06,
      "loss": 0.5635,
      "step": 3587
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7830376357780922,
      "learning_rate": 1.186457028734446e-06,
      "loss": 0.5761,
      "step": 3588
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9559946980963099,
      "learning_rate": 1.1860515880821055e-06,
      "loss": 0.6418,
      "step": 3589
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9073301422693981,
      "learning_rate": 1.1856461157471176e-06,
      "loss": 0.6128,
      "step": 3590
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7934666438398036,
      "learning_rate": 1.1852406117985303e-06,
      "loss": 0.5332,
      "step": 3591
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.609847028068369,
      "learning_rate": 1.184835076305397e-06,
      "loss": 0.5135,
      "step": 3592
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7078636384587128,
      "learning_rate": 1.1844295093367755e-06,
      "loss": 0.5601,
      "step": 3593
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8185389873708361,
      "learning_rate": 1.18402391096173e-06,
      "loss": 0.5951,
      "step": 3594
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7520725252078428,
      "learning_rate": 1.1836182812493296e-06,
      "loss": 0.5639,
      "step": 3595
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7513295419044058,
      "learning_rate": 1.1832126202686488e-06,
      "loss": 0.5551,
      "step": 3596
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6845442067765931,
      "learning_rate": 1.1828069280887676e-06,
      "loss": 0.5318,
      "step": 3597
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7088369274585685,
      "learning_rate": 1.1824012047787708e-06,
      "loss": 0.5184,
      "step": 3598
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7227816165963745,
      "learning_rate": 1.1819954504077493e-06,
      "loss": 0.5559,
      "step": 3599
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6548997655710692,
      "learning_rate": 1.1815896650447984e-06,
      "loss": 0.4961,
      "step": 3600
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9082241109198962,
      "learning_rate": 1.1811838487590195e-06,
      "loss": 0.6575,
      "step": 3601
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8929461602817208,
      "learning_rate": 1.1807780016195185e-06,
      "loss": 0.6624,
      "step": 3602
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6970893108929611,
      "learning_rate": 1.1803721236954072e-06,
      "loss": 0.54,
      "step": 3603
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8972978568092888,
      "learning_rate": 1.1799662150558026e-06,
      "loss": 0.5476,
      "step": 3604
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0023706851915388,
      "learning_rate": 1.1795602757698262e-06,
      "loss": 0.6898,
      "step": 3605
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.719470700779904,
      "learning_rate": 1.1791543059066056e-06,
      "loss": 0.592,
      "step": 3606
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.632633400204935,
      "learning_rate": 1.1787483055352732e-06,
      "loss": 0.5662,
      "step": 3607
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6344294369769692,
      "learning_rate": 1.1783422747249662e-06,
      "loss": 0.5142,
      "step": 3608
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7607594570472552,
      "learning_rate": 1.177936213544828e-06,
      "loss": 0.5877,
      "step": 3609
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8811266431526009,
      "learning_rate": 1.1775301220640064e-06,
      "loss": 0.5915,
      "step": 3610
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7426080584864317,
      "learning_rate": 1.1771240003516546e-06,
      "loss": 0.5683,
      "step": 3611
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8188080429467737,
      "learning_rate": 1.1767178484769304e-06,
      "loss": 0.6017,
      "step": 3612
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7286962636520827,
      "learning_rate": 1.176311666508998e-06,
      "loss": 0.5407,
      "step": 3613
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1358811674511573,
      "learning_rate": 1.1759054545170254e-06,
      "loss": 0.6891,
      "step": 3614
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9603309573533548,
      "learning_rate": 1.1754992125701865e-06,
      "loss": 0.6492,
      "step": 3615
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8415097107855828,
      "learning_rate": 1.1750929407376604e-06,
      "loss": 0.6394,
      "step": 3616
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5980757173338465,
      "learning_rate": 1.1746866390886302e-06,
      "loss": 0.5071,
      "step": 3617
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7323593191321959,
      "learning_rate": 1.1742803076922853e-06,
      "loss": 0.5612,
      "step": 3618
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.692883621933488,
      "learning_rate": 1.1738739466178195e-06,
      "loss": 0.5528,
      "step": 3619
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6433014908711976,
      "learning_rate": 1.1734675559344321e-06,
      "loss": 0.5776,
      "step": 3620
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.81345233531994,
      "learning_rate": 1.1730611357113275e-06,
      "loss": 0.5704,
      "step": 3621
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7058413315869119,
      "learning_rate": 1.172654686017714e-06,
      "loss": 0.5252,
      "step": 3622
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7208480001793919,
      "learning_rate": 1.1722482069228063e-06,
      "loss": 0.5394,
      "step": 3623
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6371528961380661,
      "learning_rate": 1.1718416984958234e-06,
      "loss": 0.5338,
      "step": 3624
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6905815433262681,
      "learning_rate": 1.1714351608059895e-06,
      "loss": 0.58,
      "step": 3625
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8719170090264644,
      "learning_rate": 1.1710285939225338e-06,
      "loss": 0.653,
      "step": 3626
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6932977381804137,
      "learning_rate": 1.17062199791469e-06,
      "loss": 0.5653,
      "step": 3627
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9410155657055277,
      "learning_rate": 1.1702153728516979e-06,
      "loss": 0.6247,
      "step": 3628
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7067631055995084,
      "learning_rate": 1.169808718802801e-06,
      "loss": 0.5774,
      "step": 3629
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7034274838749379,
      "learning_rate": 1.169402035837248e-06,
      "loss": 0.5672,
      "step": 3630
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7080389290734543,
      "learning_rate": 1.1689953240242933e-06,
      "loss": 0.5493,
      "step": 3631
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6269167055961934,
      "learning_rate": 1.1685885834331954e-06,
      "loss": 0.5205,
      "step": 3632
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7170609024089468,
      "learning_rate": 1.1681818141332184e-06,
      "loss": 0.5306,
      "step": 3633
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6947624802037049,
      "learning_rate": 1.1677750161936303e-06,
      "loss": 0.5197,
      "step": 3634
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5798515003428385,
      "learning_rate": 1.1673681896837048e-06,
      "loss": 0.5094,
      "step": 3635
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0250640243396019,
      "learning_rate": 1.1669613346727205e-06,
      "loss": 0.6112,
      "step": 3636
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7634876935829814,
      "learning_rate": 1.1665544512299602e-06,
      "loss": 0.5253,
      "step": 3637
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6861988645706409,
      "learning_rate": 1.166147539424712e-06,
      "loss": 0.5492,
      "step": 3638
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8551424666892626,
      "learning_rate": 1.1657405993262685e-06,
      "loss": 0.6528,
      "step": 3639
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7983300943076339,
      "learning_rate": 1.165333631003928e-06,
      "loss": 0.5504,
      "step": 3640
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7629837307304309,
      "learning_rate": 1.1649266345269925e-06,
      "loss": 0.5339,
      "step": 3641
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9712587407829955,
      "learning_rate": 1.164519609964769e-06,
      "loss": 0.6982,
      "step": 3642
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8637557019882236,
      "learning_rate": 1.16411255738657e-06,
      "loss": 0.6335,
      "step": 3643
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9013452188179716,
      "learning_rate": 1.1637054768617126e-06,
      "loss": 0.6154,
      "step": 3644
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8348139784766735,
      "learning_rate": 1.1632983684595172e-06,
      "loss": 0.5824,
      "step": 3645
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0048097682164636,
      "learning_rate": 1.1628912322493113e-06,
      "loss": 0.6535,
      "step": 3646
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8707463514413742,
      "learning_rate": 1.1624840683004255e-06,
      "loss": 0.6079,
      "step": 3647
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7311340909923244,
      "learning_rate": 1.1620768766821957e-06,
      "loss": 0.5596,
      "step": 3648
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8895525895175532,
      "learning_rate": 1.1616696574639625e-06,
      "loss": 0.5585,
      "step": 3649
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0816762534577864,
      "learning_rate": 1.1612624107150705e-06,
      "loss": 0.5964,
      "step": 3650
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7253532672108746,
      "learning_rate": 1.1608551365048702e-06,
      "loss": 0.583,
      "step": 3651
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2551727308539335,
      "learning_rate": 1.1604478349027159e-06,
      "loss": 0.6498,
      "step": 3652
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8910985828614785,
      "learning_rate": 1.1600405059779669e-06,
      "loss": 0.657,
      "step": 3653
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.777946976312628,
      "learning_rate": 1.1596331497999868e-06,
      "loss": 0.5381,
      "step": 3654
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7098297271464186,
      "learning_rate": 1.1592257664381444e-06,
      "loss": 0.5525,
      "step": 3655
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0725023480542142,
      "learning_rate": 1.158818355961813e-06,
      "loss": 0.6458,
      "step": 3656
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8757536584562327,
      "learning_rate": 1.1584109184403699e-06,
      "loss": 0.6191,
      "step": 3657
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9470950137264245,
      "learning_rate": 1.1580034539431975e-06,
      "loss": 0.6043,
      "step": 3658
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.843629809137184,
      "learning_rate": 1.1575959625396833e-06,
      "loss": 0.6378,
      "step": 3659
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9293213466033156,
      "learning_rate": 1.1571884442992185e-06,
      "loss": 0.6646,
      "step": 3660
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9115810132416571,
      "learning_rate": 1.156780899291199e-06,
      "loss": 0.6559,
      "step": 3661
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6705874144665556,
      "learning_rate": 1.1563733275850257e-06,
      "loss": 0.5367,
      "step": 3662
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8697428400624346,
      "learning_rate": 1.1559657292501042e-06,
      "loss": 0.6489,
      "step": 3663
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7948070151430111,
      "learning_rate": 1.1555581043558435e-06,
      "loss": 0.6551,
      "step": 3664
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6117085911461957,
      "learning_rate": 1.1551504529716582e-06,
      "loss": 0.5378,
      "step": 3665
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.765917003819845,
      "learning_rate": 1.1547427751669673e-06,
      "loss": 0.6041,
      "step": 3666
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.700041290568281,
      "learning_rate": 1.154335071011194e-06,
      "loss": 0.617,
      "step": 3667
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6392624973599721,
      "learning_rate": 1.1539273405737657e-06,
      "loss": 0.5471,
      "step": 3668
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8544402601481563,
      "learning_rate": 1.153519583924115e-06,
      "loss": 0.612,
      "step": 3669
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8223919630677149,
      "learning_rate": 1.1531118011316784e-06,
      "loss": 0.6013,
      "step": 3670
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8508279096714498,
      "learning_rate": 1.1527039922658973e-06,
      "loss": 0.6388,
      "step": 3671
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8950839673167871,
      "learning_rate": 1.1522961573962171e-06,
      "loss": 0.5651,
      "step": 3672
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9189110118822066,
      "learning_rate": 1.151888296592088e-06,
      "loss": 0.6199,
      "step": 3673
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7508253015712515,
      "learning_rate": 1.151480409922964e-06,
      "loss": 0.4707,
      "step": 3674
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9090710908928314,
      "learning_rate": 1.151072497458305e-06,
      "loss": 0.6338,
      "step": 3675
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8621805441555135,
      "learning_rate": 1.1506645592675728e-06,
      "loss": 0.5627,
      "step": 3676
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9268394408692671,
      "learning_rate": 1.1502565954202361e-06,
      "loss": 0.5843,
      "step": 3677
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7515184236025934,
      "learning_rate": 1.1498486059857666e-06,
      "loss": 0.537,
      "step": 3678
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6477237371280447,
      "learning_rate": 1.1494405910336407e-06,
      "loss": 0.556,
      "step": 3679
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8396348207200406,
      "learning_rate": 1.1490325506333389e-06,
      "loss": 0.5993,
      "step": 3680
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6943649408965813,
      "learning_rate": 1.1486244848543462e-06,
      "loss": 0.5007,
      "step": 3681
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6393568502865359,
      "learning_rate": 1.1482163937661522e-06,
      "loss": 0.5417,
      "step": 3682
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8266632755864813,
      "learning_rate": 1.1478082774382505e-06,
      "loss": 0.6441,
      "step": 3683
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6551815974005002,
      "learning_rate": 1.147400135940139e-06,
      "loss": 0.5375,
      "step": 3684
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8912795042690866,
      "learning_rate": 1.14699196934132e-06,
      "loss": 0.6012,
      "step": 3685
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8241691594437537,
      "learning_rate": 1.1465837777112998e-06,
      "loss": 0.6245,
      "step": 3686
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.725901301920027,
      "learning_rate": 1.1461755611195893e-06,
      "loss": 0.5508,
      "step": 3687
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6755848514912907,
      "learning_rate": 1.1457673196357038e-06,
      "loss": 0.5524,
      "step": 3688
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8828134442489497,
      "learning_rate": 1.1453590533291624e-06,
      "loss": 0.5888,
      "step": 3689
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9128717908805856,
      "learning_rate": 1.1449507622694884e-06,
      "loss": 0.6043,
      "step": 3690
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8866033268773514,
      "learning_rate": 1.1445424465262097e-06,
      "loss": 0.6216,
      "step": 3691
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8879525934127555,
      "learning_rate": 1.1441341061688584e-06,
      "loss": 0.6366,
      "step": 3692
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5915697097182501,
      "learning_rate": 1.14372574126697e-06,
      "loss": 0.4858,
      "step": 3693
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6991436830838438,
      "learning_rate": 1.1433173518900854e-06,
      "loss": 0.5721,
      "step": 3694
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9151881948612605,
      "learning_rate": 1.1429089381077489e-06,
      "loss": 0.6026,
      "step": 3695
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8006592540436487,
      "learning_rate": 1.1425004999895088e-06,
      "loss": 0.5129,
      "step": 3696
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7275436935940939,
      "learning_rate": 1.1420920376049186e-06,
      "loss": 0.485,
      "step": 3697
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7053797849964247,
      "learning_rate": 1.1416835510235343e-06,
      "loss": 0.5718,
      "step": 3698
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9048433002689582,
      "learning_rate": 1.1412750403149174e-06,
      "loss": 0.5784,
      "step": 3699
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9967803378225267,
      "learning_rate": 1.140866505548633e-06,
      "loss": 0.6206,
      "step": 3700
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3082645798244654,
      "learning_rate": 1.14045794679425e-06,
      "loss": 0.6834,
      "step": 3701
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8582512126729813,
      "learning_rate": 1.140049364121342e-06,
      "loss": 0.638,
      "step": 3702
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6361548545182607,
      "learning_rate": 1.1396407575994867e-06,
      "loss": 0.4675,
      "step": 3703
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8410226784040911,
      "learning_rate": 1.1392321272982649e-06,
      "loss": 0.6167,
      "step": 3704
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6418828525016536,
      "learning_rate": 1.138823473287262e-06,
      "loss": 0.5871,
      "step": 3705
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9008430286223994,
      "learning_rate": 1.1384147956360683e-06,
      "loss": 0.6289,
      "step": 3706
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.676416370227893,
      "learning_rate": 1.1380060944142769e-06,
      "loss": 0.534,
      "step": 3707
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6988483285968501,
      "learning_rate": 1.137597369691485e-06,
      "loss": 0.5188,
      "step": 3708
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9025668718192182,
      "learning_rate": 1.137188621537295e-06,
      "loss": 0.618,
      "step": 3709
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7008834273975737,
      "learning_rate": 1.1367798500213118e-06,
      "loss": 0.5541,
      "step": 3710
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8540132983700257,
      "learning_rate": 1.136371055213145e-06,
      "loss": 0.6122,
      "step": 3711
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7464078803435495,
      "learning_rate": 1.135962237182408e-06,
      "loss": 0.573,
      "step": 3712
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6801124148173194,
      "learning_rate": 1.1355533959987186e-06,
      "loss": 0.5426,
      "step": 3713
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7818158282964908,
      "learning_rate": 1.1351445317316982e-06,
      "loss": 0.637,
      "step": 3714
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9673976542536208,
      "learning_rate": 1.134735644450972e-06,
      "loss": 0.6118,
      "step": 3715
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.995751027227567,
      "learning_rate": 1.1343267342261689e-06,
      "loss": 0.6628,
      "step": 3716
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9881288452763789,
      "learning_rate": 1.1339178011269223e-06,
      "loss": 0.6063,
      "step": 3717
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6767177842201771,
      "learning_rate": 1.1335088452228699e-06,
      "loss": 0.5539,
      "step": 3718
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7183936169016847,
      "learning_rate": 1.1330998665836516e-06,
      "loss": 0.6567,
      "step": 3719
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6289116546719111,
      "learning_rate": 1.1326908652789127e-06,
      "loss": 0.5115,
      "step": 3720
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9004751858649493,
      "learning_rate": 1.1322818413783019e-06,
      "loss": 0.6551,
      "step": 3721
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7868643488220931,
      "learning_rate": 1.1318727949514718e-06,
      "loss": 0.5665,
      "step": 3722
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7065282969122566,
      "learning_rate": 1.1314637260680782e-06,
      "loss": 0.5059,
      "step": 3723
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7397267666147376,
      "learning_rate": 1.1310546347977818e-06,
      "loss": 0.5461,
      "step": 3724
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8862590294791912,
      "learning_rate": 1.1306455212102464e-06,
      "loss": 0.5655,
      "step": 3725
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7033322582630152,
      "learning_rate": 1.1302363853751398e-06,
      "loss": 0.5638,
      "step": 3726
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8339013442877581,
      "learning_rate": 1.1298272273621336e-06,
      "loss": 0.6519,
      "step": 3727
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9169636846092699,
      "learning_rate": 1.1294180472409032e-06,
      "loss": 0.6548,
      "step": 3728
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8498628592305902,
      "learning_rate": 1.1290088450811275e-06,
      "loss": 0.6347,
      "step": 3729
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6597006717997809,
      "learning_rate": 1.1285996209524894e-06,
      "loss": 0.5115,
      "step": 3730
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6359356228337353,
      "learning_rate": 1.1281903749246755e-06,
      "loss": 0.5767,
      "step": 3731
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9535178063760222,
      "learning_rate": 1.1277811070673764e-06,
      "loss": 0.6414,
      "step": 3732
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8990969220219319,
      "learning_rate": 1.127371817450286e-06,
      "loss": 0.6207,
      "step": 3733
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0071475212603687,
      "learning_rate": 1.126962506143102e-06,
      "loss": 0.6912,
      "step": 3734
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6583596770126955,
      "learning_rate": 1.1265531732155258e-06,
      "loss": 0.5315,
      "step": 3735
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8863981748527064,
      "learning_rate": 1.1261438187372628e-06,
      "loss": 0.6536,
      "step": 3736
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0963785519983609,
      "learning_rate": 1.1257344427780214e-06,
      "loss": 0.6762,
      "step": 3737
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7733203820016387,
      "learning_rate": 1.1253250454075147e-06,
      "loss": 0.534,
      "step": 3738
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8922322902342977,
      "learning_rate": 1.124915626695458e-06,
      "loss": 0.6192,
      "step": 3739
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6648158056979462,
      "learning_rate": 1.1245061867115716e-06,
      "loss": 0.5945,
      "step": 3740
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6830001108168379,
      "learning_rate": 1.124096725525579e-06,
      "loss": 0.5631,
      "step": 3741
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.060176801282398,
      "learning_rate": 1.1236872432072064e-06,
      "loss": 0.6603,
      "step": 3742
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7121320884043028,
      "learning_rate": 1.1232777398261849e-06,
      "loss": 0.5468,
      "step": 3743
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9011222436166068,
      "learning_rate": 1.1228682154522483e-06,
      "loss": 0.6281,
      "step": 3744
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6890188999299539,
      "learning_rate": 1.1224586701551352e-06,
      "loss": 0.5862,
      "step": 3745
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7369567946649698,
      "learning_rate": 1.1220491040045863e-06,
      "loss": 0.5816,
      "step": 3746
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8188797863087116,
      "learning_rate": 1.1216395170703464e-06,
      "loss": 0.5919,
      "step": 3747
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8599577500462381,
      "learning_rate": 1.1212299094221637e-06,
      "loss": 0.6565,
      "step": 3748
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9007165031718727,
      "learning_rate": 1.120820281129791e-06,
      "loss": 0.6178,
      "step": 3749
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.032435497903724,
      "learning_rate": 1.1204106322629825e-06,
      "loss": 0.6512,
      "step": 3750
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9283412192753755,
      "learning_rate": 1.1200009628914981e-06,
      "loss": 0.6182,
      "step": 3751
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6968683829086844,
      "learning_rate": 1.1195912730850998e-06,
      "loss": 0.5338,
      "step": 3752
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9379148262693374,
      "learning_rate": 1.1191815629135535e-06,
      "loss": 0.6586,
      "step": 3753
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.79342834666322,
      "learning_rate": 1.118771832446629e-06,
      "loss": 0.6212,
      "step": 3754
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8028522756418939,
      "learning_rate": 1.1183620817540983e-06,
      "loss": 0.5885,
      "step": 3755
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7217049043580395,
      "learning_rate": 1.1179523109057382e-06,
      "loss": 0.6321,
      "step": 3756
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9594892450996403,
      "learning_rate": 1.1175425199713288e-06,
      "loss": 0.6456,
      "step": 3757
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.79504947455154,
      "learning_rate": 1.1171327090206524e-06,
      "loss": 0.5723,
      "step": 3758
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.867307420014309,
      "learning_rate": 1.1167228781234962e-06,
      "loss": 0.6149,
      "step": 3759
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6536211299277561,
      "learning_rate": 1.1163130273496494e-06,
      "loss": 0.5224,
      "step": 3760
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6046069131717426,
      "learning_rate": 1.1159031567689064e-06,
      "loss": 0.4821,
      "step": 3761
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7578825162793845,
      "learning_rate": 1.115493266451063e-06,
      "loss": 0.5329,
      "step": 3762
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.636194825568033,
      "learning_rate": 1.1150833564659195e-06,
      "loss": 0.5063,
      "step": 3763
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7856334926588111,
      "learning_rate": 1.1146734268832792e-06,
      "loss": 0.5638,
      "step": 3764
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8652311349391903,
      "learning_rate": 1.114263477772949e-06,
      "loss": 0.5868,
      "step": 3765
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8088234401103918,
      "learning_rate": 1.113853509204739e-06,
      "loss": 0.6278,
      "step": 3766
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7222621717914747,
      "learning_rate": 1.1134435212484627e-06,
      "loss": 0.5506,
      "step": 3767
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7988033825367326,
      "learning_rate": 1.1130335139739364e-06,
      "loss": 0.6317,
      "step": 3768
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0223647643000335,
      "learning_rate": 1.1126234874509802e-06,
      "loss": 0.7092,
      "step": 3769
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7156755330820165,
      "learning_rate": 1.1122134417494176e-06,
      "loss": 0.5678,
      "step": 3770
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6528438149216544,
      "learning_rate": 1.1118033769390746e-06,
      "loss": 0.5734,
      "step": 3771
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8173070531450803,
      "learning_rate": 1.1113932930897815e-06,
      "loss": 0.607,
      "step": 3772
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8632260217466431,
      "learning_rate": 1.1109831902713708e-06,
      "loss": 0.631,
      "step": 3773
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0617871786752167,
      "learning_rate": 1.1105730685536791e-06,
      "loss": 0.6124,
      "step": 3774
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6355142294590497,
      "learning_rate": 1.1101629280065458e-06,
      "loss": 0.4945,
      "step": 3775
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7206036986520717,
      "learning_rate": 1.1097527686998139e-06,
      "loss": 0.5421,
      "step": 3776
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7392804000610621,
      "learning_rate": 1.1093425907033285e-06,
      "loss": 0.5526,
      "step": 3777
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7527049604930378,
      "learning_rate": 1.1089323940869391e-06,
      "loss": 0.6016,
      "step": 3778
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.795641520521779,
      "learning_rate": 1.1085221789204979e-06,
      "loss": 0.5911,
      "step": 3779
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8873159662634362,
      "learning_rate": 1.10811194527386e-06,
      "loss": 0.6565,
      "step": 3780
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8584678805936323,
      "learning_rate": 1.1077016932168844e-06,
      "loss": 0.6536,
      "step": 3781
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.746276777271781,
      "learning_rate": 1.1072914228194324e-06,
      "loss": 0.5504,
      "step": 3782
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.824509842356158,
      "learning_rate": 1.106881134151369e-06,
      "loss": 0.5948,
      "step": 3783
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0532894836327467,
      "learning_rate": 1.1064708272825616e-06,
      "loss": 0.6042,
      "step": 3784
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9252619387671036,
      "learning_rate": 1.1060605022828812e-06,
      "loss": 0.6352,
      "step": 3785
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5756753115479418,
      "learning_rate": 1.1056501592222028e-06,
      "loss": 0.5102,
      "step": 3786
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6600129052974246,
      "learning_rate": 1.1052397981704027e-06,
      "loss": 0.5989,
      "step": 3787
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7665406632703332,
      "learning_rate": 1.1048294191973615e-06,
      "loss": 0.6081,
      "step": 3788
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7065913862031409,
      "learning_rate": 1.1044190223729623e-06,
      "loss": 0.5694,
      "step": 3789
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8372081957596091,
      "learning_rate": 1.1040086077670914e-06,
      "loss": 0.6313,
      "step": 3790
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7802504736342917,
      "learning_rate": 1.103598175449638e-06,
      "loss": 0.6344,
      "step": 3791
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8241439835305071,
      "learning_rate": 1.1031877254904949e-06,
      "loss": 0.5173,
      "step": 3792
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7936629514530036,
      "learning_rate": 1.1027772579595569e-06,
      "loss": 0.5726,
      "step": 3793
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6707457628203407,
      "learning_rate": 1.1023667729267228e-06,
      "loss": 0.5439,
      "step": 3794
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6737119798223995,
      "learning_rate": 1.1019562704618934e-06,
      "loss": 0.5663,
      "step": 3795
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.026687685928366,
      "learning_rate": 1.1015457506349735e-06,
      "loss": 0.688,
      "step": 3796
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6902037707638047,
      "learning_rate": 1.1011352135158702e-06,
      "loss": 0.5388,
      "step": 3797
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6896284640902063,
      "learning_rate": 1.1007246591744933e-06,
      "loss": 0.5395,
      "step": 3798
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6760964269128757,
      "learning_rate": 1.1003140876807568e-06,
      "loss": 0.5612,
      "step": 3799
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8592723960805339,
      "learning_rate": 1.0999034991045759e-06,
      "loss": 0.5917,
      "step": 3800
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6579217247288897,
      "learning_rate": 1.0994928935158701e-06,
      "loss": 0.5029,
      "step": 3801
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1605091520723925,
      "learning_rate": 1.099082270984561e-06,
      "loss": 0.6257,
      "step": 3802
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7335910140892303,
      "learning_rate": 1.0986716315805734e-06,
      "loss": 0.5657,
      "step": 3803
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7177762446051024,
      "learning_rate": 1.0982609753738347e-06,
      "loss": 0.5271,
      "step": 3804
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6701885208288325,
      "learning_rate": 1.0978503024342757e-06,
      "loss": 0.5778,
      "step": 3805
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6528951719816746,
      "learning_rate": 1.0974396128318297e-06,
      "loss": 0.506,
      "step": 3806
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7807931138521967,
      "learning_rate": 1.0970289066364327e-06,
      "loss": 0.5959,
      "step": 3807
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7991688314550837,
      "learning_rate": 1.0966181839180238e-06,
      "loss": 0.6082,
      "step": 3808
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8395158315412257,
      "learning_rate": 1.0962074447465447e-06,
      "loss": 0.5399,
      "step": 3809
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.865385124994614,
      "learning_rate": 1.09579668919194e-06,
      "loss": 0.6287,
      "step": 3810
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8194872326607046,
      "learning_rate": 1.0953859173241572e-06,
      "loss": 0.6143,
      "step": 3811
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.892921925064196,
      "learning_rate": 1.0949751292131459e-06,
      "loss": 0.6113,
      "step": 3812
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6671846647685193,
      "learning_rate": 1.0945643249288602e-06,
      "loss": 0.5772,
      "step": 3813
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9540490829202073,
      "learning_rate": 1.0941535045412546e-06,
      "loss": 0.6188,
      "step": 3814
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.660394261701563,
      "learning_rate": 1.0937426681202882e-06,
      "loss": 0.5444,
      "step": 3815
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7728879872997366,
      "learning_rate": 1.0933318157359222e-06,
      "loss": 0.5977,
      "step": 3816
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1626768208170344,
      "learning_rate": 1.0929209474581203e-06,
      "loss": 0.6156,
      "step": 3817
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8493130879720228,
      "learning_rate": 1.0925100633568492e-06,
      "loss": 0.6057,
      "step": 3818
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8088094553452906,
      "learning_rate": 1.0920991635020779e-06,
      "loss": 0.5864,
      "step": 3819
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7842648992513092,
      "learning_rate": 1.0916882479637785e-06,
      "loss": 0.6019,
      "step": 3820
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.841139903055992,
      "learning_rate": 1.0912773168119259e-06,
      "loss": 0.6242,
      "step": 3821
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7308709317388855,
      "learning_rate": 1.0908663701164969e-06,
      "loss": 0.5397,
      "step": 3822
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6211828800090295,
      "learning_rate": 1.0904554079474719e-06,
      "loss": 0.492,
      "step": 3823
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7938120391719792,
      "learning_rate": 1.090044430374833e-06,
      "loss": 0.6355,
      "step": 3824
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9829076867715043,
      "learning_rate": 1.089633437468566e-06,
      "loss": 0.6416,
      "step": 3825
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2232518176506675,
      "learning_rate": 1.0892224292986583e-06,
      "loss": 0.6335,
      "step": 3826
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8599576998217255,
      "learning_rate": 1.0888114059351003e-06,
      "loss": 0.5989,
      "step": 3827
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9150294669012223,
      "learning_rate": 1.0884003674478852e-06,
      "loss": 0.6648,
      "step": 3828
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6182961202250848,
      "learning_rate": 1.0879893139070085e-06,
      "loss": 0.5696,
      "step": 3829
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8014491375053396,
      "learning_rate": 1.0875782453824681e-06,
      "loss": 0.5861,
      "step": 3830
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6952491829345054,
      "learning_rate": 1.087167161944265e-06,
      "loss": 0.5799,
      "step": 3831
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8085927162660927,
      "learning_rate": 1.0867560636624025e-06,
      "loss": 0.5638,
      "step": 3832
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7054277513052164,
      "learning_rate": 1.086344950606886e-06,
      "loss": 0.617,
      "step": 3833
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8580699861278577,
      "learning_rate": 1.0859338228477243e-06,
      "loss": 0.6003,
      "step": 3834
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0033146581137338,
      "learning_rate": 1.0855226804549273e-06,
      "loss": 0.6182,
      "step": 3835
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9612631871360799,
      "learning_rate": 1.0851115234985091e-06,
      "loss": 0.6445,
      "step": 3836
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8540633835722822,
      "learning_rate": 1.084700352048485e-06,
      "loss": 0.6564,
      "step": 3837
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8199905602636428,
      "learning_rate": 1.0842891661748734e-06,
      "loss": 0.6231,
      "step": 3838
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7222605964180555,
      "learning_rate": 1.0838779659476949e-06,
      "loss": 0.5398,
      "step": 3839
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.035798317002971,
      "learning_rate": 1.0834667514369727e-06,
      "loss": 0.605,
      "step": 3840
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7897088010826849,
      "learning_rate": 1.0830555227127321e-06,
      "loss": 0.5774,
      "step": 3841
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7246090443171848,
      "learning_rate": 1.0826442798450014e-06,
      "loss": 0.5344,
      "step": 3842
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.130234961522196,
      "learning_rate": 1.0822330229038108e-06,
      "loss": 0.6231,
      "step": 3843
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7564057590220903,
      "learning_rate": 1.081821751959193e-06,
      "loss": 0.5563,
      "step": 3844
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8510239387676737,
      "learning_rate": 1.0814104670811833e-06,
      "loss": 0.6283,
      "step": 3845
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.914314979152228,
      "learning_rate": 1.0809991683398195e-06,
      "loss": 0.6649,
      "step": 3846
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8560634215890172,
      "learning_rate": 1.080587855805141e-06,
      "loss": 0.574,
      "step": 3847
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.664576971323027,
      "learning_rate": 1.0801765295471898e-06,
      "loss": 0.4649,
      "step": 3848
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8119712580438827,
      "learning_rate": 1.0797651896360115e-06,
      "loss": 0.5963,
      "step": 3849
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6905304309678921,
      "learning_rate": 1.079353836141652e-06,
      "loss": 0.5271,
      "step": 3850
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.689858010676965,
      "learning_rate": 1.078942469134161e-06,
      "loss": 0.5499,
      "step": 3851
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8511031963719486,
      "learning_rate": 1.07853108868359e-06,
      "loss": 0.6045,
      "step": 3852
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5933672823661634,
      "learning_rate": 1.0781196948599927e-06,
      "loss": 0.5446,
      "step": 3853
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.650819812780896,
      "learning_rate": 1.0777082877334248e-06,
      "loss": 0.5657,
      "step": 3854
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8688002130109596,
      "learning_rate": 1.0772968673739456e-06,
      "loss": 0.6502,
      "step": 3855
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5891920618562846,
      "learning_rate": 1.076885433851615e-06,
      "loss": 0.5524,
      "step": 3856
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7119758778190097,
      "learning_rate": 1.076473987236496e-06,
      "loss": 0.5874,
      "step": 3857
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6809975208555901,
      "learning_rate": 1.0760625275986537e-06,
      "loss": 0.5411,
      "step": 3858
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6578472864380163,
      "learning_rate": 1.0756510550081552e-06,
      "loss": 0.5142,
      "step": 3859
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6672513708727161,
      "learning_rate": 1.0752395695350705e-06,
      "loss": 0.6074,
      "step": 3860
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6913751395916042,
      "learning_rate": 1.0748280712494706e-06,
      "loss": 0.491,
      "step": 3861
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8604718985823631,
      "learning_rate": 1.0744165602214298e-06,
      "loss": 0.5996,
      "step": 3862
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8724895646075216,
      "learning_rate": 1.0740050365210242e-06,
      "loss": 0.6199,
      "step": 3863
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7442813208318847,
      "learning_rate": 1.0735935002183318e-06,
      "loss": 0.5772,
      "step": 3864
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6709914225219399,
      "learning_rate": 1.0731819513834328e-06,
      "loss": 0.4842,
      "step": 3865
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7523437199230651,
      "learning_rate": 1.0727703900864101e-06,
      "loss": 0.5764,
      "step": 3866
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9526097681622248,
      "learning_rate": 1.072358816397348e-06,
      "loss": 0.648,
      "step": 3867
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7030348455236145,
      "learning_rate": 1.0719472303863335e-06,
      "loss": 0.5317,
      "step": 3868
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.740501434365152,
      "learning_rate": 1.071535632123455e-06,
      "loss": 0.5571,
      "step": 3869
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.033093203075489,
      "learning_rate": 1.0711240216788035e-06,
      "loss": 0.637,
      "step": 3870
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.880591638091897,
      "learning_rate": 1.0707123991224722e-06,
      "loss": 0.683,
      "step": 3871
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8529215121011173,
      "learning_rate": 1.070300764524556e-06,
      "loss": 0.6485,
      "step": 3872
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8631923330379587,
      "learning_rate": 1.0698891179551522e-06,
      "loss": 0.6825,
      "step": 3873
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6342530938842654,
      "learning_rate": 1.0694774594843594e-06,
      "loss": 0.5796,
      "step": 3874
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7491222660612251,
      "learning_rate": 1.0690657891822791e-06,
      "loss": 0.5693,
      "step": 3875
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.825898420056976,
      "learning_rate": 1.0686541071190148e-06,
      "loss": 0.5906,
      "step": 3876
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8097381829024631,
      "learning_rate": 1.068242413364671e-06,
      "loss": 0.5695,
      "step": 3877
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6421327163809314,
      "learning_rate": 1.067830707989355e-06,
      "loss": 0.5558,
      "step": 3878
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6985690310693011,
      "learning_rate": 1.0674189910631763e-06,
      "loss": 0.5093,
      "step": 3879
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7497215227335424,
      "learning_rate": 1.0670072626562458e-06,
      "loss": 0.6437,
      "step": 3880
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8081997138787704,
      "learning_rate": 1.0665955228386765e-06,
      "loss": 0.6681,
      "step": 3881
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9434005408799359,
      "learning_rate": 1.0661837716805835e-06,
      "loss": 0.648,
      "step": 3882
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5787022719610794,
      "learning_rate": 1.0657720092520834e-06,
      "loss": 0.5391,
      "step": 3883
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5873612957661662,
      "learning_rate": 1.0653602356232956e-06,
      "loss": 0.5553,
      "step": 3884
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7853388611484223,
      "learning_rate": 1.0649484508643405e-06,
      "loss": 0.6244,
      "step": 3885
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7848610768613511,
      "learning_rate": 1.064536655045341e-06,
      "loss": 0.5665,
      "step": 3886
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8953459996706609,
      "learning_rate": 1.0641248482364213e-06,
      "loss": 0.5958,
      "step": 3887
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8676432916916758,
      "learning_rate": 1.063713030507708e-06,
      "loss": 0.5839,
      "step": 3888
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.877460098847511,
      "learning_rate": 1.0633012019293295e-06,
      "loss": 0.6756,
      "step": 3889
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7433011992374953,
      "learning_rate": 1.0628893625714156e-06,
      "loss": 0.5749,
      "step": 3890
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7096798235459288,
      "learning_rate": 1.0624775125040985e-06,
      "loss": 0.5083,
      "step": 3891
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8443575004382138,
      "learning_rate": 1.062065651797512e-06,
      "loss": 0.6717,
      "step": 3892
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6452954591097366,
      "learning_rate": 1.0616537805217915e-06,
      "loss": 0.5538,
      "step": 3893
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8031244017548903,
      "learning_rate": 1.0612418987470746e-06,
      "loss": 0.6132,
      "step": 3894
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8697034853078133,
      "learning_rate": 1.0608300065435003e-06,
      "loss": 0.5416,
      "step": 3895
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6714619420844369,
      "learning_rate": 1.0604181039812095e-06,
      "loss": 0.5639,
      "step": 3896
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7370519725970414,
      "learning_rate": 1.0600061911303452e-06,
      "loss": 0.6182,
      "step": 3897
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6596369833655409,
      "learning_rate": 1.0595942680610515e-06,
      "loss": 0.5131,
      "step": 3898
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9500943696770713,
      "learning_rate": 1.0591823348434748e-06,
      "loss": 0.6657,
      "step": 3899
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8154919743641041,
      "learning_rate": 1.058770391547763e-06,
      "loss": 0.5574,
      "step": 3900
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.967953390206161,
      "learning_rate": 1.058358438244066e-06,
      "loss": 0.6574,
      "step": 3901
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.868098052130379,
      "learning_rate": 1.057946475002535e-06,
      "loss": 0.6124,
      "step": 3902
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7150064478781477,
      "learning_rate": 1.0575345018933226e-06,
      "loss": 0.5605,
      "step": 3903
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7058665895310818,
      "learning_rate": 1.0571225189865841e-06,
      "loss": 0.5152,
      "step": 3904
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.786037628469416,
      "learning_rate": 1.056710526352476e-06,
      "loss": 0.6119,
      "step": 3905
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9175916653397623,
      "learning_rate": 1.0562985240611557e-06,
      "loss": 0.5961,
      "step": 3906
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9191431536283905,
      "learning_rate": 1.0558865121827835e-06,
      "loss": 0.663,
      "step": 3907
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7306920205492291,
      "learning_rate": 1.0554744907875203e-06,
      "loss": 0.5682,
      "step": 3908
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7447064032929676,
      "learning_rate": 1.0550624599455293e-06,
      "loss": 0.5806,
      "step": 3909
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.733320319574239,
      "learning_rate": 1.054650419726975e-06,
      "loss": 0.6166,
      "step": 3910
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8592750930973613,
      "learning_rate": 1.0542383702020236e-06,
      "loss": 0.6186,
      "step": 3911
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8238014266103234,
      "learning_rate": 1.053826311440843e-06,
      "loss": 0.6188,
      "step": 3912
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7570717714256754,
      "learning_rate": 1.053414243513602e-06,
      "loss": 0.5647,
      "step": 3913
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0139280007419562,
      "learning_rate": 1.053002166490472e-06,
      "loss": 0.6609,
      "step": 3914
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7054410266181337,
      "learning_rate": 1.0525900804416253e-06,
      "loss": 0.4938,
      "step": 3915
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9347438682503897,
      "learning_rate": 1.0521779854372352e-06,
      "loss": 0.6817,
      "step": 3916
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7965423866960685,
      "learning_rate": 1.0517658815474783e-06,
      "loss": 0.6266,
      "step": 3917
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7398869969764291,
      "learning_rate": 1.051353768842531e-06,
      "loss": 0.5385,
      "step": 3918
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8850875654796503,
      "learning_rate": 1.0509416473925718e-06,
      "loss": 0.634,
      "step": 3919
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6114771403782506,
      "learning_rate": 1.0505295172677808e-06,
      "loss": 0.5273,
      "step": 3920
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8201424147336556,
      "learning_rate": 1.0501173785383395e-06,
      "loss": 0.6353,
      "step": 3921
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7482803528353174,
      "learning_rate": 1.0497052312744305e-06,
      "loss": 0.5051,
      "step": 3922
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7458468457169541,
      "learning_rate": 1.0492930755462389e-06,
      "loss": 0.5479,
      "step": 3923
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6888045191343626,
      "learning_rate": 1.0488809114239497e-06,
      "loss": 0.5482,
      "step": 3924
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8179030633433859,
      "learning_rate": 1.0484687389777506e-06,
      "loss": 0.6072,
      "step": 3925
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6732904983991791,
      "learning_rate": 1.04805655827783e-06,
      "loss": 0.5396,
      "step": 3926
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6163327353073763,
      "learning_rate": 1.0476443693943784e-06,
      "loss": 0.4855,
      "step": 3927
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7807933292132991,
      "learning_rate": 1.047232172397587e-06,
      "loss": 0.5834,
      "step": 3928
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8805661562693393,
      "learning_rate": 1.0468199673576488e-06,
      "loss": 0.6062,
      "step": 3929
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8440207367625445,
      "learning_rate": 1.0464077543447579e-06,
      "loss": 0.6368,
      "step": 3930
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5943227126422429,
      "learning_rate": 1.0459955334291097e-06,
      "loss": 0.5306,
      "step": 3931
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7896803954432793,
      "learning_rate": 1.0455833046809014e-06,
      "loss": 0.6197,
      "step": 3932
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5917989574511008,
      "learning_rate": 1.0451710681703314e-06,
      "loss": 0.4784,
      "step": 3933
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9302562793942062,
      "learning_rate": 1.0447588239675992e-06,
      "loss": 0.6343,
      "step": 3934
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7996792728413433,
      "learning_rate": 1.044346572142905e-06,
      "loss": 0.6288,
      "step": 3935
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6283572343936312,
      "learning_rate": 1.043934312766452e-06,
      "loss": 0.5228,
      "step": 3936
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1109787567597211,
      "learning_rate": 1.0435220459084431e-06,
      "loss": 0.6398,
      "step": 3937
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6257724462070454,
      "learning_rate": 1.0431097716390834e-06,
      "loss": 0.5035,
      "step": 3938
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.053415268460185,
      "learning_rate": 1.0426974900285782e-06,
      "loss": 0.6144,
      "step": 3939
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8213883921432065,
      "learning_rate": 1.0422852011471357e-06,
      "loss": 0.5746,
      "step": 3940
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.902854843586186,
      "learning_rate": 1.0418729050649638e-06,
      "loss": 0.6444,
      "step": 3941
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8087522392584068,
      "learning_rate": 1.0414606018522725e-06,
      "loss": 0.5179,
      "step": 3942
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8729807567042316,
      "learning_rate": 1.0410482915792727e-06,
      "loss": 0.5912,
      "step": 3943
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8474807240135175,
      "learning_rate": 1.0406359743161763e-06,
      "loss": 0.6741,
      "step": 3944
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9667650959501345,
      "learning_rate": 1.0402236501331968e-06,
      "loss": 0.6589,
      "step": 3945
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9905821222995113,
      "learning_rate": 1.0398113191005488e-06,
      "loss": 0.6148,
      "step": 3946
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6928597986572332,
      "learning_rate": 1.0393989812884481e-06,
      "loss": 0.5533,
      "step": 3947
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8637172684357884,
      "learning_rate": 1.038986636767111e-06,
      "loss": 0.6296,
      "step": 3948
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6622042634972852,
      "learning_rate": 1.0385742856067563e-06,
      "loss": 0.5337,
      "step": 3949
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7128074260625501,
      "learning_rate": 1.038161927877602e-06,
      "loss": 0.5365,
      "step": 3950
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7038837964538542,
      "learning_rate": 1.0377495636498694e-06,
      "loss": 0.5547,
      "step": 3951
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8381262824610105,
      "learning_rate": 1.037337192993779e-06,
      "loss": 0.6466,
      "step": 3952
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9398967536161033,
      "learning_rate": 1.036924815979554e-06,
      "loss": 0.667,
      "step": 3953
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7910159798527919,
      "learning_rate": 1.0365124326774175e-06,
      "loss": 0.6295,
      "step": 3954
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7323735986017068,
      "learning_rate": 1.0361000431575938e-06,
      "loss": 0.5664,
      "step": 3955
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7761849050641181,
      "learning_rate": 1.0356876474903091e-06,
      "loss": 0.5664,
      "step": 3956
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8865195041143753,
      "learning_rate": 1.0352752457457898e-06,
      "loss": 0.6375,
      "step": 3957
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9427649083299603,
      "learning_rate": 1.0348628379942636e-06,
      "loss": 0.6284,
      "step": 3958
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7261646227967554,
      "learning_rate": 1.0344504243059593e-06,
      "loss": 0.5974,
      "step": 3959
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7870359545237403,
      "learning_rate": 1.0340380047511067e-06,
      "loss": 0.5265,
      "step": 3960
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6049997408918022,
      "learning_rate": 1.0336255793999364e-06,
      "loss": 0.5043,
      "step": 3961
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8874125526695527,
      "learning_rate": 1.0332131483226803e-06,
      "loss": 0.671,
      "step": 3962
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8856247810297638,
      "learning_rate": 1.032800711589571e-06,
      "loss": 0.6455,
      "step": 3963
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6199453692921119,
      "learning_rate": 1.0323882692708421e-06,
      "loss": 0.561,
      "step": 3964
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7753468811998736,
      "learning_rate": 1.0319758214367284e-06,
      "loss": 0.5393,
      "step": 3965
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8264809156062267,
      "learning_rate": 1.0315633681574656e-06,
      "loss": 0.5653,
      "step": 3966
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7300175095099699,
      "learning_rate": 1.03115090950329e-06,
      "loss": 0.5652,
      "step": 3967
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7835390353198578,
      "learning_rate": 1.0307384455444388e-06,
      "loss": 0.6166,
      "step": 3968
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6322211766332422,
      "learning_rate": 1.0303259763511507e-06,
      "loss": 0.5242,
      "step": 3969
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6380481297410306,
      "learning_rate": 1.029913501993665e-06,
      "loss": 0.5632,
      "step": 3970
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9564291574942357,
      "learning_rate": 1.029501022542221e-06,
      "loss": 0.6433,
      "step": 3971
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5898622615693613,
      "learning_rate": 1.0290885380670606e-06,
      "loss": 0.4788,
      "step": 3972
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6651670926640367,
      "learning_rate": 1.0286760486384254e-06,
      "loss": 0.5336,
      "step": 3973
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6565061791168472,
      "learning_rate": 1.0282635543265576e-06,
      "loss": 0.5363,
      "step": 3974
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7270690913400814,
      "learning_rate": 1.027851055201701e-06,
      "loss": 0.6176,
      "step": 3975
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7047257352390576,
      "learning_rate": 1.0274385513340997e-06,
      "loss": 0.5408,
      "step": 3976
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.769791455769964,
      "learning_rate": 1.0270260427939993e-06,
      "loss": 0.575,
      "step": 3977
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.800814138557785,
      "learning_rate": 1.0266135296516452e-06,
      "loss": 0.5262,
      "step": 3978
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9860361085103975,
      "learning_rate": 1.0262010119772841e-06,
      "loss": 0.6296,
      "step": 3979
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.567476773407051,
      "learning_rate": 1.0257884898411637e-06,
      "loss": 0.607,
      "step": 3980
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.874877982112583,
      "learning_rate": 1.025375963313532e-06,
      "loss": 0.6855,
      "step": 3981
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6483971447857484,
      "learning_rate": 1.0249634324646383e-06,
      "loss": 0.5226,
      "step": 3982
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.798495018448395,
      "learning_rate": 1.0245508973647321e-06,
      "loss": 0.4909,
      "step": 3983
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6942577721945494,
      "learning_rate": 1.0241383580840639e-06,
      "loss": 0.5504,
      "step": 3984
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7651904961850547,
      "learning_rate": 1.0237258146928847e-06,
      "loss": 0.5483,
      "step": 3985
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6277175529287368,
      "learning_rate": 1.0233132672614466e-06,
      "loss": 0.5178,
      "step": 3986
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7883856533443289,
      "learning_rate": 1.0229007158600018e-06,
      "loss": 0.5856,
      "step": 3987
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8361812929931454,
      "learning_rate": 1.0224881605588035e-06,
      "loss": 0.5502,
      "step": 3988
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7951486772944203,
      "learning_rate": 1.0220756014281062e-06,
      "loss": 0.5573,
      "step": 3989
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7984559250100252,
      "learning_rate": 1.0216630385381637e-06,
      "loss": 0.5613,
      "step": 3990
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.797050952454509,
      "learning_rate": 1.0212504719592314e-06,
      "loss": 0.6482,
      "step": 3991
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6302645477617852,
      "learning_rate": 1.0208379017615652e-06,
      "loss": 0.5242,
      "step": 3992
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2678002733441267,
      "learning_rate": 1.0204253280154213e-06,
      "loss": 0.6062,
      "step": 3993
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1408691660851742,
      "learning_rate": 1.0200127507910568e-06,
      "loss": 0.6947,
      "step": 3994
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8093658203588632,
      "learning_rate": 1.0196001701587292e-06,
      "loss": 0.5875,
      "step": 3995
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6515160361575365,
      "learning_rate": 1.0191875861886971e-06,
      "loss": 0.5473,
      "step": 3996
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5843407811172102,
      "learning_rate": 1.0187749989512189e-06,
      "loss": 0.5166,
      "step": 3997
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6397536714736679,
      "learning_rate": 1.018362408516554e-06,
      "loss": 0.5572,
      "step": 3998
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7639246900847029,
      "learning_rate": 1.0179498149549622e-06,
      "loss": 0.6231,
      "step": 3999
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8597893018924857,
      "learning_rate": 1.017537218336704e-06,
      "loss": 0.5589,
      "step": 4000
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7144516680330706,
      "learning_rate": 1.01712461873204e-06,
      "loss": 0.596,
      "step": 4001
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2383366817385282,
      "learning_rate": 1.016712016211232e-06,
      "loss": 0.6462,
      "step": 4002
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8790867234554465,
      "learning_rate": 1.0162994108445417e-06,
      "loss": 0.5953,
      "step": 4003
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9610151138234565,
      "learning_rate": 1.0158868027022317e-06,
      "loss": 0.6477,
      "step": 4004
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6300057160524882,
      "learning_rate": 1.0154741918545645e-06,
      "loss": 0.5366,
      "step": 4005
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6970676041769038,
      "learning_rate": 1.0150615783718038e-06,
      "loss": 0.5302,
      "step": 4006
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6910440085633736,
      "learning_rate": 1.0146489623242128e-06,
      "loss": 0.5324,
      "step": 4007
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.828281284801449,
      "learning_rate": 1.0142363437820564e-06,
      "loss": 0.5874,
      "step": 4008
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7092287996038259,
      "learning_rate": 1.013823722815599e-06,
      "loss": 0.4951,
      "step": 4009
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7324447642374535,
      "learning_rate": 1.0134110994951058e-06,
      "loss": 0.542,
      "step": 4010
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7380816150657683,
      "learning_rate": 1.0129984738908416e-06,
      "loss": 0.4936,
      "step": 4011
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7478899266467467,
      "learning_rate": 1.0125858460730733e-06,
      "loss": 0.5219,
      "step": 4012
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7172115090257446,
      "learning_rate": 1.0121732161120661e-06,
      "loss": 0.5471,
      "step": 4013
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7414220473019699,
      "learning_rate": 1.011760584078087e-06,
      "loss": 0.5417,
      "step": 4014
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6672186992636696,
      "learning_rate": 1.0113479500414032e-06,
      "loss": 0.5911,
      "step": 4015
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7478806271094177,
      "learning_rate": 1.0109353140722817e-06,
      "loss": 0.5412,
      "step": 4016
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7148435444411001,
      "learning_rate": 1.01052267624099e-06,
      "loss": 0.5978,
      "step": 4017
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0211470877494204,
      "learning_rate": 1.0101100366177962e-06,
      "loss": 0.6284,
      "step": 4018
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8132487899592956,
      "learning_rate": 1.0096973952729685e-06,
      "loss": 0.6083,
      "step": 4019
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9444978934803208,
      "learning_rate": 1.0092847522767754e-06,
      "loss": 0.6183,
      "step": 4020
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8729443688646102,
      "learning_rate": 1.0088721076994857e-06,
      "loss": 0.6258,
      "step": 4021
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6231189410952798,
      "learning_rate": 1.0084594616113687e-06,
      "loss": 0.5141,
      "step": 4022
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8521687622595465,
      "learning_rate": 1.008046814082693e-06,
      "loss": 0.6297,
      "step": 4023
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6198862621132919,
      "learning_rate": 1.007634165183729e-06,
      "loss": 0.5403,
      "step": 4024
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.239050235525572,
      "learning_rate": 1.0072215149847463e-06,
      "loss": 0.6066,
      "step": 4025
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.701925588095664,
      "learning_rate": 1.0068088635560148e-06,
      "loss": 0.5522,
      "step": 4026
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.171306524508904,
      "learning_rate": 1.0063962109678048e-06,
      "loss": 0.6609,
      "step": 4027
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6334884131240638,
      "learning_rate": 1.0059835572903866e-06,
      "loss": 0.5388,
      "step": 4028
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7646913063301076,
      "learning_rate": 1.005570902594031e-06,
      "loss": 0.5719,
      "step": 4029
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7313513058847428,
      "learning_rate": 1.0051582469490088e-06,
      "loss": 0.5531,
      "step": 4030
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.925651997099406,
      "learning_rate": 1.0047455904255907e-06,
      "loss": 0.6538,
      "step": 4031
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.934075621514861,
      "learning_rate": 1.0043329330940482e-06,
      "loss": 0.6548,
      "step": 4032
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7231849018440127,
      "learning_rate": 1.003920275024652e-06,
      "loss": 0.5393,
      "step": 4033
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8357336702110518,
      "learning_rate": 1.003507616287674e-06,
      "loss": 0.6169,
      "step": 4034
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6999159288523993,
      "learning_rate": 1.0030949569533857e-06,
      "loss": 0.5442,
      "step": 4035
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7100283487511012,
      "learning_rate": 1.002682297092058e-06,
      "loss": 0.5666,
      "step": 4036
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.747897229935645,
      "learning_rate": 1.0022696367739633e-06,
      "loss": 0.5328,
      "step": 4037
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7480696118540834,
      "learning_rate": 1.0018569760693732e-06,
      "loss": 0.5495,
      "step": 4038
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8526365738518683,
      "learning_rate": 1.0014443150485593e-06,
      "loss": 0.6171,
      "step": 4039
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8455812581849723,
      "learning_rate": 1.0010316537817938e-06,
      "loss": 0.5973,
      "step": 4040
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1796952877067763,
      "learning_rate": 1.0006189923393482e-06,
      "loss": 0.5546,
      "step": 4041
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0861382639728887,
      "learning_rate": 1.0002063307914948e-06,
      "loss": 0.6262,
      "step": 4042
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8046297798717974,
      "learning_rate": 9.997936692085052e-07,
      "loss": 0.6167,
      "step": 4043
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8613331644675869,
      "learning_rate": 9.993810076606515e-07,
      "loss": 0.5786,
      "step": 4044
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7521839550768583,
      "learning_rate": 9.98968346218206e-07,
      "loss": 0.5177,
      "step": 4045
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0166766197318529,
      "learning_rate": 9.985556849514404e-07,
      "loss": 0.6709,
      "step": 4046
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7152174356688125,
      "learning_rate": 9.981430239306265e-07,
      "loss": 0.5945,
      "step": 4047
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7015662948508359,
      "learning_rate": 9.977303632260364e-07,
      "loss": 0.5405,
      "step": 4048
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7492238653394052,
      "learning_rate": 9.973177029079418e-07,
      "loss": 0.5465,
      "step": 4049
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6741582586124618,
      "learning_rate": 9.969050430466145e-07,
      "loss": 0.5227,
      "step": 4050
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7598857786946245,
      "learning_rate": 9.964923837123258e-07,
      "loss": 0.5133,
      "step": 4051
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8080740783606586,
      "learning_rate": 9.960797249753477e-07,
      "loss": 0.5836,
      "step": 4052
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.723092833199605,
      "learning_rate": 9.956670669059517e-07,
      "loss": 0.5534,
      "step": 4053
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.845508530046538,
      "learning_rate": 9.952544095744092e-07,
      "loss": 0.5422,
      "step": 4054
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.09747854274627,
      "learning_rate": 9.94841753050991e-07,
      "loss": 0.6221,
      "step": 4055
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2684147479672245,
      "learning_rate": 9.944290974059689e-07,
      "loss": 0.6907,
      "step": 4056
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7512029971245496,
      "learning_rate": 9.940164427096131e-07,
      "loss": 0.5354,
      "step": 4057
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6525944606465818,
      "learning_rate": 9.93603789032195e-07,
      "loss": 0.5526,
      "step": 4058
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6818785319605767,
      "learning_rate": 9.93191136443985e-07,
      "loss": 0.553,
      "step": 4059
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6749978606065694,
      "learning_rate": 9.927784850152536e-07,
      "loss": 0.5619,
      "step": 4060
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.88130006013368,
      "learning_rate": 9.923658348162708e-07,
      "loss": 0.5678,
      "step": 4061
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6988150931329539,
      "learning_rate": 9.919531859173067e-07,
      "loss": 0.5893,
      "step": 4062
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8273146938615004,
      "learning_rate": 9.915405383886315e-07,
      "loss": 0.6574,
      "step": 4063
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6273508340472395,
      "learning_rate": 9.911278923005143e-07,
      "loss": 0.5359,
      "step": 4064
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9238912868431238,
      "learning_rate": 9.907152477232247e-07,
      "loss": 0.6204,
      "step": 4065
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9253023417377175,
      "learning_rate": 9.903026047270315e-07,
      "loss": 0.6476,
      "step": 4066
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0025960092813542,
      "learning_rate": 9.898899633822037e-07,
      "loss": 0.6722,
      "step": 4067
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6348855971883023,
      "learning_rate": 9.8947732375901e-07,
      "loss": 0.5166,
      "step": 4068
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6545241748638911,
      "learning_rate": 9.890646859277182e-07,
      "loss": 0.5359,
      "step": 4069
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7061468852357173,
      "learning_rate": 9.886520499585967e-07,
      "loss": 0.5072,
      "step": 4070
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8161696141213848,
      "learning_rate": 9.882394159219128e-07,
      "loss": 0.6117,
      "step": 4071
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8270990711291616,
      "learning_rate": 9.87826783887934e-07,
      "loss": 0.6336,
      "step": 4072
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8166649378440293,
      "learning_rate": 9.87414153926927e-07,
      "loss": 0.615,
      "step": 4073
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7385222545312957,
      "learning_rate": 9.870015261091583e-07,
      "loss": 0.5288,
      "step": 4074
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7777795219296944,
      "learning_rate": 9.865889005048945e-07,
      "loss": 0.5727,
      "step": 4075
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6674958023494952,
      "learning_rate": 9.86176277184401e-07,
      "loss": 0.5715,
      "step": 4076
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3205863441284955,
      "learning_rate": 9.857636562179437e-07,
      "loss": 0.6632,
      "step": 4077
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9283006507110512,
      "learning_rate": 9.853510376757873e-07,
      "loss": 0.6157,
      "step": 4078
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.72809740878621,
      "learning_rate": 9.849384216281965e-07,
      "loss": 0.6113,
      "step": 4079
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3967767993163964,
      "learning_rate": 9.845258081454354e-07,
      "loss": 0.6203,
      "step": 4080
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6185487967576118,
      "learning_rate": 9.841131972977684e-07,
      "loss": 0.5354,
      "step": 4081
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8401851779178235,
      "learning_rate": 9.837005891554582e-07,
      "loss": 0.5342,
      "step": 4082
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8334001564380961,
      "learning_rate": 9.83287983788768e-07,
      "loss": 0.624,
      "step": 4083
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7278718850449575,
      "learning_rate": 9.8287538126796e-07,
      "loss": 0.5242,
      "step": 4084
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0845750589989187,
      "learning_rate": 9.824627816632961e-07,
      "loss": 0.5557,
      "step": 4085
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6759707660138367,
      "learning_rate": 9.820501850450377e-07,
      "loss": 0.5345,
      "step": 4086
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7191545147028133,
      "learning_rate": 9.816375914834461e-07,
      "loss": 0.5386,
      "step": 4087
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0861492873234466,
      "learning_rate": 9.812250010487812e-07,
      "loss": 0.6304,
      "step": 4088
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.022119259564415,
      "learning_rate": 9.808124138113028e-07,
      "loss": 0.6755,
      "step": 4089
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8835588333804869,
      "learning_rate": 9.803998298412709e-07,
      "loss": 0.595,
      "step": 4090
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7014126960393718,
      "learning_rate": 9.799872492089433e-07,
      "loss": 0.5778,
      "step": 4091
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6175034394161963,
      "learning_rate": 9.795746719845788e-07,
      "loss": 0.4809,
      "step": 4092
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9347880628166808,
      "learning_rate": 9.79162098238435e-07,
      "loss": 0.6007,
      "step": 4093
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8503609216707406,
      "learning_rate": 9.787495280407687e-07,
      "loss": 0.5999,
      "step": 4094
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6503850637885226,
      "learning_rate": 9.783369614618364e-07,
      "loss": 0.5497,
      "step": 4095
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8699228803576291,
      "learning_rate": 9.77924398571894e-07,
      "loss": 0.6623,
      "step": 4096
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7009152718132402,
      "learning_rate": 9.775118394411964e-07,
      "loss": 0.5476,
      "step": 4097
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.692576490631869,
      "learning_rate": 9.770992841399983e-07,
      "loss": 0.5777,
      "step": 4098
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7443609267917054,
      "learning_rate": 9.766867327385535e-07,
      "loss": 0.5534,
      "step": 4099
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7562628993151985,
      "learning_rate": 9.762741853071152e-07,
      "loss": 0.5819,
      "step": 4100
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.862377459986156,
      "learning_rate": 9.75861641915936e-07,
      "loss": 0.6008,
      "step": 4101
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8167814227376619,
      "learning_rate": 9.75449102635268e-07,
      "loss": 0.5602,
      "step": 4102
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8120384608859262,
      "learning_rate": 9.750365675353618e-07,
      "loss": 0.587,
      "step": 4103
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8127878281621244,
      "learning_rate": 9.74624036686468e-07,
      "loss": 0.5405,
      "step": 4104
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9096109424939246,
      "learning_rate": 9.742115101588362e-07,
      "loss": 0.5947,
      "step": 4105
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6686785476644775,
      "learning_rate": 9.73798988022716e-07,
      "loss": 0.5708,
      "step": 4106
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8810009263291421,
      "learning_rate": 9.73386470348355e-07,
      "loss": 0.5863,
      "step": 4107
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7336351978172198,
      "learning_rate": 9.729739572060009e-07,
      "loss": 0.5689,
      "step": 4108
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.639124592665011,
      "learning_rate": 9.725614486659004e-07,
      "loss": 0.4765,
      "step": 4109
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7446335712629271,
      "learning_rate": 9.721489447982992e-07,
      "loss": 0.48,
      "step": 4110
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1309245837710413,
      "learning_rate": 9.717364456734426e-07,
      "loss": 0.6334,
      "step": 4111
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.894309706664065,
      "learning_rate": 9.713239513615747e-07,
      "loss": 0.5599,
      "step": 4112
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8825534767707301,
      "learning_rate": 9.709114619329393e-07,
      "loss": 0.6155,
      "step": 4113
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5794773782431641,
      "learning_rate": 9.704989774577788e-07,
      "loss": 0.5106,
      "step": 4114
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0458752187601215,
      "learning_rate": 9.700864980063351e-07,
      "loss": 0.6638,
      "step": 4115
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8495442961341314,
      "learning_rate": 9.696740236488494e-07,
      "loss": 0.6258,
      "step": 4116
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7332250736427953,
      "learning_rate": 9.692615544555613e-07,
      "loss": 0.531,
      "step": 4117
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8433618558250839,
      "learning_rate": 9.688490904967102e-07,
      "loss": 0.6506,
      "step": 4118
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6553116035863851,
      "learning_rate": 9.684366318425345e-07,
      "loss": 0.5569,
      "step": 4119
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7376833076759458,
      "learning_rate": 9.680241785632717e-07,
      "loss": 0.5834,
      "step": 4120
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.893288830920056,
      "learning_rate": 9.67611730729158e-07,
      "loss": 0.5953,
      "step": 4121
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6819016325338135,
      "learning_rate": 9.671992884104294e-07,
      "loss": 0.5777,
      "step": 4122
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8342987725345183,
      "learning_rate": 9.6678685167732e-07,
      "loss": 0.6459,
      "step": 4123
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9615942048320508,
      "learning_rate": 9.66374420600064e-07,
      "loss": 0.6392,
      "step": 4124
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6929793217487995,
      "learning_rate": 9.659619952488935e-07,
      "loss": 0.5604,
      "step": 4125
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6482765316986931,
      "learning_rate": 9.655495756940409e-07,
      "loss": 0.5591,
      "step": 4126
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9203992084135164,
      "learning_rate": 9.651371620057365e-07,
      "loss": 0.6121,
      "step": 4127
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8632544329518762,
      "learning_rate": 9.647247542542103e-07,
      "loss": 0.6373,
      "step": 4128
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8475211859031183,
      "learning_rate": 9.643123525096912e-07,
      "loss": 0.6143,
      "step": 4129
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6294951497084769,
      "learning_rate": 9.638999568424063e-07,
      "loss": 0.5649,
      "step": 4130
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5986176438968739,
      "learning_rate": 9.634875673225826e-07,
      "loss": 0.5217,
      "step": 4131
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7860857569223271,
      "learning_rate": 9.63075184020446e-07,
      "loss": 0.6208,
      "step": 4132
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9030222268724944,
      "learning_rate": 9.626628070062208e-07,
      "loss": 0.6027,
      "step": 4133
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8001817069233831,
      "learning_rate": 9.622504363501307e-07,
      "loss": 0.523,
      "step": 4134
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7330516788618869,
      "learning_rate": 9.61838072122398e-07,
      "loss": 0.5218,
      "step": 4135
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8144771634172261,
      "learning_rate": 9.61425714393244e-07,
      "loss": 0.6025,
      "step": 4136
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9175324610900722,
      "learning_rate": 9.61013363232889e-07,
      "loss": 0.6606,
      "step": 4137
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6340844375508794,
      "learning_rate": 9.606010187115522e-07,
      "loss": 0.5168,
      "step": 4138
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7692249481244269,
      "learning_rate": 9.601886808994513e-07,
      "loss": 0.5542,
      "step": 4139
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8024708904434787,
      "learning_rate": 9.597763498668033e-07,
      "loss": 0.5289,
      "step": 4140
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7186982380169172,
      "learning_rate": 9.593640256838238e-07,
      "loss": 0.56,
      "step": 4141
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6687348417920621,
      "learning_rate": 9.589517084207276e-07,
      "loss": 0.5466,
      "step": 4142
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9579639208483804,
      "learning_rate": 9.585393981477278e-07,
      "loss": 0.6062,
      "step": 4143
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0112724464652076,
      "learning_rate": 9.581270949350365e-07,
      "loss": 0.6732,
      "step": 4144
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7240882816628017,
      "learning_rate": 9.577147988528642e-07,
      "loss": 0.5245,
      "step": 4145
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7697764037823359,
      "learning_rate": 9.573025099714217e-07,
      "loss": 0.5474,
      "step": 4146
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7279697277335438,
      "learning_rate": 9.568902283609167e-07,
      "loss": 0.5966,
      "step": 4147
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5965898761272731,
      "learning_rate": 9.564779540915568e-07,
      "loss": 0.515,
      "step": 4148
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0917314470292665,
      "learning_rate": 9.56065687233548e-07,
      "loss": 0.6098,
      "step": 4149
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8472349514126258,
      "learning_rate": 9.556534278570946e-07,
      "loss": 0.6872,
      "step": 4150
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8689391579881356,
      "learning_rate": 9.552411760324008e-07,
      "loss": 0.6189,
      "step": 4151
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7104869424942737,
      "learning_rate": 9.548289318296683e-07,
      "loss": 0.568,
      "step": 4152
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8172918081341343,
      "learning_rate": 9.544166953190983e-07,
      "loss": 0.5752,
      "step": 4153
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8921670432884192,
      "learning_rate": 9.5400446657089e-07,
      "loss": 0.6329,
      "step": 4154
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6293551508688074,
      "learning_rate": 9.535922456552421e-07,
      "loss": 0.4918,
      "step": 4155
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8739927866237936,
      "learning_rate": 9.531800326423512e-07,
      "loss": 0.6089,
      "step": 4156
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8425753363889352,
      "learning_rate": 9.527678276024128e-07,
      "loss": 0.555,
      "step": 4157
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8136250011780449,
      "learning_rate": 9.523556306056215e-07,
      "loss": 0.6575,
      "step": 4158
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7414935937144411,
      "learning_rate": 9.519434417221699e-07,
      "loss": 0.5045,
      "step": 4159
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8652014013202736,
      "learning_rate": 9.515312610222495e-07,
      "loss": 0.6066,
      "step": 4160
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.920411067000238,
      "learning_rate": 9.511190885760504e-07,
      "loss": 0.6239,
      "step": 4161
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6747263830882381,
      "learning_rate": 9.507069244537612e-07,
      "loss": 0.5323,
      "step": 4162
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.623834230786709,
      "learning_rate": 9.502947687255692e-07,
      "loss": 0.4948,
      "step": 4163
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.807972946340756,
      "learning_rate": 9.498826214616603e-07,
      "loss": 0.6457,
      "step": 4164
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9051060059517153,
      "learning_rate": 9.49470482732219e-07,
      "loss": 0.6695,
      "step": 4165
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5634847321969713,
      "learning_rate": 9.490583526074279e-07,
      "loss": 0.5021,
      "step": 4166
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.87319012061689,
      "learning_rate": 9.486462311574688e-07,
      "loss": 0.6282,
      "step": 4167
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.751955318502002,
      "learning_rate": 9.482341184525214e-07,
      "loss": 0.6161,
      "step": 4168
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8951591878594662,
      "learning_rate": 9.478220145627644e-07,
      "loss": 0.652,
      "step": 4169
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8045176229465819,
      "learning_rate": 9.474099195583749e-07,
      "loss": 0.6194,
      "step": 4170
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8309263624041435,
      "learning_rate": 9.46997833509528e-07,
      "loss": 0.6486,
      "step": 4171
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7166931918852805,
      "learning_rate": 9.465857564863979e-07,
      "loss": 0.5201,
      "step": 4172
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8190482028306955,
      "learning_rate": 9.461736885591571e-07,
      "loss": 0.672,
      "step": 4173
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.783166751335455,
      "learning_rate": 9.457616297979763e-07,
      "loss": 0.5927,
      "step": 4174
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7054808774759557,
      "learning_rate": 9.45349580273025e-07,
      "loss": 0.5199,
      "step": 4175
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9386648757027651,
      "learning_rate": 9.449375400544707e-07,
      "loss": 0.6147,
      "step": 4176
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8774987289703396,
      "learning_rate": 9.445255092124796e-07,
      "loss": 0.6254,
      "step": 4177
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7102529867092512,
      "learning_rate": 9.441134878172165e-07,
      "loss": 0.5191,
      "step": 4178
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7203824021184562,
      "learning_rate": 9.437014759388442e-07,
      "loss": 0.5382,
      "step": 4179
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.635036269412544,
      "learning_rate": 9.43289473647524e-07,
      "loss": 0.5193,
      "step": 4180
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6875585516314872,
      "learning_rate": 9.428774810134158e-07,
      "loss": 0.5239,
      "step": 4181
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9248002396459893,
      "learning_rate": 9.424654981066773e-07,
      "loss": 0.5802,
      "step": 4182
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6934972415896687,
      "learning_rate": 9.420535249974652e-07,
      "loss": 0.585,
      "step": 4183
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.612241522579423,
      "learning_rate": 9.41641561755934e-07,
      "loss": 0.6191,
      "step": 4184
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7983685601355182,
      "learning_rate": 9.41229608452237e-07,
      "loss": 0.5765,
      "step": 4185
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6549671047661653,
      "learning_rate": 9.408176651565253e-07,
      "loss": 0.5718,
      "step": 4186
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8806633235662676,
      "learning_rate": 9.404057319389487e-07,
      "loss": 0.6016,
      "step": 4187
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6274154977745044,
      "learning_rate": 9.399938088696551e-07,
      "loss": 0.5235,
      "step": 4188
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7170616100805673,
      "learning_rate": 9.395818960187906e-07,
      "loss": 0.5218,
      "step": 4189
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6319217083116779,
      "learning_rate": 9.391699934564998e-07,
      "loss": 0.5434,
      "step": 4190
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.060098937663879,
      "learning_rate": 9.387581012529254e-07,
      "loss": 0.6878,
      "step": 4191
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9027671311727884,
      "learning_rate": 9.383462194782085e-07,
      "loss": 0.6688,
      "step": 4192
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7463820968525148,
      "learning_rate": 9.379343482024879e-07,
      "loss": 0.6053,
      "step": 4193
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.755942997218373,
      "learning_rate": 9.375224874959014e-07,
      "loss": 0.5625,
      "step": 4194
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9487369744149864,
      "learning_rate": 9.371106374285844e-07,
      "loss": 0.6172,
      "step": 4195
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9615080882888112,
      "learning_rate": 9.366987980706707e-07,
      "loss": 0.6151,
      "step": 4196
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8182133658377385,
      "learning_rate": 9.36286969492292e-07,
      "loss": 0.6226,
      "step": 4197
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6246577628125036,
      "learning_rate": 9.358751517635788e-07,
      "loss": 0.5315,
      "step": 4198
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7752934716701233,
      "learning_rate": 9.354633449546592e-07,
      "loss": 0.5403,
      "step": 4199
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8099846986146835,
      "learning_rate": 9.350515491356596e-07,
      "loss": 0.5239,
      "step": 4200
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7690886537026896,
      "learning_rate": 9.346397643767045e-07,
      "loss": 0.5865,
      "step": 4201
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6886239649275603,
      "learning_rate": 9.342279907479167e-07,
      "loss": 0.5827,
      "step": 4202
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.615303133766021,
      "learning_rate": 9.338162283194165e-07,
      "loss": 0.5387,
      "step": 4203
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6035741787730857,
      "learning_rate": 9.334044771613235e-07,
      "loss": 0.5374,
      "step": 4204
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9010916903133757,
      "learning_rate": 9.329927373437542e-07,
      "loss": 0.5897,
      "step": 4205
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7559971978967951,
      "learning_rate": 9.325810089368236e-07,
      "loss": 0.5562,
      "step": 4206
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.800751156836085,
      "learning_rate": 9.321692920106449e-07,
      "loss": 0.5412,
      "step": 4207
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7687374585983546,
      "learning_rate": 9.317575866353291e-07,
      "loss": 0.5667,
      "step": 4208
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8648650869801775,
      "learning_rate": 9.313458928809853e-07,
      "loss": 0.5964,
      "step": 4209
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6996239344262282,
      "learning_rate": 9.309342108177207e-07,
      "loss": 0.5678,
      "step": 4210
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5802374983355189,
      "learning_rate": 9.305225405156405e-07,
      "loss": 0.5184,
      "step": 4211
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0814671941293699,
      "learning_rate": 9.301108820448479e-07,
      "loss": 0.6023,
      "step": 4212
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6964403863164248,
      "learning_rate": 9.29699235475444e-07,
      "loss": 0.6232,
      "step": 4213
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6588083433173392,
      "learning_rate": 9.292876008775278e-07,
      "loss": 0.4995,
      "step": 4214
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6503971182274163,
      "learning_rate": 9.288759783211966e-07,
      "loss": 0.5332,
      "step": 4215
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7419889624899512,
      "learning_rate": 9.284643678765451e-07,
      "loss": 0.5873,
      "step": 4216
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6810367432761317,
      "learning_rate": 9.280527696136665e-07,
      "loss": 0.5054,
      "step": 4217
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8504324398746961,
      "learning_rate": 9.276411836026519e-07,
      "loss": 0.5888,
      "step": 4218
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6624640151641176,
      "learning_rate": 9.272296099135899e-07,
      "loss": 0.5312,
      "step": 4219
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7081170116488755,
      "learning_rate": 9.268180486165672e-07,
      "loss": 0.5521,
      "step": 4220
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7885729177519016,
      "learning_rate": 9.264064997816684e-07,
      "loss": 0.5333,
      "step": 4221
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8245222865873894,
      "learning_rate": 9.25994963478976e-07,
      "loss": 0.6442,
      "step": 4222
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6687644250895864,
      "learning_rate": 9.255834397785703e-07,
      "loss": 0.5331,
      "step": 4223
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7559072731626507,
      "learning_rate": 9.251719287505295e-07,
      "loss": 0.5778,
      "step": 4224
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6304886966194684,
      "learning_rate": 9.247604304649298e-07,
      "loss": 0.5677,
      "step": 4225
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8002334460815927,
      "learning_rate": 9.24348944991845e-07,
      "loss": 0.561,
      "step": 4226
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9670193810656911,
      "learning_rate": 9.239374724013466e-07,
      "loss": 0.6129,
      "step": 4227
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7361565055771447,
      "learning_rate": 9.235260127635043e-07,
      "loss": 0.4944,
      "step": 4228
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7951832168100417,
      "learning_rate": 9.23114566148385e-07,
      "loss": 0.5278,
      "step": 4229
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8816036566802652,
      "learning_rate": 9.227031326260544e-07,
      "loss": 0.5968,
      "step": 4230
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.701058006739136,
      "learning_rate": 9.222917122665751e-07,
      "loss": 0.5358,
      "step": 4231
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8434419968040899,
      "learning_rate": 9.218803051400076e-07,
      "loss": 0.571,
      "step": 4232
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9220581378185699,
      "learning_rate": 9.214689113164103e-07,
      "loss": 0.6004,
      "step": 4233
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7369616029126502,
      "learning_rate": 9.210575308658393e-07,
      "loss": 0.523,
      "step": 4234
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8344457780753681,
      "learning_rate": 9.206461638583482e-07,
      "loss": 0.575,
      "step": 4235
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7136621231049342,
      "learning_rate": 9.202348103639889e-07,
      "loss": 0.5416,
      "step": 4236
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7389894331702248,
      "learning_rate": 9.198234704528102e-07,
      "loss": 0.511,
      "step": 4237
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7914182617678724,
      "learning_rate": 9.194121441948594e-07,
      "loss": 0.5527,
      "step": 4238
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6998737358799848,
      "learning_rate": 9.190008316601808e-07,
      "loss": 0.548,
      "step": 4239
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1144381728602772,
      "learning_rate": 9.185895329188168e-07,
      "loss": 0.673,
      "step": 4240
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3112793676318024,
      "learning_rate": 9.181782480408071e-07,
      "loss": 0.6057,
      "step": 4241
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1824852085945596,
      "learning_rate": 9.177669770961893e-07,
      "loss": 0.6408,
      "step": 4242
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8588036145415717,
      "learning_rate": 9.173557201549987e-07,
      "loss": 0.5546,
      "step": 4243
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6916381799373548,
      "learning_rate": 9.16944477287268e-07,
      "loss": 0.5518,
      "step": 4244
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.019248619479582,
      "learning_rate": 9.165332485630275e-07,
      "loss": 0.6275,
      "step": 4245
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8266025853741792,
      "learning_rate": 9.16122034052305e-07,
      "loss": 0.5984,
      "step": 4246
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.790795855755217,
      "learning_rate": 9.157108338251265e-07,
      "loss": 0.6308,
      "step": 4247
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8063475011404077,
      "learning_rate": 9.152996479515147e-07,
      "loss": 0.6056,
      "step": 4248
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7094044881662649,
      "learning_rate": 9.148884765014907e-07,
      "loss": 0.5366,
      "step": 4249
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8734309355566028,
      "learning_rate": 9.144773195450725e-07,
      "loss": 0.6106,
      "step": 4250
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7812895535800443,
      "learning_rate": 9.140661771522758e-07,
      "loss": 0.5527,
      "step": 4251
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9131467945721664,
      "learning_rate": 9.136550493931136e-07,
      "loss": 0.6349,
      "step": 4252
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.64972067695835,
      "learning_rate": 9.132439363375972e-07,
      "loss": 0.5359,
      "step": 4253
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6024688009271509,
      "learning_rate": 9.128328380557346e-07,
      "loss": 0.5239,
      "step": 4254
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6943213785525026,
      "learning_rate": 9.124217546175317e-07,
      "loss": 0.5244,
      "step": 4255
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6711773526033229,
      "learning_rate": 9.120106860929914e-07,
      "loss": 0.5069,
      "step": 4256
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7370365690701016,
      "learning_rate": 9.115996325521147e-07,
      "loss": 0.5793,
      "step": 4257
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9607557265109317,
      "learning_rate": 9.111885940648995e-07,
      "loss": 0.5774,
      "step": 4258
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.729675861972166,
      "learning_rate": 9.107775707013417e-07,
      "loss": 0.5849,
      "step": 4259
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6159676931679806,
      "learning_rate": 9.103665625314339e-07,
      "loss": 0.5242,
      "step": 4260
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6311393671390613,
      "learning_rate": 9.099555696251666e-07,
      "loss": 0.5439,
      "step": 4261
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.973075032003322,
      "learning_rate": 9.09544592052528e-07,
      "loss": 0.6583,
      "step": 4262
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.664693632049828,
      "learning_rate": 9.091336298835029e-07,
      "loss": 0.5571,
      "step": 4263
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6225378340117613,
      "learning_rate": 9.087226831880741e-07,
      "loss": 0.5328,
      "step": 4264
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8626233023735204,
      "learning_rate": 9.083117520362214e-07,
      "loss": 0.6063,
      "step": 4265
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7037474555819017,
      "learning_rate": 9.07900836497922e-07,
      "loss": 0.6001,
      "step": 4266
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7876191454985318,
      "learning_rate": 9.074899366431508e-07,
      "loss": 0.6011,
      "step": 4267
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8925878416249354,
      "learning_rate": 9.070790525418796e-07,
      "loss": 0.6071,
      "step": 4268
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8826752981924412,
      "learning_rate": 9.066681842640776e-07,
      "loss": 0.6402,
      "step": 4269
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8027240549731722,
      "learning_rate": 9.062573318797115e-07,
      "loss": 0.6166,
      "step": 4270
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6629788585223861,
      "learning_rate": 9.058464954587454e-07,
      "loss": 0.5439,
      "step": 4271
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7073235836323415,
      "learning_rate": 9.0543567507114e-07,
      "loss": 0.5219,
      "step": 4272
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8553963697294419,
      "learning_rate": 9.05024870786854e-07,
      "loss": 0.5751,
      "step": 4273
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9603666097145462,
      "learning_rate": 9.046140826758431e-07,
      "loss": 0.647,
      "step": 4274
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8930712127459738,
      "learning_rate": 9.042033108080601e-07,
      "loss": 0.6731,
      "step": 4275
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9915107142478033,
      "learning_rate": 9.037925552534555e-07,
      "loss": 0.6102,
      "step": 4276
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.848178591658344,
      "learning_rate": 9.033818160819763e-07,
      "loss": 0.6469,
      "step": 4277
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6897620600640135,
      "learning_rate": 9.029710933635672e-07,
      "loss": 0.5506,
      "step": 4278
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7163357573932562,
      "learning_rate": 9.025603871681703e-07,
      "loss": 0.5245,
      "step": 4279
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7488244028779203,
      "learning_rate": 9.021496975657242e-07,
      "loss": 0.5558,
      "step": 4280
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6843840578917433,
      "learning_rate": 9.017390246261653e-07,
      "loss": 0.5473,
      "step": 4281
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6955301382588726,
      "learning_rate": 9.013283684194266e-07,
      "loss": 0.5324,
      "step": 4282
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0469520452571741,
      "learning_rate": 9.00917729015439e-07,
      "loss": 0.6166,
      "step": 4283
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8937561573387722,
      "learning_rate": 9.005071064841299e-07,
      "loss": 0.6304,
      "step": 4284
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9519448065067486,
      "learning_rate": 9.00096500895424e-07,
      "loss": 0.6801,
      "step": 4285
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6698841963338981,
      "learning_rate": 8.996859123192433e-07,
      "loss": 0.5287,
      "step": 4286
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8671753597844356,
      "learning_rate": 8.992753408255066e-07,
      "loss": 0.601,
      "step": 4287
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.854077686564304,
      "learning_rate": 8.988647864841298e-07,
      "loss": 0.6285,
      "step": 4288
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9107788386640909,
      "learning_rate": 8.984542493650264e-07,
      "loss": 0.6099,
      "step": 4289
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9853097283346544,
      "learning_rate": 8.980437295381066e-07,
      "loss": 0.6807,
      "step": 4290
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9091257977303385,
      "learning_rate": 8.976332270732774e-07,
      "loss": 0.6033,
      "step": 4291
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7512386935749161,
      "learning_rate": 8.972227420404432e-07,
      "loss": 0.4975,
      "step": 4292
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.929334066789151,
      "learning_rate": 8.968122745095052e-07,
      "loss": 0.655,
      "step": 4293
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8926586015303035,
      "learning_rate": 8.964018245503619e-07,
      "loss": 0.6033,
      "step": 4294
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9385424787861008,
      "learning_rate": 8.959913922329086e-07,
      "loss": 0.6549,
      "step": 4295
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6254800364839701,
      "learning_rate": 8.955809776270378e-07,
      "loss": 0.5057,
      "step": 4296
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7901618241725072,
      "learning_rate": 8.951705808026385e-07,
      "loss": 0.6265,
      "step": 4297
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.674177411548708,
      "learning_rate": 8.947602018295972e-07,
      "loss": 0.5023,
      "step": 4298
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8154719348406759,
      "learning_rate": 8.943498407777973e-07,
      "loss": 0.5759,
      "step": 4299
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8501513334850993,
      "learning_rate": 8.939394977171188e-07,
      "loss": 0.6582,
      "step": 4300
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6169060023111056,
      "learning_rate": 8.935291727174385e-07,
      "loss": 0.4984,
      "step": 4301
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9119839941455917,
      "learning_rate": 8.931188658486313e-07,
      "loss": 0.6606,
      "step": 4302
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.172266408554823,
      "learning_rate": 8.927085771805677e-07,
      "loss": 0.6224,
      "step": 4303
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7717861737413592,
      "learning_rate": 8.922983067831156e-07,
      "loss": 0.5011,
      "step": 4304
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7034668522712864,
      "learning_rate": 8.918880547261399e-07,
      "loss": 0.4934,
      "step": 4305
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6901386263488744,
      "learning_rate": 8.914778210795022e-07,
      "loss": 0.5433,
      "step": 4306
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.115795217953946,
      "learning_rate": 8.91067605913061e-07,
      "loss": 0.657,
      "step": 4307
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.87187866267265,
      "learning_rate": 8.906574092966716e-07,
      "loss": 0.6415,
      "step": 4308
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6630197762062282,
      "learning_rate": 8.902472313001863e-07,
      "loss": 0.537,
      "step": 4309
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8456538133432309,
      "learning_rate": 8.898370719934541e-07,
      "loss": 0.6959,
      "step": 4310
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8157688806982841,
      "learning_rate": 8.894269314463209e-07,
      "loss": 0.5515,
      "step": 4311
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7714051483240109,
      "learning_rate": 8.890168097286293e-07,
      "loss": 0.5564,
      "step": 4312
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.856294607042504,
      "learning_rate": 8.886067069102188e-07,
      "loss": 0.6507,
      "step": 4313
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7001480784581334,
      "learning_rate": 8.881966230609253e-07,
      "loss": 0.5935,
      "step": 4314
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0243738644820575,
      "learning_rate": 8.877865582505826e-07,
      "loss": 0.7513,
      "step": 4315
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7980486920681704,
      "learning_rate": 8.873765125490198e-07,
      "loss": 0.6407,
      "step": 4316
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8899862037077803,
      "learning_rate": 8.869664860260638e-07,
      "loss": 0.6786,
      "step": 4317
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9578675815887929,
      "learning_rate": 8.865564787515374e-07,
      "loss": 0.6436,
      "step": 4318
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9347230672494605,
      "learning_rate": 8.86146490795261e-07,
      "loss": 0.5932,
      "step": 4319
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7594051120356744,
      "learning_rate": 8.85736522227051e-07,
      "loss": 0.6453,
      "step": 4320
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6622924231339418,
      "learning_rate": 8.853265731167209e-07,
      "loss": 0.5969,
      "step": 4321
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9221515064131514,
      "learning_rate": 8.849166435340808e-07,
      "loss": 0.68,
      "step": 4322
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.111433727786823,
      "learning_rate": 8.845067335489373e-07,
      "loss": 0.5907,
      "step": 4323
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7451623818425012,
      "learning_rate": 8.840968432310939e-07,
      "loss": 0.6289,
      "step": 4324
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8002511370589401,
      "learning_rate": 8.836869726503506e-07,
      "loss": 0.5754,
      "step": 4325
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9837739990994597,
      "learning_rate": 8.832771218765043e-07,
      "loss": 0.6249,
      "step": 4326
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7656427804748445,
      "learning_rate": 8.828672909793477e-07,
      "loss": 0.6826,
      "step": 4327
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9315810794760181,
      "learning_rate": 8.824574800286715e-07,
      "loss": 0.5702,
      "step": 4328
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9739677109187496,
      "learning_rate": 8.820476890942618e-07,
      "loss": 0.7022,
      "step": 4329
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7862195473905366,
      "learning_rate": 8.81637918245902e-07,
      "loss": 0.6509,
      "step": 4330
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6116817229526151,
      "learning_rate": 8.812281675533715e-07,
      "loss": 0.5643,
      "step": 4331
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7945559412592476,
      "learning_rate": 8.808184370864467e-07,
      "loss": 0.635,
      "step": 4332
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9161091054723991,
      "learning_rate": 8.804087269149006e-07,
      "loss": 0.618,
      "step": 4333
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6730060510387534,
      "learning_rate": 8.799990371085022e-07,
      "loss": 0.5561,
      "step": 4334
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5714905441344984,
      "learning_rate": 8.795893677370176e-07,
      "loss": 0.5101,
      "step": 4335
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8273489649008264,
      "learning_rate": 8.791797188702095e-07,
      "loss": 0.607,
      "step": 4336
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9701996840807429,
      "learning_rate": 8.787700905778363e-07,
      "loss": 0.6709,
      "step": 4337
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9998299616877546,
      "learning_rate": 8.78360482929654e-07,
      "loss": 0.7,
      "step": 4338
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6526094794044471,
      "learning_rate": 8.77950895995414e-07,
      "loss": 0.525,
      "step": 4339
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8686475812395749,
      "learning_rate": 8.775413298448647e-07,
      "loss": 0.6289,
      "step": 4340
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6598174778975326,
      "learning_rate": 8.771317845477515e-07,
      "loss": 0.5122,
      "step": 4341
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.04954213513913,
      "learning_rate": 8.767222601738154e-07,
      "loss": 0.641,
      "step": 4342
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8685400399035179,
      "learning_rate": 8.763127567927939e-07,
      "loss": 0.646,
      "step": 4343
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8192464439169792,
      "learning_rate": 8.759032744744214e-07,
      "loss": 0.6072,
      "step": 4344
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7489210133290618,
      "learning_rate": 8.754938132884285e-07,
      "loss": 0.6366,
      "step": 4345
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8562919675452393,
      "learning_rate": 8.750843733045421e-07,
      "loss": 0.5879,
      "step": 4346
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6834417499868792,
      "learning_rate": 8.746749545924856e-07,
      "loss": 0.5602,
      "step": 4347
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6921782362157067,
      "learning_rate": 8.742655572219781e-07,
      "loss": 0.5603,
      "step": 4348
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6702685895705479,
      "learning_rate": 8.73856181262737e-07,
      "loss": 0.4854,
      "step": 4349
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8258815100226267,
      "learning_rate": 8.734468267844739e-07,
      "loss": 0.5865,
      "step": 4350
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8945023333176835,
      "learning_rate": 8.730374938568978e-07,
      "loss": 0.6034,
      "step": 4351
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.953152374906618,
      "learning_rate": 8.726281825497139e-07,
      "loss": 0.6409,
      "step": 4352
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8105626421193868,
      "learning_rate": 8.722188929326235e-07,
      "loss": 0.6261,
      "step": 4353
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9839243328410233,
      "learning_rate": 8.718096250753245e-07,
      "loss": 0.6381,
      "step": 4354
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7892151402751505,
      "learning_rate": 8.714003790475106e-07,
      "loss": 0.5957,
      "step": 4355
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6128995201935047,
      "learning_rate": 8.709911549188727e-07,
      "loss": 0.4879,
      "step": 4356
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6598930583064968,
      "learning_rate": 8.70581952759097e-07,
      "loss": 0.5265,
      "step": 4357
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9934205399315076,
      "learning_rate": 8.701727726378665e-07,
      "loss": 0.6281,
      "step": 4358
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7137625407409584,
      "learning_rate": 8.6976361462486e-07,
      "loss": 0.5735,
      "step": 4359
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7889458367502884,
      "learning_rate": 8.693544787897533e-07,
      "loss": 0.5324,
      "step": 4360
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6743913393827506,
      "learning_rate": 8.68945365202218e-07,
      "loss": 0.5435,
      "step": 4361
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.754816607889499,
      "learning_rate": 8.685362739319215e-07,
      "loss": 0.5406,
      "step": 4362
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0339312428289114,
      "learning_rate": 8.681272050485282e-07,
      "loss": 0.6476,
      "step": 4363
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8828034006597015,
      "learning_rate": 8.677181586216978e-07,
      "loss": 0.5851,
      "step": 4364
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0901390978262449,
      "learning_rate": 8.67309134721087e-07,
      "loss": 0.5864,
      "step": 4365
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8774822801944981,
      "learning_rate": 8.669001334163483e-07,
      "loss": 0.6177,
      "step": 4366
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9698201288952298,
      "learning_rate": 8.6649115477713e-07,
      "loss": 0.6376,
      "step": 4367
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7167055012005153,
      "learning_rate": 8.660821988730775e-07,
      "loss": 0.6135,
      "step": 4368
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6766076268670432,
      "learning_rate": 8.65673265773831e-07,
      "loss": 0.5462,
      "step": 4369
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8601286818164767,
      "learning_rate": 8.652643555490282e-07,
      "loss": 0.6669,
      "step": 4370
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0456136896856079,
      "learning_rate": 8.648554682683019e-07,
      "loss": 0.6421,
      "step": 4371
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8218063040494518,
      "learning_rate": 8.644466040012812e-07,
      "loss": 0.5018,
      "step": 4372
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6717431055626376,
      "learning_rate": 8.640377628175919e-07,
      "loss": 0.5744,
      "step": 4373
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7887283319809039,
      "learning_rate": 8.636289447868551e-07,
      "loss": 0.5466,
      "step": 4374
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7858057501818361,
      "learning_rate": 8.632201499786884e-07,
      "loss": 0.6062,
      "step": 4375
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8673568392837707,
      "learning_rate": 8.628113784627051e-07,
      "loss": 0.6158,
      "step": 4376
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7034664482802332,
      "learning_rate": 8.624026303085149e-07,
      "loss": 0.5608,
      "step": 4377
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8124035468770514,
      "learning_rate": 8.619939055857232e-07,
      "loss": 0.6704,
      "step": 4378
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8988299765412034,
      "learning_rate": 8.615852043639317e-07,
      "loss": 0.5984,
      "step": 4379
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9003321581839285,
      "learning_rate": 8.611765267127379e-07,
      "loss": 0.5741,
      "step": 4380
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6989153265172532,
      "learning_rate": 8.607678727017354e-07,
      "loss": 0.6974,
      "step": 4381
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0373341257473043,
      "learning_rate": 8.603592424005135e-07,
      "loss": 0.698,
      "step": 4382
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9295846536302472,
      "learning_rate": 8.599506358786579e-07,
      "loss": 0.5879,
      "step": 4383
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9416069470398654,
      "learning_rate": 8.5954205320575e-07,
      "loss": 0.642,
      "step": 4384
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8115604261071249,
      "learning_rate": 8.59133494451367e-07,
      "loss": 0.6357,
      "step": 4385
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6589826222363065,
      "learning_rate": 8.587249596850826e-07,
      "loss": 0.5238,
      "step": 4386
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8303940297314657,
      "learning_rate": 8.583164489764657e-07,
      "loss": 0.5636,
      "step": 4387
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8608177759408698,
      "learning_rate": 8.579079623950815e-07,
      "loss": 0.6173,
      "step": 4388
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0461245382223046,
      "learning_rate": 8.57499500010491e-07,
      "loss": 0.6193,
      "step": 4389
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6855916166681705,
      "learning_rate": 8.570910618922512e-07,
      "loss": 0.539,
      "step": 4390
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9313322121941922,
      "learning_rate": 8.566826481099146e-07,
      "loss": 0.6586,
      "step": 4391
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.9925538461902144,
      "learning_rate": 8.5627425873303e-07,
      "loss": 0.631,
      "step": 4392
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9254096673534389,
      "learning_rate": 8.558658938311418e-07,
      "loss": 0.6216,
      "step": 4393
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6978149943255844,
      "learning_rate": 8.554575534737903e-07,
      "loss": 0.593,
      "step": 4394
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7531397854943812,
      "learning_rate": 8.550492377305117e-07,
      "loss": 0.6041,
      "step": 4395
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7344864652395828,
      "learning_rate": 8.546409466708377e-07,
      "loss": 0.5318,
      "step": 4396
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8937678320199621,
      "learning_rate": 8.542326803642964e-07,
      "loss": 0.6101,
      "step": 4397
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8261304320800287,
      "learning_rate": 8.538244388804107e-07,
      "loss": 0.6235,
      "step": 4398
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7952779299378832,
      "learning_rate": 8.534162222887001e-07,
      "loss": 0.5805,
      "step": 4399
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5873623059793163,
      "learning_rate": 8.5300803065868e-07,
      "loss": 0.4971,
      "step": 4400
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6685141322953755,
      "learning_rate": 8.525998640598609e-07,
      "loss": 0.6068,
      "step": 4401
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.025109335787052,
      "learning_rate": 8.521917225617494e-07,
      "loss": 0.6049,
      "step": 4402
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9571568649955058,
      "learning_rate": 8.517836062338477e-07,
      "loss": 0.6499,
      "step": 4403
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.76438684892713,
      "learning_rate": 8.513755151456538e-07,
      "loss": 0.5865,
      "step": 4404
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6400721371976708,
      "learning_rate": 8.509674493666613e-07,
      "loss": 0.5642,
      "step": 4405
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0628893967632864,
      "learning_rate": 8.505594089663593e-07,
      "loss": 0.5987,
      "step": 4406
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9977226661958044,
      "learning_rate": 8.501513940142334e-07,
      "loss": 0.6465,
      "step": 4407
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7521716654381636,
      "learning_rate": 8.497434045797639e-07,
      "loss": 0.5639,
      "step": 4408
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6625186403754416,
      "learning_rate": 8.493354407324271e-07,
      "loss": 0.5507,
      "step": 4409
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6740161222925183,
      "learning_rate": 8.489275025416953e-07,
      "loss": 0.5613,
      "step": 4410
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0391103162816349,
      "learning_rate": 8.485195900770359e-07,
      "loss": 0.6512,
      "step": 4411
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6406993195800389,
      "learning_rate": 8.481117034079121e-07,
      "loss": 0.5216,
      "step": 4412
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5839211063909073,
      "learning_rate": 8.477038426037828e-07,
      "loss": 0.5483,
      "step": 4413
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6870323192384473,
      "learning_rate": 8.472960077341027e-07,
      "loss": 0.5327,
      "step": 4414
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8262537355135767,
      "learning_rate": 8.468881988683216e-07,
      "loss": 0.5928,
      "step": 4415
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6417359010071221,
      "learning_rate": 8.46480416075885e-07,
      "loss": 0.4852,
      "step": 4416
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9154321503081716,
      "learning_rate": 8.460726594262344e-07,
      "loss": 0.6106,
      "step": 4417
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8010739754806401,
      "learning_rate": 8.456649289888062e-07,
      "loss": 0.6741,
      "step": 4418
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8075288073734203,
      "learning_rate": 8.452572248330327e-07,
      "loss": 0.6186,
      "step": 4419
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7889541622498499,
      "learning_rate": 8.448495470283416e-07,
      "loss": 0.5984,
      "step": 4420
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9192773921998046,
      "learning_rate": 8.444418956441564e-07,
      "loss": 0.5825,
      "step": 4421
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6886975534113211,
      "learning_rate": 8.44034270749896e-07,
      "loss": 0.5365,
      "step": 4422
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6955568301630137,
      "learning_rate": 8.436266724149742e-07,
      "loss": 0.5554,
      "step": 4423
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.749267186644043,
      "learning_rate": 8.43219100708801e-07,
      "loss": 0.5206,
      "step": 4424
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7754763325920387,
      "learning_rate": 8.428115557007816e-07,
      "loss": 0.5981,
      "step": 4425
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.904184734458228,
      "learning_rate": 8.424040374603167e-07,
      "loss": 0.6541,
      "step": 4426
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6439836598345291,
      "learning_rate": 8.419965460568024e-07,
      "loss": 0.53,
      "step": 4427
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6301531974679986,
      "learning_rate": 8.415890815596302e-07,
      "loss": 0.5378,
      "step": 4428
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7927964117573223,
      "learning_rate": 8.411816440381873e-07,
      "loss": 0.5789,
      "step": 4429
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8502649303013989,
      "learning_rate": 8.407742335618558e-07,
      "loss": 0.55,
      "step": 4430
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8651862909217385,
      "learning_rate": 8.403668502000134e-07,
      "loss": 0.6556,
      "step": 4431
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6917982362734562,
      "learning_rate": 8.399594940220335e-07,
      "loss": 0.5235,
      "step": 4432
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1022455507397788,
      "learning_rate": 8.395521650972845e-07,
      "loss": 0.5916,
      "step": 4433
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6359777780117996,
      "learning_rate": 8.391448634951302e-07,
      "loss": 0.4786,
      "step": 4434
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7952608177843232,
      "learning_rate": 8.387375892849298e-07,
      "loss": 0.6167,
      "step": 4435
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8173152019312616,
      "learning_rate": 8.38330342536038e-07,
      "loss": 0.6203,
      "step": 4436
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7995853170433503,
      "learning_rate": 8.379231233178046e-07,
      "loss": 0.5438,
      "step": 4437
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6736881049979994,
      "learning_rate": 8.375159316995745e-07,
      "loss": 0.4624,
      "step": 4438
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6326873104560141,
      "learning_rate": 8.371087677506887e-07,
      "loss": 0.5435,
      "step": 4439
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7577139148618389,
      "learning_rate": 8.367016315404829e-07,
      "loss": 0.5268,
      "step": 4440
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9494573405854854,
      "learning_rate": 8.362945231382878e-07,
      "loss": 0.6341,
      "step": 4441
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7646406077632162,
      "learning_rate": 8.358874426134301e-07,
      "loss": 0.5407,
      "step": 4442
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1090071477377337,
      "learning_rate": 8.354803900352312e-07,
      "loss": 0.6596,
      "step": 4443
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8438761177233356,
      "learning_rate": 8.35073365473008e-07,
      "loss": 0.5935,
      "step": 4444
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7812584317672604,
      "learning_rate": 8.346663689960723e-07,
      "loss": 0.5743,
      "step": 4445
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6987482571055537,
      "learning_rate": 8.342594006737316e-07,
      "loss": 0.5214,
      "step": 4446
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7976585141054953,
      "learning_rate": 8.338524605752884e-07,
      "loss": 0.6305,
      "step": 4447
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7003560386565325,
      "learning_rate": 8.334455487700401e-07,
      "loss": 0.5496,
      "step": 4448
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7185814923597355,
      "learning_rate": 8.330386653272793e-07,
      "loss": 0.5472,
      "step": 4449
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7041345390072647,
      "learning_rate": 8.326318103162948e-07,
      "loss": 0.5261,
      "step": 4450
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8159515849511303,
      "learning_rate": 8.322249838063694e-07,
      "loss": 0.5956,
      "step": 4451
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0231998869245233,
      "learning_rate": 8.318181858667814e-07,
      "loss": 0.6768,
      "step": 4452
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6276018783962658,
      "learning_rate": 8.314114165668043e-07,
      "loss": 0.5591,
      "step": 4453
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7976296813635937,
      "learning_rate": 8.310046759757066e-07,
      "loss": 0.6015,
      "step": 4454
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8951627493784508,
      "learning_rate": 8.305979641627519e-07,
      "loss": 0.677,
      "step": 4455
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.806826869495861,
      "learning_rate": 8.301912811971992e-07,
      "loss": 0.6101,
      "step": 4456
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8597374286665916,
      "learning_rate": 8.29784627148302e-07,
      "loss": 0.6084,
      "step": 4457
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8576328131619287,
      "learning_rate": 8.293780020853096e-07,
      "loss": 0.6439,
      "step": 4458
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.832415776489216,
      "learning_rate": 8.289714060774661e-07,
      "loss": 0.6707,
      "step": 4459
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8876866668275865,
      "learning_rate": 8.285648391940103e-07,
      "loss": 0.6066,
      "step": 4460
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9658130819064014,
      "learning_rate": 8.281583015041765e-07,
      "loss": 0.6451,
      "step": 4461
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8210744129302442,
      "learning_rate": 8.277517930771936e-07,
      "loss": 0.61,
      "step": 4462
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6808304848564876,
      "learning_rate": 8.27345313982286e-07,
      "loss": 0.5173,
      "step": 4463
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8424003101607799,
      "learning_rate": 8.269388642886725e-07,
      "loss": 0.6221,
      "step": 4464
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5855964935998733,
      "learning_rate": 8.265324440655676e-07,
      "loss": 0.5107,
      "step": 4465
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0715582237692367,
      "learning_rate": 8.261260533821802e-07,
      "loss": 0.5997,
      "step": 4466
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6968121242968011,
      "learning_rate": 8.257196923077146e-07,
      "loss": 0.5774,
      "step": 4467
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7909278365337222,
      "learning_rate": 8.253133609113698e-07,
      "loss": 0.6105,
      "step": 4468
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8312639291648662,
      "learning_rate": 8.249070592623397e-07,
      "loss": 0.6261,
      "step": 4469
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6832492340332178,
      "learning_rate": 8.245007874298131e-07,
      "loss": 0.5708,
      "step": 4470
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8304565094430321,
      "learning_rate": 8.240945454829743e-07,
      "loss": 0.6103,
      "step": 4471
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7051691976784179,
      "learning_rate": 8.236883334910019e-07,
      "loss": 0.5851,
      "step": 4472
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6417564627866752,
      "learning_rate": 8.232821515230694e-07,
      "loss": 0.5162,
      "step": 4473
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7511251361959094,
      "learning_rate": 8.228759996483454e-07,
      "loss": 0.4704,
      "step": 4474
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9426632185414435,
      "learning_rate": 8.224698779359935e-07,
      "loss": 0.6123,
      "step": 4475
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6315561437743511,
      "learning_rate": 8.22063786455172e-07,
      "loss": 0.4892,
      "step": 4476
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9648121440106838,
      "learning_rate": 8.216577252750338e-07,
      "loss": 0.6213,
      "step": 4477
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8710801020070927,
      "learning_rate": 8.212516944647269e-07,
      "loss": 0.5275,
      "step": 4478
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1012890650463816,
      "learning_rate": 8.208456940933946e-07,
      "loss": 0.6352,
      "step": 4479
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7390421353193901,
      "learning_rate": 8.204397242301738e-07,
      "loss": 0.5665,
      "step": 4480
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9599406650057072,
      "learning_rate": 8.200337849441976e-07,
      "loss": 0.6287,
      "step": 4481
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8878454744602954,
      "learning_rate": 8.196278763045927e-07,
      "loss": 0.65,
      "step": 4482
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6508726349286498,
      "learning_rate": 8.192219983804814e-07,
      "loss": 0.4902,
      "step": 4483
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6599468058205433,
      "learning_rate": 8.188161512409806e-07,
      "loss": 0.5485,
      "step": 4484
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9323491640592634,
      "learning_rate": 8.184103349552016e-07,
      "loss": 0.576,
      "step": 4485
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8556299488864367,
      "learning_rate": 8.180045495922508e-07,
      "loss": 0.6371,
      "step": 4486
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7877193821817254,
      "learning_rate": 8.175987952212291e-07,
      "loss": 0.5929,
      "step": 4487
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6662714781629077,
      "learning_rate": 8.171930719112326e-07,
      "loss": 0.4956,
      "step": 4488
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6089862583928608,
      "learning_rate": 8.167873797313511e-07,
      "loss": 0.563,
      "step": 4489
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9050876520646741,
      "learning_rate": 8.163817187506704e-07,
      "loss": 0.6501,
      "step": 4490
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6613344083618417,
      "learning_rate": 8.159760890382701e-07,
      "loss": 0.5279,
      "step": 4491
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7144884356138853,
      "learning_rate": 8.155704906632245e-07,
      "loss": 0.5694,
      "step": 4492
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9576235890353428,
      "learning_rate": 8.151649236946031e-07,
      "loss": 0.6127,
      "step": 4493
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8414381080095936,
      "learning_rate": 8.147593882014696e-07,
      "loss": 0.631,
      "step": 4494
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8720951839570097,
      "learning_rate": 8.143538842528824e-07,
      "loss": 0.5934,
      "step": 4495
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6435782029534615,
      "learning_rate": 8.139484119178946e-07,
      "loss": 0.5045,
      "step": 4496
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7764288454639161,
      "learning_rate": 8.135429712655541e-07,
      "loss": 0.5299,
      "step": 4497
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8911294909684478,
      "learning_rate": 8.13137562364903e-07,
      "loss": 0.6067,
      "step": 4498
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6173502374664481,
      "learning_rate": 8.127321852849784e-07,
      "loss": 0.5823,
      "step": 4499
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7245969396212082,
      "learning_rate": 8.123268400948117e-07,
      "loss": 0.5413,
      "step": 4500
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6390509947268154,
      "learning_rate": 8.119215268634289e-07,
      "loss": 0.5475,
      "step": 4501
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.631263114402124,
      "learning_rate": 8.115162456598508e-07,
      "loss": 0.5432,
      "step": 4502
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8725795895470396,
      "learning_rate": 8.111109965530924e-07,
      "loss": 0.6916,
      "step": 4503
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8605659700950471,
      "learning_rate": 8.107057796121634e-07,
      "loss": 0.5915,
      "step": 4504
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7596198436019139,
      "learning_rate": 8.103005949060682e-07,
      "loss": 0.522,
      "step": 4505
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6673171882155366,
      "learning_rate": 8.098954425038054e-07,
      "loss": 0.5278,
      "step": 4506
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6122869190020669,
      "learning_rate": 8.094903224743681e-07,
      "loss": 0.5329,
      "step": 4507
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8100072013592492,
      "learning_rate": 8.090852348867443e-07,
      "loss": 0.5895,
      "step": 4508
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1460229237500226,
      "learning_rate": 8.086801798099161e-07,
      "loss": 0.6704,
      "step": 4509
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9797330493978239,
      "learning_rate": 8.082751573128598e-07,
      "loss": 0.5954,
      "step": 4510
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.584449722409891,
      "learning_rate": 8.078701674645471e-07,
      "loss": 0.4905,
      "step": 4511
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7720810547370168,
      "learning_rate": 8.074652103339432e-07,
      "loss": 0.5545,
      "step": 4512
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7362052527979962,
      "learning_rate": 8.070602859900083e-07,
      "loss": 0.5406,
      "step": 4513
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7945417463210773,
      "learning_rate": 8.066553945016966e-07,
      "loss": 0.681,
      "step": 4514
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6260393520427715,
      "learning_rate": 8.06250535937957e-07,
      "loss": 0.5359,
      "step": 4515
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6871108136460321,
      "learning_rate": 8.058457103677327e-07,
      "loss": 0.5849,
      "step": 4516
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8361762324012048,
      "learning_rate": 8.05440917859961e-07,
      "loss": 0.5749,
      "step": 4517
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7605161584032075,
      "learning_rate": 8.050361584835743e-07,
      "loss": 0.524,
      "step": 4518
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8467409857792648,
      "learning_rate": 8.046314323074988e-07,
      "loss": 0.6454,
      "step": 4519
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8506852896050776,
      "learning_rate": 8.042267394006548e-07,
      "loss": 0.618,
      "step": 4520
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8885013934131349,
      "learning_rate": 8.038220798319576e-07,
      "loss": 0.658,
      "step": 4521
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.769174884191062,
      "learning_rate": 8.034174536703165e-07,
      "loss": 0.5353,
      "step": 4522
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7919686173296258,
      "learning_rate": 8.030128609846349e-07,
      "loss": 0.552,
      "step": 4523
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7865025008807413,
      "learning_rate": 8.026083018438109e-07,
      "loss": 0.6054,
      "step": 4524
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6988040675747746,
      "learning_rate": 8.022037763167367e-07,
      "loss": 0.4998,
      "step": 4525
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9151759860968967,
      "learning_rate": 8.017992844722989e-07,
      "loss": 0.6446,
      "step": 4526
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7666409570659523,
      "learning_rate": 8.013948263793782e-07,
      "loss": 0.5813,
      "step": 4527
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8306154507543775,
      "learning_rate": 8.009904021068493e-07,
      "loss": 0.6161,
      "step": 4528
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9117827068480757,
      "learning_rate": 8.005860117235817e-07,
      "loss": 0.6148,
      "step": 4529
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.951092141154977,
      "learning_rate": 8.001816552984389e-07,
      "loss": 0.6443,
      "step": 4530
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6181074404591639,
      "learning_rate": 7.997773329002784e-07,
      "loss": 0.491,
      "step": 4531
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8591603109391228,
      "learning_rate": 7.993730445979523e-07,
      "loss": 0.6089,
      "step": 4532
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8282227697844028,
      "learning_rate": 7.989687904603065e-07,
      "loss": 0.6108,
      "step": 4533
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8685270249658449,
      "learning_rate": 7.985645705561814e-07,
      "loss": 0.6437,
      "step": 4534
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0984010570407696,
      "learning_rate": 7.981603849544115e-07,
      "loss": 0.6324,
      "step": 4535
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6714605561062976,
      "learning_rate": 7.97756233723825e-07,
      "loss": 0.5327,
      "step": 4536
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6678223481535148,
      "learning_rate": 7.973521169332452e-07,
      "loss": 0.5127,
      "step": 4537
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6398739734070572,
      "learning_rate": 7.969480346514885e-07,
      "loss": 0.5101,
      "step": 4538
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7932789599755855,
      "learning_rate": 7.965439869473663e-07,
      "loss": 0.6047,
      "step": 4539
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8863438898874132,
      "learning_rate": 7.961399738896835e-07,
      "loss": 0.5954,
      "step": 4540
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7218634755510498,
      "learning_rate": 7.957359955472394e-07,
      "loss": 0.545,
      "step": 4541
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6818991810193427,
      "learning_rate": 7.953320519888272e-07,
      "loss": 0.5534,
      "step": 4542
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8394884998730925,
      "learning_rate": 7.949281432832343e-07,
      "loss": 0.5485,
      "step": 4543
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8388736100218905,
      "learning_rate": 7.945242694992422e-07,
      "loss": 0.6639,
      "step": 4544
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.803462496959316,
      "learning_rate": 7.941204307056263e-07,
      "loss": 0.5732,
      "step": 4545
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7103044622661596,
      "learning_rate": 7.937166269711564e-07,
      "loss": 0.5311,
      "step": 4546
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7120974461642388,
      "learning_rate": 7.933128583645958e-07,
      "loss": 0.5487,
      "step": 4547
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.609604761614603,
      "learning_rate": 7.929091249547022e-07,
      "loss": 0.4851,
      "step": 4548
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6428578128298722,
      "learning_rate": 7.92505426810227e-07,
      "loss": 0.5353,
      "step": 4549
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.740511401528985,
      "learning_rate": 7.921017639999159e-07,
      "loss": 0.5669,
      "step": 4550
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7251425042785961,
      "learning_rate": 7.916981365925085e-07,
      "loss": 0.4702,
      "step": 4551
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7004291219783643,
      "learning_rate": 7.912945446567385e-07,
      "loss": 0.5412,
      "step": 4552
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9167653477705164,
      "learning_rate": 7.908909882613334e-07,
      "loss": 0.6827,
      "step": 4553
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.676759473796595,
      "learning_rate": 7.904874674750143e-07,
      "loss": 0.489,
      "step": 4554
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7050430261469035,
      "learning_rate": 7.900839823664968e-07,
      "loss": 0.5355,
      "step": 4555
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9357232954766475,
      "learning_rate": 7.896805330044904e-07,
      "loss": 0.6487,
      "step": 4556
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9773619978800818,
      "learning_rate": 7.892771194576981e-07,
      "loss": 0.6116,
      "step": 4557
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8534888485786647,
      "learning_rate": 7.88873741794817e-07,
      "loss": 0.643,
      "step": 4558
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6260255360923213,
      "learning_rate": 7.884704000845383e-07,
      "loss": 0.5512,
      "step": 4559
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6552396664549461,
      "learning_rate": 7.880670943955467e-07,
      "loss": 0.5517,
      "step": 4560
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0272710256880404,
      "learning_rate": 7.876638247965209e-07,
      "loss": 0.5263,
      "step": 4561
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0271278113592697,
      "learning_rate": 7.872605913561339e-07,
      "loss": 0.6089,
      "step": 4562
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.940157527057518,
      "learning_rate": 7.868573941430517e-07,
      "loss": 0.6399,
      "step": 4563
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1264549596729316,
      "learning_rate": 7.864542332259347e-07,
      "loss": 0.5761,
      "step": 4564
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8460134652616151,
      "learning_rate": 7.860511086734372e-07,
      "loss": 0.5935,
      "step": 4565
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7997918553895481,
      "learning_rate": 7.856480205542071e-07,
      "loss": 0.5244,
      "step": 4566
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.041178810508615,
      "learning_rate": 7.852449689368859e-07,
      "loss": 0.6703,
      "step": 4567
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6753425393830857,
      "learning_rate": 7.848419538901088e-07,
      "loss": 0.5352,
      "step": 4568
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.83424019214742,
      "learning_rate": 7.844389754825058e-07,
      "loss": 0.697,
      "step": 4569
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8955471174197359,
      "learning_rate": 7.840360337826997e-07,
      "loss": 0.584,
      "step": 4570
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7588449936114927,
      "learning_rate": 7.83633128859307e-07,
      "loss": 0.5939,
      "step": 4571
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7058234768819842,
      "learning_rate": 7.832302607809385e-07,
      "loss": 0.5297,
      "step": 4572
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.725974614510219,
      "learning_rate": 7.828274296161981e-07,
      "loss": 0.5504,
      "step": 4573
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7989844424337375,
      "learning_rate": 7.824246354336841e-07,
      "loss": 0.6008,
      "step": 4574
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9142473221933964,
      "learning_rate": 7.820218783019879e-07,
      "loss": 0.648,
      "step": 4575
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7369134387935785,
      "learning_rate": 7.81619158289695e-07,
      "loss": 0.5695,
      "step": 4576
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0291555622531634,
      "learning_rate": 7.812164754653843e-07,
      "loss": 0.6861,
      "step": 4577
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6711406389690804,
      "learning_rate": 7.808138298976287e-07,
      "loss": 0.5435,
      "step": 4578
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.954091642335392,
      "learning_rate": 7.804112216549944e-07,
      "loss": 0.5435,
      "step": 4579
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7918542949079569,
      "learning_rate": 7.800086508060413e-07,
      "loss": 0.6621,
      "step": 4580
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6853913497865066,
      "learning_rate": 7.796061174193229e-07,
      "loss": 0.5366,
      "step": 4581
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8532256376625035,
      "learning_rate": 7.792036215633869e-07,
      "loss": 0.6248,
      "step": 4582
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6363035220611221,
      "learning_rate": 7.788011633067739e-07,
      "loss": 0.5487,
      "step": 4583
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8979061268851936,
      "learning_rate": 7.783987427180184e-07,
      "loss": 0.6894,
      "step": 4584
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7577084445961293,
      "learning_rate": 7.779963598656483e-07,
      "loss": 0.5296,
      "step": 4585
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6912127071688652,
      "learning_rate": 7.775940148181854e-07,
      "loss": 0.5352,
      "step": 4586
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6312711928244904,
      "learning_rate": 7.771917076441448e-07,
      "loss": 0.5342,
      "step": 4587
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6987860867570039,
      "learning_rate": 7.767894384120351e-07,
      "loss": 0.5754,
      "step": 4588
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9412352851156882,
      "learning_rate": 7.763872071903585e-07,
      "loss": 0.6623,
      "step": 4589
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.962573025438982,
      "learning_rate": 7.759850140476112e-07,
      "loss": 0.645,
      "step": 4590
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6116535077631905,
      "learning_rate": 7.755828590522821e-07,
      "loss": 0.5269,
      "step": 4591
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6710875214906262,
      "learning_rate": 7.751807422728541e-07,
      "loss": 0.5766,
      "step": 4592
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8557459342193208,
      "learning_rate": 7.747786637778036e-07,
      "loss": 0.5894,
      "step": 4593
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6329826791252896,
      "learning_rate": 7.743766236356001e-07,
      "loss": 0.5113,
      "step": 4594
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7782833048037182,
      "learning_rate": 7.739746219147074e-07,
      "loss": 0.5741,
      "step": 4595
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8354850734273055,
      "learning_rate": 7.735726586835817e-07,
      "loss": 0.6758,
      "step": 4596
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7251766516621277,
      "learning_rate": 7.731707340106735e-07,
      "loss": 0.5656,
      "step": 4597
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7143094153366544,
      "learning_rate": 7.72768847964426e-07,
      "loss": 0.5434,
      "step": 4598
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.679445169344833,
      "learning_rate": 7.723670006132766e-07,
      "loss": 0.5778,
      "step": 4599
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6506971650996102,
      "learning_rate": 7.719651920256554e-07,
      "loss": 0.5564,
      "step": 4600
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6894419037479782,
      "learning_rate": 7.715634222699863e-07,
      "loss": 0.5488,
      "step": 4601
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6573239256783519,
      "learning_rate": 7.711616914146867e-07,
      "loss": 0.5392,
      "step": 4602
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.705605050783217,
      "learning_rate": 7.707599995281669e-07,
      "loss": 0.5168,
      "step": 4603
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7851485071948949,
      "learning_rate": 7.70358346678831e-07,
      "loss": 0.6084,
      "step": 4604
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7146763912843771,
      "learning_rate": 7.699567329350764e-07,
      "loss": 0.517,
      "step": 4605
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.800656471178309,
      "learning_rate": 7.695551583652935e-07,
      "loss": 0.6063,
      "step": 4606
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7172055415932195,
      "learning_rate": 7.691536230378663e-07,
      "loss": 0.5792,
      "step": 4607
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8427149182820386,
      "learning_rate": 7.687521270211722e-07,
      "loss": 0.6003,
      "step": 4608
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8010817335534846,
      "learning_rate": 7.683506703835819e-07,
      "loss": 0.5523,
      "step": 4609
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8856615881364245,
      "learning_rate": 7.67949253193459e-07,
      "loss": 0.6417,
      "step": 4610
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9174482137748715,
      "learning_rate": 7.675478755191608e-07,
      "loss": 0.66,
      "step": 4611
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7656927127188602,
      "learning_rate": 7.671465374290377e-07,
      "loss": 0.5733,
      "step": 4612
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5876054403084492,
      "learning_rate": 7.667452389914335e-07,
      "loss": 0.523,
      "step": 4613
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7781226375716676,
      "learning_rate": 7.66343980274685e-07,
      "loss": 0.5764,
      "step": 4614
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8877643606292442,
      "learning_rate": 7.659427613471225e-07,
      "loss": 0.5859,
      "step": 4615
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8260360219315455,
      "learning_rate": 7.655415822770694e-07,
      "loss": 0.5186,
      "step": 4616
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6782346620800964,
      "learning_rate": 7.651404431328422e-07,
      "loss": 0.5323,
      "step": 4617
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9761799686544933,
      "learning_rate": 7.647393439827509e-07,
      "loss": 0.5492,
      "step": 4618
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.62295116664252,
      "learning_rate": 7.643382848950983e-07,
      "loss": 0.5185,
      "step": 4619
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7384212769707755,
      "learning_rate": 7.639372659381807e-07,
      "loss": 0.6122,
      "step": 4620
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6892532558942126,
      "learning_rate": 7.63536287180287e-07,
      "loss": 0.5369,
      "step": 4621
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9356120149106564,
      "learning_rate": 7.631353486897007e-07,
      "loss": 0.6643,
      "step": 4622
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.698262273672187,
      "learning_rate": 7.627344505346967e-07,
      "loss": 0.5455,
      "step": 4623
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8420032236511614,
      "learning_rate": 7.623335927835439e-07,
      "loss": 0.6816,
      "step": 4624
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6954213438632111,
      "learning_rate": 7.619327755045044e-07,
      "loss": 0.5345,
      "step": 4625
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7359233502150834,
      "learning_rate": 7.615319987658329e-07,
      "loss": 0.5563,
      "step": 4626
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8008121403321444,
      "learning_rate": 7.611312626357778e-07,
      "loss": 0.6321,
      "step": 4627
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6168772001672514,
      "learning_rate": 7.607305671825801e-07,
      "loss": 0.5193,
      "step": 4628
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.615696886486798,
      "learning_rate": 7.603299124744742e-07,
      "loss": 0.5458,
      "step": 4629
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9591415552767593,
      "learning_rate": 7.599292985796872e-07,
      "loss": 0.6429,
      "step": 4630
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8534921827141352,
      "learning_rate": 7.595287255664397e-07,
      "loss": 0.6668,
      "step": 4631
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7628529554156048,
      "learning_rate": 7.591281935029451e-07,
      "loss": 0.5185,
      "step": 4632
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.864197104032635,
      "learning_rate": 7.587277024574097e-07,
      "loss": 0.6433,
      "step": 4633
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2150934010289645,
      "learning_rate": 7.583272524980328e-07,
      "loss": 0.6201,
      "step": 4634
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6417121943460334,
      "learning_rate": 7.579268436930071e-07,
      "loss": 0.5261,
      "step": 4635
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7844758660721878,
      "learning_rate": 7.575264761105183e-07,
      "loss": 0.5145,
      "step": 4636
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.700112271081736,
      "learning_rate": 7.571261498187445e-07,
      "loss": 0.5941,
      "step": 4637
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7378502873485852,
      "learning_rate": 7.56725864885857e-07,
      "loss": 0.5766,
      "step": 4638
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6867059380167689,
      "learning_rate": 7.563256213800204e-07,
      "loss": 0.5108,
      "step": 4639
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7067113976408538,
      "learning_rate": 7.559254193693918e-07,
      "loss": 0.5378,
      "step": 4640
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7759131213204004,
      "learning_rate": 7.555252589221216e-07,
      "loss": 0.5896,
      "step": 4641
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8097277187549107,
      "learning_rate": 7.551251401063529e-07,
      "loss": 0.6013,
      "step": 4642
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6525574627584387,
      "learning_rate": 7.547250629902218e-07,
      "loss": 0.4476,
      "step": 4643
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6907440232419042,
      "learning_rate": 7.543250276418571e-07,
      "loss": 0.5555,
      "step": 4644
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8745761837626473,
      "learning_rate": 7.539250341293807e-07,
      "loss": 0.6538,
      "step": 4645
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6248851881105257,
      "learning_rate": 7.535250825209076e-07,
      "loss": 0.5661,
      "step": 4646
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7952465044865602,
      "learning_rate": 7.53125172884545e-07,
      "loss": 0.5478,
      "step": 4647
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8807436548110773,
      "learning_rate": 7.527253052883937e-07,
      "loss": 0.6771,
      "step": 4648
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7399930567277663,
      "learning_rate": 7.523254798005467e-07,
      "loss": 0.5753,
      "step": 4649
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9029797464352047,
      "learning_rate": 7.519256964890904e-07,
      "loss": 0.6368,
      "step": 4650
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8460450898321868,
      "learning_rate": 7.515259554221034e-07,
      "loss": 0.6524,
      "step": 4651
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4870350968799977,
      "learning_rate": 7.511262566676573e-07,
      "loss": 0.6217,
      "step": 4652
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7483945037386243,
      "learning_rate": 7.507266002938173e-07,
      "loss": 0.5817,
      "step": 4653
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.962709854634627,
      "learning_rate": 7.503269863686404e-07,
      "loss": 0.68,
      "step": 4654
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6005231109259365,
      "learning_rate": 7.499274149601767e-07,
      "loss": 0.546,
      "step": 4655
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8678806913637388,
      "learning_rate": 7.495278861364688e-07,
      "loss": 0.6056,
      "step": 4656
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.794653227524971,
      "learning_rate": 7.491283999655527e-07,
      "loss": 0.577,
      "step": 4657
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8479838697856121,
      "learning_rate": 7.487289565154564e-07,
      "loss": 0.5684,
      "step": 4658
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0609542765235078,
      "learning_rate": 7.483295558542011e-07,
      "loss": 0.6208,
      "step": 4659
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9296092999805887,
      "learning_rate": 7.479301980498005e-07,
      "loss": 0.6197,
      "step": 4660
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0021656146874283,
      "learning_rate": 7.475308831702613e-07,
      "loss": 0.6318,
      "step": 4661
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7611505002022495,
      "learning_rate": 7.471316112835824e-07,
      "loss": 0.5964,
      "step": 4662
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7548929860805427,
      "learning_rate": 7.467323824577556e-07,
      "loss": 0.5403,
      "step": 4663
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6718064505765661,
      "learning_rate": 7.463331967607658e-07,
      "loss": 0.565,
      "step": 4664
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8805517177677572,
      "learning_rate": 7.459340542605898e-07,
      "loss": 0.5443,
      "step": 4665
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9371012558059065,
      "learning_rate": 7.455349550251973e-07,
      "loss": 0.6207,
      "step": 4666
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9337285013399496,
      "learning_rate": 7.451358991225511e-07,
      "loss": 0.5988,
      "step": 4667
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8131459584864631,
      "learning_rate": 7.447368866206062e-07,
      "loss": 0.5527,
      "step": 4668
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6709459238126295,
      "learning_rate": 7.443379175873101e-07,
      "loss": 0.5699,
      "step": 4669
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8232073721970536,
      "learning_rate": 7.439389920906034e-07,
      "loss": 0.6442,
      "step": 4670
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3781458671367686,
      "learning_rate": 7.435401101984184e-07,
      "loss": 0.6534,
      "step": 4671
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8325234841060969,
      "learning_rate": 7.431412719786812e-07,
      "loss": 0.6457,
      "step": 4672
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8054547754562412,
      "learning_rate": 7.427424774993092e-07,
      "loss": 0.5343,
      "step": 4673
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7575849771574459,
      "learning_rate": 7.423437268282133e-07,
      "loss": 0.5335,
      "step": 4674
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7529255085358358,
      "learning_rate": 7.419450200332964e-07,
      "loss": 0.5509,
      "step": 4675
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8247884849635866,
      "learning_rate": 7.415463571824543e-07,
      "loss": 0.5778,
      "step": 4676
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8288704393201374,
      "learning_rate": 7.411477383435749e-07,
      "loss": 0.6264,
      "step": 4677
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0334082185780522,
      "learning_rate": 7.407491635845391e-07,
      "loss": 0.6183,
      "step": 4678
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6588801952902797,
      "learning_rate": 7.403506329732196e-07,
      "loss": 0.5509,
      "step": 4679
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5830229812751486,
      "learning_rate": 7.399521465774825e-07,
      "loss": 0.5126,
      "step": 4680
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.688490175027201,
      "learning_rate": 7.395537044651856e-07,
      "loss": 0.5249,
      "step": 4681
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6536797243682454,
      "learning_rate": 7.391553067041796e-07,
      "loss": 0.5658,
      "step": 4682
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.568285200373215,
      "learning_rate": 7.387569533623075e-07,
      "loss": 0.4703,
      "step": 4683
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6606576995329159,
      "learning_rate": 7.383586445074045e-07,
      "loss": 0.5009,
      "step": 4684
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7512373699871159,
      "learning_rate": 7.379603802072985e-07,
      "loss": 0.5387,
      "step": 4685
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6581974072856829,
      "learning_rate": 7.3756216052981e-07,
      "loss": 0.5654,
      "step": 4686
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.882176911414058,
      "learning_rate": 7.371639855427515e-07,
      "loss": 0.5697,
      "step": 4687
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.695208999370907,
      "learning_rate": 7.36765855313928e-07,
      "loss": 0.5922,
      "step": 4688
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.762618526729417,
      "learning_rate": 7.363677699111371e-07,
      "loss": 0.5934,
      "step": 4689
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.634117474220558,
      "learning_rate": 7.359697294021683e-07,
      "loss": 0.5492,
      "step": 4690
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7495758296103276,
      "learning_rate": 7.35571733854804e-07,
      "loss": 0.5764,
      "step": 4691
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7248532616668805,
      "learning_rate": 7.351737833368186e-07,
      "loss": 0.555,
      "step": 4692
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8142070068131667,
      "learning_rate": 7.34775877915979e-07,
      "loss": 0.5717,
      "step": 4693
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6086909421132517,
      "learning_rate": 7.343780176600443e-07,
      "loss": 0.5201,
      "step": 4694
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7676643504626551,
      "learning_rate": 7.339802026367659e-07,
      "loss": 0.5438,
      "step": 4695
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6492795722392808,
      "learning_rate": 7.335824329138877e-07,
      "loss": 0.5132,
      "step": 4696
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9600232142678801,
      "learning_rate": 7.331847085591456e-07,
      "loss": 0.6632,
      "step": 4697
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7928907216319171,
      "learning_rate": 7.32787029640268e-07,
      "loss": 0.5906,
      "step": 4698
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6618419610396009,
      "learning_rate": 7.323893962249756e-07,
      "loss": 0.5465,
      "step": 4699
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8211597659624463,
      "learning_rate": 7.319918083809809e-07,
      "loss": 0.6704,
      "step": 4700
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6174588663405822,
      "learning_rate": 7.315942661759893e-07,
      "loss": 0.5175,
      "step": 4701
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8403476956720377,
      "learning_rate": 7.311967696776977e-07,
      "loss": 0.5595,
      "step": 4702
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5922708820270163,
      "learning_rate": 7.30799318953796e-07,
      "loss": 0.497,
      "step": 4703
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8824279673640503,
      "learning_rate": 7.304019140719657e-07,
      "loss": 0.641,
      "step": 4704
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7772768815676877,
      "learning_rate": 7.300045550998808e-07,
      "loss": 0.6031,
      "step": 4705
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7863181926651701,
      "learning_rate": 7.296072421052073e-07,
      "loss": 0.5813,
      "step": 4706
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8142065329853736,
      "learning_rate": 7.292099751556037e-07,
      "loss": 0.6466,
      "step": 4707
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8227180633025091,
      "learning_rate": 7.288127543187201e-07,
      "loss": 0.569,
      "step": 4708
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8522694641556555,
      "learning_rate": 7.284155796621993e-07,
      "loss": 0.6085,
      "step": 4709
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8688279956537864,
      "learning_rate": 7.280184512536761e-07,
      "loss": 0.6169,
      "step": 4710
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.952854827398654,
      "learning_rate": 7.276213691607771e-07,
      "loss": 0.627,
      "step": 4711
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8552010284755536,
      "learning_rate": 7.272243334511212e-07,
      "loss": 0.6254,
      "step": 4712
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6604314546165456,
      "learning_rate": 7.268273441923196e-07,
      "loss": 0.5416,
      "step": 4713
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6551079047775132,
      "learning_rate": 7.264304014519754e-07,
      "loss": 0.5188,
      "step": 4714
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8703541187488316,
      "learning_rate": 7.260335052976838e-07,
      "loss": 0.5851,
      "step": 4715
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.757046968976391,
      "learning_rate": 7.256366557970321e-07,
      "loss": 0.5838,
      "step": 4716
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0515175950519395,
      "learning_rate": 7.252398530175996e-07,
      "loss": 0.6763,
      "step": 4717
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6271552253514056,
      "learning_rate": 7.248430970269575e-07,
      "loss": 0.5292,
      "step": 4718
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9782847485924131,
      "learning_rate": 7.244463878926696e-07,
      "loss": 0.648,
      "step": 4719
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7581041034417082,
      "learning_rate": 7.240497256822913e-07,
      "loss": 0.5712,
      "step": 4720
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0719275935078403,
      "learning_rate": 7.236531104633697e-07,
      "loss": 0.673,
      "step": 4721
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8758944997911009,
      "learning_rate": 7.232565423034446e-07,
      "loss": 0.6491,
      "step": 4722
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6586421684348878,
      "learning_rate": 7.22860021270047e-07,
      "loss": 0.533,
      "step": 4723
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0152947338005016,
      "learning_rate": 7.224635474307008e-07,
      "loss": 0.6698,
      "step": 4724
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8251852441354635,
      "learning_rate": 7.22067120852921e-07,
      "loss": 0.582,
      "step": 4725
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8094211623621679,
      "learning_rate": 7.216707416042151e-07,
      "loss": 0.6151,
      "step": 4726
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7105081712664875,
      "learning_rate": 7.212744097520823e-07,
      "loss": 0.5295,
      "step": 4727
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8330324682726733,
      "learning_rate": 7.208781253640137e-07,
      "loss": 0.6153,
      "step": 4728
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6253953210825905,
      "learning_rate": 7.204818885074923e-07,
      "loss": 0.4871,
      "step": 4729
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.659277464914254,
      "learning_rate": 7.200856992499936e-07,
      "loss": 0.5435,
      "step": 4730
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8791191992573567,
      "learning_rate": 7.196895576589839e-07,
      "loss": 0.6294,
      "step": 4731
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7303186214955407,
      "learning_rate": 7.192934638019225e-07,
      "loss": 0.5325,
      "step": 4732
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8121226973761069,
      "learning_rate": 7.188974177462598e-07,
      "loss": 0.6303,
      "step": 4733
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6876573230811017,
      "learning_rate": 7.185014195594383e-07,
      "loss": 0.5492,
      "step": 4734
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.041398486557105,
      "learning_rate": 7.181054693088925e-07,
      "loss": 0.6473,
      "step": 4735
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9008191068244247,
      "learning_rate": 7.177095670620484e-07,
      "loss": 0.6194,
      "step": 4736
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6668835663744821,
      "learning_rate": 7.173137128863242e-07,
      "loss": 0.5479,
      "step": 4737
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1132586446348012,
      "learning_rate": 7.169179068491298e-07,
      "loss": 0.5977,
      "step": 4738
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8519859757653813,
      "learning_rate": 7.165221490178667e-07,
      "loss": 0.6078,
      "step": 4739
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8054190747440693,
      "learning_rate": 7.161264394599282e-07,
      "loss": 0.5944,
      "step": 4740
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8187173000516479,
      "learning_rate": 7.157307782427e-07,
      "loss": 0.6244,
      "step": 4741
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8127632953418372,
      "learning_rate": 7.153351654335585e-07,
      "loss": 0.6508,
      "step": 4742
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9956703648143062,
      "learning_rate": 7.149396010998728e-07,
      "loss": 0.6022,
      "step": 4743
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.088198199748241,
      "learning_rate": 7.145440853090033e-07,
      "loss": 0.6578,
      "step": 4744
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5976694021071232,
      "learning_rate": 7.14148618128302e-07,
      "loss": 0.551,
      "step": 4745
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7343452996291056,
      "learning_rate": 7.137531996251133e-07,
      "loss": 0.5526,
      "step": 4746
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.78515709508061,
      "learning_rate": 7.133578298667727e-07,
      "loss": 0.5978,
      "step": 4747
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.797112240208171,
      "learning_rate": 7.129625089206071e-07,
      "loss": 0.6731,
      "step": 4748
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9425807633858226,
      "learning_rate": 7.125672368539362e-07,
      "loss": 0.6574,
      "step": 4749
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0512058925097378,
      "learning_rate": 7.121720137340704e-07,
      "loss": 0.6929,
      "step": 4750
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.750028257637845,
      "learning_rate": 7.11776839628312e-07,
      "loss": 0.5504,
      "step": 4751
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7176662080854478,
      "learning_rate": 7.113817146039552e-07,
      "loss": 0.4925,
      "step": 4752
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7821620274320258,
      "learning_rate": 7.109866387282855e-07,
      "loss": 0.6034,
      "step": 4753
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6721927883594091,
      "learning_rate": 7.105916120685799e-07,
      "loss": 0.5977,
      "step": 4754
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6895200844556199,
      "learning_rate": 7.101966346921083e-07,
      "loss": 0.5626,
      "step": 4755
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9564520497941171,
      "learning_rate": 7.098017066661303e-07,
      "loss": 0.6268,
      "step": 4756
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6292146662297082,
      "learning_rate": 7.094068280578988e-07,
      "loss": 0.4715,
      "step": 4757
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0769687087604,
      "learning_rate": 7.090119989346568e-07,
      "loss": 0.6413,
      "step": 4758
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.633486954765076,
      "learning_rate": 7.086172193636401e-07,
      "loss": 0.5259,
      "step": 4759
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7853610452488977,
      "learning_rate": 7.082224894120752e-07,
      "loss": 0.5467,
      "step": 4760
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7624862105817894,
      "learning_rate": 7.078278091471808e-07,
      "loss": 0.5271,
      "step": 4761
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8148066972309069,
      "learning_rate": 7.074331786361665e-07,
      "loss": 0.6085,
      "step": 4762
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7895519103863339,
      "learning_rate": 7.070385979462338e-07,
      "loss": 0.5312,
      "step": 4763
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.692040207682597,
      "learning_rate": 7.066440671445762e-07,
      "loss": 0.5558,
      "step": 4764
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.74430700688589,
      "learning_rate": 7.062495862983775e-07,
      "loss": 0.582,
      "step": 4765
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7387533240209501,
      "learning_rate": 7.058551554748142e-07,
      "loss": 0.5714,
      "step": 4766
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5849698414425039,
      "learning_rate": 7.054607747410535e-07,
      "loss": 0.4758,
      "step": 4767
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1246617157719698,
      "learning_rate": 7.050664441642543e-07,
      "loss": 0.7112,
      "step": 4768
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7486763369816976,
      "learning_rate": 7.046721638115672e-07,
      "loss": 0.5703,
      "step": 4769
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7241175110856931,
      "learning_rate": 7.042779337501336e-07,
      "loss": 0.57,
      "step": 4770
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7401943771307347,
      "learning_rate": 7.038837540470874e-07,
      "loss": 0.5706,
      "step": 4771
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7587195794397982,
      "learning_rate": 7.034896247695528e-07,
      "loss": 0.5416,
      "step": 4772
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7072404208831705,
      "learning_rate": 7.030955459846462e-07,
      "loss": 0.5857,
      "step": 4773
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7705257992585705,
      "learning_rate": 7.027015177594748e-07,
      "loss": 0.6023,
      "step": 4774
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6617289582873193,
      "learning_rate": 7.023075401611379e-07,
      "loss": 0.5765,
      "step": 4775
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.835679920633192,
      "learning_rate": 7.019136132567255e-07,
      "loss": 0.5522,
      "step": 4776
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7879134341223859,
      "learning_rate": 7.01519737113319e-07,
      "loss": 0.6415,
      "step": 4777
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6502323819788756,
      "learning_rate": 7.011259117979922e-07,
      "loss": 0.5162,
      "step": 4778
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0055612498664552,
      "learning_rate": 7.007321373778088e-07,
      "loss": 0.6376,
      "step": 4779
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7660119665206292,
      "learning_rate": 7.003384139198245e-07,
      "loss": 0.5454,
      "step": 4780
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9900353074170458,
      "learning_rate": 6.999447414910867e-07,
      "loss": 0.649,
      "step": 4781
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8163926150169506,
      "learning_rate": 6.995511201586334e-07,
      "loss": 0.5574,
      "step": 4782
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7709304368712685,
      "learning_rate": 6.991575499894941e-07,
      "loss": 0.6046,
      "step": 4783
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8009888505157126,
      "learning_rate": 6.987640310506901e-07,
      "loss": 0.5392,
      "step": 4784
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7185691835158405,
      "learning_rate": 6.983705634092334e-07,
      "loss": 0.5595,
      "step": 4785
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6107605902040845,
      "learning_rate": 6.979771471321273e-07,
      "loss": 0.5033,
      "step": 4786
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6487810679107059,
      "learning_rate": 6.975837822863666e-07,
      "loss": 0.4997,
      "step": 4787
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7144406443356014,
      "learning_rate": 6.971904689389372e-07,
      "loss": 0.5008,
      "step": 4788
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6349709191191786,
      "learning_rate": 6.967972071568165e-07,
      "loss": 0.5675,
      "step": 4789
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.753590452373471,
      "learning_rate": 6.964039970069721e-07,
      "loss": 0.5336,
      "step": 4790
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8046679245658495,
      "learning_rate": 6.960108385563645e-07,
      "loss": 0.5679,
      "step": 4791
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8426606403966166,
      "learning_rate": 6.956177318719443e-07,
      "loss": 0.5877,
      "step": 4792
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8815655198821202,
      "learning_rate": 6.952246770206534e-07,
      "loss": 0.6334,
      "step": 4793
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8660118505291564,
      "learning_rate": 6.948316740694246e-07,
      "loss": 0.6019,
      "step": 4794
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7808875781669933,
      "learning_rate": 6.944387230851827e-07,
      "loss": 0.5575,
      "step": 4795
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8112020505434123,
      "learning_rate": 6.940458241348428e-07,
      "loss": 0.6398,
      "step": 4796
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7635212708499578,
      "learning_rate": 6.936529772853117e-07,
      "loss": 0.5612,
      "step": 4797
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8859448764387569,
      "learning_rate": 6.932601826034871e-07,
      "loss": 0.6575,
      "step": 4798
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8304377529016366,
      "learning_rate": 6.928674401562578e-07,
      "loss": 0.5844,
      "step": 4799
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8629948719835527,
      "learning_rate": 6.924747500105037e-07,
      "loss": 0.6188,
      "step": 4800
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6749040285598692,
      "learning_rate": 6.92082112233096e-07,
      "loss": 0.5453,
      "step": 4801
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6510188861081363,
      "learning_rate": 6.916895268908967e-07,
      "loss": 0.5841,
      "step": 4802
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7898676989975746,
      "learning_rate": 6.912969940507587e-07,
      "loss": 0.6037,
      "step": 4803
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7154769251012439,
      "learning_rate": 6.909045137795268e-07,
      "loss": 0.5556,
      "step": 4804
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8761612356477939,
      "learning_rate": 6.905120861440362e-07,
      "loss": 0.5936,
      "step": 4805
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.906458135120129,
      "learning_rate": 6.90119711211113e-07,
      "loss": 0.6344,
      "step": 4806
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9520359562907782,
      "learning_rate": 6.897273890475746e-07,
      "loss": 0.5961,
      "step": 4807
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7054099333791667,
      "learning_rate": 6.893351197202296e-07,
      "loss": 0.5583,
      "step": 4808
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8439252488244353,
      "learning_rate": 6.889429032958772e-07,
      "loss": 0.6299,
      "step": 4809
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9203848960347051,
      "learning_rate": 6.885507398413077e-07,
      "loss": 0.6462,
      "step": 4810
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7672046552640451,
      "learning_rate": 6.881586294233027e-07,
      "loss": 0.5989,
      "step": 4811
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7684940879176732,
      "learning_rate": 6.877665721086343e-07,
      "loss": 0.5589,
      "step": 4812
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.635692612464444,
      "learning_rate": 6.873745679640659e-07,
      "loss": 0.6498,
      "step": 4813
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7161497112484605,
      "learning_rate": 6.869826170563515e-07,
      "loss": 0.5284,
      "step": 4814
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7283410519946485,
      "learning_rate": 6.865907194522365e-07,
      "loss": 0.5661,
      "step": 4815
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0361594151789284,
      "learning_rate": 6.861988752184567e-07,
      "loss": 0.6555,
      "step": 4816
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7136941108175854,
      "learning_rate": 6.858070844217396e-07,
      "loss": 0.544,
      "step": 4817
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7853751720583463,
      "learning_rate": 6.854153471288027e-07,
      "loss": 0.5413,
      "step": 4818
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8406235118737051,
      "learning_rate": 6.850236634063548e-07,
      "loss": 0.6425,
      "step": 4819
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.716756228798767,
      "learning_rate": 6.846320333210957e-07,
      "loss": 0.5209,
      "step": 4820
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7147562991430463,
      "learning_rate": 6.84240456939716e-07,
      "loss": 0.5155,
      "step": 4821
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7887131612775877,
      "learning_rate": 6.838489343288967e-07,
      "loss": 0.5537,
      "step": 4822
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6479113876271091,
      "learning_rate": 6.834574655553102e-07,
      "loss": 0.6061,
      "step": 4823
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9503540453255233,
      "learning_rate": 6.830660506856199e-07,
      "loss": 0.637,
      "step": 4824
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7032096319455334,
      "learning_rate": 6.82674689786479e-07,
      "loss": 0.5231,
      "step": 4825
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6644745911897748,
      "learning_rate": 6.822833829245329e-07,
      "loss": 0.5296,
      "step": 4826
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7283875314178863,
      "learning_rate": 6.818921301664165e-07,
      "loss": 0.5618,
      "step": 4827
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.617711306475877,
      "learning_rate": 6.815009315787564e-07,
      "loss": 0.4938,
      "step": 4828
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7196728513342804,
      "learning_rate": 6.811097872281694e-07,
      "loss": 0.5741,
      "step": 4829
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7230316673128611,
      "learning_rate": 6.807186971812635e-07,
      "loss": 0.552,
      "step": 4830
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8262021710428504,
      "learning_rate": 6.803276615046374e-07,
      "loss": 0.6414,
      "step": 4831
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8585626083512728,
      "learning_rate": 6.7993668026488e-07,
      "loss": 0.5918,
      "step": 4832
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7077791245103183,
      "learning_rate": 6.795457535285718e-07,
      "loss": 0.5925,
      "step": 4833
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7523436891900062,
      "learning_rate": 6.791548813622833e-07,
      "loss": 0.5262,
      "step": 4834
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8780827062911793,
      "learning_rate": 6.787640638325758e-07,
      "loss": 0.636,
      "step": 4835
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7275887987262222,
      "learning_rate": 6.783733010060017e-07,
      "loss": 0.5634,
      "step": 4836
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9700338728247637,
      "learning_rate": 6.779825929491038e-07,
      "loss": 0.6412,
      "step": 4837
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8826238063417386,
      "learning_rate": 6.775919397284156e-07,
      "loss": 0.6449,
      "step": 4838
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8440895353053162,
      "learning_rate": 6.77201341410461e-07,
      "loss": 0.6341,
      "step": 4839
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7871904032268768,
      "learning_rate": 6.768107980617552e-07,
      "loss": 0.5403,
      "step": 4840
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7824694377949004,
      "learning_rate": 6.764203097488036e-07,
      "loss": 0.6156,
      "step": 4841
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6602031235662502,
      "learning_rate": 6.760298765381021e-07,
      "loss": 0.5598,
      "step": 4842
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8206943267582267,
      "learning_rate": 6.756394984961375e-07,
      "loss": 0.6576,
      "step": 4843
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8191598842750761,
      "learning_rate": 6.752491756893872e-07,
      "loss": 0.6043,
      "step": 4844
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1280344224488659,
      "learning_rate": 6.748589081843191e-07,
      "loss": 0.6174,
      "step": 4845
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6563546862316217,
      "learning_rate": 6.744686960473916e-07,
      "loss": 0.5476,
      "step": 4846
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7288512080022654,
      "learning_rate": 6.740785393450539e-07,
      "loss": 0.5211,
      "step": 4847
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8494058124002419,
      "learning_rate": 6.736884381437455e-07,
      "loss": 0.6217,
      "step": 4848
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7171345994963593,
      "learning_rate": 6.732983925098965e-07,
      "loss": 0.5246,
      "step": 4849
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6451486662683338,
      "learning_rate": 6.729084025099278e-07,
      "loss": 0.5247,
      "step": 4850
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8416736189261517,
      "learning_rate": 6.725184682102505e-07,
      "loss": 0.5529,
      "step": 4851
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8257004682909737,
      "learning_rate": 6.721285896772664e-07,
      "loss": 0.6359,
      "step": 4852
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7284664332304899,
      "learning_rate": 6.717387669773678e-07,
      "loss": 0.5252,
      "step": 4853
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6978963816922141,
      "learning_rate": 6.713490001769373e-07,
      "loss": 0.5532,
      "step": 4854
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6712188116042048,
      "learning_rate": 6.709592893423477e-07,
      "loss": 0.508,
      "step": 4855
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8599403718104743,
      "learning_rate": 6.705696345399638e-07,
      "loss": 0.5752,
      "step": 4856
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0296835087056169,
      "learning_rate": 6.70180035836139e-07,
      "loss": 0.5951,
      "step": 4857
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8870984777002201,
      "learning_rate": 6.697904932972182e-07,
      "loss": 0.6199,
      "step": 4858
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6961652486444441,
      "learning_rate": 6.694010069895361e-07,
      "loss": 0.6272,
      "step": 4859
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7107906739428895,
      "learning_rate": 6.690115769794186e-07,
      "loss": 0.5466,
      "step": 4860
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9138769885240373,
      "learning_rate": 6.68622203333181e-07,
      "loss": 0.6471,
      "step": 4861
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7926008244441692,
      "learning_rate": 6.6823288611713e-07,
      "loss": 0.6957,
      "step": 4862
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7272846279357286,
      "learning_rate": 6.678436253975624e-07,
      "loss": 0.571,
      "step": 4863
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6896564498174491,
      "learning_rate": 6.674544212407649e-07,
      "loss": 0.5562,
      "step": 4864
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9890947250370119,
      "learning_rate": 6.670652737130149e-07,
      "loss": 0.6269,
      "step": 4865
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8642369579729864,
      "learning_rate": 6.666761828805804e-07,
      "loss": 0.592,
      "step": 4866
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.774674201419634,
      "learning_rate": 6.662871488097195e-07,
      "loss": 0.5547,
      "step": 4867
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8643518231961775,
      "learning_rate": 6.658981715666805e-07,
      "loss": 0.5964,
      "step": 4868
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6936260213035784,
      "learning_rate": 6.655092512177023e-07,
      "loss": 0.5712,
      "step": 4869
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6547179077665197,
      "learning_rate": 6.651203878290138e-07,
      "loss": 0.5679,
      "step": 4870
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.057312308194285,
      "learning_rate": 6.647315814668347e-07,
      "loss": 0.6252,
      "step": 4871
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.666867936718001,
      "learning_rate": 6.643428321973742e-07,
      "loss": 0.5501,
      "step": 4872
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9389527106922586,
      "learning_rate": 6.639541400868327e-07,
      "loss": 0.6276,
      "step": 4873
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8852747374209617,
      "learning_rate": 6.635655052014e-07,
      "loss": 0.5932,
      "step": 4874
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.771451574434832,
      "learning_rate": 6.63176927607257e-07,
      "loss": 0.5992,
      "step": 4875
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9028245080355353,
      "learning_rate": 6.627884073705742e-07,
      "loss": 0.6082,
      "step": 4876
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9110975698470604,
      "learning_rate": 6.623999445575125e-07,
      "loss": 0.6536,
      "step": 4877
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0452207657694201,
      "learning_rate": 6.620115392342234e-07,
      "loss": 0.6379,
      "step": 4878
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7547747845777661,
      "learning_rate": 6.616231914668479e-07,
      "loss": 0.5126,
      "step": 4879
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9623014156166316,
      "learning_rate": 6.612349013215177e-07,
      "loss": 0.6469,
      "step": 4880
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6467433454366135,
      "learning_rate": 6.608466688643545e-07,
      "loss": 0.5522,
      "step": 4881
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8681980473138912,
      "learning_rate": 6.604584941614705e-07,
      "loss": 0.6493,
      "step": 4882
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8694519271810779,
      "learning_rate": 6.600703772789675e-07,
      "loss": 0.5946,
      "step": 4883
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6406747073233817,
      "learning_rate": 6.596823182829379e-07,
      "loss": 0.5606,
      "step": 4884
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7468883034562007,
      "learning_rate": 6.592943172394642e-07,
      "loss": 0.6015,
      "step": 4885
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8233223464111684,
      "learning_rate": 6.589063742146187e-07,
      "loss": 0.6341,
      "step": 4886
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9797785284570254,
      "learning_rate": 6.585184892744642e-07,
      "loss": 0.6879,
      "step": 4887
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8533240303983688,
      "learning_rate": 6.581306624850534e-07,
      "loss": 0.6323,
      "step": 4888
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7621165448330751,
      "learning_rate": 6.577428939124292e-07,
      "loss": 0.5355,
      "step": 4889
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.840352662520139,
      "learning_rate": 6.573551836226246e-07,
      "loss": 0.659,
      "step": 4890
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0078896780483122,
      "learning_rate": 6.569675316816627e-07,
      "loss": 0.601,
      "step": 4891
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7549521867020064,
      "learning_rate": 6.565799381555564e-07,
      "loss": 0.5707,
      "step": 4892
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6325902417172217,
      "learning_rate": 6.561924031103088e-07,
      "loss": 0.5759,
      "step": 4893
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6447953323478757,
      "learning_rate": 6.558049266119133e-07,
      "loss": 0.5188,
      "step": 4894
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6147895339031084,
      "learning_rate": 6.55417508726353e-07,
      "loss": 0.5047,
      "step": 4895
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6786957387957474,
      "learning_rate": 6.550301495196011e-07,
      "loss": 0.5613,
      "step": 4896
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7580510879088299,
      "learning_rate": 6.54642849057621e-07,
      "loss": 0.5675,
      "step": 4897
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6332224947528619,
      "learning_rate": 6.542556074063657e-07,
      "loss": 0.5337,
      "step": 4898
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1718889512545676,
      "learning_rate": 6.538684246317784e-07,
      "loss": 0.6109,
      "step": 4899
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8399563557643595,
      "learning_rate": 6.534813007997925e-07,
      "loss": 0.6011,
      "step": 4900
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9099523307497173,
      "learning_rate": 6.530942359763309e-07,
      "loss": 0.6492,
      "step": 4901
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6976258901635441,
      "learning_rate": 6.527072302273071e-07,
      "loss": 0.5559,
      "step": 4902
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7967638308160917,
      "learning_rate": 6.523202836186239e-07,
      "loss": 0.5806,
      "step": 4903
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6629516413739844,
      "learning_rate": 6.519333962161741e-07,
      "loss": 0.5277,
      "step": 4904
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7550426941653392,
      "learning_rate": 6.51546568085841e-07,
      "loss": 0.5795,
      "step": 4905
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8871271406604001,
      "learning_rate": 6.511597992934972e-07,
      "loss": 0.6128,
      "step": 4906
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9368843154016564,
      "learning_rate": 6.507730899050054e-07,
      "loss": 0.6358,
      "step": 4907
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7608914456432945,
      "learning_rate": 6.503864399862181e-07,
      "loss": 0.5634,
      "step": 4908
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.788032965909177,
      "learning_rate": 6.49999849602978e-07,
      "loss": 0.5283,
      "step": 4909
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.347956943566713,
      "learning_rate": 6.49613318821117e-07,
      "loss": 0.6361,
      "step": 4910
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7137491502381899,
      "learning_rate": 6.492268477064575e-07,
      "loss": 0.562,
      "step": 4911
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.783047037516308,
      "learning_rate": 6.488404363248117e-07,
      "loss": 0.5507,
      "step": 4912
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1501506603744205,
      "learning_rate": 6.484540847419811e-07,
      "loss": 0.6462,
      "step": 4913
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8478163000494211,
      "learning_rate": 6.480677930237572e-07,
      "loss": 0.6522,
      "step": 4914
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6116167156423842,
      "learning_rate": 6.476815612359221e-07,
      "loss": 0.5725,
      "step": 4915
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6469411895445523,
      "learning_rate": 6.472953894442468e-07,
      "loss": 0.5538,
      "step": 4916
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8415444045244135,
      "learning_rate": 6.46909277714492e-07,
      "loss": 0.5466,
      "step": 4917
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.605308570927544,
      "learning_rate": 6.465232261124088e-07,
      "loss": 0.599,
      "step": 4918
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8598605706008651,
      "learning_rate": 6.461372347037377e-07,
      "loss": 0.5505,
      "step": 4919
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0261045172423413,
      "learning_rate": 6.457513035542091e-07,
      "loss": 0.6164,
      "step": 4920
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8466212245841493,
      "learning_rate": 6.453654327295429e-07,
      "loss": 0.545,
      "step": 4921
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6431198835841655,
      "learning_rate": 6.449796222954489e-07,
      "loss": 0.5003,
      "step": 4922
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0649376284705676,
      "learning_rate": 6.445938723176267e-07,
      "loss": 0.6223,
      "step": 4923
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8248291366932888,
      "learning_rate": 6.442081828617655e-07,
      "loss": 0.5315,
      "step": 4924
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7572796617768951,
      "learning_rate": 6.438225539935441e-07,
      "loss": 0.5838,
      "step": 4925
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8449101231554983,
      "learning_rate": 6.434369857786312e-07,
      "loss": 0.6347,
      "step": 4926
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.859352111189483,
      "learning_rate": 6.430514782826845e-07,
      "loss": 0.6348,
      "step": 4927
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6878526838899733,
      "learning_rate": 6.426660315713529e-07,
      "loss": 0.5251,
      "step": 4928
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7421909660845593,
      "learning_rate": 6.422806457102734e-07,
      "loss": 0.57,
      "step": 4929
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6469185857412609,
      "learning_rate": 6.418953207650731e-07,
      "loss": 0.5521,
      "step": 4930
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.782972449613917,
      "learning_rate": 6.415100568013691e-07,
      "loss": 0.542,
      "step": 4931
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9030844618435762,
      "learning_rate": 6.411248538847676e-07,
      "loss": 0.6509,
      "step": 4932
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8168228495247508,
      "learning_rate": 6.407397120808649e-07,
      "loss": 0.5653,
      "step": 4933
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9057969670312863,
      "learning_rate": 6.403546314552463e-07,
      "loss": 0.5937,
      "step": 4934
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8205666056281299,
      "learning_rate": 6.399696120734872e-07,
      "loss": 0.5691,
      "step": 4935
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8057065321119412,
      "learning_rate": 6.395846540011526e-07,
      "loss": 0.7093,
      "step": 4936
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6656842535810251,
      "learning_rate": 6.391997573037965e-07,
      "loss": 0.5145,
      "step": 4937
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6596052612510509,
      "learning_rate": 6.38814922046963e-07,
      "loss": 0.5813,
      "step": 4938
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7166697856992714,
      "learning_rate": 6.384301482961854e-07,
      "loss": 0.5288,
      "step": 4939
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9365063315004525,
      "learning_rate": 6.380454361169866e-07,
      "loss": 0.6492,
      "step": 4940
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.665813560245457,
      "learning_rate": 6.376607855748794e-07,
      "loss": 0.5745,
      "step": 4941
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.169198048318818,
      "learning_rate": 6.372761967353655e-07,
      "loss": 0.6489,
      "step": 4942
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8428087752392551,
      "learning_rate": 6.368916696639366e-07,
      "loss": 0.6564,
      "step": 4943
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8990641256495119,
      "learning_rate": 6.365072044260735e-07,
      "loss": 0.6098,
      "step": 4944
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7594257757747117,
      "learning_rate": 6.361228010872466e-07,
      "loss": 0.5333,
      "step": 4945
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6563218401053665,
      "learning_rate": 6.357384597129158e-07,
      "loss": 0.5534,
      "step": 4946
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7406002029461745,
      "learning_rate": 6.353541803685305e-07,
      "loss": 0.5951,
      "step": 4947
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8165278684328335,
      "learning_rate": 6.349699631195295e-07,
      "loss": 0.5904,
      "step": 4948
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.813183338067519,
      "learning_rate": 6.345858080313409e-07,
      "loss": 0.6175,
      "step": 4949
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.709216320269669,
      "learning_rate": 6.342017151693822e-07,
      "loss": 0.5362,
      "step": 4950
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7170581577683568,
      "learning_rate": 6.338176845990607e-07,
      "loss": 0.6163,
      "step": 4951
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9079859049636873,
      "learning_rate": 6.334337163857725e-07,
      "loss": 0.7018,
      "step": 4952
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0218457930170581,
      "learning_rate": 6.330498105949036e-07,
      "loss": 0.609,
      "step": 4953
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8245608196816369,
      "learning_rate": 6.326659672918291e-07,
      "loss": 0.6108,
      "step": 4954
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9290067189320299,
      "learning_rate": 6.322821865419135e-07,
      "loss": 0.6427,
      "step": 4955
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9281370575539485,
      "learning_rate": 6.318984684105108e-07,
      "loss": 0.6065,
      "step": 4956
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8067568147951699,
      "learning_rate": 6.315148129629638e-07,
      "loss": 0.5724,
      "step": 4957
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7161480404545053,
      "learning_rate": 6.311312202646055e-07,
      "loss": 0.565,
      "step": 4958
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9334238781309483,
      "learning_rate": 6.307476903807573e-07,
      "loss": 0.64,
      "step": 4959
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7613104224188724,
      "learning_rate": 6.303642233767308e-07,
      "loss": 0.5323,
      "step": 4960
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.89251144490296,
      "learning_rate": 6.299808193178262e-07,
      "loss": 0.6009,
      "step": 4961
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8289484968645028,
      "learning_rate": 6.295974782693334e-07,
      "loss": 0.626,
      "step": 4962
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8509765562880285,
      "learning_rate": 6.292142002965311e-07,
      "loss": 0.6412,
      "step": 4963
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8514468617566323,
      "learning_rate": 6.288309854646876e-07,
      "loss": 0.6503,
      "step": 4964
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.780924284027837,
      "learning_rate": 6.284478338390606e-07,
      "loss": 0.5875,
      "step": 4965
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.905755235832951,
      "learning_rate": 6.280647454848966e-07,
      "loss": 0.6757,
      "step": 4966
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0500278098327973,
      "learning_rate": 6.276817204674319e-07,
      "loss": 0.6809,
      "step": 4967
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9029279716988909,
      "learning_rate": 6.272987588518911e-07,
      "loss": 0.6644,
      "step": 4968
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8283752539344065,
      "learning_rate": 6.26915860703489e-07,
      "loss": 0.6099,
      "step": 4969
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8716947743952852,
      "learning_rate": 6.26533026087429e-07,
      "loss": 0.6019,
      "step": 4970
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7360810923821216,
      "learning_rate": 6.26150255068904e-07,
      "loss": 0.5001,
      "step": 4971
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7178246439773184,
      "learning_rate": 6.257675477130957e-07,
      "loss": 0.5755,
      "step": 4972
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8299638203459027,
      "learning_rate": 6.253849040851753e-07,
      "loss": 0.5972,
      "step": 4973
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8508951565266676,
      "learning_rate": 6.250023242503031e-07,
      "loss": 0.5596,
      "step": 4974
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6500818153881901,
      "learning_rate": 6.246198082736285e-07,
      "loss": 0.5384,
      "step": 4975
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7187411393553,
      "learning_rate": 6.242373562202897e-07,
      "loss": 0.546,
      "step": 4976
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6473034671658946,
      "learning_rate": 6.238549681554147e-07,
      "loss": 0.5537,
      "step": 4977
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6640803537389897,
      "learning_rate": 6.234726441441197e-07,
      "loss": 0.5025,
      "step": 4978
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9227965725247756,
      "learning_rate": 6.230903842515109e-07,
      "loss": 0.6559,
      "step": 4979
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7673545910840528,
      "learning_rate": 6.22708188542683e-07,
      "loss": 0.5565,
      "step": 4980
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.756460568659084,
      "learning_rate": 6.2232605708272e-07,
      "loss": 0.554,
      "step": 4981
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8739971322497097,
      "learning_rate": 6.219439899366949e-07,
      "loss": 0.644,
      "step": 4982
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8567515583734208,
      "learning_rate": 6.215619871696698e-07,
      "loss": 0.5921,
      "step": 4983
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7556644032725468,
      "learning_rate": 6.211800488466957e-07,
      "loss": 0.5824,
      "step": 4984
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.914646205473272,
      "learning_rate": 6.207981750328126e-07,
      "loss": 0.621,
      "step": 4985
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6417673288921504,
      "learning_rate": 6.2041636579305e-07,
      "loss": 0.4747,
      "step": 4986
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6618414921234277,
      "learning_rate": 6.200346211924259e-07,
      "loss": 0.5547,
      "step": 4987
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6902203006156109,
      "learning_rate": 6.196529412959472e-07,
      "loss": 0.5667,
      "step": 4988
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7550065684709346,
      "learning_rate": 6.192713261686102e-07,
      "loss": 0.6165,
      "step": 4989
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8216213827326314,
      "learning_rate": 6.188897758754001e-07,
      "loss": 0.5624,
      "step": 4990
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8093315922316112,
      "learning_rate": 6.185082904812906e-07,
      "loss": 0.5545,
      "step": 4991
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8742182036929619,
      "learning_rate": 6.18126870051245e-07,
      "loss": 0.6409,
      "step": 4992
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7978051546813751,
      "learning_rate": 6.177455146502151e-07,
      "loss": 0.6093,
      "step": 4993
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6897747425776496,
      "learning_rate": 6.173642243431417e-07,
      "loss": 0.5288,
      "step": 4994
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7532305702078202,
      "learning_rate": 6.169829991949546e-07,
      "loss": 0.5767,
      "step": 4995
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6569957616707532,
      "learning_rate": 6.166018392705726e-07,
      "loss": 0.5476,
      "step": 4996
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7957341683337777,
      "learning_rate": 6.162207446349031e-07,
      "loss": 0.5657,
      "step": 4997
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9004763465437087,
      "learning_rate": 6.158397153528427e-07,
      "loss": 0.6882,
      "step": 4998
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7772164644087306,
      "learning_rate": 6.154587514892761e-07,
      "loss": 0.5861,
      "step": 4999
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7687490873669001,
      "learning_rate": 6.150778531090785e-07,
      "loss": 0.5211,
      "step": 5000
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8093706322265606,
      "learning_rate": 6.146970202771124e-07,
      "loss": 0.5507,
      "step": 5001
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9918926436562822,
      "learning_rate": 6.143162530582297e-07,
      "loss": 0.6053,
      "step": 5002
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8130443721276802,
      "learning_rate": 6.13935551517271e-07,
      "loss": 0.5955,
      "step": 5003
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6471742389513562,
      "learning_rate": 6.135549157190659e-07,
      "loss": 0.5063,
      "step": 5004
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0437946907972142,
      "learning_rate": 6.131743457284328e-07,
      "loss": 0.634,
      "step": 5005
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0687083116021114,
      "learning_rate": 6.127938416101786e-07,
      "loss": 0.6465,
      "step": 5006
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8612141361521637,
      "learning_rate": 6.124134034290993e-07,
      "loss": 0.5909,
      "step": 5007
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7759203963871272,
      "learning_rate": 6.120330312499795e-07,
      "loss": 0.519,
      "step": 5008
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8009859750770079,
      "learning_rate": 6.116527251375927e-07,
      "loss": 0.6054,
      "step": 5009
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8065844323118677,
      "learning_rate": 6.11272485156701e-07,
      "loss": 0.5771,
      "step": 5010
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8211173251871965,
      "learning_rate": 6.108923113720554e-07,
      "loss": 0.6547,
      "step": 5011
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6755481365742355,
      "learning_rate": 6.105122038483954e-07,
      "loss": 0.5352,
      "step": 5012
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8660748235591168,
      "learning_rate": 6.101321626504494e-07,
      "loss": 0.6249,
      "step": 5013
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8637177658083318,
      "learning_rate": 6.097521878429345e-07,
      "loss": 0.592,
      "step": 5014
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9253391958662451,
      "learning_rate": 6.093722794905567e-07,
      "loss": 0.6361,
      "step": 5015
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9166015021716615,
      "learning_rate": 6.0899243765801e-07,
      "loss": 0.6264,
      "step": 5016
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6983658180805641,
      "learning_rate": 6.086126624099777e-07,
      "loss": 0.5922,
      "step": 5017
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9835948279146213,
      "learning_rate": 6.082329538111315e-07,
      "loss": 0.6904,
      "step": 5018
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9284813364489757,
      "learning_rate": 6.078533119261318e-07,
      "loss": 0.5865,
      "step": 5019
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7306505092383813,
      "learning_rate": 6.074737368196279e-07,
      "loss": 0.5766,
      "step": 5020
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8444092087241589,
      "learning_rate": 6.070942285562572e-07,
      "loss": 0.6053,
      "step": 5021
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8143710418037718,
      "learning_rate": 6.067147872006461e-07,
      "loss": 0.6022,
      "step": 5022
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6834113062292975,
      "learning_rate": 6.063354128174096e-07,
      "loss": 0.517,
      "step": 5023
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.654444081468434,
      "learning_rate": 6.059561054711512e-07,
      "loss": 0.5208,
      "step": 5024
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6829282752172614,
      "learning_rate": 6.055768652264625e-07,
      "loss": 0.5196,
      "step": 5025
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8858204863458103,
      "learning_rate": 6.051976921479249e-07,
      "loss": 0.6152,
      "step": 5026
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8260474743105054,
      "learning_rate": 6.048185863001074e-07,
      "loss": 0.6457,
      "step": 5027
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.704128804470987,
      "learning_rate": 6.044395477475676e-07,
      "loss": 0.5256,
      "step": 5028
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9848551422363379,
      "learning_rate": 6.040605765548519e-07,
      "loss": 0.6342,
      "step": 5029
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7628276912534608,
      "learning_rate": 6.03681672786495e-07,
      "loss": 0.5759,
      "step": 5030
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8110916594432857,
      "learning_rate": 6.033028365070207e-07,
      "loss": 0.5471,
      "step": 5031
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9016975759148083,
      "learning_rate": 6.029240677809404e-07,
      "loss": 0.6218,
      "step": 5032
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7183808734749543,
      "learning_rate": 6.025453666727547e-07,
      "loss": 0.5283,
      "step": 5033
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4390614234386085,
      "learning_rate": 6.021667332469523e-07,
      "loss": 0.6238,
      "step": 5034
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7554507786921595,
      "learning_rate": 6.017881675680107e-07,
      "loss": 0.6051,
      "step": 5035
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6358237871152954,
      "learning_rate": 6.014096697003957e-07,
      "loss": 0.5695,
      "step": 5036
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6377939608862446,
      "learning_rate": 6.010312397085613e-07,
      "loss": 0.5701,
      "step": 5037
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7695221234903264,
      "learning_rate": 6.006528776569504e-07,
      "loss": 0.5821,
      "step": 5038
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7715144763859078,
      "learning_rate": 6.002745836099941e-07,
      "loss": 0.6339,
      "step": 5039
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7857716793280495,
      "learning_rate": 5.998963576321119e-07,
      "loss": 0.5451,
      "step": 5040
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0522187748236873,
      "learning_rate": 5.995181997877117e-07,
      "loss": 0.5672,
      "step": 5041
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6669504917647237,
      "learning_rate": 5.991401101411899e-07,
      "loss": 0.5319,
      "step": 5042
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.80251396948831,
      "learning_rate": 5.987620887569313e-07,
      "loss": 0.5986,
      "step": 5043
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7015073834912915,
      "learning_rate": 5.983841356993089e-07,
      "loss": 0.5751,
      "step": 5044
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6475561378305018,
      "learning_rate": 5.980062510326842e-07,
      "loss": 0.5207,
      "step": 5045
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7527385662977498,
      "learning_rate": 5.97628434821407e-07,
      "loss": 0.5953,
      "step": 5046
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8501566073388109,
      "learning_rate": 5.972506871298156e-07,
      "loss": 0.5161,
      "step": 5047
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9160663091141346,
      "learning_rate": 5.968730080222361e-07,
      "loss": 0.6188,
      "step": 5048
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.861499366523754,
      "learning_rate": 5.964953975629836e-07,
      "loss": 0.5615,
      "step": 5049
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6664628322734598,
      "learning_rate": 5.961178558163614e-07,
      "loss": 0.582,
      "step": 5050
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1579646048558891,
      "learning_rate": 5.957403828466605e-07,
      "loss": 0.6524,
      "step": 5051
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.821458669601864,
      "learning_rate": 5.95362978718161e-07,
      "loss": 0.5755,
      "step": 5052
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5985831094136993,
      "learning_rate": 5.949856434951308e-07,
      "loss": 0.511,
      "step": 5053
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9497938554733488,
      "learning_rate": 5.94608377241826e-07,
      "loss": 0.6587,
      "step": 5054
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1859720698253693,
      "learning_rate": 5.942311800224913e-07,
      "loss": 0.608,
      "step": 5055
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0771315826328354,
      "learning_rate": 5.938540519013594e-07,
      "loss": 0.5927,
      "step": 5056
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8184816624990666,
      "learning_rate": 5.934769929426512e-07,
      "loss": 0.5614,
      "step": 5057
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7813791802459399,
      "learning_rate": 5.931000032105757e-07,
      "loss": 0.6085,
      "step": 5058
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.751256812824723,
      "learning_rate": 5.927230827693308e-07,
      "loss": 0.5623,
      "step": 5059
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6000530065662966,
      "learning_rate": 5.92346231683102e-07,
      "loss": 0.566,
      "step": 5060
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7739495245918401,
      "learning_rate": 5.919694500160631e-07,
      "loss": 0.5726,
      "step": 5061
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7772571431207527,
      "learning_rate": 5.91592737832376e-07,
      "loss": 0.5335,
      "step": 5062
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8113011841768453,
      "learning_rate": 5.912160951961908e-07,
      "loss": 0.5996,
      "step": 5063
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7983879658872886,
      "learning_rate": 5.90839522171646e-07,
      "loss": 0.5792,
      "step": 5064
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8743015732055807,
      "learning_rate": 5.90463018822868e-07,
      "loss": 0.5926,
      "step": 5065
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7830507508463779,
      "learning_rate": 5.900865852139713e-07,
      "loss": 0.5693,
      "step": 5066
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9189113356999796,
      "learning_rate": 5.897102214090587e-07,
      "loss": 0.6091,
      "step": 5067
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8739337434649757,
      "learning_rate": 5.893339274722211e-07,
      "loss": 0.637,
      "step": 5068
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.681804251152822,
      "learning_rate": 5.889577034675373e-07,
      "loss": 0.5645,
      "step": 5069
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6555366560265268,
      "learning_rate": 5.885815494590742e-07,
      "loss": 0.4481,
      "step": 5070
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7614095732517614,
      "learning_rate": 5.882054655108874e-07,
      "loss": 0.6166,
      "step": 5071
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7326413647073123,
      "learning_rate": 5.878294516870196e-07,
      "loss": 0.5576,
      "step": 5072
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8186075662888059,
      "learning_rate": 5.874535080515024e-07,
      "loss": 0.5208,
      "step": 5073
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7574887958945209,
      "learning_rate": 5.870776346683548e-07,
      "loss": 0.5781,
      "step": 5074
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7830121607743807,
      "learning_rate": 5.867018316015842e-07,
      "loss": 0.6194,
      "step": 5075
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.732968374536864,
      "learning_rate": 5.863260989151859e-07,
      "loss": 0.5147,
      "step": 5076
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7162454519792573,
      "learning_rate": 5.859504366731435e-07,
      "loss": 0.5325,
      "step": 5077
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.911551691518996,
      "learning_rate": 5.855748449394281e-07,
      "loss": 0.6274,
      "step": 5078
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7245923035418435,
      "learning_rate": 5.851993237779992e-07,
      "loss": 0.5346,
      "step": 5079
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6876521023441245,
      "learning_rate": 5.848238732528041e-07,
      "loss": 0.5554,
      "step": 5080
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6324907612905983,
      "learning_rate": 5.84448493427778e-07,
      "loss": 0.5138,
      "step": 5081
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6885048684524788,
      "learning_rate": 5.840731843668444e-07,
      "loss": 0.4976,
      "step": 5082
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8005902867831588,
      "learning_rate": 5.836979461339142e-07,
      "loss": 0.5747,
      "step": 5083
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8322618149027909,
      "learning_rate": 5.833227787928869e-07,
      "loss": 0.6141,
      "step": 5084
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.708675636533902,
      "learning_rate": 5.829476824076496e-07,
      "loss": 0.5216,
      "step": 5085
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7270248432377032,
      "learning_rate": 5.825726570420771e-07,
      "loss": 0.5796,
      "step": 5086
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1813317279481148,
      "learning_rate": 5.821977027600322e-07,
      "loss": 0.6608,
      "step": 5087
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8512731345911401,
      "learning_rate": 5.818228196253661e-07,
      "loss": 0.6379,
      "step": 5088
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8179900658273415,
      "learning_rate": 5.814480077019173e-07,
      "loss": 0.5787,
      "step": 5089
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8794787436230713,
      "learning_rate": 5.810732670535122e-07,
      "loss": 0.6216,
      "step": 5090
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6985198773689165,
      "learning_rate": 5.806985977439654e-07,
      "loss": 0.509,
      "step": 5091
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8681988780824225,
      "learning_rate": 5.803239998370793e-07,
      "loss": 0.6763,
      "step": 5092
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8750881326770189,
      "learning_rate": 5.799494733966437e-07,
      "loss": 0.6112,
      "step": 5093
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6328604194952325,
      "learning_rate": 5.795750184864368e-07,
      "loss": 0.6096,
      "step": 5094
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8601784461554801,
      "learning_rate": 5.792006351702244e-07,
      "loss": 0.6246,
      "step": 5095
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6784051214222683,
      "learning_rate": 5.788263235117598e-07,
      "loss": 0.4941,
      "step": 5096
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.945907718371338,
      "learning_rate": 5.784520835747846e-07,
      "loss": 0.6286,
      "step": 5097
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8524267272989058,
      "learning_rate": 5.78077915423028e-07,
      "loss": 0.6033,
      "step": 5098
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8962315033504842,
      "learning_rate": 5.777038191202067e-07,
      "loss": 0.6304,
      "step": 5099
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9873022625942054,
      "learning_rate": 5.773297947300257e-07,
      "loss": 0.6585,
      "step": 5100
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7584336395440215,
      "learning_rate": 5.769558423161771e-07,
      "loss": 0.6081,
      "step": 5101
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9249639892367475,
      "learning_rate": 5.765819619423415e-07,
      "loss": 0.6676,
      "step": 5102
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.868116745146901,
      "learning_rate": 5.762081536721862e-07,
      "loss": 0.6122,
      "step": 5103
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.654113464241392,
      "learning_rate": 5.758344175693677e-07,
      "loss": 0.5753,
      "step": 5104
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7902854325303073,
      "learning_rate": 5.75460753697529e-07,
      "loss": 0.567,
      "step": 5105
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.423628550163955,
      "learning_rate": 5.75087162120301e-07,
      "loss": 0.6036,
      "step": 5106
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6393862123543143,
      "learning_rate": 5.747136429013028e-07,
      "loss": 0.5266,
      "step": 5107
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6791805419557542,
      "learning_rate": 5.743401961041404e-07,
      "loss": 0.5425,
      "step": 5108
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6584652165860774,
      "learning_rate": 5.739668217924083e-07,
      "loss": 0.5533,
      "step": 5109
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8580499077340181,
      "learning_rate": 5.735935200296879e-07,
      "loss": 0.6582,
      "step": 5110
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6511144807862034,
      "learning_rate": 5.732202908795488e-07,
      "loss": 0.5189,
      "step": 5111
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.636381576170471,
      "learning_rate": 5.728471344055481e-07,
      "loss": 0.5052,
      "step": 5112
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9835570522882756,
      "learning_rate": 5.724740506712302e-07,
      "loss": 0.6209,
      "step": 5113
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6064867154013767,
      "learning_rate": 5.721010397401277e-07,
      "loss": 0.5286,
      "step": 5114
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3811986984629432,
      "learning_rate": 5.717281016757603e-07,
      "loss": 0.6718,
      "step": 5115
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8258201055263272,
      "learning_rate": 5.713552365416354e-07,
      "loss": 0.5629,
      "step": 5116
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6500044924096365,
      "learning_rate": 5.70982444401248e-07,
      "loss": 0.5105,
      "step": 5117
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0653608392245684,
      "learning_rate": 5.70609725318081e-07,
      "loss": 0.6806,
      "step": 5118
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.02720684755916,
      "learning_rate": 5.702370793556046e-07,
      "loss": 0.6554,
      "step": 5119
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6860602899612561,
      "learning_rate": 5.698645065772762e-07,
      "loss": 0.5561,
      "step": 5120
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9025652293740645,
      "learning_rate": 5.694920070465411e-07,
      "loss": 0.5812,
      "step": 5121
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9436602663030286,
      "learning_rate": 5.691195808268323e-07,
      "loss": 0.6488,
      "step": 5122
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.633609690431565,
      "learning_rate": 5.687472279815699e-07,
      "loss": 0.5565,
      "step": 5123
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6022063065860346,
      "learning_rate": 5.68374948574162e-07,
      "loss": 0.4931,
      "step": 5124
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6385614404213904,
      "learning_rate": 5.680027426680036e-07,
      "loss": 0.5405,
      "step": 5125
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9114009286591812,
      "learning_rate": 5.676306103264777e-07,
      "loss": 0.6692,
      "step": 5126
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7093953761933804,
      "learning_rate": 5.672585516129543e-07,
      "loss": 0.5969,
      "step": 5127
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6452944119401267,
      "learning_rate": 5.668865665907914e-07,
      "loss": 0.5911,
      "step": 5128
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8516414229812703,
      "learning_rate": 5.665146553233338e-07,
      "loss": 0.6173,
      "step": 5129
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8136173473983309,
      "learning_rate": 5.661428178739145e-07,
      "loss": 0.5986,
      "step": 5130
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8592251324604326,
      "learning_rate": 5.657710543058536e-07,
      "loss": 0.5955,
      "step": 5131
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6570834221119015,
      "learning_rate": 5.653993646824584e-07,
      "loss": 0.5254,
      "step": 5132
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8413622070655353,
      "learning_rate": 5.650277490670237e-07,
      "loss": 0.6109,
      "step": 5133
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6899290267214773,
      "learning_rate": 5.646562075228319e-07,
      "loss": 0.5049,
      "step": 5134
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.920818851600754,
      "learning_rate": 5.642847401131525e-07,
      "loss": 0.6255,
      "step": 5135
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7166500348282763,
      "learning_rate": 5.639133469012426e-07,
      "loss": 0.5297,
      "step": 5136
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6843371088937003,
      "learning_rate": 5.635420279503467e-07,
      "loss": 0.5405,
      "step": 5137
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9024191480953071,
      "learning_rate": 5.631707833236962e-07,
      "loss": 0.6381,
      "step": 5138
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1764222110411937,
      "learning_rate": 5.627996130845107e-07,
      "loss": 0.6383,
      "step": 5139
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7319410664791259,
      "learning_rate": 5.624285172959964e-07,
      "loss": 0.5676,
      "step": 5140
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7858930832094569,
      "learning_rate": 5.620574960213469e-07,
      "loss": 0.6339,
      "step": 5141
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7286445804589572,
      "learning_rate": 5.616865493237434e-07,
      "loss": 0.5778,
      "step": 5142
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.020132381465057,
      "learning_rate": 5.613156772663542e-07,
      "loss": 0.6351,
      "step": 5143
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7152149051029734,
      "learning_rate": 5.609448799123351e-07,
      "loss": 0.5828,
      "step": 5144
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7854618541545233,
      "learning_rate": 5.605741573248288e-07,
      "loss": 0.5715,
      "step": 5145
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8407183333760758,
      "learning_rate": 5.602035095669656e-07,
      "loss": 0.5505,
      "step": 5146
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7089945351036089,
      "learning_rate": 5.59832936701863e-07,
      "loss": 0.5229,
      "step": 5147
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8258125813082188,
      "learning_rate": 5.594624387926257e-07,
      "loss": 0.5526,
      "step": 5148
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.843474521147449,
      "learning_rate": 5.590920159023455e-07,
      "loss": 0.6266,
      "step": 5149
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9479985000773808,
      "learning_rate": 5.587216680941016e-07,
      "loss": 0.6262,
      "step": 5150
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7077055844087151,
      "learning_rate": 5.583513954309604e-07,
      "loss": 0.5375,
      "step": 5151
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8657708405175799,
      "learning_rate": 5.579811979759757e-07,
      "loss": 0.6395,
      "step": 5152
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6971206919020668,
      "learning_rate": 5.576110757921879e-07,
      "loss": 0.5451,
      "step": 5153
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9949717184285992,
      "learning_rate": 5.572410289426252e-07,
      "loss": 0.6733,
      "step": 5154
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6800846574139368,
      "learning_rate": 5.568710574903026e-07,
      "loss": 0.5601,
      "step": 5155
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9761555239442732,
      "learning_rate": 5.56501161498222e-07,
      "loss": 0.5988,
      "step": 5156
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6113763756320756,
      "learning_rate": 5.561313410293738e-07,
      "loss": 0.4914,
      "step": 5157
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8399259385106103,
      "learning_rate": 5.557615961467338e-07,
      "loss": 0.6606,
      "step": 5158
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1852452293414182,
      "learning_rate": 5.553919269132661e-07,
      "loss": 0.6251,
      "step": 5159
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8663631238133971,
      "learning_rate": 5.550223333919213e-07,
      "loss": 0.6486,
      "step": 5160
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.890695778807739,
      "learning_rate": 5.546528156456369e-07,
      "loss": 0.6247,
      "step": 5161
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7970453546045762,
      "learning_rate": 5.542833737373389e-07,
      "loss": 0.6115,
      "step": 5162
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8987296081236028,
      "learning_rate": 5.539140077299389e-07,
      "loss": 0.6107,
      "step": 5163
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8065912801456753,
      "learning_rate": 5.535447176863362e-07,
      "loss": 0.5893,
      "step": 5164
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7072572019971983,
      "learning_rate": 5.531755036694168e-07,
      "loss": 0.5346,
      "step": 5165
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7469512045447022,
      "learning_rate": 5.528063657420542e-07,
      "loss": 0.5323,
      "step": 5166
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6837402980708217,
      "learning_rate": 5.524373039671087e-07,
      "loss": 0.5098,
      "step": 5167
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.021774197352233,
      "learning_rate": 5.520683184074275e-07,
      "loss": 0.6121,
      "step": 5168
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7043376218330065,
      "learning_rate": 5.516994091258453e-07,
      "loss": 0.5924,
      "step": 5169
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9033220579349501,
      "learning_rate": 5.513305761851834e-07,
      "loss": 0.6169,
      "step": 5170
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8364284660030101,
      "learning_rate": 5.509618196482501e-07,
      "loss": 0.6166,
      "step": 5171
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8296727124169542,
      "learning_rate": 5.50593139577841e-07,
      "loss": 0.5872,
      "step": 5172
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6929753337300463,
      "learning_rate": 5.502245360367382e-07,
      "loss": 0.5299,
      "step": 5173
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7934575827250357,
      "learning_rate": 5.498560090877111e-07,
      "loss": 0.5614,
      "step": 5174
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6700536830112204,
      "learning_rate": 5.494875587935162e-07,
      "loss": 0.5411,
      "step": 5175
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6993977583203144,
      "learning_rate": 5.491191852168966e-07,
      "loss": 0.5175,
      "step": 5176
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9524299668903423,
      "learning_rate": 5.487508884205825e-07,
      "loss": 0.6005,
      "step": 5177
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8444431556505653,
      "learning_rate": 5.483826684672912e-07,
      "loss": 0.6565,
      "step": 5178
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8737859785765094,
      "learning_rate": 5.480145254197264e-07,
      "loss": 0.633,
      "step": 5179
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6588048936046896,
      "learning_rate": 5.476464593405792e-07,
      "loss": 0.5359,
      "step": 5180
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7902504023352743,
      "learning_rate": 5.472784702925272e-07,
      "loss": 0.5842,
      "step": 5181
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6387543013069655,
      "learning_rate": 5.469105583382356e-07,
      "loss": 0.5498,
      "step": 5182
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7674972225753357,
      "learning_rate": 5.465427235403555e-07,
      "loss": 0.5356,
      "step": 5183
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8145385097341103,
      "learning_rate": 5.461749659615255e-07,
      "loss": 0.5962,
      "step": 5184
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8783139324471076,
      "learning_rate": 5.458072856643709e-07,
      "loss": 0.6601,
      "step": 5185
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.725858528279023,
      "learning_rate": 5.454396827115037e-07,
      "loss": 0.5142,
      "step": 5186
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8066255320758969,
      "learning_rate": 5.450721571655232e-07,
      "loss": 0.6002,
      "step": 5187
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8642919131376363,
      "learning_rate": 5.447047090890145e-07,
      "loss": 0.5816,
      "step": 5188
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6848398998053352,
      "learning_rate": 5.443373385445509e-07,
      "loss": 0.4815,
      "step": 5189
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7775814435855888,
      "learning_rate": 5.439700455946914e-07,
      "loss": 0.5717,
      "step": 5190
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7960884124749111,
      "learning_rate": 5.436028303019824e-07,
      "loss": 0.613,
      "step": 5191
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7021504380318427,
      "learning_rate": 5.432356927289565e-07,
      "loss": 0.5207,
      "step": 5192
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6856621254639056,
      "learning_rate": 5.428686329381338e-07,
      "loss": 0.5331,
      "step": 5193
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6291448066646765,
      "learning_rate": 5.425016509920205e-07,
      "loss": 0.5126,
      "step": 5194
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8292705863694717,
      "learning_rate": 5.421347469531099e-07,
      "loss": 0.6414,
      "step": 5195
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8525337103530422,
      "learning_rate": 5.417679208838817e-07,
      "loss": 0.6,
      "step": 5196
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8952269229248355,
      "learning_rate": 5.414011728468029e-07,
      "loss": 0.6451,
      "step": 5197
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7685862903210215,
      "learning_rate": 5.410345029043268e-07,
      "loss": 0.5807,
      "step": 5198
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6384502248765729,
      "learning_rate": 5.406679111188932e-07,
      "loss": 0.5162,
      "step": 5199
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8052630061141106,
      "learning_rate": 5.403013975529292e-07,
      "loss": 0.5488,
      "step": 5200
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8783642939826254,
      "learning_rate": 5.399349622688478e-07,
      "loss": 0.6469,
      "step": 5201
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.675450997167393,
      "learning_rate": 5.395686053290498e-07,
      "loss": 0.4977,
      "step": 5202
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7347055482177096,
      "learning_rate": 5.392023267959214e-07,
      "loss": 0.5329,
      "step": 5203
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8199308797303028,
      "learning_rate": 5.38836126731836e-07,
      "loss": 0.5848,
      "step": 5204
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.771495710274613,
      "learning_rate": 5.384700051991541e-07,
      "loss": 0.579,
      "step": 5205
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7051338569643383,
      "learning_rate": 5.38103962260222e-07,
      "loss": 0.5147,
      "step": 5206
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7373923577448827,
      "learning_rate": 5.377379979773732e-07,
      "loss": 0.619,
      "step": 5207
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8119409401938465,
      "learning_rate": 5.373721124129276e-07,
      "loss": 0.5838,
      "step": 5208
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9924237958524367,
      "learning_rate": 5.370063056291915e-07,
      "loss": 0.647,
      "step": 5209
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7252596210894041,
      "learning_rate": 5.366405776884582e-07,
      "loss": 0.5346,
      "step": 5210
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7913602694784511,
      "learning_rate": 5.362749286530073e-07,
      "loss": 0.5421,
      "step": 5211
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.167464005039399,
      "learning_rate": 5.359093585851049e-07,
      "loss": 0.6329,
      "step": 5212
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8985184017865717,
      "learning_rate": 5.355438675470039e-07,
      "loss": 0.6151,
      "step": 5213
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.862180951670977,
      "learning_rate": 5.351784556009432e-07,
      "loss": 0.5768,
      "step": 5214
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6506968274593046,
      "learning_rate": 5.348131228091495e-07,
      "loss": 0.5149,
      "step": 5215
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7131402508909719,
      "learning_rate": 5.344478692338346e-07,
      "loss": 0.5561,
      "step": 5216
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8794935801556957,
      "learning_rate": 5.340826949371975e-07,
      "loss": 0.6572,
      "step": 5217
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9366247061925059,
      "learning_rate": 5.337175999814236e-07,
      "loss": 0.6555,
      "step": 5218
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6770348353006898,
      "learning_rate": 5.333525844286846e-07,
      "loss": 0.5645,
      "step": 5219
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9267644799894419,
      "learning_rate": 5.32987648341139e-07,
      "loss": 0.6636,
      "step": 5220
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.220326052728481,
      "learning_rate": 5.326227917809318e-07,
      "loss": 0.6033,
      "step": 5221
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.3252353418415628,
      "learning_rate": 5.322580148101939e-07,
      "loss": 0.614,
      "step": 5222
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7738683498856262,
      "learning_rate": 5.318933174910433e-07,
      "loss": 0.6084,
      "step": 5223
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8526850884592909,
      "learning_rate": 5.315286998855839e-07,
      "loss": 0.6476,
      "step": 5224
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7076687168973954,
      "learning_rate": 5.311641620559064e-07,
      "loss": 0.5804,
      "step": 5225
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7663700577450718,
      "learning_rate": 5.307997040640879e-07,
      "loss": 0.5296,
      "step": 5226
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.745538053676213,
      "learning_rate": 5.304353259721917e-07,
      "loss": 0.5772,
      "step": 5227
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6682047573668682,
      "learning_rate": 5.300710278422675e-07,
      "loss": 0.5327,
      "step": 5228
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9022911408941751,
      "learning_rate": 5.297068097363517e-07,
      "loss": 0.6274,
      "step": 5229
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9149487331492179,
      "learning_rate": 5.293426717164666e-07,
      "loss": 0.6526,
      "step": 5230
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9033151642430932,
      "learning_rate": 5.289786138446213e-07,
      "loss": 0.6279,
      "step": 5231
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7831631700441274,
      "learning_rate": 5.286146361828109e-07,
      "loss": 0.5918,
      "step": 5232
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6958329192529397,
      "learning_rate": 5.282507387930172e-07,
      "loss": 0.5249,
      "step": 5233
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7075252222219387,
      "learning_rate": 5.278869217372082e-07,
      "loss": 0.5515,
      "step": 5234
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9183978303526275,
      "learning_rate": 5.275231850773378e-07,
      "loss": 0.6541,
      "step": 5235
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8780136239883257,
      "learning_rate": 5.271595288753468e-07,
      "loss": 0.6591,
      "step": 5236
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7353623876622807,
      "learning_rate": 5.267959531931619e-07,
      "loss": 0.5891,
      "step": 5237
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9051689094153474,
      "learning_rate": 5.264324580926964e-07,
      "loss": 0.6401,
      "step": 5238
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6730220712735507,
      "learning_rate": 5.260690436358497e-07,
      "loss": 0.5547,
      "step": 5239
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9176290991194557,
      "learning_rate": 5.257057098845072e-07,
      "loss": 0.6345,
      "step": 5240
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7813820128151073,
      "learning_rate": 5.253424569005415e-07,
      "loss": 0.57,
      "step": 5241
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6420388475196485,
      "learning_rate": 5.249792847458105e-07,
      "loss": 0.4953,
      "step": 5242
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6832168349309564,
      "learning_rate": 5.246161934821585e-07,
      "loss": 0.5252,
      "step": 5243
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7940851752621456,
      "learning_rate": 5.242531831714161e-07,
      "loss": 0.6698,
      "step": 5244
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8741667047660675,
      "learning_rate": 5.238902538754005e-07,
      "loss": 0.6437,
      "step": 5245
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8216352739697113,
      "learning_rate": 5.235274056559146e-07,
      "loss": 0.7138,
      "step": 5246
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.84412337834813,
      "learning_rate": 5.231646385747476e-07,
      "loss": 0.6224,
      "step": 5247
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.154907958338507,
      "learning_rate": 5.22801952693675e-07,
      "loss": 0.6757,
      "step": 5248
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7255701556653747,
      "learning_rate": 5.224393480744584e-07,
      "loss": 0.5331,
      "step": 5249
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.618183486053049,
      "learning_rate": 5.220768247788458e-07,
      "loss": 0.538,
      "step": 5250
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6478060961208488,
      "learning_rate": 5.217143828685709e-07,
      "loss": 0.5424,
      "step": 5251
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5914186487450298,
      "learning_rate": 5.213520224053538e-07,
      "loss": 0.5292,
      "step": 5252
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.753169827567143,
      "learning_rate": 5.209897434509008e-07,
      "loss": 0.6128,
      "step": 5253
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7355045094771601,
      "learning_rate": 5.206275460669041e-07,
      "loss": 0.5514,
      "step": 5254
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.010414235637921,
      "learning_rate": 5.202654303150423e-07,
      "loss": 0.6564,
      "step": 5255
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7013839497595362,
      "learning_rate": 5.199033962569799e-07,
      "loss": 0.5068,
      "step": 5256
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0835941442264907,
      "learning_rate": 5.195414439543675e-07,
      "loss": 0.6619,
      "step": 5257
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6667410188714661,
      "learning_rate": 5.191795734688418e-07,
      "loss": 0.5756,
      "step": 5258
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7830396229887109,
      "learning_rate": 5.188177848620257e-07,
      "loss": 0.5894,
      "step": 5259
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6798527671521377,
      "learning_rate": 5.184560781955278e-07,
      "loss": 0.5363,
      "step": 5260
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8589500006781942,
      "learning_rate": 5.180944535309428e-07,
      "loss": 0.5931,
      "step": 5261
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7886108360564799,
      "learning_rate": 5.177329109298522e-07,
      "loss": 0.55,
      "step": 5262
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6966495938930823,
      "learning_rate": 5.173714504538227e-07,
      "loss": 0.5222,
      "step": 5263
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.866548546216134,
      "learning_rate": 5.170100721644071e-07,
      "loss": 0.602,
      "step": 5264
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7165204682880972,
      "learning_rate": 5.166487761231446e-07,
      "loss": 0.56,
      "step": 5265
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8978838793850341,
      "learning_rate": 5.1628756239156e-07,
      "loss": 0.6409,
      "step": 5266
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8754657190017986,
      "learning_rate": 5.159264310311641e-07,
      "loss": 0.588,
      "step": 5267
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.719354666719648,
      "learning_rate": 5.15565382103454e-07,
      "loss": 0.5615,
      "step": 5268
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9280298812015718,
      "learning_rate": 5.152044156699126e-07,
      "loss": 0.6659,
      "step": 5269
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2946603685013434,
      "learning_rate": 5.148435317920086e-07,
      "loss": 0.6875,
      "step": 5270
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8132268787898511,
      "learning_rate": 5.144827305311968e-07,
      "loss": 0.578,
      "step": 5271
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7560382621948439,
      "learning_rate": 5.141220119489176e-07,
      "loss": 0.5722,
      "step": 5272
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8388697796158662,
      "learning_rate": 5.137613761065982e-07,
      "loss": 0.6233,
      "step": 5273
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7212878676237563,
      "learning_rate": 5.134008230656509e-07,
      "loss": 0.6134,
      "step": 5274
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0102500376145254,
      "learning_rate": 5.130403528874741e-07,
      "loss": 0.5896,
      "step": 5275
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8897316528675647,
      "learning_rate": 5.126799656334521e-07,
      "loss": 0.6537,
      "step": 5276
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8505618394093455,
      "learning_rate": 5.123196613649551e-07,
      "loss": 0.6015,
      "step": 5277
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9282130023848788,
      "learning_rate": 5.119594401433391e-07,
      "loss": 0.6375,
      "step": 5278
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8378939228950745,
      "learning_rate": 5.11599302029946e-07,
      "loss": 0.6331,
      "step": 5279
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8671395913515828,
      "learning_rate": 5.112392470861038e-07,
      "loss": 0.6107,
      "step": 5280
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7353977111902207,
      "learning_rate": 5.108792753731259e-07,
      "loss": 0.5415,
      "step": 5281
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.615257343846587,
      "learning_rate": 5.105193869523119e-07,
      "loss": 0.5139,
      "step": 5282
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.878878376143998,
      "learning_rate": 5.101595818849467e-07,
      "loss": 0.6632,
      "step": 5283
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7426325358489514,
      "learning_rate": 5.097998602323017e-07,
      "loss": 0.5937,
      "step": 5284
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0063001703405317,
      "learning_rate": 5.094402220556336e-07,
      "loss": 0.7121,
      "step": 5285
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.629877317027908,
      "learning_rate": 5.090806674161852e-07,
      "loss": 0.5293,
      "step": 5286
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.4359764263321377,
      "learning_rate": 5.087211963751845e-07,
      "loss": 0.6293,
      "step": 5287
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9735182744330695,
      "learning_rate": 5.083618089938461e-07,
      "loss": 0.6462,
      "step": 5288
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7410989473893468,
      "learning_rate": 5.080025053333699e-07,
      "loss": 0.601,
      "step": 5289
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9191921987929599,
      "learning_rate": 5.076432854549413e-07,
      "loss": 0.6169,
      "step": 5290
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7102658769122203,
      "learning_rate": 5.072841494197319e-07,
      "loss": 0.5737,
      "step": 5291
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7545552669266572,
      "learning_rate": 5.069250972888987e-07,
      "loss": 0.6293,
      "step": 5292
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.737364114686024,
      "learning_rate": 5.065661291235848e-07,
      "loss": 0.5635,
      "step": 5293
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.24965071666077,
      "learning_rate": 5.062072449849184e-07,
      "loss": 0.6541,
      "step": 5294
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7110705014516703,
      "learning_rate": 5.058484449340139e-07,
      "loss": 0.5581,
      "step": 5295
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.685611848078359,
      "learning_rate": 5.054897290319713e-07,
      "loss": 0.5915,
      "step": 5296
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8795104117308582,
      "learning_rate": 5.05131097339876e-07,
      "loss": 0.5966,
      "step": 5297
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.969993802229717,
      "learning_rate": 5.047725499187994e-07,
      "loss": 0.6101,
      "step": 5298
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7949357695213908,
      "learning_rate": 5.044140868297977e-07,
      "loss": 0.6163,
      "step": 5299
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8837592275707001,
      "learning_rate": 5.040557081339147e-07,
      "loss": 0.5009,
      "step": 5300
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8843526084793552,
      "learning_rate": 5.036974138921777e-07,
      "loss": 0.615,
      "step": 5301
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6142618186009866,
      "learning_rate": 5.033392041656006e-07,
      "loss": 0.5127,
      "step": 5302
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6937589805017076,
      "learning_rate": 5.02981079015183e-07,
      "loss": 0.5385,
      "step": 5303
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7304270496164742,
      "learning_rate": 5.026230385019098e-07,
      "loss": 0.5652,
      "step": 5304
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9875031822363577,
      "learning_rate": 5.022650826867513e-07,
      "loss": 0.604,
      "step": 5305
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6684967370460357,
      "learning_rate": 5.019072116306641e-07,
      "loss": 0.5338,
      "step": 5306
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7063319387154257,
      "learning_rate": 5.015494253945896e-07,
      "loss": 0.538,
      "step": 5307
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8665107843552544,
      "learning_rate": 5.01191724039455e-07,
      "loss": 0.6361,
      "step": 5308
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7686115630357471,
      "learning_rate": 5.008341076261733e-07,
      "loss": 0.4826,
      "step": 5309
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.887435457854863,
      "learning_rate": 5.004765762156429e-07,
      "loss": 0.5868,
      "step": 5310
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8869826586975291,
      "learning_rate": 5.001191298687475e-07,
      "loss": 0.6357,
      "step": 5311
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7425486620191694,
      "learning_rate": 4.997617686463565e-07,
      "loss": 0.5726,
      "step": 5312
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7858603556155882,
      "learning_rate": 4.994044926093249e-07,
      "loss": 0.5464,
      "step": 5313
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7577009504435037,
      "learning_rate": 4.990473018184931e-07,
      "loss": 0.6019,
      "step": 5314
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8668237698065206,
      "learning_rate": 4.986901963346869e-07,
      "loss": 0.6207,
      "step": 5315
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7245318167686288,
      "learning_rate": 4.983331762187176e-07,
      "loss": 0.524,
      "step": 5316
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.683395776996338,
      "learning_rate": 4.97976241531382e-07,
      "loss": 0.5644,
      "step": 5317
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7571299879501692,
      "learning_rate": 4.976193923334626e-07,
      "loss": 0.5495,
      "step": 5318
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7300483211339173,
      "learning_rate": 4.972626286857268e-07,
      "loss": 0.5593,
      "step": 5319
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7047413234461785,
      "learning_rate": 4.96905950648928e-07,
      "loss": 0.5551,
      "step": 5320
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6813292877537857,
      "learning_rate": 4.965493582838044e-07,
      "loss": 0.5423,
      "step": 5321
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8411840294161926,
      "learning_rate": 4.961928516510803e-07,
      "loss": 0.636,
      "step": 5322
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8450954306169717,
      "learning_rate": 4.958364308114649e-07,
      "loss": 0.6756,
      "step": 5323
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7771242857376862,
      "learning_rate": 4.95480095825653e-07,
      "loss": 0.5848,
      "step": 5324
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6992460123020512,
      "learning_rate": 4.951238467543244e-07,
      "loss": 0.5164,
      "step": 5325
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0010501165629846,
      "learning_rate": 4.947676836581452e-07,
      "loss": 0.5568,
      "step": 5326
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9998962142215632,
      "learning_rate": 4.94411606597766e-07,
      "loss": 0.635,
      "step": 5327
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.74514198843349,
      "learning_rate": 4.940556156338231e-07,
      "loss": 0.5368,
      "step": 5328
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6677414549393945,
      "learning_rate": 4.936997108269381e-07,
      "loss": 0.5323,
      "step": 5329
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6885715516513323,
      "learning_rate": 4.933438922377176e-07,
      "loss": 0.5661,
      "step": 5330
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.56358116654663,
      "learning_rate": 4.929881599267539e-07,
      "loss": 0.6491,
      "step": 5331
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6476102502359745,
      "learning_rate": 4.926325139546246e-07,
      "loss": 0.5714,
      "step": 5332
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6232819097893236,
      "learning_rate": 4.922769543818924e-07,
      "loss": 0.5471,
      "step": 5333
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7244851355407912,
      "learning_rate": 4.919214812691056e-07,
      "loss": 0.6057,
      "step": 5334
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9873731326871554,
      "learning_rate": 4.915660946767972e-07,
      "loss": 0.5867,
      "step": 5335
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7993046286504325,
      "learning_rate": 4.912107946654862e-07,
      "loss": 0.5991,
      "step": 5336
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7952390025935714,
      "learning_rate": 4.908555812956764e-07,
      "loss": 0.5952,
      "step": 5337
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6855435946785916,
      "learning_rate": 4.905004546278567e-07,
      "loss": 0.5639,
      "step": 5338
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8821705625877165,
      "learning_rate": 4.901454147225018e-07,
      "loss": 0.6427,
      "step": 5339
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7911148304304257,
      "learning_rate": 4.897904616400711e-07,
      "loss": 0.6082,
      "step": 5340
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7955060654750401,
      "learning_rate": 4.894355954410095e-07,
      "loss": 0.5583,
      "step": 5341
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.731755458288174,
      "learning_rate": 4.890808161857468e-07,
      "loss": 0.5424,
      "step": 5342
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8486989880107244,
      "learning_rate": 4.887261239346986e-07,
      "loss": 0.5924,
      "step": 5343
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0712876951691928,
      "learning_rate": 4.88371518748265e-07,
      "loss": 0.6255,
      "step": 5344
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6678311957511555,
      "learning_rate": 4.880170006868316e-07,
      "loss": 0.5236,
      "step": 5345
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6848651438621817,
      "learning_rate": 4.876625698107691e-07,
      "loss": 0.5932,
      "step": 5346
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7594326502880127,
      "learning_rate": 4.873082261804336e-07,
      "loss": 0.5756,
      "step": 5347
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8984224184727355,
      "learning_rate": 4.869539698561661e-07,
      "loss": 0.6246,
      "step": 5348
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7187588824413186,
      "learning_rate": 4.865998008982926e-07,
      "loss": 0.5678,
      "step": 5349
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.650651644887821,
      "learning_rate": 4.862457193671243e-07,
      "loss": 0.5338,
      "step": 5350
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7043223100513321,
      "learning_rate": 4.858917253229574e-07,
      "loss": 0.5058,
      "step": 5351
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6202547253881282,
      "learning_rate": 4.855378188260743e-07,
      "loss": 0.5137,
      "step": 5352
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7706827222787114,
      "learning_rate": 4.851839999367407e-07,
      "loss": 0.5549,
      "step": 5353
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6466321952568665,
      "learning_rate": 4.848302687152089e-07,
      "loss": 0.5025,
      "step": 5354
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8410774590558495,
      "learning_rate": 4.844766252217152e-07,
      "loss": 0.6136,
      "step": 5355
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7369558673607465,
      "learning_rate": 4.841230695164815e-07,
      "loss": 0.5931,
      "step": 5356
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8479113600681114,
      "learning_rate": 4.837696016597148e-07,
      "loss": 0.622,
      "step": 5357
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6616251657493748,
      "learning_rate": 4.834162217116069e-07,
      "loss": 0.5081,
      "step": 5358
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9767221266654141,
      "learning_rate": 4.830629297323346e-07,
      "loss": 0.6177,
      "step": 5359
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8424822991394464,
      "learning_rate": 4.827097257820601e-07,
      "loss": 0.5938,
      "step": 5360
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7770916356224592,
      "learning_rate": 4.823566099209301e-07,
      "loss": 0.5845,
      "step": 5361
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6222341504716099,
      "learning_rate": 4.820035822090767e-07,
      "loss": 0.485,
      "step": 5362
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6942105915552753,
      "learning_rate": 4.816506427066164e-07,
      "loss": 0.5289,
      "step": 5363
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6426510202102209,
      "learning_rate": 4.812977914736519e-07,
      "loss": 0.5284,
      "step": 5364
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6379116141685525,
      "learning_rate": 4.809450285702697e-07,
      "loss": 0.5211,
      "step": 5365
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7268946608784657,
      "learning_rate": 4.805923540565417e-07,
      "loss": 0.5552,
      "step": 5366
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.70876899929458,
      "learning_rate": 4.802397679925246e-07,
      "loss": 0.5821,
      "step": 5367
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7440894802423124,
      "learning_rate": 4.798872704382602e-07,
      "loss": 0.6089,
      "step": 5368
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.914052763542029,
      "learning_rate": 4.795348614537752e-07,
      "loss": 0.6564,
      "step": 5369
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8626181854178324,
      "learning_rate": 4.79182541099081e-07,
      "loss": 0.6446,
      "step": 5370
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6466558367009497,
      "learning_rate": 4.788303094341744e-07,
      "loss": 0.546,
      "step": 5371
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6993432650760381,
      "learning_rate": 4.784781665190365e-07,
      "loss": 0.5231,
      "step": 5372
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6505623513086882,
      "learning_rate": 4.781261124136337e-07,
      "loss": 0.5454,
      "step": 5373
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7258466623182719,
      "learning_rate": 4.777741471779171e-07,
      "loss": 0.5641,
      "step": 5374
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7849937377412379,
      "learning_rate": 4.774222708718226e-07,
      "loss": 0.6144,
      "step": 5375
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6637375345104587,
      "learning_rate": 4.770704835552713e-07,
      "loss": 0.5251,
      "step": 5376
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9213384543008514,
      "learning_rate": 4.767187852881689e-07,
      "loss": 0.6607,
      "step": 5377
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6176858218836799,
      "learning_rate": 4.763671761304058e-07,
      "loss": 0.5049,
      "step": 5378
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6585776331949535,
      "learning_rate": 4.7601565614185733e-07,
      "loss": 0.5306,
      "step": 5379
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.826581327533941,
      "learning_rate": 4.7566422538238396e-07,
      "loss": 0.5775,
      "step": 5380
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7015746641579432,
      "learning_rate": 4.7531288391183045e-07,
      "loss": 0.5218,
      "step": 5381
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9698192069115962,
      "learning_rate": 4.7496163179002657e-07,
      "loss": 0.5877,
      "step": 5382
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6367217678719059,
      "learning_rate": 4.746104690767867e-07,
      "loss": 0.4858,
      "step": 5383
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8053692197725493,
      "learning_rate": 4.7425939583191086e-07,
      "loss": 0.6054,
      "step": 5384
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.841630921104746,
      "learning_rate": 4.739084121151827e-07,
      "loss": 0.5425,
      "step": 5385
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.825536499366827,
      "learning_rate": 4.7355751798637125e-07,
      "loss": 0.6113,
      "step": 5386
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.731044592470668,
      "learning_rate": 4.7320671350523014e-07,
      "loss": 0.5158,
      "step": 5387
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7928247003372195,
      "learning_rate": 4.728559987314974e-07,
      "loss": 0.5618,
      "step": 5388
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8854642415833501,
      "learning_rate": 4.725053737248965e-07,
      "loss": 0.6243,
      "step": 5389
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6138392983921646,
      "learning_rate": 4.7215483854513503e-07,
      "loss": 0.4653,
      "step": 5390
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8706432280004185,
      "learning_rate": 4.7180439325190546e-07,
      "loss": 0.6316,
      "step": 5391
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9340318775890795,
      "learning_rate": 4.71454037904885e-07,
      "loss": 0.6265,
      "step": 5392
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7660005863966213,
      "learning_rate": 4.711037725637356e-07,
      "loss": 0.5089,
      "step": 5393
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7385013963002642,
      "learning_rate": 4.7075359728810373e-07,
      "loss": 0.5637,
      "step": 5394
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9181219030284592,
      "learning_rate": 4.704035121376205e-07,
      "loss": 0.6872,
      "step": 5395
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6058211069189818,
      "learning_rate": 4.70053517171902e-07,
      "loss": 0.537,
      "step": 5396
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7788836708133637,
      "learning_rate": 4.697036124505485e-07,
      "loss": 0.6061,
      "step": 5397
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6522445791673998,
      "learning_rate": 4.693537980331451e-07,
      "loss": 0.5134,
      "step": 5398
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6574731077984831,
      "learning_rate": 4.690040739792619e-07,
      "loss": 0.5385,
      "step": 5399
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6340058946059042,
      "learning_rate": 4.686544403484528e-07,
      "loss": 0.5056,
      "step": 5400
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8216824012672439,
      "learning_rate": 4.683048972002571e-07,
      "loss": 0.645,
      "step": 5401
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.985222828142204,
      "learning_rate": 4.6795544459419833e-07,
      "loss": 0.6752,
      "step": 5402
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8274199509625679,
      "learning_rate": 4.676060825897844e-07,
      "loss": 0.5486,
      "step": 5403
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7895062043621063,
      "learning_rate": 4.672568112465083e-07,
      "loss": 0.517,
      "step": 5404
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.710469230499792,
      "learning_rate": 4.669076306238471e-07,
      "loss": 0.5158,
      "step": 5405
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7545364115552494,
      "learning_rate": 4.6655854078126255e-07,
      "loss": 0.545,
      "step": 5406
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8152647831343788,
      "learning_rate": 4.662095417782014e-07,
      "loss": 0.5832,
      "step": 5407
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7791371657582434,
      "learning_rate": 4.658606336740941e-07,
      "loss": 0.6179,
      "step": 5408
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8236075044117962,
      "learning_rate": 4.6551181652835635e-07,
      "loss": 0.6195,
      "step": 5409
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8621228278848556,
      "learning_rate": 4.651630904003876e-07,
      "loss": 0.6138,
      "step": 5410
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6768093849047949,
      "learning_rate": 4.648144553495731e-07,
      "loss": 0.5115,
      "step": 5411
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7283713505965954,
      "learning_rate": 4.6446591143528125e-07,
      "loss": 0.5281,
      "step": 5412
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7995898360124369,
      "learning_rate": 4.6411745871686547e-07,
      "loss": 0.6187,
      "step": 5413
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7457992718961622,
      "learning_rate": 4.6376909725366385e-07,
      "loss": 0.5287,
      "step": 5414
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8039543359968708,
      "learning_rate": 4.634208271049984e-07,
      "loss": 0.5951,
      "step": 5415
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.682833332912845,
      "learning_rate": 4.6307264833017614e-07,
      "loss": 0.5317,
      "step": 5416
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8726562211047633,
      "learning_rate": 4.627245609884882e-07,
      "loss": 0.6236,
      "step": 5417
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6876557906121669,
      "learning_rate": 4.623765651392102e-07,
      "loss": 0.5487,
      "step": 5418
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8346329905768244,
      "learning_rate": 4.620286608416023e-07,
      "loss": 0.6494,
      "step": 5419
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8903808785599667,
      "learning_rate": 4.616808481549088e-07,
      "loss": 0.6894,
      "step": 5420
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8858233931578466,
      "learning_rate": 4.6133312713835884e-07,
      "loss": 0.654,
      "step": 5421
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.709617619202685,
      "learning_rate": 4.609854978511654e-07,
      "loss": 0.618,
      "step": 5422
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6515773590469973,
      "learning_rate": 4.606379603525263e-07,
      "loss": 0.5169,
      "step": 5423
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7719779062568372,
      "learning_rate": 4.602905147016235e-07,
      "loss": 0.6225,
      "step": 5424
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.674537950314378,
      "learning_rate": 4.599431609576234e-07,
      "loss": 0.5827,
      "step": 5425
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8868170888717785,
      "learning_rate": 4.5959589917967677e-07,
      "loss": 0.6377,
      "step": 5426
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8509949939718056,
      "learning_rate": 4.592487294269186e-07,
      "loss": 0.6024,
      "step": 5427
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8585723877774838,
      "learning_rate": 4.589016517584683e-07,
      "loss": 0.5923,
      "step": 5428
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9555101806054861,
      "learning_rate": 4.5855466623342965e-07,
      "loss": 0.5156,
      "step": 5429
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7065212174768765,
      "learning_rate": 4.5820777291089054e-07,
      "loss": 0.5895,
      "step": 5430
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1458470540580727,
      "learning_rate": 4.578609718499233e-07,
      "loss": 0.6076,
      "step": 5431
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6417300910768817,
      "learning_rate": 4.5751426310958465e-07,
      "loss": 0.525,
      "step": 5432
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7317180485591868,
      "learning_rate": 4.571676467489154e-07,
      "loss": 0.5767,
      "step": 5433
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8181700208681539,
      "learning_rate": 4.568211228269407e-07,
      "loss": 0.6143,
      "step": 5434
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9302081049439518,
      "learning_rate": 4.5647469140267e-07,
      "loss": 0.6227,
      "step": 5435
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8865776394814306,
      "learning_rate": 4.5612835253509665e-07,
      "loss": 0.6302,
      "step": 5436
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6732611085819743,
      "learning_rate": 4.5578210628319915e-07,
      "loss": 0.5563,
      "step": 5437
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8192906377847118,
      "learning_rate": 4.554359527059394e-07,
      "loss": 0.6171,
      "step": 5438
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8966213269191172,
      "learning_rate": 4.5508989186226356e-07,
      "loss": 0.6291,
      "step": 5439
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8168979636280663,
      "learning_rate": 4.547439238111024e-07,
      "loss": 0.5361,
      "step": 5440
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6535345477706126,
      "learning_rate": 4.5439804861137055e-07,
      "loss": 0.5414,
      "step": 5441
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8215570384535594,
      "learning_rate": 4.540522663219669e-07,
      "loss": 0.6539,
      "step": 5442
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8151970345566367,
      "learning_rate": 4.537065770017747e-07,
      "loss": 0.6144,
      "step": 5443
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8846753475701035,
      "learning_rate": 4.5336098070966124e-07,
      "loss": 0.6216,
      "step": 5444
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.729705915871295,
      "learning_rate": 4.5301547750447786e-07,
      "loss": 0.572,
      "step": 5445
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7217828875465638,
      "learning_rate": 4.526700674450603e-07,
      "loss": 0.5095,
      "step": 5446
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9496949765365021,
      "learning_rate": 4.523247505902282e-07,
      "loss": 0.6538,
      "step": 5447
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9126392855265738,
      "learning_rate": 4.519795269987854e-07,
      "loss": 0.6182,
      "step": 5448
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8371751953861961,
      "learning_rate": 4.5163439672951987e-07,
      "loss": 0.5764,
      "step": 5449
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7579529939217367,
      "learning_rate": 4.5128935984120384e-07,
      "loss": 0.604,
      "step": 5450
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6854682027542485,
      "learning_rate": 4.509444163925934e-07,
      "loss": 0.5506,
      "step": 5451
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7390644556190168,
      "learning_rate": 4.505995664424287e-07,
      "loss": 0.5809,
      "step": 5452
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7566369743146946,
      "learning_rate": 4.5025481004943435e-07,
      "loss": 0.5219,
      "step": 5453
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7995271557726705,
      "learning_rate": 4.499101472723186e-07,
      "loss": 0.538,
      "step": 5454
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8391448800061801,
      "learning_rate": 4.49565578169774e-07,
      "loss": 0.5668,
      "step": 5455
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6280776698791654,
      "learning_rate": 4.4922110280047697e-07,
      "loss": 0.5291,
      "step": 5456
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.725429061941008,
      "learning_rate": 4.488767212230883e-07,
      "loss": 0.4987,
      "step": 5457
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.784989057575474,
      "learning_rate": 4.4853243349625235e-07,
      "loss": 0.5771,
      "step": 5458
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8681769671128385,
      "learning_rate": 4.4818823967859786e-07,
      "loss": 0.6138,
      "step": 5459
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6212673666159269,
      "learning_rate": 4.4784413982873735e-07,
      "loss": 0.5826,
      "step": 5460
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8093977730511368,
      "learning_rate": 4.4750013400526755e-07,
      "loss": 0.6426,
      "step": 5461
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7435666575191491,
      "learning_rate": 4.4715622226676873e-07,
      "loss": 0.5513,
      "step": 5462
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9319125983948559,
      "learning_rate": 4.46812404671806e-07,
      "loss": 0.6438,
      "step": 5463
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9092075390017121,
      "learning_rate": 4.4646868127892777e-07,
      "loss": 0.6443,
      "step": 5464
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8003580713638484,
      "learning_rate": 4.461250521466665e-07,
      "loss": 0.598,
      "step": 5465
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7505898516711969,
      "learning_rate": 4.4578151733353866e-07,
      "loss": 0.5437,
      "step": 5466
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9278691649570422,
      "learning_rate": 4.4543807689804445e-07,
      "loss": 0.6703,
      "step": 5467
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6251566013535413,
      "learning_rate": 4.450947308986681e-07,
      "loss": 0.5165,
      "step": 5468
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7784289964980218,
      "learning_rate": 4.4475147939387845e-07,
      "loss": 0.5551,
      "step": 5469
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7046830964982311,
      "learning_rate": 4.444083224421274e-07,
      "loss": 0.5774,
      "step": 5470
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.007966619535289,
      "learning_rate": 4.440652601018509e-07,
      "loss": 0.6578,
      "step": 5471
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8001453062216861,
      "learning_rate": 4.4372229243146885e-07,
      "loss": 0.6104,
      "step": 5472
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.777214701326824,
      "learning_rate": 4.433794194893852e-07,
      "loss": 0.5306,
      "step": 5473
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8694563725406029,
      "learning_rate": 4.4303664133398755e-07,
      "loss": 0.6012,
      "step": 5474
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7123246974548884,
      "learning_rate": 4.4269395802364753e-07,
      "loss": 0.5779,
      "step": 5475
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8319572261155206,
      "learning_rate": 4.423513696167204e-07,
      "loss": 0.5454,
      "step": 5476
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6065292101247821,
      "learning_rate": 4.420088761715455e-07,
      "loss": 0.5134,
      "step": 5477
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8560047090675773,
      "learning_rate": 4.416664777464459e-07,
      "loss": 0.6129,
      "step": 5478
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7535040239512761,
      "learning_rate": 4.413241743997285e-07,
      "loss": 0.6001,
      "step": 5479
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9068458815192074,
      "learning_rate": 4.409819661896839e-07,
      "loss": 0.5758,
      "step": 5480
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7593317337484493,
      "learning_rate": 4.406398531745866e-07,
      "loss": 0.5953,
      "step": 5481
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.833624812597557,
      "learning_rate": 4.4029783541269507e-07,
      "loss": 0.6255,
      "step": 5482
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8526117079446215,
      "learning_rate": 4.399559129622512e-07,
      "loss": 0.5308,
      "step": 5483
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8410899081206291,
      "learning_rate": 4.396140858814809e-07,
      "loss": 0.6614,
      "step": 5484
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0075001997508735,
      "learning_rate": 4.3927235422859355e-07,
      "loss": 0.6677,
      "step": 5485
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7921228782738354,
      "learning_rate": 4.389307180617827e-07,
      "loss": 0.5428,
      "step": 5486
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7628654098416533,
      "learning_rate": 4.3858917743922554e-07,
      "loss": 0.5724,
      "step": 5487
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6754606242335749,
      "learning_rate": 4.3824773241908255e-07,
      "loss": 0.5401,
      "step": 5488
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8325025122674288,
      "learning_rate": 4.3790638305949855e-07,
      "loss": 0.5876,
      "step": 5489
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7374963133273117,
      "learning_rate": 4.375651294186016e-07,
      "loss": 0.6136,
      "step": 5490
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.738609331673596,
      "learning_rate": 4.3722397155450376e-07,
      "loss": 0.5348,
      "step": 5491
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8320151306912408,
      "learning_rate": 4.368829095253006e-07,
      "loss": 0.6434,
      "step": 5492
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7718077296972778,
      "learning_rate": 4.3654194338907137e-07,
      "loss": 0.541,
      "step": 5493
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6513105141051528,
      "learning_rate": 4.362010732038788e-07,
      "loss": 0.5369,
      "step": 5494
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7277128432539303,
      "learning_rate": 4.358602990277702e-07,
      "loss": 0.5627,
      "step": 5495
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8881761845421416,
      "learning_rate": 4.355196209187755e-07,
      "loss": 0.6361,
      "step": 5496
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.763204549274549,
      "learning_rate": 4.351790389349087e-07,
      "loss": 0.5723,
      "step": 5497
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7201401145253701,
      "learning_rate": 4.348385531341673e-07,
      "loss": 0.5773,
      "step": 5498
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9060093358781092,
      "learning_rate": 4.344981635745324e-07,
      "loss": 0.6483,
      "step": 5499
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5965228711429188,
      "learning_rate": 4.34157870313969e-07,
      "loss": 0.5326,
      "step": 5500
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.004376062188023,
      "learning_rate": 4.338176734104253e-07,
      "loss": 0.6063,
      "step": 5501
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9089965818285728,
      "learning_rate": 4.334775729218334e-07,
      "loss": 0.6732,
      "step": 5502
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.800569772158204,
      "learning_rate": 4.331375689061089e-07,
      "loss": 0.6087,
      "step": 5503
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6734584341516239,
      "learning_rate": 4.327976614211508e-07,
      "loss": 0.5426,
      "step": 5504
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8396604902951967,
      "learning_rate": 4.32457850524842e-07,
      "loss": 0.6049,
      "step": 5505
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7400415508096883,
      "learning_rate": 4.3211813627504843e-07,
      "loss": 0.5582,
      "step": 5506
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8380971393410184,
      "learning_rate": 4.3177851872962035e-07,
      "loss": 0.6492,
      "step": 5507
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8236259890371715,
      "learning_rate": 4.314389979463907e-07,
      "loss": 0.6377,
      "step": 5508
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8305802797963977,
      "learning_rate": 4.3109957398317653e-07,
      "loss": 0.5864,
      "step": 5509
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1018208532116782,
      "learning_rate": 4.3076024689777813e-07,
      "loss": 0.6848,
      "step": 5510
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7469160916261077,
      "learning_rate": 4.3042101674797937e-07,
      "loss": 0.5668,
      "step": 5511
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.753334978112115,
      "learning_rate": 4.3008188359154764e-07,
      "loss": 0.5494,
      "step": 5512
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7018113143756353,
      "learning_rate": 4.297428474862337e-07,
      "loss": 0.5151,
      "step": 5513
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7211559822408069,
      "learning_rate": 4.2940390848977203e-07,
      "loss": 0.5586,
      "step": 5514
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6394960443270393,
      "learning_rate": 4.290650666598803e-07,
      "loss": 0.5642,
      "step": 5515
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6831244535396285,
      "learning_rate": 4.287263220542597e-07,
      "loss": 0.5809,
      "step": 5516
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8256276629619962,
      "learning_rate": 4.283876747305951e-07,
      "loss": 0.5735,
      "step": 5517
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8141503761402799,
      "learning_rate": 4.2804912474655443e-07,
      "loss": 0.584,
      "step": 5518
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6692471322025692,
      "learning_rate": 4.277106721597892e-07,
      "loss": 0.5567,
      "step": 5519
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.735135640863164,
      "learning_rate": 4.2737231702793455e-07,
      "loss": 0.5081,
      "step": 5520
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6381807141610216,
      "learning_rate": 4.2703405940860826e-07,
      "loss": 0.5094,
      "step": 5521
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6404782407849322,
      "learning_rate": 4.2669589935941284e-07,
      "loss": 0.534,
      "step": 5522
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9221702588493129,
      "learning_rate": 4.2635783693793324e-07,
      "loss": 0.5956,
      "step": 5523
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9063132128079254,
      "learning_rate": 4.2601987220173785e-07,
      "loss": 0.5846,
      "step": 5524
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7819918722991914,
      "learning_rate": 4.256820052083785e-07,
      "loss": 0.5443,
      "step": 5525
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8391613121642005,
      "learning_rate": 4.2534423601539047e-07,
      "loss": 0.6487,
      "step": 5526
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9948903162400459,
      "learning_rate": 4.250065646802923e-07,
      "loss": 0.6315,
      "step": 5527
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8522128341197969,
      "learning_rate": 4.24668991260586e-07,
      "loss": 0.6446,
      "step": 5528
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8546929810890971,
      "learning_rate": 4.243315158137566e-07,
      "loss": 0.5616,
      "step": 5529
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6860412444730833,
      "learning_rate": 4.2399413839727296e-07,
      "loss": 0.6015,
      "step": 5530
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7878612177956198,
      "learning_rate": 4.236568590685866e-07,
      "loss": 0.5659,
      "step": 5531
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7275369402110485,
      "learning_rate": 4.233196778851329e-07,
      "loss": 0.5439,
      "step": 5532
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.840614643868386,
      "learning_rate": 4.229825949043303e-07,
      "loss": 0.5312,
      "step": 5533
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7060702643029093,
      "learning_rate": 4.2264561018358047e-07,
      "loss": 0.5083,
      "step": 5534
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8970001489253682,
      "learning_rate": 4.223087237802683e-07,
      "loss": 0.541,
      "step": 5535
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8826539833368219,
      "learning_rate": 4.2197193575176217e-07,
      "loss": 0.6611,
      "step": 5536
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6435859316896598,
      "learning_rate": 4.216352461554136e-07,
      "loss": 0.5372,
      "step": 5537
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8131539739702812,
      "learning_rate": 4.212986550485571e-07,
      "loss": 0.6648,
      "step": 5538
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.896726324143933,
      "learning_rate": 4.209621624885109e-07,
      "loss": 0.6451,
      "step": 5539
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6782656362241735,
      "learning_rate": 4.20625768532576e-07,
      "loss": 0.5786,
      "step": 5540
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7944266615404366,
      "learning_rate": 4.2028947323803686e-07,
      "loss": 0.6402,
      "step": 5541
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8766927667934332,
      "learning_rate": 4.199532766621611e-07,
      "loss": 0.592,
      "step": 5542
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8541194918668754,
      "learning_rate": 4.1961717886219926e-07,
      "loss": 0.531,
      "step": 5543
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8632312428897135,
      "learning_rate": 4.192811798953857e-07,
      "loss": 0.668,
      "step": 5544
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7723574711996681,
      "learning_rate": 4.1894527981893713e-07,
      "loss": 0.5253,
      "step": 5545
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7540666300392412,
      "learning_rate": 4.1860947869005416e-07,
      "loss": 0.5755,
      "step": 5546
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.576732516607456,
      "learning_rate": 4.182737765659197e-07,
      "loss": 0.6129,
      "step": 5547
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1285833861803778,
      "learning_rate": 4.17938173503701e-07,
      "loss": 0.5671,
      "step": 5548
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9817668805690377,
      "learning_rate": 4.176026695605476e-07,
      "loss": 0.6565,
      "step": 5549
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.731479778961776,
      "learning_rate": 4.172672647935921e-07,
      "loss": 0.5108,
      "step": 5550
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6402725864570282,
      "learning_rate": 4.169319592599505e-07,
      "loss": 0.5348,
      "step": 5551
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6377211230997982,
      "learning_rate": 4.165967530167219e-07,
      "loss": 0.522,
      "step": 5552
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6580860332714137,
      "learning_rate": 4.162616461209886e-07,
      "loss": 0.552,
      "step": 5553
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8534743055359045,
      "learning_rate": 4.1592663862981534e-07,
      "loss": 0.5554,
      "step": 5554
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.762920555317595,
      "learning_rate": 4.1559173060025086e-07,
      "loss": 0.5212,
      "step": 5555
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7513913814383733,
      "learning_rate": 4.152569220893264e-07,
      "loss": 0.4937,
      "step": 5556
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9096179858363339,
      "learning_rate": 4.149222131540562e-07,
      "loss": 0.6915,
      "step": 5557
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8283759306999988,
      "learning_rate": 4.145876038514378e-07,
      "loss": 0.5996,
      "step": 5558
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7923393508116303,
      "learning_rate": 4.142530942384517e-07,
      "loss": 0.6166,
      "step": 5559
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4006509383864563,
      "learning_rate": 4.139186843720613e-07,
      "loss": 0.5845,
      "step": 5560
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7081437891227408,
      "learning_rate": 4.1358437430921336e-07,
      "loss": 0.5648,
      "step": 5561
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8103297699357376,
      "learning_rate": 4.132501641068371e-07,
      "loss": 0.6215,
      "step": 5562
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0271387867929571,
      "learning_rate": 4.1291605382184525e-07,
      "loss": 0.6089,
      "step": 5563
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8478144195009228,
      "learning_rate": 4.1258204351113325e-07,
      "loss": 0.6546,
      "step": 5564
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7643265149216378,
      "learning_rate": 4.122481332315795e-07,
      "loss": 0.5944,
      "step": 5565
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0342018426447956,
      "learning_rate": 4.119143230400451e-07,
      "loss": 0.6221,
      "step": 5566
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6330615775923993,
      "learning_rate": 4.1158061299337534e-07,
      "loss": 0.5793,
      "step": 5567
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9337947529072448,
      "learning_rate": 4.112470031483971e-07,
      "loss": 0.6478,
      "step": 5568
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7803883003424512,
      "learning_rate": 4.1091349356192053e-07,
      "loss": 0.6011,
      "step": 5569
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6224534916134425,
      "learning_rate": 4.10580084290739e-07,
      "loss": 0.4646,
      "step": 5570
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6904933503810439,
      "learning_rate": 4.1024677539162855e-07,
      "loss": 0.5069,
      "step": 5571
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8430228862862325,
      "learning_rate": 4.0991356692134824e-07,
      "loss": 0.6169,
      "step": 5572
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7251585423754022,
      "learning_rate": 4.0958045893664005e-07,
      "loss": 0.559,
      "step": 5573
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.787634014497346,
      "learning_rate": 4.092474514942288e-07,
      "loss": 0.5764,
      "step": 5574
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9395665278061174,
      "learning_rate": 4.089145446508221e-07,
      "loss": 0.6621,
      "step": 5575
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.631691205020156,
      "learning_rate": 4.0858173846311053e-07,
      "loss": 0.5216,
      "step": 5576
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8114871879530118,
      "learning_rate": 4.082490329877676e-07,
      "loss": 0.5746,
      "step": 5577
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7738383667642961,
      "learning_rate": 4.0791642828144944e-07,
      "loss": 0.5957,
      "step": 5578
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8160265639013825,
      "learning_rate": 4.07583924400795e-07,
      "loss": 0.6197,
      "step": 5579
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6608551580797347,
      "learning_rate": 4.072515214024268e-07,
      "loss": 0.6036,
      "step": 5580
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7443339107534718,
      "learning_rate": 4.0691921934294937e-07,
      "loss": 0.5794,
      "step": 5581
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0708714061671945,
      "learning_rate": 4.0658701827895016e-07,
      "loss": 0.5967,
      "step": 5582
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8255139039207315,
      "learning_rate": 4.0625491826699953e-07,
      "loss": 0.5086,
      "step": 5583
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7097246845586394,
      "learning_rate": 4.0592291936365086e-07,
      "loss": 0.5201,
      "step": 5584
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7349182839850993,
      "learning_rate": 4.055910216254399e-07,
      "loss": 0.5859,
      "step": 5585
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9690631065422957,
      "learning_rate": 4.0525922510888546e-07,
      "loss": 0.6849,
      "step": 5586
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7225041063573439,
      "learning_rate": 4.0492752987048893e-07,
      "loss": 0.5281,
      "step": 5587
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6843323170409429,
      "learning_rate": 4.045959359667348e-07,
      "loss": 0.5059,
      "step": 5588
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.011963930556679,
      "learning_rate": 4.0426444345408983e-07,
      "loss": 0.5892,
      "step": 5589
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7712460039729854,
      "learning_rate": 4.0393305238900374e-07,
      "loss": 0.5907,
      "step": 5590
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.86221860675215,
      "learning_rate": 4.0360176282790925e-07,
      "loss": 0.6344,
      "step": 5591
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8832338119075592,
      "learning_rate": 4.0327057482722117e-07,
      "loss": 0.5771,
      "step": 5592
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6865511323517142,
      "learning_rate": 4.029394884433375e-07,
      "loss": 0.5319,
      "step": 5593
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6615191354260711,
      "learning_rate": 4.0260850373263887e-07,
      "loss": 0.5508,
      "step": 5594
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7890917712144613,
      "learning_rate": 4.022776207514884e-07,
      "loss": 0.5576,
      "step": 5595
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8532376133816559,
      "learning_rate": 4.0194683955623223e-07,
      "loss": 0.5468,
      "step": 5596
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7520894444895001,
      "learning_rate": 4.0161616020319866e-07,
      "loss": 0.4907,
      "step": 5597
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7219070237325335,
      "learning_rate": 4.012855827486992e-07,
      "loss": 0.5953,
      "step": 5598
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6857148720499368,
      "learning_rate": 4.009551072490276e-07,
      "loss": 0.5449,
      "step": 5599
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6500294845000614,
      "learning_rate": 4.006247337604605e-07,
      "loss": 0.5143,
      "step": 5600
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.623466335763457,
      "learning_rate": 4.002944623392568e-07,
      "loss": 0.5327,
      "step": 5601
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9025375596739125,
      "learning_rate": 3.9996429304165856e-07,
      "loss": 0.6474,
      "step": 5602
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9046501915525249,
      "learning_rate": 3.996342259238902e-07,
      "loss": 0.652,
      "step": 5603
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7039723057634765,
      "learning_rate": 3.993042610421585e-07,
      "loss": 0.5535,
      "step": 5604
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6334470707881604,
      "learning_rate": 3.989743984526528e-07,
      "loss": 0.508,
      "step": 5605
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.82091195721671,
      "learning_rate": 3.98644638211546e-07,
      "loss": 0.6107,
      "step": 5606
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8624439606559575,
      "learning_rate": 3.9831498037499243e-07,
      "loss": 0.6108,
      "step": 5607
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9094038244399927,
      "learning_rate": 3.9798542499912936e-07,
      "loss": 0.6297,
      "step": 5608
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0521058320113514,
      "learning_rate": 3.976559721400766e-07,
      "loss": 0.6478,
      "step": 5609
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8231717023383446,
      "learning_rate": 3.9732662185393674e-07,
      "loss": 0.6199,
      "step": 5610
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7926876986400526,
      "learning_rate": 3.969973741967946e-07,
      "loss": 0.5665,
      "step": 5611
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.835191320991798,
      "learning_rate": 3.966682292247174e-07,
      "loss": 0.6034,
      "step": 5612
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8296815347919673,
      "learning_rate": 3.9633918699375544e-07,
      "loss": 0.6052,
      "step": 5613
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7360078426614185,
      "learning_rate": 3.9601024755994104e-07,
      "loss": 0.5072,
      "step": 5614
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8824243689174313,
      "learning_rate": 3.956814109792892e-07,
      "loss": 0.6522,
      "step": 5615
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6793761013810344,
      "learning_rate": 3.953526773077972e-07,
      "loss": 0.5409,
      "step": 5616
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.642444997815709,
      "learning_rate": 3.950240466014452e-07,
      "loss": 0.4993,
      "step": 5617
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8816442534590868,
      "learning_rate": 3.9469551891619536e-07,
      "loss": 0.5663,
      "step": 5618
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6929419192053166,
      "learning_rate": 3.9436709430799277e-07,
      "loss": 0.5232,
      "step": 5619
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9947276098428101,
      "learning_rate": 3.9403877283276446e-07,
      "loss": 0.633,
      "step": 5620
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7480748335773182,
      "learning_rate": 3.937105545464203e-07,
      "loss": 0.5277,
      "step": 5621
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6698008638046612,
      "learning_rate": 3.933824395048525e-07,
      "loss": 0.5263,
      "step": 5622
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6977147704052601,
      "learning_rate": 3.930544277639355e-07,
      "loss": 0.5666,
      "step": 5623
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9917816910174254,
      "learning_rate": 3.9272651937952626e-07,
      "loss": 0.7003,
      "step": 5624
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8498837195360109,
      "learning_rate": 3.923987144074644e-07,
      "loss": 0.6543,
      "step": 5625
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6713859259844839,
      "learning_rate": 3.9207101290357145e-07,
      "loss": 0.5373,
      "step": 5626
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7737901209783398,
      "learning_rate": 3.917434149236516e-07,
      "loss": 0.5723,
      "step": 5627
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7753213536762295,
      "learning_rate": 3.9141592052349147e-07,
      "loss": 0.5504,
      "step": 5628
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6852210620905105,
      "learning_rate": 3.910885297588599e-07,
      "loss": 0.5383,
      "step": 5629
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6460473619354119,
      "learning_rate": 3.9076124268550815e-07,
      "loss": 0.5231,
      "step": 5630
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7334996676346494,
      "learning_rate": 3.9043405935916975e-07,
      "loss": 0.5924,
      "step": 5631
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1855916115819223,
      "learning_rate": 3.901069798355602e-07,
      "loss": 0.6142,
      "step": 5632
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9231902141472927,
      "learning_rate": 3.897800041703785e-07,
      "loss": 0.6079,
      "step": 5633
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6605595845266886,
      "learning_rate": 3.89453132419305e-07,
      "loss": 0.4868,
      "step": 5634
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7437356844070493,
      "learning_rate": 3.891263646380023e-07,
      "loss": 0.5119,
      "step": 5635
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9299997125364499,
      "learning_rate": 3.887997008821158e-07,
      "loss": 0.6432,
      "step": 5636
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7290486746367228,
      "learning_rate": 3.884731412072726e-07,
      "loss": 0.512,
      "step": 5637
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6625466735837077,
      "learning_rate": 3.881466856690828e-07,
      "loss": 0.5427,
      "step": 5638
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7947228523207731,
      "learning_rate": 3.878203343231381e-07,
      "loss": 0.5534,
      "step": 5639
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7598809309543411,
      "learning_rate": 3.8749408722501286e-07,
      "loss": 0.584,
      "step": 5640
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6799863980299155,
      "learning_rate": 3.8716794443026346e-07,
      "loss": 0.5211,
      "step": 5641
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7177061591114916,
      "learning_rate": 3.8684190599442877e-07,
      "loss": 0.5389,
      "step": 5642
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6774474055425095,
      "learning_rate": 3.8651597197302957e-07,
      "loss": 0.5144,
      "step": 5643
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7936497786347725,
      "learning_rate": 3.861901424215691e-07,
      "loss": 0.6249,
      "step": 5644
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7193171634884387,
      "learning_rate": 3.8586441739553286e-07,
      "loss": 0.5401,
      "step": 5645
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6407225657827852,
      "learning_rate": 3.8553879695038825e-07,
      "loss": 0.5302,
      "step": 5646
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7594564268589244,
      "learning_rate": 3.8521328114158514e-07,
      "loss": 0.5709,
      "step": 5647
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7235973825228644,
      "learning_rate": 3.848878700245555e-07,
      "loss": 0.5508,
      "step": 5648
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7082406274337057,
      "learning_rate": 3.845625636547133e-07,
      "loss": 0.5594,
      "step": 5649
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6849050190232984,
      "learning_rate": 3.8423736208745495e-07,
      "loss": 0.5544,
      "step": 5650
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7083126181878967,
      "learning_rate": 3.8391226537815893e-07,
      "loss": 0.5608,
      "step": 5651
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9165264121721303,
      "learning_rate": 3.8358727358218567e-07,
      "loss": 0.6017,
      "step": 5652
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6792139031008324,
      "learning_rate": 3.8326238675487796e-07,
      "loss": 0.5051,
      "step": 5653
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8783079974285088,
      "learning_rate": 3.829376049515608e-07,
      "loss": 0.6141,
      "step": 5654
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6910445849666734,
      "learning_rate": 3.826129282275409e-07,
      "loss": 0.5792,
      "step": 5655
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6264432886792263,
      "learning_rate": 3.8228835663810744e-07,
      "loss": 0.5001,
      "step": 5656
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6886425668490055,
      "learning_rate": 3.819638902385316e-07,
      "loss": 0.5326,
      "step": 5657
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7083541693510764,
      "learning_rate": 3.816395290840663e-07,
      "loss": 0.5718,
      "step": 5658
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8706110717209241,
      "learning_rate": 3.813152732299474e-07,
      "loss": 0.5923,
      "step": 5659
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6481811781065423,
      "learning_rate": 3.809911227313921e-07,
      "loss": 0.5564,
      "step": 5660
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8860276958758795,
      "learning_rate": 3.8066707764359984e-07,
      "loss": 0.6638,
      "step": 5661
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7065248016366197,
      "learning_rate": 3.803431380217521e-07,
      "loss": 0.5159,
      "step": 5662
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8184495735106552,
      "learning_rate": 3.800193039210123e-07,
      "loss": 0.5614,
      "step": 5663
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7456130862153767,
      "learning_rate": 3.796955753965263e-07,
      "loss": 0.5629,
      "step": 5664
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6494410600876533,
      "learning_rate": 3.7937195250342146e-07,
      "loss": 0.4446,
      "step": 5665
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8292052811304128,
      "learning_rate": 3.790484352968074e-07,
      "loss": 0.5864,
      "step": 5666
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.943749969158664,
      "learning_rate": 3.787250238317755e-07,
      "loss": 0.6446,
      "step": 5667
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1305559908320273,
      "learning_rate": 3.7840171816339995e-07,
      "loss": 0.6887,
      "step": 5668
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9048990157465916,
      "learning_rate": 3.780785183467361e-07,
      "loss": 0.5743,
      "step": 5669
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0013800266330322,
      "learning_rate": 3.777554244368214e-07,
      "loss": 0.6507,
      "step": 5670
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6408286246853133,
      "learning_rate": 3.774324364886754e-07,
      "loss": 0.545,
      "step": 5671
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6987290941933705,
      "learning_rate": 3.7710955455729965e-07,
      "loss": 0.5518,
      "step": 5672
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6554024374966796,
      "learning_rate": 3.767867786976775e-07,
      "loss": 0.5338,
      "step": 5673
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6239037578107334,
      "learning_rate": 3.7646410896477434e-07,
      "loss": 0.5475,
      "step": 5674
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7884867587550346,
      "learning_rate": 3.761415454135375e-07,
      "loss": 0.5515,
      "step": 5675
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6726928476493904,
      "learning_rate": 3.758190880988962e-07,
      "loss": 0.5805,
      "step": 5676
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7442757405552903,
      "learning_rate": 3.7549673707576134e-07,
      "loss": 0.5993,
      "step": 5677
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9335524228198817,
      "learning_rate": 3.751744923990263e-07,
      "loss": 0.6677,
      "step": 5678
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8306350597170582,
      "learning_rate": 3.7485235412356587e-07,
      "loss": 0.5326,
      "step": 5679
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6374638959772995,
      "learning_rate": 3.7453032230423667e-07,
      "loss": 0.597,
      "step": 5680
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6991136645641728,
      "learning_rate": 3.7420839699587746e-07,
      "loss": 0.5833,
      "step": 5681
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6509100616241281,
      "learning_rate": 3.7388657825330896e-07,
      "loss": 0.538,
      "step": 5682
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.741819997366788,
      "learning_rate": 3.735648661313332e-07,
      "loss": 0.5591,
      "step": 5683
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7237088359086035,
      "learning_rate": 3.7324326068473465e-07,
      "loss": 0.6578,
      "step": 5684
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6856475418533206,
      "learning_rate": 3.729217619682794e-07,
      "loss": 0.5173,
      "step": 5685
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9904712018620103,
      "learning_rate": 3.7260037003671507e-07,
      "loss": 0.5813,
      "step": 5686
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9435615051526234,
      "learning_rate": 3.7227908494477166e-07,
      "loss": 0.6004,
      "step": 5687
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9248242407688326,
      "learning_rate": 3.7195790674716043e-07,
      "loss": 0.6485,
      "step": 5688
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3706062483485777,
      "learning_rate": 3.7163683549857483e-07,
      "loss": 0.6197,
      "step": 5689
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8803934424241523,
      "learning_rate": 3.7131587125368957e-07,
      "loss": 0.6409,
      "step": 5690
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8302924323683479,
      "learning_rate": 3.7099501406716213e-07,
      "loss": 0.6899,
      "step": 5691
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6412812985269206,
      "learning_rate": 3.70674263993631e-07,
      "loss": 0.5179,
      "step": 5692
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7041495309602617,
      "learning_rate": 3.703536210877163e-07,
      "loss": 0.5592,
      "step": 5693
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7343601360002561,
      "learning_rate": 3.7003308540402043e-07,
      "loss": 0.623,
      "step": 5694
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.265452661775937,
      "learning_rate": 3.6971265699712707e-07,
      "loss": 0.6617,
      "step": 5695
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6367484623399495,
      "learning_rate": 3.69392335921602e-07,
      "loss": 0.6298,
      "step": 5696
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6541144746138948,
      "learning_rate": 3.6907212223199245e-07,
      "loss": 0.5352,
      "step": 5697
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.887057917053404,
      "learning_rate": 3.687520159828277e-07,
      "loss": 0.6013,
      "step": 5698
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7097280276076654,
      "learning_rate": 3.6843201722861815e-07,
      "loss": 0.5318,
      "step": 5699
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7642529179001789,
      "learning_rate": 3.6811212602385643e-07,
      "loss": 0.6056,
      "step": 5700
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7825105960744196,
      "learning_rate": 3.6779234242301684e-07,
      "loss": 0.4952,
      "step": 5701
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6646143998436757,
      "learning_rate": 3.6747266648055486e-07,
      "loss": 0.5654,
      "step": 5702
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7662180896345064,
      "learning_rate": 3.671530982509082e-07,
      "loss": 0.514,
      "step": 5703
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8918002016201372,
      "learning_rate": 3.6683363778849606e-07,
      "loss": 0.6056,
      "step": 5704
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6431338243137317,
      "learning_rate": 3.665142851477191e-07,
      "loss": 0.4868,
      "step": 5705
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7079225303547374,
      "learning_rate": 3.661950403829597e-07,
      "loss": 0.5162,
      "step": 5706
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.693117849949462,
      "learning_rate": 3.65875903548582e-07,
      "loss": 0.5862,
      "step": 5707
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9528997946336036,
      "learning_rate": 3.655568746989317e-07,
      "loss": 0.563,
      "step": 5708
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8025995367302733,
      "learning_rate": 3.652379538883361e-07,
      "loss": 0.5413,
      "step": 5709
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9338997104986322,
      "learning_rate": 3.64919141171104e-07,
      "loss": 0.6396,
      "step": 5710
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9213992574521623,
      "learning_rate": 3.6460043660152597e-07,
      "loss": 0.6392,
      "step": 5711
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7497322852113424,
      "learning_rate": 3.64281840233874e-07,
      "loss": 0.5543,
      "step": 5712
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9883316551361626,
      "learning_rate": 3.639633521224018e-07,
      "loss": 0.6072,
      "step": 5713
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6022824292986073,
      "learning_rate": 3.636449723213445e-07,
      "loss": 0.5175,
      "step": 5714
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6988706816024985,
      "learning_rate": 3.63326700884919e-07,
      "loss": 0.5022,
      "step": 5715
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6423306824324638,
      "learning_rate": 3.630085378673231e-07,
      "loss": 0.552,
      "step": 5716
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7070433458716213,
      "learning_rate": 3.626904833227374e-07,
      "loss": 0.524,
      "step": 5717
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9211856589031331,
      "learning_rate": 3.62372537305323e-07,
      "loss": 0.5994,
      "step": 5718
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8412038039125522,
      "learning_rate": 3.6205469986922276e-07,
      "loss": 0.6806,
      "step": 5719
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9619305359062121,
      "learning_rate": 3.6173697106856105e-07,
      "loss": 0.5893,
      "step": 5720
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6670960280214673,
      "learning_rate": 3.614193509574438e-07,
      "loss": 0.4929,
      "step": 5721
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8134727560394401,
      "learning_rate": 3.611018395899584e-07,
      "loss": 0.5222,
      "step": 5722
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9100486479867437,
      "learning_rate": 3.607844370201737e-07,
      "loss": 0.6301,
      "step": 5723
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9654131959319229,
      "learning_rate": 3.604671433021401e-07,
      "loss": 0.6433,
      "step": 5724
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8745255679086882,
      "learning_rate": 3.601499584898894e-07,
      "loss": 0.6353,
      "step": 5725
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7874900499305247,
      "learning_rate": 3.5983288263743484e-07,
      "loss": 0.6016,
      "step": 5726
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.92974620714117,
      "learning_rate": 3.5951591579877116e-07,
      "loss": 0.6301,
      "step": 5727
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.095512968931211,
      "learning_rate": 3.5919905802787444e-07,
      "loss": 0.6404,
      "step": 5728
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6936820973871652,
      "learning_rate": 3.5888230937870235e-07,
      "loss": 0.4504,
      "step": 5729
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.885761147455906,
      "learning_rate": 3.5856566990519386e-07,
      "loss": 0.6426,
      "step": 5730
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7578193556243144,
      "learning_rate": 3.582491396612695e-07,
      "loss": 0.5432,
      "step": 5731
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7692919709012693,
      "learning_rate": 3.5793271870083076e-07,
      "loss": 0.5671,
      "step": 5732
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.705146091921639,
      "learning_rate": 3.576164070777611e-07,
      "loss": 0.5412,
      "step": 5733
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8251223016221925,
      "learning_rate": 3.5730020484592503e-07,
      "loss": 0.5869,
      "step": 5734
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8413417025311235,
      "learning_rate": 3.569841120591684e-07,
      "loss": 0.6398,
      "step": 5735
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7767730569925216,
      "learning_rate": 3.566681287713187e-07,
      "loss": 0.6751,
      "step": 5736
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9218267478738325,
      "learning_rate": 3.5635225503618436e-07,
      "loss": 0.5885,
      "step": 5737
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9371008718996421,
      "learning_rate": 3.560364909075556e-07,
      "loss": 0.6263,
      "step": 5738
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8505065826853668,
      "learning_rate": 3.557208364392037e-07,
      "loss": 0.6352,
      "step": 5739
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9971196099513305,
      "learning_rate": 3.554052916848813e-07,
      "loss": 0.6141,
      "step": 5740
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8669741693440715,
      "learning_rate": 3.550898566983224e-07,
      "loss": 0.5501,
      "step": 5741
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.734184248733248,
      "learning_rate": 3.547745315332422e-07,
      "loss": 0.4984,
      "step": 5742
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9132682328556705,
      "learning_rate": 3.54459316243337e-07,
      "loss": 0.637,
      "step": 5743
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7293704124754273,
      "learning_rate": 3.541442108822855e-07,
      "loss": 0.5558,
      "step": 5744
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6583476480373075,
      "learning_rate": 3.538292155037463e-07,
      "loss": 0.5578,
      "step": 5745
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8273508700945823,
      "learning_rate": 3.5351433016136e-07,
      "loss": 0.6368,
      "step": 5746
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0801050890692236,
      "learning_rate": 3.5319955490874833e-07,
      "loss": 0.6997,
      "step": 5747
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6923928977206866,
      "learning_rate": 3.5288488979951405e-07,
      "loss": 0.5793,
      "step": 5748
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6475577225395311,
      "learning_rate": 3.525703348872414e-07,
      "loss": 0.5399,
      "step": 5749
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.717391738841494,
      "learning_rate": 3.5225589022549587e-07,
      "loss": 0.5385,
      "step": 5750
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.619170186234015,
      "learning_rate": 3.5194155586782405e-07,
      "loss": 0.4734,
      "step": 5751
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6865834786064441,
      "learning_rate": 3.5162733186775395e-07,
      "loss": 0.5155,
      "step": 5752
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8425316295286494,
      "learning_rate": 3.5131321827879445e-07,
      "loss": 0.6992,
      "step": 5753
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7089700642503015,
      "learning_rate": 3.50999215154436e-07,
      "loss": 0.5359,
      "step": 5754
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7548519832301605,
      "learning_rate": 3.5068532254814985e-07,
      "loss": 0.5743,
      "step": 5755
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8162225543428423,
      "learning_rate": 3.503715405133888e-07,
      "loss": 0.5814,
      "step": 5756
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6999851248255272,
      "learning_rate": 3.5005786910358657e-07,
      "loss": 0.6528,
      "step": 5757
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0286441846532688,
      "learning_rate": 3.497443083721583e-07,
      "loss": 0.6493,
      "step": 5758
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6056998477752776,
      "learning_rate": 3.4943085837249996e-07,
      "loss": 0.488,
      "step": 5759
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7336988272175029,
      "learning_rate": 3.4911751915798884e-07,
      "loss": 0.5099,
      "step": 5760
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7082676119891784,
      "learning_rate": 3.4880429078198336e-07,
      "loss": 0.541,
      "step": 5761
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6961277029557731,
      "learning_rate": 3.4849117329782307e-07,
      "loss": 0.6181,
      "step": 5762
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6360665241105075,
      "learning_rate": 3.481781667588286e-07,
      "loss": 0.5429,
      "step": 5763
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6546843052751988,
      "learning_rate": 3.478652712183017e-07,
      "loss": 0.5436,
      "step": 5764
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7534481495545311,
      "learning_rate": 3.475524867295251e-07,
      "loss": 0.532,
      "step": 5765
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9141756088033309,
      "learning_rate": 3.47239813345763e-07,
      "loss": 0.5733,
      "step": 5766
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6552895499153839,
      "learning_rate": 3.4692725112026023e-07,
      "loss": 0.5349,
      "step": 5767
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8989269405108575,
      "learning_rate": 3.46614800106243e-07,
      "loss": 0.6189,
      "step": 5768
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8157002812079178,
      "learning_rate": 3.4630246035691803e-07,
      "loss": 0.6272,
      "step": 5769
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.765591443916113,
      "learning_rate": 3.4599023192547416e-07,
      "loss": 0.5348,
      "step": 5770
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6610493503221078,
      "learning_rate": 3.4567811486508036e-07,
      "loss": 0.5578,
      "step": 5771
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6154499474453796,
      "learning_rate": 3.45366109228887e-07,
      "loss": 0.5131,
      "step": 5772
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.736466065843992,
      "learning_rate": 3.450542150700253e-07,
      "loss": 0.5905,
      "step": 5773
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8641074444330841,
      "learning_rate": 3.447424324416076e-07,
      "loss": 0.6174,
      "step": 5774
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9036901292081694,
      "learning_rate": 3.4443076139672686e-07,
      "loss": 0.5486,
      "step": 5775
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.777909857157713,
      "learning_rate": 3.441192019884581e-07,
      "loss": 0.5217,
      "step": 5776
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6953340673619705,
      "learning_rate": 3.438077542698564e-07,
      "loss": 0.522,
      "step": 5777
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7925712093410265,
      "learning_rate": 3.434964182939579e-07,
      "loss": 0.5432,
      "step": 5778
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8698399505396617,
      "learning_rate": 3.4318519411378e-07,
      "loss": 0.6053,
      "step": 5779
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6562668687938756,
      "learning_rate": 3.4287408178232093e-07,
      "loss": 0.5674,
      "step": 5780
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6393016710646782,
      "learning_rate": 3.4256308135255986e-07,
      "loss": 0.5374,
      "step": 5781
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8482827901332342,
      "learning_rate": 3.422521928774569e-07,
      "loss": 0.6118,
      "step": 5782
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8137188698815405,
      "learning_rate": 3.41941416409953e-07,
      "loss": 0.5832,
      "step": 5783
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7175818979624727,
      "learning_rate": 3.4163075200297044e-07,
      "loss": 0.5306,
      "step": 5784
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6499453413626055,
      "learning_rate": 3.413201997094118e-07,
      "loss": 0.5078,
      "step": 5785
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8887045495644119,
      "learning_rate": 3.410097595821613e-07,
      "loss": 0.6071,
      "step": 5786
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8068510664159982,
      "learning_rate": 3.406994316740832e-07,
      "loss": 0.6095,
      "step": 5787
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5978211233773183,
      "learning_rate": 3.4038921603802353e-07,
      "loss": 0.5437,
      "step": 5788
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6397747946298286,
      "learning_rate": 3.4007911272680845e-07,
      "loss": 0.5287,
      "step": 5789
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.743824166198371,
      "learning_rate": 3.397691217932456e-07,
      "loss": 0.5567,
      "step": 5790
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8610524241864733,
      "learning_rate": 3.39459243290123e-07,
      "loss": 0.5567,
      "step": 5791
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8834732216972989,
      "learning_rate": 3.3914947727020983e-07,
      "loss": 0.6245,
      "step": 5792
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7323000852870963,
      "learning_rate": 3.3883982378625587e-07,
      "loss": 0.5299,
      "step": 5793
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1178825406788375,
      "learning_rate": 3.385302828909922e-07,
      "loss": 0.6723,
      "step": 5794
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6663765502003052,
      "learning_rate": 3.3822085463713e-07,
      "loss": 0.5355,
      "step": 5795
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7275387942290371,
      "learning_rate": 3.37911539077362e-07,
      "loss": 0.5456,
      "step": 5796
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.793264902488691,
      "learning_rate": 3.3760233626436143e-07,
      "loss": 0.6138,
      "step": 5797
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6357927591353114,
      "learning_rate": 3.3729324625078204e-07,
      "loss": 0.5409,
      "step": 5798
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6616743283827022,
      "learning_rate": 3.369842690892588e-07,
      "loss": 0.5123,
      "step": 5799
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6557755226080524,
      "learning_rate": 3.366754048324073e-07,
      "loss": 0.5258,
      "step": 5800
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7016171342251507,
      "learning_rate": 3.3636665353282356e-07,
      "loss": 0.5419,
      "step": 5801
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8264927178926433,
      "learning_rate": 3.360580152430853e-07,
      "loss": 0.5457,
      "step": 5802
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6757933884449984,
      "learning_rate": 3.3574949001575024e-07,
      "loss": 0.5131,
      "step": 5803
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6978315581337292,
      "learning_rate": 3.3544107790335686e-07,
      "loss": 0.5245,
      "step": 5804
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.168345711241714,
      "learning_rate": 3.3513277895842464e-07,
      "loss": 0.5766,
      "step": 5805
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7072733393663226,
      "learning_rate": 3.3482459323345357e-07,
      "loss": 0.626,
      "step": 5806
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.87929478218828,
      "learning_rate": 3.345165207809246e-07,
      "loss": 0.6377,
      "step": 5807
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7302395341845647,
      "learning_rate": 3.3420856165329914e-07,
      "loss": 0.5043,
      "step": 5808
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7562458096444351,
      "learning_rate": 3.339007159030195e-07,
      "loss": 0.6078,
      "step": 5809
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9176892728148279,
      "learning_rate": 3.3359298358250856e-07,
      "loss": 0.6179,
      "step": 5810
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0168648141318495,
      "learning_rate": 3.332853647441699e-07,
      "loss": 0.6196,
      "step": 5811
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8792832325335851,
      "learning_rate": 3.329778594403879e-07,
      "loss": 0.61,
      "step": 5812
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.620818500133194,
      "learning_rate": 3.3267046772352736e-07,
      "loss": 0.5233,
      "step": 5813
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7956865772804173,
      "learning_rate": 3.32363189645934e-07,
      "loss": 0.5224,
      "step": 5814
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9712212396831633,
      "learning_rate": 3.3205602525993414e-07,
      "loss": 0.575,
      "step": 5815
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.687942536881875,
      "learning_rate": 3.317489746178346e-07,
      "loss": 0.5423,
      "step": 5816
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.790649606600395,
      "learning_rate": 3.314420377719228e-07,
      "loss": 0.5504,
      "step": 5817
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8911346697113687,
      "learning_rate": 3.3113521477446694e-07,
      "loss": 0.633,
      "step": 5818
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8655680726192567,
      "learning_rate": 3.308285056777159e-07,
      "loss": 0.6088,
      "step": 5819
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6209669470859536,
      "learning_rate": 3.30521910533899e-07,
      "loss": 0.5294,
      "step": 5820
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.884016614082206,
      "learning_rate": 3.3021542939522604e-07,
      "loss": 0.5896,
      "step": 5821
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8238501434643635,
      "learning_rate": 3.299090623138878e-07,
      "loss": 0.6253,
      "step": 5822
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5901895001899615,
      "learning_rate": 3.296028093420553e-07,
      "loss": 0.5076,
      "step": 5823
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7141574395921789,
      "learning_rate": 3.2929667053188016e-07,
      "loss": 0.5554,
      "step": 5824
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8327205427871258,
      "learning_rate": 3.2899064593549473e-07,
      "loss": 0.5562,
      "step": 5825
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7349262171116876,
      "learning_rate": 3.2868473560501173e-07,
      "loss": 0.5904,
      "step": 5826
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8749755200595951,
      "learning_rate": 3.2837893959252427e-07,
      "loss": 0.5949,
      "step": 5827
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8812178432126165,
      "learning_rate": 3.2807325795010675e-07,
      "loss": 0.5829,
      "step": 5828
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6641614595818213,
      "learning_rate": 3.2776769072981334e-07,
      "loss": 0.5149,
      "step": 5829
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7229082763539239,
      "learning_rate": 3.27462237983679e-07,
      "loss": 0.5831,
      "step": 5830
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8685170072285386,
      "learning_rate": 3.2715689976371905e-07,
      "loss": 0.5929,
      "step": 5831
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8442555346061793,
      "learning_rate": 3.268516761219294e-07,
      "loss": 0.6283,
      "step": 5832
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9047127824991924,
      "learning_rate": 3.265465671102865e-07,
      "loss": 0.5944,
      "step": 5833
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7487992377221222,
      "learning_rate": 3.2624157278074725e-07,
      "loss": 0.6051,
      "step": 5834
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7440641737492663,
      "learning_rate": 3.2593669318524906e-07,
      "loss": 0.4919,
      "step": 5835
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.699072335321257,
      "learning_rate": 3.256319283757096e-07,
      "loss": 0.6298,
      "step": 5836
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7779862057038283,
      "learning_rate": 3.253272784040273e-07,
      "loss": 0.5882,
      "step": 5837
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7093812299805684,
      "learning_rate": 3.2502274332208066e-07,
      "loss": 0.54,
      "step": 5838
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7921287549656686,
      "learning_rate": 3.247183231817291e-07,
      "loss": 0.5281,
      "step": 5839
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7431927855377036,
      "learning_rate": 3.2441401803481204e-07,
      "loss": 0.5662,
      "step": 5840
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7988834614200662,
      "learning_rate": 3.2410982793314943e-07,
      "loss": 0.5267,
      "step": 5841
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7196114429735521,
      "learning_rate": 3.2380575292854185e-07,
      "loss": 0.5484,
      "step": 5842
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9493327805246935,
      "learning_rate": 3.2350179307277003e-07,
      "loss": 0.6253,
      "step": 5843
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.923116914420983,
      "learning_rate": 3.231979484175951e-07,
      "loss": 0.6609,
      "step": 5844
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7335184992438261,
      "learning_rate": 3.2289421901475866e-07,
      "loss": 0.5611,
      "step": 5845
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7473007903135083,
      "learning_rate": 3.2259060491598265e-07,
      "loss": 0.6196,
      "step": 5846
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6626600084839624,
      "learning_rate": 3.222871061729694e-07,
      "loss": 0.5469,
      "step": 5847
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9307967898959475,
      "learning_rate": 3.2198372283740174e-07,
      "loss": 0.6419,
      "step": 5848
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1239732910627858,
      "learning_rate": 3.216804549609424e-07,
      "loss": 0.6662,
      "step": 5849
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9860592383288433,
      "learning_rate": 3.21377302595235e-07,
      "loss": 0.6305,
      "step": 5850
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7037699287905361,
      "learning_rate": 3.2107426579190313e-07,
      "loss": 0.5379,
      "step": 5851
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8041110153243058,
      "learning_rate": 3.2077134460255085e-07,
      "loss": 0.6147,
      "step": 5852
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9277156991105695,
      "learning_rate": 3.2046853907876226e-07,
      "loss": 0.6315,
      "step": 5853
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6645947637164459,
      "learning_rate": 3.20165849272102e-07,
      "loss": 0.526,
      "step": 5854
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.630829437090905,
      "learning_rate": 3.198632752341154e-07,
      "loss": 0.5276,
      "step": 5855
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9585301332652266,
      "learning_rate": 3.1956081701632744e-07,
      "loss": 0.6518,
      "step": 5856
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8075130004812773,
      "learning_rate": 3.192584746702435e-07,
      "loss": 0.5808,
      "step": 5857
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.896658701161667,
      "learning_rate": 3.1895624824734944e-07,
      "loss": 0.6567,
      "step": 5858
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6993892189305252,
      "learning_rate": 3.186541377991113e-07,
      "loss": 0.572,
      "step": 5859
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9111677077086835,
      "learning_rate": 3.183521433769752e-07,
      "loss": 0.6346,
      "step": 5860
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8241060621320211,
      "learning_rate": 3.1805026503236786e-07,
      "loss": 0.5864,
      "step": 5861
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8453980519869486,
      "learning_rate": 3.1774850281669574e-07,
      "loss": 0.5797,
      "step": 5862
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6837030792523212,
      "learning_rate": 3.174468567813461e-07,
      "loss": 0.539,
      "step": 5863
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7145268606034808,
      "learning_rate": 3.171453269776858e-07,
      "loss": 0.523,
      "step": 5864
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.873029284981752,
      "learning_rate": 3.168439134570625e-07,
      "loss": 0.602,
      "step": 5865
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9084347916469823,
      "learning_rate": 3.165426162708036e-07,
      "loss": 0.6118,
      "step": 5866
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7668492368718475,
      "learning_rate": 3.16241435470217e-07,
      "loss": 0.5235,
      "step": 5867
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9148247845735581,
      "learning_rate": 3.1594037110659055e-07,
      "loss": 0.6085,
      "step": 5868
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7505909157839912,
      "learning_rate": 3.156394232311924e-07,
      "loss": 0.5847,
      "step": 5869
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.972669951420328,
      "learning_rate": 3.1533859189527056e-07,
      "loss": 0.6468,
      "step": 5870
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7073165524131723,
      "learning_rate": 3.1503787715005413e-07,
      "loss": 0.5613,
      "step": 5871
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7364884884168592,
      "learning_rate": 3.147372790467513e-07,
      "loss": 0.5884,
      "step": 5872
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8198146454217182,
      "learning_rate": 3.1443679763655085e-07,
      "loss": 0.6002,
      "step": 5873
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.640736354809276,
      "learning_rate": 3.141364329706215e-07,
      "loss": 0.5498,
      "step": 5874
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8282411289472325,
      "learning_rate": 3.1383618510011256e-07,
      "loss": 0.6634,
      "step": 5875
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6898672165851946,
      "learning_rate": 3.135360540761527e-07,
      "loss": 0.543,
      "step": 5876
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7772202242358836,
      "learning_rate": 3.1323603994985136e-07,
      "loss": 0.484,
      "step": 5877
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7160851177015373,
      "learning_rate": 3.1293614277229783e-07,
      "loss": 0.5489,
      "step": 5878
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.470634077254529,
      "learning_rate": 3.126363625945614e-07,
      "loss": 0.6202,
      "step": 5879
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6962478631085542,
      "learning_rate": 3.1233669946769146e-07,
      "loss": 0.5654,
      "step": 5880
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2372474509010094,
      "learning_rate": 3.1203715344271763e-07,
      "loss": 0.5885,
      "step": 5881
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6905374522776376,
      "learning_rate": 3.117377245706494e-07,
      "loss": 0.5601,
      "step": 5882
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6446096421491658,
      "learning_rate": 3.114384129024763e-07,
      "loss": 0.5369,
      "step": 5883
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8597559308851518,
      "learning_rate": 3.1113921848916824e-07,
      "loss": 0.5895,
      "step": 5884
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8772365787598089,
      "learning_rate": 3.108401413816747e-07,
      "loss": 0.6113,
      "step": 5885
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9534231117900916,
      "learning_rate": 3.105411816309252e-07,
      "loss": 0.6511,
      "step": 5886
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6426617550519783,
      "learning_rate": 3.1024233928783007e-07,
      "loss": 0.5189,
      "step": 5887
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7991458684372903,
      "learning_rate": 3.0994361440327865e-07,
      "loss": 0.6336,
      "step": 5888
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7863653407208486,
      "learning_rate": 3.096450070281408e-07,
      "loss": 0.559,
      "step": 5889
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8139991693851578,
      "learning_rate": 3.093465172132661e-07,
      "loss": 0.6072,
      "step": 5890
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7531112597954773,
      "learning_rate": 3.090481450094844e-07,
      "loss": 0.5563,
      "step": 5891
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7928777938975712,
      "learning_rate": 3.0874989046760537e-07,
      "loss": 0.6599,
      "step": 5892
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8305517739709349,
      "learning_rate": 3.0845175363841857e-07,
      "loss": 0.6036,
      "step": 5893
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8707600837671162,
      "learning_rate": 3.0815373457269354e-07,
      "loss": 0.5195,
      "step": 5894
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6826618122513075,
      "learning_rate": 3.0785583332118005e-07,
      "loss": 0.566,
      "step": 5895
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9580070604257871,
      "learning_rate": 3.075580499346073e-07,
      "loss": 0.6318,
      "step": 5896
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9024557428755139,
      "learning_rate": 3.07260384463685e-07,
      "loss": 0.643,
      "step": 5897
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9010132672597385,
      "learning_rate": 3.069628369591022e-07,
      "loss": 0.6662,
      "step": 5898
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8653428609624886,
      "learning_rate": 3.0666540747152825e-07,
      "loss": 0.6094,
      "step": 5899
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7567622702802866,
      "learning_rate": 3.063680960516123e-07,
      "loss": 0.5387,
      "step": 5900
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7058286972326075,
      "learning_rate": 3.060709027499835e-07,
      "loss": 0.5688,
      "step": 5901
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7099711315485172,
      "learning_rate": 3.0577382761725057e-07,
      "loss": 0.5554,
      "step": 5902
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3374451369285574,
      "learning_rate": 3.054768707040024e-07,
      "loss": 0.6636,
      "step": 5903
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7678088930422041,
      "learning_rate": 3.0518003206080764e-07,
      "loss": 0.5435,
      "step": 5904
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8057889743666774,
      "learning_rate": 3.048833117382149e-07,
      "loss": 0.6084,
      "step": 5905
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.789402270116011,
      "learning_rate": 3.045867097867525e-07,
      "loss": 0.6166,
      "step": 5906
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6010092065464289,
      "learning_rate": 3.0429022625692855e-07,
      "loss": 0.521,
      "step": 5907
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7685883267794263,
      "learning_rate": 3.039938611992313e-07,
      "loss": 0.5646,
      "step": 5908
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7271137045588643,
      "learning_rate": 3.036976146641286e-07,
      "loss": 0.5217,
      "step": 5909
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8723033513420435,
      "learning_rate": 3.03401486702068e-07,
      "loss": 0.6198,
      "step": 5910
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6420631709705109,
      "learning_rate": 3.0310547736347716e-07,
      "loss": 0.5385,
      "step": 5911
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7285856950327538,
      "learning_rate": 3.02809586698763e-07,
      "loss": 0.508,
      "step": 5912
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9364729073239833,
      "learning_rate": 3.025138147583133e-07,
      "loss": 0.6238,
      "step": 5913
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.672512677048132,
      "learning_rate": 3.022181615924945e-07,
      "loss": 0.5078,
      "step": 5914
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7286426086039016,
      "learning_rate": 3.0192262725165343e-07,
      "loss": 0.5476,
      "step": 5915
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6051222656158866,
      "learning_rate": 3.016272117861164e-07,
      "loss": 0.5213,
      "step": 5916
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7458263641943669,
      "learning_rate": 3.013319152461895e-07,
      "loss": 0.5218,
      "step": 5917
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5798087896752275,
      "learning_rate": 3.0103673768215885e-07,
      "loss": 0.5144,
      "step": 5918
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8593932169101195,
      "learning_rate": 3.0074167914429003e-07,
      "loss": 0.6413,
      "step": 5919
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1249442734507533,
      "learning_rate": 3.004467396828285e-07,
      "loss": 0.6472,
      "step": 5920
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1032764200996437,
      "learning_rate": 3.001519193479992e-07,
      "loss": 0.6142,
      "step": 5921
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7960672483162087,
      "learning_rate": 2.9985721819000707e-07,
      "loss": 0.5517,
      "step": 5922
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8905330011675561,
      "learning_rate": 2.9956263625903675e-07,
      "loss": 0.6225,
      "step": 5923
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7375109057317656,
      "learning_rate": 2.9926817360525226e-07,
      "loss": 0.5631,
      "step": 5924
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7315892574651534,
      "learning_rate": 2.989738302787977e-07,
      "loss": 0.5022,
      "step": 5925
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8902411162171869,
      "learning_rate": 2.986796063297966e-07,
      "loss": 0.6337,
      "step": 5926
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6648039128691491,
      "learning_rate": 2.983855018083522e-07,
      "loss": 0.5424,
      "step": 5927
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7350085097037302,
      "learning_rate": 2.980915167645476e-07,
      "loss": 0.5902,
      "step": 5928
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8940501390351756,
      "learning_rate": 2.977976512484451e-07,
      "loss": 0.6454,
      "step": 5929
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6993363739562425,
      "learning_rate": 2.9750390531008716e-07,
      "loss": 0.54,
      "step": 5930
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6200667335529357,
      "learning_rate": 2.972102789994956e-07,
      "loss": 0.5284,
      "step": 5931
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7822235653729906,
      "learning_rate": 2.969167723666719e-07,
      "loss": 0.5475,
      "step": 5932
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8330225515108399,
      "learning_rate": 2.966233854615973e-07,
      "loss": 0.6172,
      "step": 5933
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6636067570718531,
      "learning_rate": 2.963301183342323e-07,
      "loss": 0.5242,
      "step": 5934
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9121945330320309,
      "learning_rate": 2.960369710345174e-07,
      "loss": 0.6105,
      "step": 5935
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8167515517254055,
      "learning_rate": 2.9574394361237256e-07,
      "loss": 0.5645,
      "step": 5936
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.812513448596765,
      "learning_rate": 2.9545103611769715e-07,
      "loss": 0.6111,
      "step": 5937
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9495300615855655,
      "learning_rate": 2.9515824860037007e-07,
      "loss": 0.616,
      "step": 5938
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6913689970084194,
      "learning_rate": 2.948655811102506e-07,
      "loss": 0.5371,
      "step": 5939
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8802937277625881,
      "learning_rate": 2.945730336971767e-07,
      "loss": 0.6405,
      "step": 5940
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7373584911815954,
      "learning_rate": 2.9428060641096597e-07,
      "loss": 0.514,
      "step": 5941
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7069808947771954,
      "learning_rate": 2.939882993014158e-07,
      "loss": 0.5611,
      "step": 5942
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8262990620670041,
      "learning_rate": 2.9369611241830317e-07,
      "loss": 0.6475,
      "step": 5943
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9805695536166312,
      "learning_rate": 2.934040458113843e-07,
      "loss": 0.6657,
      "step": 5944
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8416006490534759,
      "learning_rate": 2.931120995303953e-07,
      "loss": 0.6377,
      "step": 5945
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7024144332158757,
      "learning_rate": 2.928202736250514e-07,
      "loss": 0.561,
      "step": 5946
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8380097521976135,
      "learning_rate": 2.9252856814504744e-07,
      "loss": 0.6261,
      "step": 5947
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6676750358608934,
      "learning_rate": 2.9223698314005795e-07,
      "loss": 0.6036,
      "step": 5948
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7503804444841885,
      "learning_rate": 2.9194551865973693e-07,
      "loss": 0.5397,
      "step": 5949
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7185406873018652,
      "learning_rate": 2.9165417475371755e-07,
      "loss": 0.5518,
      "step": 5950
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9515600478127305,
      "learning_rate": 2.913629514716127e-07,
      "loss": 0.6256,
      "step": 5951
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8155721731434047,
      "learning_rate": 2.910718488630147e-07,
      "loss": 0.5559,
      "step": 5952
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8346433566380455,
      "learning_rate": 2.907808669774954e-07,
      "loss": 0.5992,
      "step": 5953
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8681765162946844,
      "learning_rate": 2.904900058646057e-07,
      "loss": 0.6027,
      "step": 5954
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8985961344220704,
      "learning_rate": 2.9019926557387655e-07,
      "loss": 0.6296,
      "step": 5955
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8250808935886176,
      "learning_rate": 2.899086461548177e-07,
      "loss": 0.6135,
      "step": 5956
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8543350021824618,
      "learning_rate": 2.896181476569187e-07,
      "loss": 0.5992,
      "step": 5957
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6290483916150993,
      "learning_rate": 2.8932777012964847e-07,
      "loss": 0.5285,
      "step": 5958
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7897683100273575,
      "learning_rate": 2.890375136224553e-07,
      "loss": 0.5494,
      "step": 5959
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6985844366504481,
      "learning_rate": 2.887473781847668e-07,
      "loss": 0.5162,
      "step": 5960
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8868476607320602,
      "learning_rate": 2.8845736386598996e-07,
      "loss": 0.6152,
      "step": 5961
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.661612283398571,
      "learning_rate": 2.881674707155114e-07,
      "loss": 0.5509,
      "step": 5962
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6592782315845923,
      "learning_rate": 2.8787769878269663e-07,
      "loss": 0.5782,
      "step": 5963
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6621554586034782,
      "learning_rate": 2.87588048116891e-07,
      "loss": 0.5321,
      "step": 5964
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6302506449920465,
      "learning_rate": 2.8729851876741853e-07,
      "loss": 0.5387,
      "step": 5965
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7597814475514004,
      "learning_rate": 2.870091107835838e-07,
      "loss": 0.5877,
      "step": 5966
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8048621459678099,
      "learning_rate": 2.8671982421466955e-07,
      "loss": 0.5961,
      "step": 5967
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8260896335850648,
      "learning_rate": 2.864306591099385e-07,
      "loss": 0.6144,
      "step": 5968
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6679263097156994,
      "learning_rate": 2.8614161551863216e-07,
      "loss": 0.5776,
      "step": 5969
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8429044437378839,
      "learning_rate": 2.858526934899718e-07,
      "loss": 0.5991,
      "step": 5970
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6192099758622432,
      "learning_rate": 2.8556389307315777e-07,
      "loss": 0.5455,
      "step": 5971
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6656384729197151,
      "learning_rate": 2.8527521431736956e-07,
      "loss": 0.5413,
      "step": 5972
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8303735985887707,
      "learning_rate": 2.849866572717666e-07,
      "loss": 0.5686,
      "step": 5973
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6814074621153496,
      "learning_rate": 2.846982219854871e-07,
      "loss": 0.5108,
      "step": 5974
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9248419250513557,
      "learning_rate": 2.8440990850764837e-07,
      "loss": 0.6285,
      "step": 5975
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7171224594582833,
      "learning_rate": 2.841217168873472e-07,
      "loss": 0.5253,
      "step": 5976
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3774248509425957,
      "learning_rate": 2.8383364717365977e-07,
      "loss": 0.5233,
      "step": 5977
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8276614614456784,
      "learning_rate": 2.835456994156412e-07,
      "loss": 0.6004,
      "step": 5978
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7087019241257078,
      "learning_rate": 2.832578736623261e-07,
      "loss": 0.5144,
      "step": 5979
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6276359430334197,
      "learning_rate": 2.8297016996272815e-07,
      "loss": 0.5452,
      "step": 5980
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8217087958089072,
      "learning_rate": 2.8268258836584026e-07,
      "loss": 0.532,
      "step": 5981
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7237063868456727,
      "learning_rate": 2.8239512892063465e-07,
      "loss": 0.553,
      "step": 5982
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7978746153924494,
      "learning_rate": 2.821077916760626e-07,
      "loss": 0.511,
      "step": 5983
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.801993817746113,
      "learning_rate": 2.8182057668105477e-07,
      "loss": 0.5745,
      "step": 5984
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9009017945670311,
      "learning_rate": 2.815334839845207e-07,
      "loss": 0.601,
      "step": 5985
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6382229369215264,
      "learning_rate": 2.8124651363534937e-07,
      "loss": 0.5638,
      "step": 5986
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6531849683090609,
      "learning_rate": 2.809596656824088e-07,
      "loss": 0.5314,
      "step": 5987
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.858003054363273,
      "learning_rate": 2.8067294017454634e-07,
      "loss": 0.606,
      "step": 5988
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6534372121008329,
      "learning_rate": 2.803863371605883e-07,
      "loss": 0.5379,
      "step": 5989
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6780509901106644,
      "learning_rate": 2.800998566893401e-07,
      "loss": 0.567,
      "step": 5990
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7245963483016915,
      "learning_rate": 2.7981349880958647e-07,
      "loss": 0.6026,
      "step": 5991
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6834542212246184,
      "learning_rate": 2.7952726357009117e-07,
      "loss": 0.5741,
      "step": 5992
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7117522101772877,
      "learning_rate": 2.79241151019597e-07,
      "loss": 0.5602,
      "step": 5993
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7829435136749733,
      "learning_rate": 2.7895516120682604e-07,
      "loss": 0.5894,
      "step": 5994
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8552278471823893,
      "learning_rate": 2.7866929418047924e-07,
      "loss": 0.5855,
      "step": 5995
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7876571787609605,
      "learning_rate": 2.7838354998923696e-07,
      "loss": 0.6259,
      "step": 5996
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7468719952193309,
      "learning_rate": 2.7809792868175806e-07,
      "loss": 0.6015,
      "step": 5997
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6653389532943565,
      "learning_rate": 2.7781243030668146e-07,
      "loss": 0.5129,
      "step": 5998
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7302850748586067,
      "learning_rate": 2.775270549126243e-07,
      "loss": 0.5613,
      "step": 5999
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8012297022188196,
      "learning_rate": 2.772418025481831e-07,
      "loss": 0.5791,
      "step": 6000
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7712472965593518,
      "learning_rate": 2.7695667326193315e-07,
      "loss": 0.6132,
      "step": 6001
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9609163877266168,
      "learning_rate": 2.7667166710242917e-07,
      "loss": 0.621,
      "step": 6002
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7093871269858573,
      "learning_rate": 2.763867841182047e-07,
      "loss": 0.5344,
      "step": 6003
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8373493860580146,
      "learning_rate": 2.761020243577724e-07,
      "loss": 0.6002,
      "step": 6004
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8757701649812755,
      "learning_rate": 2.7581738786962383e-07,
      "loss": 0.6883,
      "step": 6005
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8850009422470035,
      "learning_rate": 2.755328747022296e-07,
      "loss": 0.6528,
      "step": 6006
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6119258812233408,
      "learning_rate": 2.752484849040394e-07,
      "loss": 0.5361,
      "step": 6007
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9181877129155992,
      "learning_rate": 2.7496421852348174e-07,
      "loss": 0.6422,
      "step": 6008
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9390595934923341,
      "learning_rate": 2.7468007560896436e-07,
      "loss": 0.6581,
      "step": 6009
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.599311594049132,
      "learning_rate": 2.7439605620887373e-07,
      "loss": 0.4629,
      "step": 6010
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6323790413306916,
      "learning_rate": 2.7411216037157547e-07,
      "loss": 0.5128,
      "step": 6011
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8333163902664827,
      "learning_rate": 2.738283881454141e-07,
      "loss": 0.6086,
      "step": 6012
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7831915336004608,
      "learning_rate": 2.7354473957871296e-07,
      "loss": 0.6011,
      "step": 6013
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0086546721777832,
      "learning_rate": 2.7326121471977446e-07,
      "loss": 0.6588,
      "step": 6014
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6512889296463563,
      "learning_rate": 2.7297781361688013e-07,
      "loss": 0.5406,
      "step": 6015
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7056414248507044,
      "learning_rate": 2.7269453631828997e-07,
      "loss": 0.5633,
      "step": 6016
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.685780711374309,
      "learning_rate": 2.724113828722433e-07,
      "loss": 0.5203,
      "step": 6017
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.95818693964581,
      "learning_rate": 2.7212835332695816e-07,
      "loss": 0.6883,
      "step": 6018
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1733131071209184,
      "learning_rate": 2.7184544773063155e-07,
      "loss": 0.6331,
      "step": 6019
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8648006787176854,
      "learning_rate": 2.715626661314393e-07,
      "loss": 0.618,
      "step": 6020
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9174846417979671,
      "learning_rate": 2.712800085775362e-07,
      "loss": 0.6733,
      "step": 6021
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9044695136020101,
      "learning_rate": 2.7099747511705596e-07,
      "loss": 0.6042,
      "step": 6022
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7249793149771113,
      "learning_rate": 2.707150657981107e-07,
      "loss": 0.5477,
      "step": 6023
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9120084937602344,
      "learning_rate": 2.704327806687923e-07,
      "loss": 0.5806,
      "step": 6024
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7179680252188136,
      "learning_rate": 2.7015061977717077e-07,
      "loss": 0.5615,
      "step": 6025
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8767730419977466,
      "learning_rate": 2.698685831712952e-07,
      "loss": 0.5966,
      "step": 6026
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6030170582314912,
      "learning_rate": 2.695866708991936e-07,
      "loss": 0.5098,
      "step": 6027
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7998701878114364,
      "learning_rate": 2.6930488300887245e-07,
      "loss": 0.6114,
      "step": 6028
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5926202499380043,
      "learning_rate": 2.690232195483174e-07,
      "loss": 0.5378,
      "step": 6029
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7452987961358355,
      "learning_rate": 2.6874168056549287e-07,
      "loss": 0.5214,
      "step": 6030
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8779067594455152,
      "learning_rate": 2.6846026610834184e-07,
      "loss": 0.6497,
      "step": 6031
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8918113440799582,
      "learning_rate": 2.6817897622478635e-07,
      "loss": 0.6204,
      "step": 6032
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6555849519456907,
      "learning_rate": 2.678978109627272e-07,
      "loss": 0.5411,
      "step": 6033
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.82334238102189,
      "learning_rate": 2.676167703700439e-07,
      "loss": 0.6289,
      "step": 6034
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9163059483655831,
      "learning_rate": 2.673358544945946e-07,
      "loss": 0.619,
      "step": 6035
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7993175819126311,
      "learning_rate": 2.670550633842166e-07,
      "loss": 0.5785,
      "step": 6036
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6472885092900198,
      "learning_rate": 2.6677439708672544e-07,
      "loss": 0.5259,
      "step": 6037
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.673458237520553,
      "learning_rate": 2.664938556499158e-07,
      "loss": 0.5284,
      "step": 6038
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.72758721732903,
      "learning_rate": 2.662134391215608e-07,
      "loss": 0.585,
      "step": 6039
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8546450319004737,
      "learning_rate": 2.6593314754941276e-07,
      "loss": 0.609,
      "step": 6040
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7430057513295224,
      "learning_rate": 2.6565298098120213e-07,
      "loss": 0.4881,
      "step": 6041
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0513023731653521,
      "learning_rate": 2.6537293946463846e-07,
      "loss": 0.6306,
      "step": 6042
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7072360416962058,
      "learning_rate": 2.6509302304741e-07,
      "loss": 0.5034,
      "step": 6043
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7029103184322868,
      "learning_rate": 2.648132317771834e-07,
      "loss": 0.5379,
      "step": 6044
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6968776175618545,
      "learning_rate": 2.6453356570160445e-07,
      "loss": 0.5981,
      "step": 6045
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9282120789079186,
      "learning_rate": 2.6425402486829706e-07,
      "loss": 0.5668,
      "step": 6046
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7024100666034648,
      "learning_rate": 2.639746093248644e-07,
      "loss": 0.5308,
      "step": 6047
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7797537546210808,
      "learning_rate": 2.636953191188879e-07,
      "loss": 0.6172,
      "step": 6048
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8477346694084857,
      "learning_rate": 2.634161542979275e-07,
      "loss": 0.6261,
      "step": 6049
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6226113146710814,
      "learning_rate": 2.631371149095226e-07,
      "loss": 0.5089,
      "step": 6050
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6862058312423902,
      "learning_rate": 2.628582010011905e-07,
      "loss": 0.5606,
      "step": 6051
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6596465176514164,
      "learning_rate": 2.625794126204273e-07,
      "loss": 0.5599,
      "step": 6052
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.706557169153166,
      "learning_rate": 2.623007498147076e-07,
      "loss": 0.5908,
      "step": 6053
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2337620003495469,
      "learning_rate": 2.620222126314851e-07,
      "loss": 0.6388,
      "step": 6054
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8286792519104608,
      "learning_rate": 2.6174380111819136e-07,
      "loss": 0.6585,
      "step": 6055
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6660888805168419,
      "learning_rate": 2.614655153222374e-07,
      "loss": 0.5667,
      "step": 6056
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.84946454983359,
      "learning_rate": 2.611873552910121e-07,
      "loss": 0.6307,
      "step": 6057
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6954618567056445,
      "learning_rate": 2.6090932107188334e-07,
      "loss": 0.567,
      "step": 6058
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6363999654044589,
      "learning_rate": 2.6063141271219733e-07,
      "loss": 0.5084,
      "step": 6059
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8863282394633892,
      "learning_rate": 2.603536302592791e-07,
      "loss": 0.6457,
      "step": 6060
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7716979753587137,
      "learning_rate": 2.6007597376043213e-07,
      "loss": 0.6087,
      "step": 6061
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8491321496740809,
      "learning_rate": 2.597984432629382e-07,
      "loss": 0.5783,
      "step": 6062
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7575038963855081,
      "learning_rate": 2.5952103881405816e-07,
      "loss": 0.5343,
      "step": 6063
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8790303573683372,
      "learning_rate": 2.592437604610309e-07,
      "loss": 0.5781,
      "step": 6064
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7175466448177668,
      "learning_rate": 2.589666082510741e-07,
      "loss": 0.5735,
      "step": 6065
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9008488225721069,
      "learning_rate": 2.586895822313838e-07,
      "loss": 0.6768,
      "step": 6066
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7813207459569818,
      "learning_rate": 2.584126824491348e-07,
      "loss": 0.5979,
      "step": 6067
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7536466706599362,
      "learning_rate": 2.581359089514802e-07,
      "loss": 0.5722,
      "step": 6068
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6098455091943951,
      "learning_rate": 2.578592617855516e-07,
      "loss": 0.5599,
      "step": 6069
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7379375632818016,
      "learning_rate": 2.5758274099845924e-07,
      "loss": 0.5646,
      "step": 6070
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1268492324067776,
      "learning_rate": 2.5730634663729155e-07,
      "loss": 0.5671,
      "step": 6071
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9876355433443764,
      "learning_rate": 2.5703007874911566e-07,
      "loss": 0.5809,
      "step": 6072
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8503826355065556,
      "learning_rate": 2.5675393738097695e-07,
      "loss": 0.5519,
      "step": 6073
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7633957151592401,
      "learning_rate": 2.564779225798999e-07,
      "loss": 0.5927,
      "step": 6074
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7498349488460568,
      "learning_rate": 2.5620203439288654e-07,
      "loss": 0.5553,
      "step": 6075
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7646240192835654,
      "learning_rate": 2.55926272866918e-07,
      "loss": 0.5602,
      "step": 6076
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6923670154849167,
      "learning_rate": 2.556506380489536e-07,
      "loss": 0.5248,
      "step": 6077
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5882822932529663,
      "learning_rate": 2.553751299859308e-07,
      "loss": 0.5038,
      "step": 6078
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7184045870360668,
      "learning_rate": 2.550997487247659e-07,
      "loss": 0.614,
      "step": 6079
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7396086174753792,
      "learning_rate": 2.5482449431235367e-07,
      "loss": 0.5546,
      "step": 6080
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.839498982699552,
      "learning_rate": 2.5454936679556637e-07,
      "loss": 0.6348,
      "step": 6081
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2844836248116014,
      "learning_rate": 2.542743662212563e-07,
      "loss": 0.6028,
      "step": 6082
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8626961173870437,
      "learning_rate": 2.5399949263625264e-07,
      "loss": 0.6127,
      "step": 6083
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7169707515493922,
      "learning_rate": 2.537247460873637e-07,
      "loss": 0.5375,
      "step": 6084
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9177461478876875,
      "learning_rate": 2.5345012662137576e-07,
      "loss": 0.6239,
      "step": 6085
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.86032458799023,
      "learning_rate": 2.5317563428505384e-07,
      "loss": 0.5523,
      "step": 6086
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6310966666878378,
      "learning_rate": 2.529012691251411e-07,
      "loss": 0.5294,
      "step": 6087
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8590898962341832,
      "learning_rate": 2.5262703118835893e-07,
      "loss": 0.6141,
      "step": 6088
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6990846512341045,
      "learning_rate": 2.5235292052140724e-07,
      "loss": 0.539,
      "step": 6089
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.711640847688368,
      "learning_rate": 2.520789371709643e-07,
      "loss": 0.5281,
      "step": 6090
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9508934633138567,
      "learning_rate": 2.518050811836865e-07,
      "loss": 0.6899,
      "step": 6091
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8414401434256702,
      "learning_rate": 2.5153135260620884e-07,
      "loss": 0.5984,
      "step": 6092
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8030659867717226,
      "learning_rate": 2.512577514851443e-07,
      "loss": 0.5571,
      "step": 6093
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.682602018310188,
      "learning_rate": 2.509842778670843e-07,
      "loss": 0.5477,
      "step": 6094
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7733376679539254,
      "learning_rate": 2.507109317985986e-07,
      "loss": 0.5766,
      "step": 6095
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.653894245541389,
      "learning_rate": 2.504377133262352e-07,
      "loss": 0.5307,
      "step": 6096
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8025611776222085,
      "learning_rate": 2.501646224965204e-07,
      "loss": 0.5452,
      "step": 6097
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9064154088879603,
      "learning_rate": 2.4989165935595856e-07,
      "loss": 0.6348,
      "step": 6098
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9701378869557915,
      "learning_rate": 2.496188239510326e-07,
      "loss": 0.6102,
      "step": 6099
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8636391009582799,
      "learning_rate": 2.4934611632820355e-07,
      "loss": 0.6453,
      "step": 6100
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6444961434363945,
      "learning_rate": 2.490735365339106e-07,
      "loss": 0.5063,
      "step": 6101
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7861813193236913,
      "learning_rate": 2.488010846145713e-07,
      "loss": 0.5004,
      "step": 6102
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9246001674383162,
      "learning_rate": 2.4852876061658135e-07,
      "loss": 0.6255,
      "step": 6103
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6998986128582865,
      "learning_rate": 2.482565645863146e-07,
      "loss": 0.5556,
      "step": 6104
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7750874413452558,
      "learning_rate": 2.4798449657012346e-07,
      "loss": 0.5391,
      "step": 6105
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7330613515598562,
      "learning_rate": 2.47712556614338e-07,
      "loss": 0.6398,
      "step": 6106
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6800382695291657,
      "learning_rate": 2.474407447652669e-07,
      "loss": 0.5456,
      "step": 6107
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7945426239907315,
      "learning_rate": 2.4716906106919655e-07,
      "loss": 0.6407,
      "step": 6108
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0273469437952154,
      "learning_rate": 2.4689750557239254e-07,
      "loss": 0.6423,
      "step": 6109
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6693935614946517,
      "learning_rate": 2.466260783210975e-07,
      "loss": 0.5235,
      "step": 6110
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6390362376226462,
      "learning_rate": 2.463547793615327e-07,
      "loss": 0.6077,
      "step": 6111
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6704570518970413,
      "learning_rate": 2.460836087398975e-07,
      "loss": 0.4968,
      "step": 6112
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8912108939109336,
      "learning_rate": 2.4581256650236957e-07,
      "loss": 0.6156,
      "step": 6113
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8322900410594176,
      "learning_rate": 2.4554165269510464e-07,
      "loss": 0.6619,
      "step": 6114
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8037544824383916,
      "learning_rate": 2.452708673642362e-07,
      "loss": 0.5208,
      "step": 6115
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7923775818520773,
      "learning_rate": 2.450002105558763e-07,
      "loss": 0.5446,
      "step": 6116
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6541664350567258,
      "learning_rate": 2.4472968231611514e-07,
      "loss": 0.5657,
      "step": 6117
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7507058070968536,
      "learning_rate": 2.444592826910207e-07,
      "loss": 0.5654,
      "step": 6118
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8131987818553688,
      "learning_rate": 2.441890117266392e-07,
      "loss": 0.5368,
      "step": 6119
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8194160494509573,
      "learning_rate": 2.43918869468995e-07,
      "loss": 0.6116,
      "step": 6120
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5885947732279796,
      "learning_rate": 2.436488559640906e-07,
      "loss": 0.5304,
      "step": 6121
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.912678274163651,
      "learning_rate": 2.433789712579064e-07,
      "loss": 0.6646,
      "step": 6122
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.697436056537751,
      "learning_rate": 2.4310921539640096e-07,
      "loss": 0.5468,
      "step": 6123
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.917269640770019,
      "learning_rate": 2.428395884255109e-07,
      "loss": 0.6514,
      "step": 6124
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6277692097972063,
      "learning_rate": 2.4257009039115086e-07,
      "loss": 0.5016,
      "step": 6125
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8851186465441103,
      "learning_rate": 2.4230072133921363e-07,
      "loss": 0.6733,
      "step": 6126
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7167326259486761,
      "learning_rate": 2.4203148131556994e-07,
      "loss": 0.5695,
      "step": 6127
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.795848265873327,
      "learning_rate": 2.417623703660685e-07,
      "loss": 0.5531,
      "step": 6128
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8395441831281,
      "learning_rate": 2.414933885365361e-07,
      "loss": 0.6273,
      "step": 6129
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9083910954991953,
      "learning_rate": 2.412245358727776e-07,
      "loss": 0.6536,
      "step": 6130
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8073624926381848,
      "learning_rate": 2.4095581242057574e-07,
      "loss": 0.5776,
      "step": 6131
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.966493862579685,
      "learning_rate": 2.406872182256913e-07,
      "loss": 0.6228,
      "step": 6132
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3744782102181836,
      "learning_rate": 2.404187533338632e-07,
      "loss": 0.6135,
      "step": 6133
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7040664936061835,
      "learning_rate": 2.401504177908079e-07,
      "loss": 0.4891,
      "step": 6134
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9007136033087073,
      "learning_rate": 2.3988221164222057e-07,
      "loss": 0.6053,
      "step": 6135
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8825921920360752,
      "learning_rate": 2.396141349337738e-07,
      "loss": 0.6231,
      "step": 6136
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8061405324686548,
      "learning_rate": 2.3934618771111827e-07,
      "loss": 0.6048,
      "step": 6137
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.896511943011456,
      "learning_rate": 2.390783700198825e-07,
      "loss": 0.6417,
      "step": 6138
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6907283545552877,
      "learning_rate": 2.3881068190567313e-07,
      "loss": 0.5087,
      "step": 6139
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8643050289739771,
      "learning_rate": 2.3854312341407456e-07,
      "loss": 0.6739,
      "step": 6140
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1313963787201058,
      "learning_rate": 2.3827569459064921e-07,
      "loss": 0.6172,
      "step": 6141
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6790436340164293,
      "learning_rate": 2.3800839548093766e-07,
      "loss": 0.536,
      "step": 6142
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9025010247593205,
      "learning_rate": 2.3774122613045788e-07,
      "loss": 0.6232,
      "step": 6143
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6983096215804301,
      "learning_rate": 2.374741865847061e-07,
      "loss": 0.5871,
      "step": 6144
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8720582820253614,
      "learning_rate": 2.3720727688915643e-07,
      "loss": 0.6272,
      "step": 6145
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8103668726560552,
      "learning_rate": 2.369404970892609e-07,
      "loss": 0.6021,
      "step": 6146
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8380919811573346,
      "learning_rate": 2.3667384723044915e-07,
      "loss": 0.6051,
      "step": 6147
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7051538427291505,
      "learning_rate": 2.3640732735812884e-07,
      "loss": 0.5595,
      "step": 6148
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6753894229103087,
      "learning_rate": 2.361409375176857e-07,
      "loss": 0.603,
      "step": 6149
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8540110218099938,
      "learning_rate": 2.3587467775448312e-07,
      "loss": 0.6301,
      "step": 6150
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6839159268351425,
      "learning_rate": 2.3560854811386233e-07,
      "loss": 0.5673,
      "step": 6151
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.849517657035319,
      "learning_rate": 2.3534254864114234e-07,
      "loss": 0.6076,
      "step": 6152
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.601423182989811,
      "learning_rate": 2.3507667938162025e-07,
      "loss": 0.539,
      "step": 6153
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7347950639052361,
      "learning_rate": 2.348109403805707e-07,
      "loss": 0.6093,
      "step": 6154
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6234778762735239,
      "learning_rate": 2.345453316832463e-07,
      "loss": 0.5314,
      "step": 6155
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7080346633341409,
      "learning_rate": 2.3427985333487755e-07,
      "loss": 0.631,
      "step": 6156
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8627958933750759,
      "learning_rate": 2.3401450538067246e-07,
      "loss": 0.5575,
      "step": 6157
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7698302461478834,
      "learning_rate": 2.3374928786581717e-07,
      "loss": 0.6538,
      "step": 6158
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8191399467609689,
      "learning_rate": 2.3348420083547549e-07,
      "loss": 0.6162,
      "step": 6159
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7978954900859313,
      "learning_rate": 2.3321924433478845e-07,
      "loss": 0.58,
      "step": 6160
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0096037009488432,
      "learning_rate": 2.3295441840887632e-07,
      "loss": 0.6583,
      "step": 6161
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8077684486627513,
      "learning_rate": 2.3268972310283552e-07,
      "loss": 0.557,
      "step": 6162
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9059330619770233,
      "learning_rate": 2.3242515846174125e-07,
      "loss": 0.6176,
      "step": 6163
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8456246308072842,
      "learning_rate": 2.3216072453064583e-07,
      "loss": 0.6181,
      "step": 6164
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6662105081644466,
      "learning_rate": 2.3189642135457987e-07,
      "loss": 0.4938,
      "step": 6165
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6412952348471704,
      "learning_rate": 2.3163224897855116e-07,
      "loss": 0.4717,
      "step": 6166
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.658051029100701,
      "learning_rate": 2.313682074475458e-07,
      "loss": 0.5605,
      "step": 6167
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6546807765105235,
      "learning_rate": 2.311042968065271e-07,
      "loss": 0.5024,
      "step": 6168
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7221144963752282,
      "learning_rate": 2.308405171004363e-07,
      "loss": 0.5603,
      "step": 6169
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6648297769418202,
      "learning_rate": 2.3057686837419243e-07,
      "loss": 0.5343,
      "step": 6170
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.345693928460443,
      "learning_rate": 2.3031335067269208e-07,
      "loss": 0.6107,
      "step": 6171
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7682834688878059,
      "learning_rate": 2.3004996404080957e-07,
      "loss": 0.5876,
      "step": 6172
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9228110819314236,
      "learning_rate": 2.2978670852339698e-07,
      "loss": 0.6508,
      "step": 6173
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8414256310721255,
      "learning_rate": 2.2952358416528383e-07,
      "loss": 0.6273,
      "step": 6174
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.825416034353555,
      "learning_rate": 2.2926059101127727e-07,
      "loss": 0.6615,
      "step": 6175
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7903043909293808,
      "learning_rate": 2.2899772910616276e-07,
      "loss": 0.6043,
      "step": 6176
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6344609516988309,
      "learning_rate": 2.287349984947028e-07,
      "loss": 0.5433,
      "step": 6177
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7786275728995095,
      "learning_rate": 2.2847239922163763e-07,
      "loss": 0.6226,
      "step": 6178
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8085348294502698,
      "learning_rate": 2.282099313316852e-07,
      "loss": 0.6045,
      "step": 6179
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8402381436845174,
      "learning_rate": 2.2794759486954095e-07,
      "loss": 0.6763,
      "step": 6180
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.784085074016798,
      "learning_rate": 2.2768538987987808e-07,
      "loss": 0.6018,
      "step": 6181
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7082911759927489,
      "learning_rate": 2.2742331640734745e-07,
      "loss": 0.5078,
      "step": 6182
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6877394593846569,
      "learning_rate": 2.2716137449657734e-07,
      "loss": 0.5292,
      "step": 6183
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7131794318013999,
      "learning_rate": 2.268995641921738e-07,
      "loss": 0.5109,
      "step": 6184
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8070935178112008,
      "learning_rate": 2.2663788553872042e-07,
      "loss": 0.6405,
      "step": 6185
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8510749910399822,
      "learning_rate": 2.263763385807783e-07,
      "loss": 0.6401,
      "step": 6186
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7421056235210514,
      "learning_rate": 2.2611492336288617e-07,
      "loss": 0.5731,
      "step": 6187
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8394451344147188,
      "learning_rate": 2.258536399295603e-07,
      "loss": 0.6112,
      "step": 6188
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8608248262112019,
      "learning_rate": 2.2559248832529453e-07,
      "loss": 0.649,
      "step": 6189
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8231386492152521,
      "learning_rate": 2.2533146859456032e-07,
      "loss": 0.586,
      "step": 6190
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8864386376330439,
      "learning_rate": 2.2507058078180652e-07,
      "loss": 0.6393,
      "step": 6191
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8523150774911349,
      "learning_rate": 2.248098249314594e-07,
      "loss": 0.5791,
      "step": 6192
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7245519485811553,
      "learning_rate": 2.2454920108792352e-07,
      "loss": 0.5142,
      "step": 6193
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6229072817668845,
      "learning_rate": 2.2428870929558007e-07,
      "loss": 0.4774,
      "step": 6194
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.615961047832533,
      "learning_rate": 2.2402834959878813e-07,
      "loss": 0.544,
      "step": 6195
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8719132548508182,
      "learning_rate": 2.2376812204188423e-07,
      "loss": 0.6309,
      "step": 6196
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6511790523559398,
      "learning_rate": 2.2350802666918245e-07,
      "loss": 0.5465,
      "step": 6197
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.129917346950309,
      "learning_rate": 2.2324806352497427e-07,
      "loss": 0.5943,
      "step": 6198
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5643226897481541,
      "learning_rate": 2.2298823265352873e-07,
      "loss": 0.4993,
      "step": 6199
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5828763068031434,
      "learning_rate": 2.227285340990924e-07,
      "loss": 0.4714,
      "step": 6200
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8251071262766166,
      "learning_rate": 2.224689679058891e-07,
      "loss": 0.5888,
      "step": 6201
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8262014776427893,
      "learning_rate": 2.2220953411812026e-07,
      "loss": 0.6359,
      "step": 6202
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7236075754734544,
      "learning_rate": 2.2195023277996482e-07,
      "loss": 0.5711,
      "step": 6203
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7054642196908039,
      "learning_rate": 2.2169106393557914e-07,
      "loss": 0.5468,
      "step": 6204
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8547135533385697,
      "learning_rate": 2.2143202762909686e-07,
      "loss": 0.5628,
      "step": 6205
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8491345267036173,
      "learning_rate": 2.2117312390462916e-07,
      "loss": 0.6379,
      "step": 6206
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0459942755065024,
      "learning_rate": 2.209143528062648e-07,
      "loss": 0.6973,
      "step": 6207
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.739616360893179,
      "learning_rate": 2.2065571437806962e-07,
      "loss": 0.5401,
      "step": 6208
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.817552372591894,
      "learning_rate": 2.2039720866408717e-07,
      "loss": 0.6163,
      "step": 6209
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0250304468731926,
      "learning_rate": 2.2013883570833823e-07,
      "loss": 0.6374,
      "step": 6210
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7444594755446832,
      "learning_rate": 2.1988059555482098e-07,
      "loss": 0.5512,
      "step": 6211
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6644078161501535,
      "learning_rate": 2.1962248824751118e-07,
      "loss": 0.518,
      "step": 6212
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6448817598128918,
      "learning_rate": 2.1936451383036168e-07,
      "loss": 0.5224,
      "step": 6213
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7767852020030122,
      "learning_rate": 2.1910667234730284e-07,
      "loss": 0.5148,
      "step": 6214
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6387847716845088,
      "learning_rate": 2.188489638422425e-07,
      "loss": 0.5277,
      "step": 6215
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6310793953834269,
      "learning_rate": 2.1859138835906553e-07,
      "loss": 0.5051,
      "step": 6216
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.654122115748672,
      "learning_rate": 2.1833394594163458e-07,
      "loss": 0.5397,
      "step": 6217
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8739973972298009,
      "learning_rate": 2.1807663663378918e-07,
      "loss": 0.6393,
      "step": 6218
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8104825042986271,
      "learning_rate": 2.178194604793463e-07,
      "loss": 0.5591,
      "step": 6219
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6499605015633183,
      "learning_rate": 2.1756241752210092e-07,
      "loss": 0.4538,
      "step": 6220
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.019532995564683,
      "learning_rate": 2.1730550780582446e-07,
      "loss": 0.6201,
      "step": 6221
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7135091890960624,
      "learning_rate": 2.1704873137426593e-07,
      "loss": 0.5264,
      "step": 6222
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6391255040093908,
      "learning_rate": 2.1679208827115169e-07,
      "loss": 0.5279,
      "step": 6223
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7775088685578503,
      "learning_rate": 2.1653557854018555e-07,
      "loss": 0.6043,
      "step": 6224
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6772384287864399,
      "learning_rate": 2.1627920222504826e-07,
      "loss": 0.5552,
      "step": 6225
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.771834559345236,
      "learning_rate": 2.1602295936939796e-07,
      "loss": 0.6346,
      "step": 6226
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8337009838122383,
      "learning_rate": 2.1576685001687045e-07,
      "loss": 0.6111,
      "step": 6227
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6756951104199925,
      "learning_rate": 2.1551087421107817e-07,
      "loss": 0.5199,
      "step": 6228
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8064353383610573,
      "learning_rate": 2.1525503199561135e-07,
      "loss": 0.6212,
      "step": 6229
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7793518603617083,
      "learning_rate": 2.1499932341403705e-07,
      "loss": 0.6158,
      "step": 6230
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6559695483889093,
      "learning_rate": 2.1474374850989996e-07,
      "loss": 0.4771,
      "step": 6231
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8343644003109523,
      "learning_rate": 2.1448830732672185e-07,
      "loss": 0.6676,
      "step": 6232
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7788103340204466,
      "learning_rate": 2.1423299990800149e-07,
      "loss": 0.5985,
      "step": 6233
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7170363534888302,
      "learning_rate": 2.1397782629721517e-07,
      "loss": 0.6061,
      "step": 6234
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.806468822073873,
      "learning_rate": 2.137227865378164e-07,
      "loss": 0.6236,
      "step": 6235
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8513196987582275,
      "learning_rate": 2.1346788067323563e-07,
      "loss": 0.5906,
      "step": 6236
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8289682413831649,
      "learning_rate": 2.1321310874688082e-07,
      "loss": 0.5969,
      "step": 6237
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3137590899314386,
      "learning_rate": 2.129584708021368e-07,
      "loss": 0.6239,
      "step": 6238
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0789711622266045,
      "learning_rate": 2.1270396688236593e-07,
      "loss": 0.5637,
      "step": 6239
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6990102676105047,
      "learning_rate": 2.1244959703090748e-07,
      "loss": 0.5226,
      "step": 6240
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8457214667465854,
      "learning_rate": 2.1219536129107807e-07,
      "loss": 0.6199,
      "step": 6241
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.797665763783231,
      "learning_rate": 2.1194125970617128e-07,
      "loss": 0.6065,
      "step": 6242
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8932407104696183,
      "learning_rate": 2.1168729231945814e-07,
      "loss": 0.6087,
      "step": 6243
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6819998509438194,
      "learning_rate": 2.1143345917418643e-07,
      "loss": 0.4935,
      "step": 6244
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.937928650863166,
      "learning_rate": 2.111797603135811e-07,
      "loss": 0.5585,
      "step": 6245
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5516065324835596,
      "learning_rate": 2.1092619578084504e-07,
      "loss": 0.503,
      "step": 6246
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9594721420673439,
      "learning_rate": 2.1067276561915736e-07,
      "loss": 0.6298,
      "step": 6247
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5980685326717005,
      "learning_rate": 2.1041946987167447e-07,
      "loss": 0.5303,
      "step": 6248
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6756724552520457,
      "learning_rate": 2.1016630858153016e-07,
      "loss": 0.5465,
      "step": 6249
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7983669310966843,
      "learning_rate": 2.0991328179183498e-07,
      "loss": 0.6209,
      "step": 6250
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7678742555934108,
      "learning_rate": 2.0966038954567688e-07,
      "loss": 0.5971,
      "step": 6251
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9154015018958837,
      "learning_rate": 2.0940763188612077e-07,
      "loss": 0.6669,
      "step": 6252
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8220317850433128,
      "learning_rate": 2.0915500885620862e-07,
      "loss": 0.5726,
      "step": 6253
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6589395038463134,
      "learning_rate": 2.089025204989594e-07,
      "loss": 0.5232,
      "step": 6254
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6709481000249065,
      "learning_rate": 2.0865016685736947e-07,
      "loss": 0.5982,
      "step": 6255
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7788342113952555,
      "learning_rate": 2.083979479744118e-07,
      "loss": 0.5763,
      "step": 6256
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3512401019032951,
      "learning_rate": 2.0814586389303678e-07,
      "loss": 0.6627,
      "step": 6257
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6488245785651335,
      "learning_rate": 2.0789391465617168e-07,
      "loss": 0.5433,
      "step": 6258
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6617934735702057,
      "learning_rate": 2.0764210030672081e-07,
      "loss": 0.5228,
      "step": 6259
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7233697790640223,
      "learning_rate": 2.0739042088756553e-07,
      "loss": 0.587,
      "step": 6260
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9254888381235377,
      "learning_rate": 2.0713887644156424e-07,
      "loss": 0.5935,
      "step": 6261
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9100774967680485,
      "learning_rate": 2.0688746701155236e-07,
      "loss": 0.6123,
      "step": 6262
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6436387665579886,
      "learning_rate": 2.0663619264034214e-07,
      "loss": 0.5319,
      "step": 6263
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8590345538005468,
      "learning_rate": 2.063850533707232e-07,
      "loss": 0.6104,
      "step": 6264
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8463191713603788,
      "learning_rate": 2.0613404924546184e-07,
      "loss": 0.5961,
      "step": 6265
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8085234682305709,
      "learning_rate": 2.0588318030730144e-07,
      "loss": 0.5678,
      "step": 6266
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6914137412706343,
      "learning_rate": 2.0563244659896238e-07,
      "loss": 0.5121,
      "step": 6267
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6843283864341307,
      "learning_rate": 2.0538184816314208e-07,
      "loss": 0.5124,
      "step": 6268
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8946948998774575,
      "learning_rate": 2.051313850425147e-07,
      "loss": 0.6132,
      "step": 6269
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7428266860491495,
      "learning_rate": 2.0488105727973148e-07,
      "loss": 0.5607,
      "step": 6270
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8730111544453218,
      "learning_rate": 2.0463086491742055e-07,
      "loss": 0.6008,
      "step": 6271
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7252408618930705,
      "learning_rate": 2.0438080799818746e-07,
      "loss": 0.5374,
      "step": 6272
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8438369637444263,
      "learning_rate": 2.0413088656461408e-07,
      "loss": 0.5972,
      "step": 6273
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8131843149474692,
      "learning_rate": 2.0388110065925944e-07,
      "loss": 0.6137,
      "step": 6274
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8415116289003903,
      "learning_rate": 2.0363145032465944e-07,
      "loss": 0.6101,
      "step": 6275
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7714995326644103,
      "learning_rate": 2.0338193560332695e-07,
      "loss": 0.5445,
      "step": 6276
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7327847152386519,
      "learning_rate": 2.0313255653775141e-07,
      "loss": 0.5384,
      "step": 6277
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.629290614789581,
      "learning_rate": 2.0288331317040008e-07,
      "loss": 0.5305,
      "step": 6278
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6621278271819153,
      "learning_rate": 2.0263420554371623e-07,
      "loss": 0.5472,
      "step": 6279
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7942750294566222,
      "learning_rate": 2.023852337001203e-07,
      "loss": 0.6171,
      "step": 6280
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6458812080589901,
      "learning_rate": 2.0213639768200953e-07,
      "loss": 0.5199,
      "step": 6281
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8938848159028191,
      "learning_rate": 2.0188769753175815e-07,
      "loss": 0.6954,
      "step": 6282
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8522091108964585,
      "learning_rate": 2.0163913329171722e-07,
      "loss": 0.5363,
      "step": 6283
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8090130223193264,
      "learning_rate": 2.0139070500421462e-07,
      "loss": 0.5808,
      "step": 6284
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8329550738956516,
      "learning_rate": 2.011424127115552e-07,
      "loss": 0.5909,
      "step": 6285
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6936498490632294,
      "learning_rate": 2.008942564560203e-07,
      "loss": 0.5191,
      "step": 6286
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7242559743663319,
      "learning_rate": 2.0064623627986865e-07,
      "loss": 0.5806,
      "step": 6287
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6389128592749685,
      "learning_rate": 2.003983522253352e-07,
      "loss": 0.5405,
      "step": 6288
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7194923453893851,
      "learning_rate": 2.0015060433463227e-07,
      "loss": 0.5223,
      "step": 6289
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7754058787819915,
      "learning_rate": 1.999029926499487e-07,
      "loss": 0.5452,
      "step": 6290
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7336404798764289,
      "learning_rate": 1.996555172134501e-07,
      "loss": 0.538,
      "step": 6291
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9305867929066064,
      "learning_rate": 1.9940817806727895e-07,
      "loss": 0.6457,
      "step": 6292
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7816832650370892,
      "learning_rate": 1.9916097525355458e-07,
      "loss": 0.6481,
      "step": 6293
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8856471665991816,
      "learning_rate": 1.9891390881437299e-07,
      "loss": 0.6572,
      "step": 6294
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7133195386681009,
      "learning_rate": 1.9866697879180715e-07,
      "loss": 0.5318,
      "step": 6295
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.013253942911983,
      "learning_rate": 1.984201852279066e-07,
      "loss": 0.7022,
      "step": 6296
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8727206688441004,
      "learning_rate": 1.981735281646977e-07,
      "loss": 0.5987,
      "step": 6297
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7742667803597295,
      "learning_rate": 1.9792700764418345e-07,
      "loss": 0.5594,
      "step": 6298
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.745606837374919,
      "learning_rate": 1.9768062370834405e-07,
      "loss": 0.5461,
      "step": 6299
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0833835682966506,
      "learning_rate": 1.9743437639913585e-07,
      "loss": 0.6158,
      "step": 6300
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7640600021136525,
      "learning_rate": 1.9718826575849224e-07,
      "loss": 0.6632,
      "step": 6301
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9434436905733974,
      "learning_rate": 1.9694229182832346e-07,
      "loss": 0.6293,
      "step": 6302
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6272984304251428,
      "learning_rate": 1.9669645465051589e-07,
      "loss": 0.4862,
      "step": 6303
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8349542597411452,
      "learning_rate": 1.9645075426693348e-07,
      "loss": 0.5976,
      "step": 6304
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7185382204153398,
      "learning_rate": 1.9620519071941643e-07,
      "loss": 0.5265,
      "step": 6305
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8431576955594597,
      "learning_rate": 1.9595976404978143e-07,
      "loss": 0.6224,
      "step": 6306
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8780822050580305,
      "learning_rate": 1.9571447429982224e-07,
      "loss": 0.6449,
      "step": 6307
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7396418814743158,
      "learning_rate": 1.9546932151130912e-07,
      "loss": 0.5831,
      "step": 6308
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7519152428086103,
      "learning_rate": 1.9522430572598892e-07,
      "loss": 0.5698,
      "step": 6309
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6843386158412604,
      "learning_rate": 1.9497942698558546e-07,
      "loss": 0.5245,
      "step": 6310
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7895614587897429,
      "learning_rate": 1.9473468533179892e-07,
      "loss": 0.6306,
      "step": 6311
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7587816135022525,
      "learning_rate": 1.9449008080630624e-07,
      "loss": 0.503,
      "step": 6312
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7672071734838551,
      "learning_rate": 1.9424561345076108e-07,
      "loss": 0.5462,
      "step": 6313
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9808399568330299,
      "learning_rate": 1.9400128330679355e-07,
      "loss": 0.6619,
      "step": 6314
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8715896927010516,
      "learning_rate": 1.9375709041601074e-07,
      "loss": 0.6793,
      "step": 6315
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8521171782075534,
      "learning_rate": 1.9351303481999614e-07,
      "loss": 0.6833,
      "step": 6316
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7482531927697557,
      "learning_rate": 1.9326911656030965e-07,
      "loss": 0.512,
      "step": 6317
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8575181625791382,
      "learning_rate": 1.9302533567848823e-07,
      "loss": 0.5831,
      "step": 6318
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6566070350060423,
      "learning_rate": 1.9278169221604513e-07,
      "loss": 0.523,
      "step": 6319
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9529699050001621,
      "learning_rate": 1.9253818621447026e-07,
      "loss": 0.6402,
      "step": 6320
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0139588522930159,
      "learning_rate": 1.9229481771523027e-07,
      "loss": 0.6214,
      "step": 6321
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7329070842248678,
      "learning_rate": 1.9205158675976808e-07,
      "loss": 0.5349,
      "step": 6322
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6931646080031605,
      "learning_rate": 1.9180849338950366e-07,
      "loss": 0.5474,
      "step": 6323
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8263718797189621,
      "learning_rate": 1.9156553764583305e-07,
      "loss": 0.6064,
      "step": 6324
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7814413757375455,
      "learning_rate": 1.9132271957012923e-07,
      "loss": 0.6105,
      "step": 6325
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8070805576527864,
      "learning_rate": 1.910800392037415e-07,
      "loss": 0.5949,
      "step": 6326
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6912477101000122,
      "learning_rate": 1.9083749658799585e-07,
      "loss": 0.5715,
      "step": 6327
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7748718444805102,
      "learning_rate": 1.9059509176419475e-07,
      "loss": 0.5671,
      "step": 6328
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8287448142592743,
      "learning_rate": 1.9035282477361714e-07,
      "loss": 0.5249,
      "step": 6329
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.806782163536333,
      "learning_rate": 1.9011069565751836e-07,
      "loss": 0.638,
      "step": 6330
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9420943436691243,
      "learning_rate": 1.898687044571311e-07,
      "loss": 0.675,
      "step": 6331
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7196863345732111,
      "learning_rate": 1.896268512136635e-07,
      "loss": 0.5583,
      "step": 6332
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.880855053144274,
      "learning_rate": 1.8938513596830073e-07,
      "loss": 0.6334,
      "step": 6333
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9545686648502021,
      "learning_rate": 1.8914355876220445e-07,
      "loss": 0.6407,
      "step": 6334
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7281200473455491,
      "learning_rate": 1.8890211963651258e-07,
      "loss": 0.5606,
      "step": 6335
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8711095709007972,
      "learning_rate": 1.8866081863233985e-07,
      "loss": 0.6629,
      "step": 6336
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7852680421749125,
      "learning_rate": 1.8841965579077724e-07,
      "loss": 0.5847,
      "step": 6337
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7730199525005151,
      "learning_rate": 1.8817863115289222e-07,
      "loss": 0.6129,
      "step": 6338
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6427158482924086,
      "learning_rate": 1.879377447597289e-07,
      "loss": 0.5347,
      "step": 6339
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6581761014726824,
      "learning_rate": 1.8769699665230754e-07,
      "loss": 0.5525,
      "step": 6340
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9869996411358212,
      "learning_rate": 1.8745638687162523e-07,
      "loss": 0.6665,
      "step": 6341
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9526859933328118,
      "learning_rate": 1.8721591545865512e-07,
      "loss": 0.6288,
      "step": 6342
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8641573246907084,
      "learning_rate": 1.869755824543472e-07,
      "loss": 0.6347,
      "step": 6343
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6036746107142822,
      "learning_rate": 1.867353878996275e-07,
      "loss": 0.4964,
      "step": 6344
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7271420038311592,
      "learning_rate": 1.8649533183539867e-07,
      "loss": 0.5888,
      "step": 6345
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8457423292929522,
      "learning_rate": 1.8625541430253988e-07,
      "loss": 0.62,
      "step": 6346
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8524668566132301,
      "learning_rate": 1.8601563534190645e-07,
      "loss": 0.6276,
      "step": 6347
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6489201474489246,
      "learning_rate": 1.8577599499433027e-07,
      "loss": 0.5255,
      "step": 6348
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7334057555958982,
      "learning_rate": 1.855364933006197e-07,
      "loss": 0.586,
      "step": 6349
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6961730046553802,
      "learning_rate": 1.8529713030155926e-07,
      "loss": 0.5236,
      "step": 6350
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7147010474141334,
      "learning_rate": 1.8505790603790993e-07,
      "loss": 0.5408,
      "step": 6351
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6164425135836364,
      "learning_rate": 1.848188205504093e-07,
      "loss": 0.5392,
      "step": 6352
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0597653165636454,
      "learning_rate": 1.8457987387977103e-07,
      "loss": 0.6279,
      "step": 6353
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.624523829496321,
      "learning_rate": 1.843410660666852e-07,
      "loss": 0.5621,
      "step": 6354
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9573255030824929,
      "learning_rate": 1.841023971518184e-07,
      "loss": 0.6869,
      "step": 6355
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7037267213184512,
      "learning_rate": 1.8386386717581315e-07,
      "loss": 0.5891,
      "step": 6356
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7800208927624231,
      "learning_rate": 1.8362547617928915e-07,
      "loss": 0.5133,
      "step": 6357
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6584307138097006,
      "learning_rate": 1.833872242028416e-07,
      "loss": 0.5452,
      "step": 6358
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7216430296428064,
      "learning_rate": 1.8314911128704223e-07,
      "loss": 0.5622,
      "step": 6359
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6251242607662475,
      "learning_rate": 1.8291113747243947e-07,
      "loss": 0.5292,
      "step": 6360
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7087409223720543,
      "learning_rate": 1.8267330279955762e-07,
      "loss": 0.5252,
      "step": 6361
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.867110843805794,
      "learning_rate": 1.824356073088974e-07,
      "loss": 0.6113,
      "step": 6362
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7110211789926288,
      "learning_rate": 1.82198051040936e-07,
      "loss": 0.5981,
      "step": 6363
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8016313328242031,
      "learning_rate": 1.8196063403612672e-07,
      "loss": 0.5634,
      "step": 6364
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.01569137624723,
      "learning_rate": 1.817233563348991e-07,
      "loss": 0.6485,
      "step": 6365
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6591051211191011,
      "learning_rate": 1.8148621797765917e-07,
      "loss": 0.5895,
      "step": 6366
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7030828793386561,
      "learning_rate": 1.812492190047892e-07,
      "loss": 0.5617,
      "step": 6367
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6738638626550979,
      "learning_rate": 1.8101235945664738e-07,
      "loss": 0.5458,
      "step": 6368
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.942073762382475,
      "learning_rate": 1.8077563937356876e-07,
      "loss": 0.6222,
      "step": 6369
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7136713015062975,
      "learning_rate": 1.8053905879586406e-07,
      "loss": 0.5724,
      "step": 6370
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.591526390254191,
      "learning_rate": 1.803026177638205e-07,
      "loss": 0.4962,
      "step": 6371
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7220390551766422,
      "learning_rate": 1.8006631631770165e-07,
      "loss": 0.555,
      "step": 6372
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7602859269961497,
      "learning_rate": 1.7983015449774718e-07,
      "loss": 0.5702,
      "step": 6373
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6320896953743237,
      "learning_rate": 1.7959413234417287e-07,
      "loss": 0.4979,
      "step": 6374
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9495571312759755,
      "learning_rate": 1.7935824989717096e-07,
      "loss": 0.6251,
      "step": 6375
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7477847773011265,
      "learning_rate": 1.7912250719690958e-07,
      "loss": 0.5071,
      "step": 6376
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7600333163371802,
      "learning_rate": 1.7888690428353348e-07,
      "loss": 0.57,
      "step": 6377
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8884002058168086,
      "learning_rate": 1.7865144119716303e-07,
      "loss": 0.6447,
      "step": 6378
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6977790288722401,
      "learning_rate": 1.7841611797789567e-07,
      "loss": 0.5408,
      "step": 6379
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8262005274786918,
      "learning_rate": 1.7818093466580418e-07,
      "loss": 0.6134,
      "step": 6380
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9694590862128035,
      "learning_rate": 1.7794589130093784e-07,
      "loss": 0.6135,
      "step": 6381
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9078416562536212,
      "learning_rate": 1.7771098792332216e-07,
      "loss": 0.5155,
      "step": 6382
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7790278931943109,
      "learning_rate": 1.7747622457295864e-07,
      "loss": 0.4999,
      "step": 6383
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8656281436552976,
      "learning_rate": 1.7724160128982514e-07,
      "loss": 0.6482,
      "step": 6384
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8495498071544636,
      "learning_rate": 1.7700711811387548e-07,
      "loss": 0.5911,
      "step": 6385
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6620275459228138,
      "learning_rate": 1.7677277508503975e-07,
      "loss": 0.4804,
      "step": 6386
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1846555280971403,
      "learning_rate": 1.765385722432241e-07,
      "loss": 0.6611,
      "step": 6387
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.927148551253297,
      "learning_rate": 1.7630450962831057e-07,
      "loss": 0.6511,
      "step": 6388
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8331121006933375,
      "learning_rate": 1.760705872801581e-07,
      "loss": 0.5948,
      "step": 6389
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.639205670609675,
      "learning_rate": 1.75836805238601e-07,
      "loss": 0.5146,
      "step": 6390
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0216423260893694,
      "learning_rate": 1.7560316354344985e-07,
      "loss": 0.6268,
      "step": 6391
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6681394359426978,
      "learning_rate": 1.7536966223449157e-07,
      "loss": 0.5128,
      "step": 6392
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9974760248514055,
      "learning_rate": 1.751363013514887e-07,
      "loss": 0.5872,
      "step": 6393
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6670593548664926,
      "learning_rate": 1.749030809341805e-07,
      "loss": 0.5307,
      "step": 6394
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7544519465139703,
      "learning_rate": 1.746700010222817e-07,
      "loss": 0.5412,
      "step": 6395
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7160350778658963,
      "learning_rate": 1.7443706165548357e-07,
      "loss": 0.5142,
      "step": 6396
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6549785071593615,
      "learning_rate": 1.7420426287345314e-07,
      "loss": 0.5015,
      "step": 6397
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6642450313320617,
      "learning_rate": 1.7397160471583372e-07,
      "loss": 0.4854,
      "step": 6398
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7636945659730447,
      "learning_rate": 1.737390872222445e-07,
      "loss": 0.6147,
      "step": 6399
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8260526902724561,
      "learning_rate": 1.7350671043228072e-07,
      "loss": 0.6781,
      "step": 6400
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7191454276398482,
      "learning_rate": 1.732744743855139e-07,
      "loss": 0.5649,
      "step": 6401
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8297999983852089,
      "learning_rate": 1.7304237912149133e-07,
      "loss": 0.5537,
      "step": 6402
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9557534527789039,
      "learning_rate": 1.728104246797364e-07,
      "loss": 0.6836,
      "step": 6403
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9206662740081434,
      "learning_rate": 1.7257861109974858e-07,
      "loss": 0.5252,
      "step": 6404
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0026819383098646,
      "learning_rate": 1.7234693842100324e-07,
      "loss": 0.6116,
      "step": 6405
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7924675594308621,
      "learning_rate": 1.7211540668295178e-07,
      "loss": 0.6099,
      "step": 6406
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8181930154286083,
      "learning_rate": 1.7188401592502177e-07,
      "loss": 0.5612,
      "step": 6407
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8815405876265957,
      "learning_rate": 1.716527661866166e-07,
      "loss": 0.6692,
      "step": 6408
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9286025048840973,
      "learning_rate": 1.7142165750711578e-07,
      "loss": 0.6278,
      "step": 6409
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6329255217331994,
      "learning_rate": 1.7119068992587459e-07,
      "loss": 0.5136,
      "step": 6410
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7731308199990283,
      "learning_rate": 1.7095986348222435e-07,
      "loss": 0.6067,
      "step": 6411
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8612889388826246,
      "learning_rate": 1.707291782154725e-07,
      "loss": 0.6371,
      "step": 6412
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7644036995048867,
      "learning_rate": 1.704986341649024e-07,
      "loss": 0.5106,
      "step": 6413
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9343442730788272,
      "learning_rate": 1.7026823136977286e-07,
      "loss": 0.5797,
      "step": 6414
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7196132490374512,
      "learning_rate": 1.7003796986931973e-07,
      "loss": 0.5818,
      "step": 6415
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8957309653844695,
      "learning_rate": 1.6980784970275387e-07,
      "loss": 0.6632,
      "step": 6416
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8440408274405804,
      "learning_rate": 1.6957787090926236e-07,
      "loss": 0.6011,
      "step": 6417
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.93792552823572,
      "learning_rate": 1.6934803352800808e-07,
      "loss": 0.6583,
      "step": 6418
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9322096866592263,
      "learning_rate": 1.691183375981301e-07,
      "loss": 0.5917,
      "step": 6419
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6503567276753359,
      "learning_rate": 1.688887831587431e-07,
      "loss": 0.5503,
      "step": 6420
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8316738841530569,
      "learning_rate": 1.6865937024893795e-07,
      "loss": 0.6626,
      "step": 6421
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6956798547432238,
      "learning_rate": 1.6843009890778114e-07,
      "loss": 0.5684,
      "step": 6422
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9562807844716208,
      "learning_rate": 1.6820096917431526e-07,
      "loss": 0.6184,
      "step": 6423
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7183818880857306,
      "learning_rate": 1.6797198108755872e-07,
      "loss": 0.589,
      "step": 6424
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.66546701926384,
      "learning_rate": 1.677431346865057e-07,
      "loss": 0.4897,
      "step": 6425
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9082990924901301,
      "learning_rate": 1.6751443001012654e-07,
      "loss": 0.6684,
      "step": 6426
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6905442657991575,
      "learning_rate": 1.6728586709736726e-07,
      "loss": 0.5521,
      "step": 6427
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9785544715746913,
      "learning_rate": 1.6705744598714955e-07,
      "loss": 0.7008,
      "step": 6428
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9293403069099524,
      "learning_rate": 1.6682916671837122e-07,
      "loss": 0.6076,
      "step": 6429
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8678694414405309,
      "learning_rate": 1.6660102932990594e-07,
      "loss": 0.6377,
      "step": 6430
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7844578098871634,
      "learning_rate": 1.663730338606032e-07,
      "loss": 0.5752,
      "step": 6431
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8202003274152232,
      "learning_rate": 1.6614518034928793e-07,
      "loss": 0.5561,
      "step": 6432
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9173399870710137,
      "learning_rate": 1.6591746883476155e-07,
      "loss": 0.4989,
      "step": 6433
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8762911447371399,
      "learning_rate": 1.656898993558007e-07,
      "loss": 0.6048,
      "step": 6434
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7340270326810643,
      "learning_rate": 1.6546247195115835e-07,
      "loss": 0.5098,
      "step": 6435
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9272776908177466,
      "learning_rate": 1.6523518665956282e-07,
      "loss": 0.6135,
      "step": 6436
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6040889990067481,
      "learning_rate": 1.6500804351971841e-07,
      "loss": 0.4972,
      "step": 6437
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.661345947261647,
      "learning_rate": 1.6478104257030534e-07,
      "loss": 0.559,
      "step": 6438
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.649652803496156,
      "learning_rate": 1.6455418384997954e-07,
      "loss": 0.5231,
      "step": 6439
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7995945062024263,
      "learning_rate": 1.6432746739737258e-07,
      "loss": 0.5921,
      "step": 6440
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8492162415656811,
      "learning_rate": 1.6410089325109167e-07,
      "loss": 0.5771,
      "step": 6441
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9078490728373156,
      "learning_rate": 1.638744614497206e-07,
      "loss": 0.6423,
      "step": 6442
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7413654105852527,
      "learning_rate": 1.6364817203181802e-07,
      "loss": 0.5211,
      "step": 6443
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8239655648294082,
      "learning_rate": 1.6342202503591873e-07,
      "loss": 0.6064,
      "step": 6444
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6670570384021737,
      "learning_rate": 1.6319602050053315e-07,
      "loss": 0.5126,
      "step": 6445
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7648876064152663,
      "learning_rate": 1.6297015846414753e-07,
      "loss": 0.59,
      "step": 6446
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8368277373518951,
      "learning_rate": 1.627444389652237e-07,
      "loss": 0.5981,
      "step": 6447
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5976086122367198,
      "learning_rate": 1.6251886204219956e-07,
      "loss": 0.5213,
      "step": 6448
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8258949888052496,
      "learning_rate": 1.6229342773348842e-07,
      "loss": 0.5766,
      "step": 6449
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7984499235273745,
      "learning_rate": 1.6206813607747938e-07,
      "loss": 0.5326,
      "step": 6450
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8518211429643289,
      "learning_rate": 1.6184298711253718e-07,
      "loss": 0.5654,
      "step": 6451
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6397122071272141,
      "learning_rate": 1.616179808770024e-07,
      "loss": 0.5184,
      "step": 6452
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.630496127055483,
      "learning_rate": 1.6139311740919137e-07,
      "loss": 0.4745,
      "step": 6453
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7713550866422154,
      "learning_rate": 1.6116839674739578e-07,
      "loss": 0.5725,
      "step": 6454
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9324817635530495,
      "learning_rate": 1.609438189298835e-07,
      "loss": 0.5728,
      "step": 6455
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9914071622430473,
      "learning_rate": 1.607193839948975e-07,
      "loss": 0.6457,
      "step": 6456
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.655805911409877,
      "learning_rate": 1.6049509198065692e-07,
      "loss": 0.4996,
      "step": 6457
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6184139576260695,
      "learning_rate": 1.6027094292535625e-07,
      "loss": 0.5512,
      "step": 6458
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7214272013046492,
      "learning_rate": 1.6004693686716574e-07,
      "loss": 0.5581,
      "step": 6459
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8772221301785722,
      "learning_rate": 1.5982307384423132e-07,
      "loss": 0.5602,
      "step": 6460
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.854963003663283,
      "learning_rate": 1.5959935389467448e-07,
      "loss": 0.6465,
      "step": 6461
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8809321236222114,
      "learning_rate": 1.5937577705659254e-07,
      "loss": 0.6256,
      "step": 6462
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7255865413800601,
      "learning_rate": 1.5915234336805815e-07,
      "loss": 0.5249,
      "step": 6463
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7992754982921081,
      "learning_rate": 1.589290528671199e-07,
      "loss": 0.6538,
      "step": 6464
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.873305641435741,
      "learning_rate": 1.5870590559180164e-07,
      "loss": 0.6024,
      "step": 6465
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.838081913665831,
      "learning_rate": 1.5848290158010313e-07,
      "loss": 0.6351,
      "step": 6466
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7088100619406694,
      "learning_rate": 1.5826004086999932e-07,
      "loss": 0.5718,
      "step": 6467
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8293577427819205,
      "learning_rate": 1.5803732349944166e-07,
      "loss": 0.5441,
      "step": 6468
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9314062763345221,
      "learning_rate": 1.5781474950635633e-07,
      "loss": 0.5891,
      "step": 6469
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6529738743956238,
      "learning_rate": 1.5759231892864533e-07,
      "loss": 0.5498,
      "step": 6470
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7630417774586676,
      "learning_rate": 1.5737003180418628e-07,
      "loss": 0.5154,
      "step": 6471
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6866345975506298,
      "learning_rate": 1.5714788817083235e-07,
      "loss": 0.5353,
      "step": 6472
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6653141842795357,
      "learning_rate": 1.5692588806641228e-07,
      "loss": 0.583,
      "step": 6473
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8473097530843544,
      "learning_rate": 1.5670403152873036e-07,
      "loss": 0.6193,
      "step": 6474
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7435850532859277,
      "learning_rate": 1.564823185955665e-07,
      "loss": 0.5124,
      "step": 6475
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7896586640866586,
      "learning_rate": 1.5626074930467615e-07,
      "loss": 0.5009,
      "step": 6476
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6976975356759486,
      "learning_rate": 1.5603932369379002e-07,
      "loss": 0.5352,
      "step": 6477
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8325220740811106,
      "learning_rate": 1.5581804180061487e-07,
      "loss": 0.6602,
      "step": 6478
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9408500210225593,
      "learning_rate": 1.555969036628324e-07,
      "loss": 0.6169,
      "step": 6479
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8963494881602688,
      "learning_rate": 1.553759093181004e-07,
      "loss": 0.646,
      "step": 6480
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9014680548878518,
      "learning_rate": 1.551550588040519e-07,
      "loss": 0.5986,
      "step": 6481
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6472376954796372,
      "learning_rate": 1.549343521582953e-07,
      "loss": 0.4855,
      "step": 6482
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.681807805793271,
      "learning_rate": 1.5471378941841474e-07,
      "loss": 0.5761,
      "step": 6483
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7122307063924425,
      "learning_rate": 1.5449337062196975e-07,
      "loss": 0.5398,
      "step": 6484
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.842584534863215,
      "learning_rate": 1.5427309580649527e-07,
      "loss": 0.5483,
      "step": 6485
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6091861360422927,
      "learning_rate": 1.5405296500950182e-07,
      "loss": 0.4722,
      "step": 6486
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.687036081919355,
      "learning_rate": 1.538329782684754e-07,
      "loss": 0.5438,
      "step": 6487
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8466948947588971,
      "learning_rate": 1.5361313562087764e-07,
      "loss": 0.6259,
      "step": 6488
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6992040618921059,
      "learning_rate": 1.5339343710414508e-07,
      "loss": 0.5537,
      "step": 6489
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8639102897517498,
      "learning_rate": 1.5317388275569044e-07,
      "loss": 0.6229,
      "step": 6490
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.649975156685462,
      "learning_rate": 1.529544726129014e-07,
      "loss": 0.5119,
      "step": 6491
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.755327241357558,
      "learning_rate": 1.527352067131411e-07,
      "loss": 0.5791,
      "step": 6492
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7029121462507827,
      "learning_rate": 1.5251608509374847e-07,
      "loss": 0.5323,
      "step": 6493
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8877752542191736,
      "learning_rate": 1.5229710779203742e-07,
      "loss": 0.6282,
      "step": 6494
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9334873840229414,
      "learning_rate": 1.5207827484529768e-07,
      "loss": 0.6141,
      "step": 6495
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6687585141221396,
      "learning_rate": 1.518595862907942e-07,
      "loss": 0.5169,
      "step": 6496
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.792020114956357,
      "learning_rate": 1.5164104216576723e-07,
      "loss": 0.5918,
      "step": 6497
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7956780566363902,
      "learning_rate": 1.5142264250743265e-07,
      "loss": 0.6191,
      "step": 6498
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8872366557514223,
      "learning_rate": 1.5120438735298135e-07,
      "loss": 0.6191,
      "step": 6499
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6983377532530025,
      "learning_rate": 1.5098627673958054e-07,
      "loss": 0.535,
      "step": 6500
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8204140662095852,
      "learning_rate": 1.507683107043718e-07,
      "loss": 0.5886,
      "step": 6501
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8971501688744877,
      "learning_rate": 1.5055048928447256e-07,
      "loss": 0.6626,
      "step": 6502
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8648254248438829,
      "learning_rate": 1.503328125169754e-07,
      "loss": 0.7023,
      "step": 6503
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8068553009722405,
      "learning_rate": 1.501152804389485e-07,
      "loss": 0.5585,
      "step": 6504
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6309762813601055,
      "learning_rate": 1.498978930874354e-07,
      "loss": 0.6116,
      "step": 6505
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6222580135887993,
      "learning_rate": 1.4968065049945478e-07,
      "loss": 0.4785,
      "step": 6506
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8481866242584207,
      "learning_rate": 1.4946355271200074e-07,
      "loss": 0.6318,
      "step": 6507
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9584963075347593,
      "learning_rate": 1.492465997620429e-07,
      "loss": 0.6556,
      "step": 6508
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6162561214865427,
      "learning_rate": 1.4902979168652607e-07,
      "loss": 0.5706,
      "step": 6509
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7648698987514884,
      "learning_rate": 1.4881312852237026e-07,
      "loss": 0.5659,
      "step": 6510
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7030439748636484,
      "learning_rate": 1.485966103064712e-07,
      "loss": 0.5465,
      "step": 6511
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.667857675456393,
      "learning_rate": 1.4838023707569946e-07,
      "loss": 0.5491,
      "step": 6512
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7100249374472007,
      "learning_rate": 1.4816400886690127e-07,
      "loss": 0.5057,
      "step": 6513
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7514819565522591,
      "learning_rate": 1.4794792571689797e-07,
      "loss": 0.5868,
      "step": 6514
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7257981876811831,
      "learning_rate": 1.477319876624864e-07,
      "loss": 0.5318,
      "step": 6515
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8036673745491099,
      "learning_rate": 1.4751619474043843e-07,
      "loss": 0.6503,
      "step": 6516
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6741845464808413,
      "learning_rate": 1.4730054698750126e-07,
      "loss": 0.5379,
      "step": 6517
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7783101949147241,
      "learning_rate": 1.470850444403977e-07,
      "loss": 0.5397,
      "step": 6518
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8422940959410613,
      "learning_rate": 1.4686968713582549e-07,
      "loss": 0.5795,
      "step": 6519
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.872014422020499,
      "learning_rate": 1.4665447511045758e-07,
      "loss": 0.7028,
      "step": 6520
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7123893456551749,
      "learning_rate": 1.464394084009426e-07,
      "loss": 0.5459,
      "step": 6521
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8817654664448331,
      "learning_rate": 1.46224487043904e-07,
      "loss": 0.6513,
      "step": 6522
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9435918283860236,
      "learning_rate": 1.4600971107594074e-07,
      "loss": 0.6788,
      "step": 6523
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9313283743227634,
      "learning_rate": 1.4579508053362677e-07,
      "loss": 0.6509,
      "step": 6524
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7585167595958451,
      "learning_rate": 1.4558059545351142e-07,
      "loss": 0.585,
      "step": 6525
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7899481298766209,
      "learning_rate": 1.4536625587211947e-07,
      "loss": 0.6273,
      "step": 6526
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8742309760882022,
      "learning_rate": 1.4515206182595073e-07,
      "loss": 0.5328,
      "step": 6527
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7876055280872883,
      "learning_rate": 1.4493801335148015e-07,
      "loss": 0.5582,
      "step": 6528
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.777145513648088,
      "learning_rate": 1.4472411048515786e-07,
      "loss": 0.5226,
      "step": 6529
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.736675628086677,
      "learning_rate": 1.4451035326340943e-07,
      "loss": 0.5286,
      "step": 6530
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7234362508432582,
      "learning_rate": 1.4429674172263538e-07,
      "loss": 0.5426,
      "step": 6531
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.943305132926063,
      "learning_rate": 1.4408327589921155e-07,
      "loss": 0.6382,
      "step": 6532
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7693478465087069,
      "learning_rate": 1.438699558294889e-07,
      "loss": 0.5437,
      "step": 6533
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.925944123753023,
      "learning_rate": 1.4365678154979376e-07,
      "loss": 0.6312,
      "step": 6534
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7165278339926276,
      "learning_rate": 1.4344375309642742e-07,
      "loss": 0.5477,
      "step": 6535
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8518564301080275,
      "learning_rate": 1.4323087050566628e-07,
      "loss": 0.6002,
      "step": 6536
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.982185612662202,
      "learning_rate": 1.4301813381376226e-07,
      "loss": 0.6417,
      "step": 6537
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8347147269926177,
      "learning_rate": 1.4280554305694203e-07,
      "loss": 0.5943,
      "step": 6538
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.814560850972755,
      "learning_rate": 1.4259309827140764e-07,
      "loss": 0.6051,
      "step": 6539
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.113341492585925,
      "learning_rate": 1.4238079949333615e-07,
      "loss": 0.6408,
      "step": 6540
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8238855606112827,
      "learning_rate": 1.4216864675887997e-07,
      "loss": 0.5631,
      "step": 6541
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0253125235743323,
      "learning_rate": 1.4195664010416641e-07,
      "loss": 0.6234,
      "step": 6542
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.649662714718655,
      "learning_rate": 1.4174477956529795e-07,
      "loss": 0.5368,
      "step": 6543
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8377916733057882,
      "learning_rate": 1.415330651783523e-07,
      "loss": 0.5323,
      "step": 6544
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9085764746768034,
      "learning_rate": 1.4132149697938221e-07,
      "loss": 0.6741,
      "step": 6545
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7606017187112979,
      "learning_rate": 1.4111007500441564e-07,
      "loss": 0.5638,
      "step": 6546
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.789828875096549,
      "learning_rate": 1.4089879928945536e-07,
      "loss": 0.6151,
      "step": 6547
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8561631290638471,
      "learning_rate": 1.4068766987047952e-07,
      "loss": 0.6275,
      "step": 6548
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8636281879678054,
      "learning_rate": 1.4047668678344125e-07,
      "loss": 0.6171,
      "step": 6549
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7077571834976603,
      "learning_rate": 1.4026585006426884e-07,
      "loss": 0.5741,
      "step": 6550
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.345115247163805,
      "learning_rate": 1.400551597488654e-07,
      "loss": 0.658,
      "step": 6551
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.63129245352723,
      "learning_rate": 1.3984461587310925e-07,
      "loss": 0.4978,
      "step": 6552
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0860823177194958,
      "learning_rate": 1.3963421847285417e-07,
      "loss": 0.6435,
      "step": 6553
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8203381484315194,
      "learning_rate": 1.394239675839285e-07,
      "loss": 0.6694,
      "step": 6554
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6810333823363631,
      "learning_rate": 1.392138632421358e-07,
      "loss": 0.5704,
      "step": 6555
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7186016607021293,
      "learning_rate": 1.3900390548325447e-07,
      "loss": 0.5295,
      "step": 6556
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6262430513637797,
      "learning_rate": 1.3879409434303825e-07,
      "loss": 0.5101,
      "step": 6557
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9365582301978889,
      "learning_rate": 1.3858442985721586e-07,
      "loss": 0.6707,
      "step": 6558
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6925915340299651,
      "learning_rate": 1.3837491206149088e-07,
      "loss": 0.51,
      "step": 6559
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6342401201483552,
      "learning_rate": 1.3816554099154198e-07,
      "loss": 0.5631,
      "step": 6560
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6756196714090805,
      "learning_rate": 1.37956316683023e-07,
      "loss": 0.5132,
      "step": 6561
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8126883442128703,
      "learning_rate": 1.377472391715625e-07,
      "loss": 0.6157,
      "step": 6562
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8588731426247326,
      "learning_rate": 1.3753830849276428e-07,
      "loss": 0.6847,
      "step": 6563
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.809003457832747,
      "learning_rate": 1.3732952468220716e-07,
      "loss": 0.6098,
      "step": 6564
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6594047518370252,
      "learning_rate": 1.3712088777544462e-07,
      "loss": 0.5286,
      "step": 6565
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8988460711217129,
      "learning_rate": 1.3691239780800557e-07,
      "loss": 0.6244,
      "step": 6566
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7264885947235417,
      "learning_rate": 1.367040548153936e-07,
      "loss": 0.594,
      "step": 6567
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6736899763373626,
      "learning_rate": 1.3649585883308734e-07,
      "loss": 0.5066,
      "step": 6568
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7554583611591057,
      "learning_rate": 1.362878098965403e-07,
      "loss": 0.4949,
      "step": 6569
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6796017835448974,
      "learning_rate": 1.360799080411813e-07,
      "loss": 0.6228,
      "step": 6570
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7803923320094418,
      "learning_rate": 1.3587215330241365e-07,
      "loss": 0.585,
      "step": 6571
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8553407627882796,
      "learning_rate": 1.3566454571561582e-07,
      "loss": 0.6361,
      "step": 6572
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8012576759831905,
      "learning_rate": 1.354570853161413e-07,
      "loss": 0.6001,
      "step": 6573
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.731128933531591,
      "learning_rate": 1.3524977213931843e-07,
      "loss": 0.5697,
      "step": 6574
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7872161738737213,
      "learning_rate": 1.350426062204505e-07,
      "loss": 0.6165,
      "step": 6575
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6933003117401341,
      "learning_rate": 1.3483558759481561e-07,
      "loss": 0.5444,
      "step": 6576
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6910999712129957,
      "learning_rate": 1.3462871629766692e-07,
      "loss": 0.5722,
      "step": 6577
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0973078411389914,
      "learning_rate": 1.3442199236423225e-07,
      "loss": 0.5955,
      "step": 6578
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6472127559721503,
      "learning_rate": 1.3421541582971497e-07,
      "loss": 0.5018,
      "step": 6579
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8270865354894193,
      "learning_rate": 1.3400898672929272e-07,
      "loss": 0.6506,
      "step": 6580
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7976147252189003,
      "learning_rate": 1.338027050981183e-07,
      "loss": 0.5201,
      "step": 6581
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7405648149171674,
      "learning_rate": 1.3359657097131905e-07,
      "loss": 0.5778,
      "step": 6582
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6801264794094015,
      "learning_rate": 1.3339058438399752e-07,
      "loss": 0.5348,
      "step": 6583
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.646340423713667,
      "learning_rate": 1.3318474537123137e-07,
      "loss": 0.567,
      "step": 6584
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7109740501792439,
      "learning_rate": 1.3297905396807273e-07,
      "loss": 0.5903,
      "step": 6585
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0232668221402046,
      "learning_rate": 1.3277351020954864e-07,
      "loss": 0.588,
      "step": 6586
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6259289048602668,
      "learning_rate": 1.3256811413066104e-07,
      "loss": 0.5054,
      "step": 6587
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.791669412735896,
      "learning_rate": 1.3236286576638667e-07,
      "loss": 0.622,
      "step": 6588
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7194127365164742,
      "learning_rate": 1.3215776515167743e-07,
      "loss": 0.5322,
      "step": 6589
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9017032117481077,
      "learning_rate": 1.3195281232145961e-07,
      "loss": 0.6096,
      "step": 6590
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9121874485161451,
      "learning_rate": 1.317480073106345e-07,
      "loss": 0.6584,
      "step": 6591
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7457618736352661,
      "learning_rate": 1.3154335015407847e-07,
      "loss": 0.5421,
      "step": 6592
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9045701643052891,
      "learning_rate": 1.313388408866424e-07,
      "loss": 0.6081,
      "step": 6593
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9323097142890862,
      "learning_rate": 1.3113447954315205e-07,
      "loss": 0.6649,
      "step": 6594
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8931746042883778,
      "learning_rate": 1.3093026615840809e-07,
      "loss": 0.6516,
      "step": 6595
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7415344973990833,
      "learning_rate": 1.307262007671859e-07,
      "loss": 0.5884,
      "step": 6596
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7973027568681745,
      "learning_rate": 1.3052228340423565e-07,
      "loss": 0.5734,
      "step": 6597
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7061783740926056,
      "learning_rate": 1.303185141042824e-07,
      "loss": 0.5592,
      "step": 6598
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7744447772824435,
      "learning_rate": 1.3011489290202593e-07,
      "loss": 0.6237,
      "step": 6599
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6878177373660196,
      "learning_rate": 1.2991141983214072e-07,
      "loss": 0.5688,
      "step": 6600
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7995798396023718,
      "learning_rate": 1.2970809492927626e-07,
      "loss": 0.5606,
      "step": 6601
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.902353775933625,
      "learning_rate": 1.2950491822805665e-07,
      "loss": 0.6286,
      "step": 6602
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7842112889824372,
      "learning_rate": 1.2930188976308065e-07,
      "loss": 0.5938,
      "step": 6603
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1880893590675192,
      "learning_rate": 1.2909900956892195e-07,
      "loss": 0.6885,
      "step": 6604
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7075721410512502,
      "learning_rate": 1.28896277680129e-07,
      "loss": 0.568,
      "step": 6605
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8686764695268824,
      "learning_rate": 1.286936941312249e-07,
      "loss": 0.615,
      "step": 6606
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6797875534778155,
      "learning_rate": 1.2849125895670732e-07,
      "loss": 0.539,
      "step": 6607
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9561620123645299,
      "learning_rate": 1.2828897219104918e-07,
      "loss": 0.6255,
      "step": 6608
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6576725219393901,
      "learning_rate": 1.2808683386869757e-07,
      "loss": 0.5053,
      "step": 6609
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8602697108693139,
      "learning_rate": 1.278848440240744e-07,
      "loss": 0.6381,
      "step": 6610
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7007164457034769,
      "learning_rate": 1.2768300269157694e-07,
      "loss": 0.5356,
      "step": 6611
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6900494935389696,
      "learning_rate": 1.2748130990557638e-07,
      "loss": 0.5185,
      "step": 6612
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9019610861536197,
      "learning_rate": 1.2727976570041888e-07,
      "loss": 0.5171,
      "step": 6613
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9003308091236472,
      "learning_rate": 1.2707837011042545e-07,
      "loss": 0.6327,
      "step": 6614
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6655901590600044,
      "learning_rate": 1.268771231698914e-07,
      "loss": 0.524,
      "step": 6615
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6637540631482273,
      "learning_rate": 1.2667602491308727e-07,
      "loss": 0.5023,
      "step": 6616
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8584162978628107,
      "learning_rate": 1.2647507537425784e-07,
      "loss": 0.6208,
      "step": 6617
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.9482405812779877,
      "learning_rate": 1.2627427458762275e-07,
      "loss": 0.6399,
      "step": 6618
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8778367638183109,
      "learning_rate": 1.260736225873762e-07,
      "loss": 0.6317,
      "step": 6619
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7334119359015416,
      "learning_rate": 1.258731194076873e-07,
      "loss": 0.577,
      "step": 6620
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8178849640619678,
      "learning_rate": 1.2567276508269965e-07,
      "loss": 0.6157,
      "step": 6621
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7374329202968979,
      "learning_rate": 1.2547255964653137e-07,
      "loss": 0.5201,
      "step": 6622
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7011504105362326,
      "learning_rate": 1.2527250313327543e-07,
      "loss": 0.5415,
      "step": 6623
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7193264607394868,
      "learning_rate": 1.2507259557699933e-07,
      "loss": 0.5295,
      "step": 6624
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7737404606027619,
      "learning_rate": 1.2487283701174522e-07,
      "loss": 0.549,
      "step": 6625
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7787757083245738,
      "learning_rate": 1.2467322747153008e-07,
      "loss": 0.6091,
      "step": 6626
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7599066282218115,
      "learning_rate": 1.2447376699034518e-07,
      "loss": 0.5127,
      "step": 6627
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6251734130236115,
      "learning_rate": 1.2427445560215654e-07,
      "loss": 0.5376,
      "step": 6628
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9058772093256446,
      "learning_rate": 1.2407529334090494e-07,
      "loss": 0.6395,
      "step": 6629
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6487502377371414,
      "learning_rate": 1.2387628024050556e-07,
      "loss": 0.551,
      "step": 6630
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7243707741245649,
      "learning_rate": 1.236774163348482e-07,
      "loss": 0.51,
      "step": 6631
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8853909691891717,
      "learning_rate": 1.2347870165779739e-07,
      "loss": 0.6221,
      "step": 6632
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6470700735591604,
      "learning_rate": 1.2328013624319223e-07,
      "loss": 0.482,
      "step": 6633
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5932581835428871,
      "learning_rate": 1.2308172012484618e-07,
      "loss": 0.4663,
      "step": 6634
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7669032840159081,
      "learning_rate": 1.2288345333654758e-07,
      "loss": 0.5212,
      "step": 6635
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8486069859351337,
      "learning_rate": 1.2268533591205889e-07,
      "loss": 0.6351,
      "step": 6636
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7400250037729184,
      "learning_rate": 1.2248736788511794e-07,
      "loss": 0.5091,
      "step": 6637
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8382119795140021,
      "learning_rate": 1.2228954928943634e-07,
      "loss": 0.6198,
      "step": 6638
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.889488081623315,
      "learning_rate": 1.2209188015870064e-07,
      "loss": 0.6795,
      "step": 6639
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.953436924447045,
      "learning_rate": 1.2189436052657164e-07,
      "loss": 0.5815,
      "step": 6640
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7016326675850533,
      "learning_rate": 1.2169699042668514e-07,
      "loss": 0.5207,
      "step": 6641
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7439956264510875,
      "learning_rate": 1.2149976989265097e-07,
      "loss": 0.5763,
      "step": 6642
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9074550608377534,
      "learning_rate": 1.213026989580539e-07,
      "loss": 0.6486,
      "step": 6643
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8841208838313453,
      "learning_rate": 1.2110577765645303e-07,
      "loss": 0.6226,
      "step": 6644
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7875208315192488,
      "learning_rate": 1.2090900602138198e-07,
      "loss": 0.4861,
      "step": 6645
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7897848812241941,
      "learning_rate": 1.2071238408634875e-07,
      "loss": 0.5597,
      "step": 6646
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6734837743998734,
      "learning_rate": 1.2051591188483633e-07,
      "loss": 0.5519,
      "step": 6647
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6516856111343156,
      "learning_rate": 1.2031958945030164e-07,
      "loss": 0.5175,
      "step": 6648
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9139426363160901,
      "learning_rate": 1.2012341681617643e-07,
      "loss": 0.6112,
      "step": 6649
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8736849144145844,
      "learning_rate": 1.1992739401586682e-07,
      "loss": 0.6887,
      "step": 6650
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6141920765886194,
      "learning_rate": 1.1973152108275352e-07,
      "loss": 0.5306,
      "step": 6651
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.954838920264758,
      "learning_rate": 1.1953579805019165e-07,
      "loss": 0.6632,
      "step": 6652
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0271555591763564,
      "learning_rate": 1.1934022495151065e-07,
      "loss": 0.6309,
      "step": 6653
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6437111020352964,
      "learning_rate": 1.1914480182001475e-07,
      "loss": 0.5644,
      "step": 6654
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6737005848950873,
      "learning_rate": 1.1894952868898245e-07,
      "loss": 0.5591,
      "step": 6655
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6195736235668116,
      "learning_rate": 1.1875440559166672e-07,
      "loss": 0.513,
      "step": 6656
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0443924884149405,
      "learning_rate": 1.1855943256129485e-07,
      "loss": 0.577,
      "step": 6657
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9480243516281474,
      "learning_rate": 1.1836460963106887e-07,
      "loss": 0.7036,
      "step": 6658
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8349313421226502,
      "learning_rate": 1.181699368341651e-07,
      "loss": 0.6673,
      "step": 6659
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6178351783855008,
      "learning_rate": 1.1797541420373424e-07,
      "loss": 0.5053,
      "step": 6660
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.839456525068892,
      "learning_rate": 1.1778104177290149e-07,
      "loss": 0.5853,
      "step": 6661
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0201704392513333,
      "learning_rate": 1.1758681957476623e-07,
      "loss": 0.5908,
      "step": 6662
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8626420618788531,
      "learning_rate": 1.1739274764240282e-07,
      "loss": 0.5979,
      "step": 6663
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7453625560484749,
      "learning_rate": 1.1719882600885967e-07,
      "loss": 0.5746,
      "step": 6664
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6674614786355566,
      "learning_rate": 1.1700505470715938e-07,
      "loss": 0.5521,
      "step": 6665
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7203034476137303,
      "learning_rate": 1.1681143377029934e-07,
      "loss": 0.5599,
      "step": 6666
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9124766647135,
      "learning_rate": 1.1661796323125117e-07,
      "loss": 0.625,
      "step": 6667
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7487601393714207,
      "learning_rate": 1.1642464312296074e-07,
      "loss": 0.5108,
      "step": 6668
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6626124685098754,
      "learning_rate": 1.1623147347834871e-07,
      "loss": 0.5573,
      "step": 6669
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7639055273102708,
      "learning_rate": 1.1603845433030967e-07,
      "loss": 0.5677,
      "step": 6670
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7295418202806963,
      "learning_rate": 1.1584558571171277e-07,
      "loss": 0.5922,
      "step": 6671
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5666244146704841,
      "learning_rate": 1.1565286765540161e-07,
      "loss": 0.4978,
      "step": 6672
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6774337091699196,
      "learning_rate": 1.1546030019419396e-07,
      "loss": 0.5196,
      "step": 6673
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8641139739513589,
      "learning_rate": 1.1526788336088222e-07,
      "loss": 0.5155,
      "step": 6674
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8363027669486846,
      "learning_rate": 1.1507561718823277e-07,
      "loss": 0.6075,
      "step": 6675
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8787427159601146,
      "learning_rate": 1.1488350170898675e-07,
      "loss": 0.6419,
      "step": 6676
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8447805169335512,
      "learning_rate": 1.146915369558592e-07,
      "loss": 0.6523,
      "step": 6677
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8384500708112046,
      "learning_rate": 1.1449972296153987e-07,
      "loss": 0.6211,
      "step": 6678
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8155321378313181,
      "learning_rate": 1.1430805975869262e-07,
      "loss": 0.6251,
      "step": 6679
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.911429919984612,
      "learning_rate": 1.1411654737995579e-07,
      "loss": 0.5734,
      "step": 6680
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.746567988582566,
      "learning_rate": 1.1392518585794186e-07,
      "loss": 0.5465,
      "step": 6681
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6976705153159293,
      "learning_rate": 1.1373397522523742e-07,
      "loss": 0.5425,
      "step": 6682
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.712378668253382,
      "learning_rate": 1.1354291551440431e-07,
      "loss": 0.5517,
      "step": 6683
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.9491721928659422,
      "learning_rate": 1.1335200675797751e-07,
      "loss": 0.6098,
      "step": 6684
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9076368822175975,
      "learning_rate": 1.1316124898846702e-07,
      "loss": 0.5922,
      "step": 6685
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7155415984720567,
      "learning_rate": 1.1297064223835672e-07,
      "loss": 0.5271,
      "step": 6686
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7594468891339862,
      "learning_rate": 1.1278018654010513e-07,
      "loss": 0.5371,
      "step": 6687
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0190801826453102,
      "learning_rate": 1.1258988192614471e-07,
      "loss": 0.5833,
      "step": 6688
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7339736667083511,
      "learning_rate": 1.1239972842888246e-07,
      "loss": 0.5326,
      "step": 6689
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8338657461801464,
      "learning_rate": 1.1220972608069946e-07,
      "loss": 0.6,
      "step": 6690
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7041980556719474,
      "learning_rate": 1.1201987491395126e-07,
      "loss": 0.5693,
      "step": 6691
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8706531457370236,
      "learning_rate": 1.1183017496096736e-07,
      "loss": 0.6453,
      "step": 6692
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6890933788648442,
      "learning_rate": 1.1164062625405191e-07,
      "loss": 0.5388,
      "step": 6693
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8628548936245998,
      "learning_rate": 1.1145122882548263e-07,
      "loss": 0.6328,
      "step": 6694
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7744261864043575,
      "learning_rate": 1.1126198270751241e-07,
      "loss": 0.5908,
      "step": 6695
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7732611108477915,
      "learning_rate": 1.110728879323678e-07,
      "loss": 0.5533,
      "step": 6696
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6085845797745635,
      "learning_rate": 1.108839445322497e-07,
      "loss": 0.5499,
      "step": 6697
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8152828133194631,
      "learning_rate": 1.1069515253933292e-07,
      "loss": 0.5925,
      "step": 6698
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8523757693168615,
      "learning_rate": 1.1050651198576711e-07,
      "loss": 0.6044,
      "step": 6699
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9048013996429554,
      "learning_rate": 1.1031802290367553e-07,
      "loss": 0.6124,
      "step": 6700
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6768827886691475,
      "learning_rate": 1.1012968532515598e-07,
      "loss": 0.5274,
      "step": 6701
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7129570815656142,
      "learning_rate": 1.0994149928228047e-07,
      "loss": 0.5526,
      "step": 6702
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.680973949174689,
      "learning_rate": 1.0975346480709512e-07,
      "loss": 0.5859,
      "step": 6703
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8397369873679229,
      "learning_rate": 1.0956558193162014e-07,
      "loss": 0.6224,
      "step": 6704
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9432661576334819,
      "learning_rate": 1.093778506878501e-07,
      "loss": 0.6018,
      "step": 6705
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6928328038403601,
      "learning_rate": 1.0919027110775358e-07,
      "loss": 0.4614,
      "step": 6706
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9027244651066589,
      "learning_rate": 1.0900284322327358e-07,
      "loss": 0.6275,
      "step": 6707
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8266825092053888,
      "learning_rate": 1.08815567066327e-07,
      "loss": 0.5362,
      "step": 6708
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8748823078427032,
      "learning_rate": 1.086284426688051e-07,
      "loss": 0.6195,
      "step": 6709
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6129085600296452,
      "learning_rate": 1.084414700625731e-07,
      "loss": 0.5331,
      "step": 6710
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9133326556204208,
      "learning_rate": 1.0825464927947059e-07,
      "loss": 0.6493,
      "step": 6711
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7360747300532534,
      "learning_rate": 1.0806798035131126e-07,
      "loss": 0.5375,
      "step": 6712
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8988102722728267,
      "learning_rate": 1.0788146330988269e-07,
      "loss": 0.651,
      "step": 6713
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7607042658100939,
      "learning_rate": 1.0769509818694689e-07,
      "loss": 0.5714,
      "step": 6714
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8142816991598115,
      "learning_rate": 1.0750888501423994e-07,
      "loss": 0.5754,
      "step": 6715
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6766259696647785,
      "learning_rate": 1.0732282382347202e-07,
      "loss": 0.517,
      "step": 6716
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9160529022852956,
      "learning_rate": 1.0713691464632735e-07,
      "loss": 0.6545,
      "step": 6717
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8144254014278861,
      "learning_rate": 1.0695115751446426e-07,
      "loss": 0.5743,
      "step": 6718
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7499536822694304,
      "learning_rate": 1.0676555245951535e-07,
      "loss": 0.5662,
      "step": 6719
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8724351031869012,
      "learning_rate": 1.0658009951308722e-07,
      "loss": 0.6212,
      "step": 6720
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9128834690329203,
      "learning_rate": 1.0639479870676038e-07,
      "loss": 0.6287,
      "step": 6721
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6103191872658422,
      "learning_rate": 1.0620965007208993e-07,
      "loss": 0.4862,
      "step": 6722
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6551613737967841,
      "learning_rate": 1.0602465364060475e-07,
      "loss": 0.5272,
      "step": 6723
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8371605329167151,
      "learning_rate": 1.0583980944380755e-07,
      "loss": 0.5837,
      "step": 6724
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7277728065936244,
      "learning_rate": 1.0565511751317558e-07,
      "loss": 0.529,
      "step": 6725
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7643517966513762,
      "learning_rate": 1.054705778801599e-07,
      "loss": 0.5615,
      "step": 6726
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9436863124961647,
      "learning_rate": 1.0528619057618571e-07,
      "loss": 0.6281,
      "step": 6727
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.828605408227628,
      "learning_rate": 1.0510195563265212e-07,
      "loss": 0.5317,
      "step": 6728
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7401904632986045,
      "learning_rate": 1.0491787308093248e-07,
      "loss": 0.5529,
      "step": 6729
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7270841622546429,
      "learning_rate": 1.0473394295237426e-07,
      "loss": 0.5062,
      "step": 6730
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7055885270528672,
      "learning_rate": 1.0455016527829863e-07,
      "loss": 0.5582,
      "step": 6731
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7020935718856278,
      "learning_rate": 1.0436654009000123e-07,
      "loss": 0.544,
      "step": 6732
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6334362518632649,
      "learning_rate": 1.0418306741875138e-07,
      "loss": 0.5319,
      "step": 6733
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6615210119243968,
      "learning_rate": 1.0399974729579253e-07,
      "loss": 0.4932,
      "step": 6734
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0173080969906356,
      "learning_rate": 1.0381657975234237e-07,
      "loss": 0.5955,
      "step": 6735
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0273283002421458,
      "learning_rate": 1.0363356481959218e-07,
      "loss": 0.6401,
      "step": 6736
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8691917665086982,
      "learning_rate": 1.0345070252870769e-07,
      "loss": 0.6366,
      "step": 6737
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6528316075955349,
      "learning_rate": 1.0326799291082832e-07,
      "loss": 0.5443,
      "step": 6738
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.009417440100602,
      "learning_rate": 1.0308543599706776e-07,
      "loss": 0.5954,
      "step": 6739
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7135821122723486,
      "learning_rate": 1.0290303181851345e-07,
      "loss": 0.5412,
      "step": 6740
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6629733017127923,
      "learning_rate": 1.027207804062269e-07,
      "loss": 0.5674,
      "step": 6741
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8109402686849455,
      "learning_rate": 1.025386817912437e-07,
      "loss": 0.6101,
      "step": 6742
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8884085208743441,
      "learning_rate": 1.0235673600457317e-07,
      "loss": 0.5541,
      "step": 6743
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7224869856825186,
      "learning_rate": 1.0217494307719909e-07,
      "loss": 0.5261,
      "step": 6744
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.839018281235952,
      "learning_rate": 1.0199330304007859e-07,
      "loss": 0.5994,
      "step": 6745
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6687246683527613,
      "learning_rate": 1.0181181592414334e-07,
      "loss": 0.464,
      "step": 6746
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8836726903950467,
      "learning_rate": 1.0163048176029831e-07,
      "loss": 0.6691,
      "step": 6747
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8297961842309421,
      "learning_rate": 1.0144930057942336e-07,
      "loss": 0.646,
      "step": 6748
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6105724368319058,
      "learning_rate": 1.0126827241237134e-07,
      "loss": 0.5387,
      "step": 6749
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.849690045802289,
      "learning_rate": 1.0108739728996973e-07,
      "loss": 0.6008,
      "step": 6750
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7515886197845002,
      "learning_rate": 1.0090667524301954e-07,
      "loss": 0.5837,
      "step": 6751
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6302776493671874,
      "learning_rate": 1.0072610630229583e-07,
      "loss": 0.5255,
      "step": 6752
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6942068817321367,
      "learning_rate": 1.0054569049854766e-07,
      "loss": 0.5685,
      "step": 6753
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8936073299368944,
      "learning_rate": 1.0036542786249802e-07,
      "loss": 0.6345,
      "step": 6754
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.153845869124329,
      "learning_rate": 1.0018531842484368e-07,
      "loss": 0.6331,
      "step": 6755
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8380349440536838,
      "learning_rate": 1.0000536221625532e-07,
      "loss": 0.6178,
      "step": 6756
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7862464877506053,
      "learning_rate": 9.98255592673778e-08,
      "loss": 0.6437,
      "step": 6757
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9286609372073289,
      "learning_rate": 9.964590960882946e-08,
      "loss": 0.6507,
      "step": 6758
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6256742970285631,
      "learning_rate": 9.946641327120297e-08,
      "loss": 0.5031,
      "step": 6759
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6209429785739257,
      "learning_rate": 9.928707028506466e-08,
      "loss": 0.5255,
      "step": 6760
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8562337446806013,
      "learning_rate": 9.910788068095466e-08,
      "loss": 0.5645,
      "step": 6761
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9810535114074306,
      "learning_rate": 9.892884448938709e-08,
      "loss": 0.5959,
      "step": 6762
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8862165087264969,
      "learning_rate": 9.874996174085015e-08,
      "loss": 0.5996,
      "step": 6763
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8918420413990427,
      "learning_rate": 9.857123246580545e-08,
      "loss": 0.6678,
      "step": 6764
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7868093801462644,
      "learning_rate": 9.839265669468899e-08,
      "loss": 0.6026,
      "step": 6765
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8363795999689583,
      "learning_rate": 9.821423445791011e-08,
      "loss": 0.6248,
      "step": 6766
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6860195080010946,
      "learning_rate": 9.803596578585238e-08,
      "loss": 0.5215,
      "step": 6767
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.854664158450517,
      "learning_rate": 9.785785070887309e-08,
      "loss": 0.6111,
      "step": 6768
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7564539389942752,
      "learning_rate": 9.767988925730331e-08,
      "loss": 0.4769,
      "step": 6769
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7328477552225422,
      "learning_rate": 9.750208146144822e-08,
      "loss": 0.5716,
      "step": 6770
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8957478061560773,
      "learning_rate": 9.732442735158641e-08,
      "loss": 0.6395,
      "step": 6771
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6577035334055743,
      "learning_rate": 9.714692695797066e-08,
      "loss": 0.5609,
      "step": 6772
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5869134142808156,
      "learning_rate": 9.696958031082714e-08,
      "loss": 0.5027,
      "step": 6773
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7181504617935077,
      "learning_rate": 9.679238744035656e-08,
      "loss": 0.5586,
      "step": 6774
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8398668452246594,
      "learning_rate": 9.661534837673291e-08,
      "loss": 0.5303,
      "step": 6775
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8391012936346779,
      "learning_rate": 9.643846315010406e-08,
      "loss": 0.6083,
      "step": 6776
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8112297265116198,
      "learning_rate": 9.626173179059171e-08,
      "loss": 0.6048,
      "step": 6777
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8721029422727922,
      "learning_rate": 9.608515432829134e-08,
      "loss": 0.6883,
      "step": 6778
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7728073598942554,
      "learning_rate": 9.590873079327222e-08,
      "loss": 0.6255,
      "step": 6779
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8533586624456085,
      "learning_rate": 9.573246121557754e-08,
      "loss": 0.5756,
      "step": 6780
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6284498892168462,
      "learning_rate": 9.555634562522408e-08,
      "loss": 0.5617,
      "step": 6781
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9408383467899323,
      "learning_rate": 9.538038405220262e-08,
      "loss": 0.6014,
      "step": 6782
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7705686234462577,
      "learning_rate": 9.52045765264774e-08,
      "loss": 0.5751,
      "step": 6783
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8099089307706886,
      "learning_rate": 9.50289230779866e-08,
      "loss": 0.6579,
      "step": 6784
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6284434496973932,
      "learning_rate": 9.485342373664241e-08,
      "loss": 0.5404,
      "step": 6785
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6383037304909641,
      "learning_rate": 9.467807853233046e-08,
      "loss": 0.5221,
      "step": 6786
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7100400672729043,
      "learning_rate": 9.450288749491008e-08,
      "loss": 0.604,
      "step": 6787
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7752669299867923,
      "learning_rate": 9.432785065421467e-08,
      "loss": 0.5676,
      "step": 6788
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6708949231566096,
      "learning_rate": 9.415296804005101e-08,
      "loss": 0.5785,
      "step": 6789
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6875327835454713,
      "learning_rate": 9.397823968219976e-08,
      "loss": 0.4991,
      "step": 6790
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7659104269082397,
      "learning_rate": 9.380366561041552e-08,
      "loss": 0.5743,
      "step": 6791
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5971162799968365,
      "learning_rate": 9.362924585442633e-08,
      "loss": 0.5374,
      "step": 6792
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6941701316312453,
      "learning_rate": 9.345498044393408e-08,
      "loss": 0.5088,
      "step": 6793
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7270743750489037,
      "learning_rate": 9.328086940861435e-08,
      "loss": 0.5094,
      "step": 6794
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7172624570677782,
      "learning_rate": 9.310691277811644e-08,
      "loss": 0.5975,
      "step": 6795
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6851460501216999,
      "learning_rate": 9.293311058206332e-08,
      "loss": 0.5365,
      "step": 6796
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9029254320862996,
      "learning_rate": 9.275946285005165e-08,
      "loss": 0.574,
      "step": 6797
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6902533354083347,
      "learning_rate": 9.258596961165199e-08,
      "loss": 0.5551,
      "step": 6798
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8687318222840329,
      "learning_rate": 9.241263089640827e-08,
      "loss": 0.5519,
      "step": 6799
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8312866618744909,
      "learning_rate": 9.223944673383832e-08,
      "loss": 0.6318,
      "step": 6800
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7180592716626668,
      "learning_rate": 9.206641715343367e-08,
      "loss": 0.5728,
      "step": 6801
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.672076819008067,
      "learning_rate": 9.189354218465938e-08,
      "loss": 0.5678,
      "step": 6802
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1059381931601633,
      "learning_rate": 9.172082185695429e-08,
      "loss": 0.6648,
      "step": 6803
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7191162566734296,
      "learning_rate": 9.154825619973083e-08,
      "loss": 0.5593,
      "step": 6804
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9737008990219779,
      "learning_rate": 9.137584524237496e-08,
      "loss": 0.6572,
      "step": 6805
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6693208582118326,
      "learning_rate": 9.120358901424685e-08,
      "loss": 0.5508,
      "step": 6806
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.684606707735644,
      "learning_rate": 9.10314875446797e-08,
      "loss": 0.5768,
      "step": 6807
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6354787066206922,
      "learning_rate": 9.085954086298076e-08,
      "loss": 0.5536,
      "step": 6808
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.832107147400525,
      "learning_rate": 9.068774899843057e-08,
      "loss": 0.5872,
      "step": 6809
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.733975620543821,
      "learning_rate": 9.051611198028364e-08,
      "loss": 0.5637,
      "step": 6810
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8387209005960767,
      "learning_rate": 9.034462983776791e-08,
      "loss": 0.6311,
      "step": 6811
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7780183170063538,
      "learning_rate": 9.017330260008493e-08,
      "loss": 0.573,
      "step": 6812
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7907845579268973,
      "learning_rate": 9.000213029641012e-08,
      "loss": 0.6162,
      "step": 6813
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6715061782753139,
      "learning_rate": 8.983111295589218e-08,
      "loss": 0.5145,
      "step": 6814
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6940391696848187,
      "learning_rate": 8.966025060765359e-08,
      "loss": 0.4955,
      "step": 6815
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7961286683178959,
      "learning_rate": 8.948954328079051e-08,
      "loss": 0.6167,
      "step": 6816
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7465720521570403,
      "learning_rate": 8.931899100437257e-08,
      "loss": 0.608,
      "step": 6817
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7664232984714339,
      "learning_rate": 8.914859380744311e-08,
      "loss": 0.5918,
      "step": 6818
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8825795416165956,
      "learning_rate": 8.89783517190189e-08,
      "loss": 0.6424,
      "step": 6819
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9360786729769102,
      "learning_rate": 8.880826476809034e-08,
      "loss": 0.5909,
      "step": 6820
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7633767306266623,
      "learning_rate": 8.863833298362167e-08,
      "loss": 0.5401,
      "step": 6821
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9749441124753508,
      "learning_rate": 8.846855639455042e-08,
      "loss": 0.6215,
      "step": 6822
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.867090895690543,
      "learning_rate": 8.829893502978769e-08,
      "loss": 0.6568,
      "step": 6823
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7151972195190742,
      "learning_rate": 8.812946891821838e-08,
      "loss": 0.5422,
      "step": 6824
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7712009868711653,
      "learning_rate": 8.796015808870061e-08,
      "loss": 0.6041,
      "step": 6825
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6215407183641615,
      "learning_rate": 8.779100257006633e-08,
      "loss": 0.5468,
      "step": 6826
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6780762773168175,
      "learning_rate": 8.762200239112105e-08,
      "loss": 0.5593,
      "step": 6827
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7676743870246849,
      "learning_rate": 8.745315758064375e-08,
      "loss": 0.5213,
      "step": 6828
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8277305592600002,
      "learning_rate": 8.728446816738677e-08,
      "loss": 0.5945,
      "step": 6829
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8501984783227978,
      "learning_rate": 8.711593418007624e-08,
      "loss": 0.6046,
      "step": 6830
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8214772176370492,
      "learning_rate": 8.694755564741185e-08,
      "loss": 0.5359,
      "step": 6831
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9015815903611316,
      "learning_rate": 8.677933259806635e-08,
      "loss": 0.601,
      "step": 6832
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7653392459857699,
      "learning_rate": 8.661126506068695e-08,
      "loss": 0.6486,
      "step": 6833
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.820007074501415,
      "learning_rate": 8.644335306389339e-08,
      "loss": 0.5817,
      "step": 6834
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6975463357538378,
      "learning_rate": 8.627559663627948e-08,
      "loss": 0.5298,
      "step": 6835
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6850010282215753,
      "learning_rate": 8.610799580641237e-08,
      "loss": 0.5475,
      "step": 6836
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7917599700550572,
      "learning_rate": 8.594055060283267e-08,
      "loss": 0.5761,
      "step": 6837
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.21340983869454,
      "learning_rate": 8.577326105405458e-08,
      "loss": 0.6429,
      "step": 6838
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7849072325192115,
      "learning_rate": 8.560612718856586e-08,
      "loss": 0.565,
      "step": 6839
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7349181219805981,
      "learning_rate": 8.543914903482752e-08,
      "loss": 0.5591,
      "step": 6840
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7777222731112763,
      "learning_rate": 8.527232662127426e-08,
      "loss": 0.5123,
      "step": 6841
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8321514468362095,
      "learning_rate": 8.510565997631425e-08,
      "loss": 0.5706,
      "step": 6842
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0468228365570798,
      "learning_rate": 8.49391491283291e-08,
      "loss": 0.6007,
      "step": 6843
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8740319734554856,
      "learning_rate": 8.477279410567384e-08,
      "loss": 0.569,
      "step": 6844
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8861879983603689,
      "learning_rate": 8.460659493667687e-08,
      "loss": 0.662,
      "step": 6845
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9551669969880263,
      "learning_rate": 8.444055164964025e-08,
      "loss": 0.6242,
      "step": 6846
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6677460159827141,
      "learning_rate": 8.427466427283957e-08,
      "loss": 0.5507,
      "step": 6847
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8121108930165598,
      "learning_rate": 8.410893283452359e-08,
      "loss": 0.5806,
      "step": 6848
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.703116842796417,
      "learning_rate": 8.394335736291459e-08,
      "loss": 0.592,
      "step": 6849
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8266183317464724,
      "learning_rate": 8.377793788620847e-08,
      "loss": 0.6302,
      "step": 6850
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8936519535434424,
      "learning_rate": 8.361267443257448e-08,
      "loss": 0.5631,
      "step": 6851
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6829197600005379,
      "learning_rate": 8.34475670301551e-08,
      "loss": 0.5748,
      "step": 6852
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8986446533029359,
      "learning_rate": 8.32826157070664e-08,
      "loss": 0.687,
      "step": 6853
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8186926384201073,
      "learning_rate": 8.311782049139804e-08,
      "loss": 0.6157,
      "step": 6854
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.64727726731569,
      "learning_rate": 8.295318141121288e-08,
      "loss": 0.5528,
      "step": 6855
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7666734035984534,
      "learning_rate": 8.278869849454717e-08,
      "loss": 0.5799,
      "step": 6856
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7176478842717808,
      "learning_rate": 8.262437176941062e-08,
      "loss": 0.5072,
      "step": 6857
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6066560301355026,
      "learning_rate": 8.246020126378628e-08,
      "loss": 0.5247,
      "step": 6858
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7680876047187593,
      "learning_rate": 8.22961870056309e-08,
      "loss": 0.5427,
      "step": 6859
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8654548931119209,
      "learning_rate": 8.213232902287438e-08,
      "loss": 0.588,
      "step": 6860
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.870994139970731,
      "learning_rate": 8.196862734341992e-08,
      "loss": 0.5837,
      "step": 6861
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7705763922204295,
      "learning_rate": 8.180508199514424e-08,
      "loss": 0.5334,
      "step": 6862
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7066147013413556,
      "learning_rate": 8.164169300589752e-08,
      "loss": 0.5706,
      "step": 6863
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6249259219526876,
      "learning_rate": 8.147846040350292e-08,
      "loss": 0.5289,
      "step": 6864
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8670330036495717,
      "learning_rate": 8.131538421575757e-08,
      "loss": 0.5499,
      "step": 6865
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7392985231468201,
      "learning_rate": 8.115246447043144e-08,
      "loss": 0.6194,
      "step": 6866
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7235977291327412,
      "learning_rate": 8.098970119526816e-08,
      "loss": 0.5477,
      "step": 6867
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6186889907820781,
      "learning_rate": 8.082709441798452e-08,
      "loss": 0.4982,
      "step": 6868
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6065518503945571,
      "learning_rate": 8.066464416627084e-08,
      "loss": 0.5618,
      "step": 6869
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8429031935760314,
      "learning_rate": 8.050235046779074e-08,
      "loss": 0.6179,
      "step": 6870
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1464909020852825,
      "learning_rate": 8.034021335018104e-08,
      "loss": 0.6275,
      "step": 6871
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6162559446109809,
      "learning_rate": 8.017823284105207e-08,
      "loss": 0.49,
      "step": 6872
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0452705994828997,
      "learning_rate": 8.001640896798744e-08,
      "loss": 0.6896,
      "step": 6873
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8626459048987661,
      "learning_rate": 7.985474175854401e-08,
      "loss": 0.6274,
      "step": 6874
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8007173766631358,
      "learning_rate": 7.969323124025207e-08,
      "loss": 0.5676,
      "step": 6875
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8178685703215661,
      "learning_rate": 7.953187744061517e-08,
      "loss": 0.6273,
      "step": 6876
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.794724527709687,
      "learning_rate": 7.937068038711025e-08,
      "loss": 0.5413,
      "step": 6877
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7981443349812419,
      "learning_rate": 7.920964010718733e-08,
      "loss": 0.6435,
      "step": 6878
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8169239414073827,
      "learning_rate": 7.904875662826993e-08,
      "loss": 0.5827,
      "step": 6879
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8469438538860539,
      "learning_rate": 7.888802997775501e-08,
      "loss": 0.5744,
      "step": 6880
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9210301163912626,
      "learning_rate": 7.872746018301235e-08,
      "loss": 0.6209,
      "step": 6881
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7361719091158877,
      "learning_rate": 7.856704727138563e-08,
      "loss": 0.5594,
      "step": 6882
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.866455375895114,
      "learning_rate": 7.840679127019123e-08,
      "loss": 0.6465,
      "step": 6883
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6579234938884513,
      "learning_rate": 7.824669220671909e-08,
      "loss": 0.5257,
      "step": 6884
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9855475943417021,
      "learning_rate": 7.808675010823262e-08,
      "loss": 0.6268,
      "step": 6885
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6715333014422652,
      "learning_rate": 7.792696500196816e-08,
      "loss": 0.5347,
      "step": 6886
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.715290090561249,
      "learning_rate": 7.77673369151356e-08,
      "loss": 0.5348,
      "step": 6887
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9966692315370363,
      "learning_rate": 7.760786587491763e-08,
      "loss": 0.6532,
      "step": 6888
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6739805185981467,
      "learning_rate": 7.744855190847089e-08,
      "loss": 0.5344,
      "step": 6889
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1403019993341226,
      "learning_rate": 7.728939504292442e-08,
      "loss": 0.6209,
      "step": 6890
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8204945819940481,
      "learning_rate": 7.713039530538135e-08,
      "loss": 0.6173,
      "step": 6891
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7066961701393989,
      "learning_rate": 7.697155272291767e-08,
      "loss": 0.5282,
      "step": 6892
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3334245427667666,
      "learning_rate": 7.681286732258251e-08,
      "loss": 0.636,
      "step": 6893
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8053545368900017,
      "learning_rate": 7.66543391313983e-08,
      "loss": 0.6726,
      "step": 6894
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.729481934191792,
      "learning_rate": 7.649596817636084e-08,
      "loss": 0.5351,
      "step": 6895
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6659848902818687,
      "learning_rate": 7.633775448443903e-08,
      "loss": 0.5576,
      "step": 6896
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9400457311828868,
      "learning_rate": 7.617969808257485e-08,
      "loss": 0.6463,
      "step": 6897
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9545068535770227,
      "learning_rate": 7.602179899768391e-08,
      "loss": 0.5893,
      "step": 6898
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.981148363417466,
      "learning_rate": 7.586405725665468e-08,
      "loss": 0.6496,
      "step": 6899
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.032086894043614,
      "learning_rate": 7.57064728863489e-08,
      "loss": 0.6571,
      "step": 6900
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.786972767534941,
      "learning_rate": 7.554904591360156e-08,
      "loss": 0.6192,
      "step": 6901
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0593063026666267,
      "learning_rate": 7.539177636522077e-08,
      "loss": 0.6161,
      "step": 6902
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8062060563216312,
      "learning_rate": 7.523466426798808e-08,
      "loss": 0.5461,
      "step": 6903
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6133803446117289,
      "learning_rate": 7.507770964865789e-08,
      "loss": 0.5395,
      "step": 6904
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6903782765967106,
      "learning_rate": 7.492091253395793e-08,
      "loss": 0.5662,
      "step": 6905
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6716531382689335,
      "learning_rate": 7.476427295058918e-08,
      "loss": 0.4571,
      "step": 6906
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7151594756951869,
      "learning_rate": 7.460779092522573e-08,
      "loss": 0.5715,
      "step": 6907
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9219911491229557,
      "learning_rate": 7.445146648451483e-08,
      "loss": 0.6418,
      "step": 6908
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8375210506370196,
      "learning_rate": 7.429529965507686e-08,
      "loss": 0.6223,
      "step": 6909
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9321159735628902,
      "learning_rate": 7.413929046350542e-08,
      "loss": 0.6681,
      "step": 6910
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8602580715954953,
      "learning_rate": 7.398343893636738e-08,
      "loss": 0.6061,
      "step": 6911
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6394954138442802,
      "learning_rate": 7.38277451002024e-08,
      "loss": 0.5277,
      "step": 6912
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.66555279040408,
      "learning_rate": 7.367220898152371e-08,
      "loss": 0.5411,
      "step": 6913
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9494891599745592,
      "learning_rate": 7.351683060681734e-08,
      "loss": 0.6659,
      "step": 6914
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9262798418832864,
      "learning_rate": 7.336161000254281e-08,
      "loss": 0.6533,
      "step": 6915
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9229669373747947,
      "learning_rate": 7.32065471951322e-08,
      "loss": 0.5959,
      "step": 6916
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7179226066448882,
      "learning_rate": 7.305164221099147e-08,
      "loss": 0.5648,
      "step": 6917
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.650157866450246,
      "learning_rate": 7.289689507649921e-08,
      "loss": 0.528,
      "step": 6918
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.648120408204685,
      "learning_rate": 7.274230581800722e-08,
      "loss": 0.54,
      "step": 6919
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7868036011208724,
      "learning_rate": 7.258787446184055e-08,
      "loss": 0.5845,
      "step": 6920
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8290873009807476,
      "learning_rate": 7.243360103429707e-08,
      "loss": 0.6417,
      "step": 6921
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7578942271529542,
      "learning_rate": 7.227948556164798e-08,
      "loss": 0.57,
      "step": 6922
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8918249702654337,
      "learning_rate": 7.212552807013761e-08,
      "loss": 0.6151,
      "step": 6923
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0562440358099203,
      "learning_rate": 7.197172858598333e-08,
      "loss": 0.5901,
      "step": 6924
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9166551971054305,
      "learning_rate": 7.181808713537552e-08,
      "loss": 0.6726,
      "step": 6925
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.833281185013969,
      "learning_rate": 7.16646037444777e-08,
      "loss": 0.6054,
      "step": 6926
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8724139484318465,
      "learning_rate": 7.151127843942661e-08,
      "loss": 0.6755,
      "step": 6927
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8033149169612629,
      "learning_rate": 7.135811124633185e-08,
      "loss": 0.4989,
      "step": 6928
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9029011618830645,
      "learning_rate": 7.120510219127618e-08,
      "loss": 0.5228,
      "step": 6929
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6433894776261151,
      "learning_rate": 7.105225130031544e-08,
      "loss": 0.5469,
      "step": 6930
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7362709223353763,
      "learning_rate": 7.089955859947872e-08,
      "loss": 0.5882,
      "step": 6931
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7247743627710322,
      "learning_rate": 7.074702411476785e-08,
      "loss": 0.5566,
      "step": 6932
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6466446013557478,
      "learning_rate": 7.059464787215786e-08,
      "loss": 0.5219,
      "step": 6933
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6515325887976694,
      "learning_rate": 7.044242989759685e-08,
      "loss": 0.5661,
      "step": 6934
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7577275332162173,
      "learning_rate": 7.02903702170059e-08,
      "loss": 0.5311,
      "step": 6935
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6932903478452594,
      "learning_rate": 7.013846885627939e-08,
      "loss": 0.5552,
      "step": 6936
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7037976890395362,
      "learning_rate": 6.99867258412844e-08,
      "loss": 0.5222,
      "step": 6937
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7013182816015341,
      "learning_rate": 6.983514119786104e-08,
      "loss": 0.5652,
      "step": 6938
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9648116059149787,
      "learning_rate": 6.96837149518229e-08,
      "loss": 0.5932,
      "step": 6939
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8665016930562945,
      "learning_rate": 6.953244712895612e-08,
      "loss": 0.6361,
      "step": 6940
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7298476186018755,
      "learning_rate": 6.938133775501998e-08,
      "loss": 0.5444,
      "step": 6941
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8048790705659395,
      "learning_rate": 6.9230386855747e-08,
      "loss": 0.6153,
      "step": 6942
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9725276862344931,
      "learning_rate": 6.90795944568422e-08,
      "loss": 0.6696,
      "step": 6943
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6531194821576215,
      "learning_rate": 6.892896058398435e-08,
      "loss": 0.5341,
      "step": 6944
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5720547383127168,
      "learning_rate": 6.877848526282482e-08,
      "loss": 0.4684,
      "step": 6945
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6424773152354798,
      "learning_rate": 6.862816851898778e-08,
      "loss": 0.5028,
      "step": 6946
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.732354076579313,
      "learning_rate": 6.847801037807066e-08,
      "loss": 0.5809,
      "step": 6947
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9890253360174299,
      "learning_rate": 6.832801086564377e-08,
      "loss": 0.578,
      "step": 6948
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9257025794569517,
      "learning_rate": 6.817817000725068e-08,
      "loss": 0.6644,
      "step": 6949
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8020415586958842,
      "learning_rate": 6.802848782840754e-08,
      "loss": 0.5971,
      "step": 6950
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8301822486951606,
      "learning_rate": 6.787896435460372e-08,
      "loss": 0.6396,
      "step": 6951
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.68628990903234,
      "learning_rate": 6.772959961130153e-08,
      "loss": 0.5739,
      "step": 6952
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7055082842241646,
      "learning_rate": 6.758039362393609e-08,
      "loss": 0.4978,
      "step": 6953
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8264713731605237,
      "learning_rate": 6.743134641791581e-08,
      "loss": 0.6464,
      "step": 6954
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1122096246422692,
      "learning_rate": 6.728245801862187e-08,
      "loss": 0.7371,
      "step": 6955
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6698898490741051,
      "learning_rate": 6.71337284514083e-08,
      "loss": 0.4681,
      "step": 6956
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.793084449984343,
      "learning_rate": 6.698515774160219e-08,
      "loss": 0.6508,
      "step": 6957
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.950195459429806,
      "learning_rate": 6.683674591450372e-08,
      "loss": 0.5983,
      "step": 6958
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7011857713541121,
      "learning_rate": 6.66884929953857e-08,
      "loss": 0.5585,
      "step": 6959
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7370790662525254,
      "learning_rate": 6.654039900949426e-08,
      "loss": 0.6015,
      "step": 6960
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6586092218759189,
      "learning_rate": 6.639246398204812e-08,
      "loss": 0.5576,
      "step": 6961
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7761813259055178,
      "learning_rate": 6.624468793823912e-08,
      "loss": 0.5801,
      "step": 6962
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7705640854811164,
      "learning_rate": 6.60970709032319e-08,
      "loss": 0.5624,
      "step": 6963
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7753959052012914,
      "learning_rate": 6.594961290216427e-08,
      "loss": 0.5676,
      "step": 6964
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.791167667817404,
      "learning_rate": 6.580231396014657e-08,
      "loss": 0.59,
      "step": 6965
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7919152950924248,
      "learning_rate": 6.565517410226251e-08,
      "loss": 0.5922,
      "step": 6966
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8313115429330323,
      "learning_rate": 6.550819335356839e-08,
      "loss": 0.6168,
      "step": 6967
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9077003998401888,
      "learning_rate": 6.53613717390934e-08,
      "loss": 0.6841,
      "step": 6968
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9032696126034437,
      "learning_rate": 6.521470928383965e-08,
      "loss": 0.5138,
      "step": 6969
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9068482657213722,
      "learning_rate": 6.506820601278262e-08,
      "loss": 0.6309,
      "step": 6970
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9267133061026424,
      "learning_rate": 6.492186195087002e-08,
      "loss": 0.602,
      "step": 6971
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8797253382735752,
      "learning_rate": 6.477567712302267e-08,
      "loss": 0.5649,
      "step": 6972
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7487443713547458,
      "learning_rate": 6.462965155413457e-08,
      "loss": 0.5517,
      "step": 6973
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7508024197982183,
      "learning_rate": 6.448378526907206e-08,
      "loss": 0.5297,
      "step": 6974
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8968118617383457,
      "learning_rate": 6.43380782926749e-08,
      "loss": 0.6843,
      "step": 6975
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8169930457748107,
      "learning_rate": 6.419253064975527e-08,
      "loss": 0.4929,
      "step": 6976
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6906877405065984,
      "learning_rate": 6.404714236509845e-08,
      "loss": 0.5425,
      "step": 6977
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5982670614112038,
      "learning_rate": 6.390191346346263e-08,
      "loss": 0.5457,
      "step": 6978
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6819306750645038,
      "learning_rate": 6.375684396957881e-08,
      "loss": 0.5602,
      "step": 6979
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6665020075968046,
      "learning_rate": 6.361193390815078e-08,
      "loss": 0.539,
      "step": 6980
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7364039475920091,
      "learning_rate": 6.346718330385503e-08,
      "loss": 0.5585,
      "step": 6981
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7455926190571361,
      "learning_rate": 6.33225921813414e-08,
      "loss": 0.5917,
      "step": 6982
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6836527738813388,
      "learning_rate": 6.317816056523195e-08,
      "loss": 0.5856,
      "step": 6983
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8991476596377391,
      "learning_rate": 6.303388848012215e-08,
      "loss": 0.6005,
      "step": 6984
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8854265293043774,
      "learning_rate": 6.288977595057975e-08,
      "loss": 0.6581,
      "step": 6985
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9572188658011584,
      "learning_rate": 6.274582300114584e-08,
      "loss": 0.5685,
      "step": 6986
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6462710448571821,
      "learning_rate": 6.260202965633387e-08,
      "loss": 0.5496,
      "step": 6987
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.781748665267685,
      "learning_rate": 6.245839594063074e-08,
      "loss": 0.5519,
      "step": 6988
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8598895717648117,
      "learning_rate": 6.231492187849541e-08,
      "loss": 0.619,
      "step": 6989
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6034370443589493,
      "learning_rate": 6.217160749436023e-08,
      "loss": 0.4898,
      "step": 6990
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6773241990681657,
      "learning_rate": 6.202845281263014e-08,
      "loss": 0.5306,
      "step": 6991
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6741899411891884,
      "learning_rate": 6.188545785768274e-08,
      "loss": 0.5343,
      "step": 6992
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7714288834477202,
      "learning_rate": 6.174262265386865e-08,
      "loss": 0.5628,
      "step": 6993
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8146217862817768,
      "learning_rate": 6.159994722551121e-08,
      "loss": 0.57,
      "step": 6994
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7105403885842833,
      "learning_rate": 6.145743159690653e-08,
      "loss": 0.5453,
      "step": 6995
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6904829573549789,
      "learning_rate": 6.131507579232364e-08,
      "loss": 0.5643,
      "step": 6996
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8801054327049488,
      "learning_rate": 6.117287983600427e-08,
      "loss": 0.6681,
      "step": 6997
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8619721623384892,
      "learning_rate": 6.103084375216271e-08,
      "loss": 0.5772,
      "step": 6998
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9398662087432044,
      "learning_rate": 6.08889675649864e-08,
      "loss": 0.638,
      "step": 6999
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6969350419724655,
      "learning_rate": 6.074725129863534e-08,
      "loss": 0.5578,
      "step": 7000
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7070171214536989,
      "learning_rate": 6.060569497724222e-08,
      "loss": 0.596,
      "step": 7001
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7023650513547002,
      "learning_rate": 6.046429862491276e-08,
      "loss": 0.561,
      "step": 7002
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6639650128132786,
      "learning_rate": 6.032306226572525e-08,
      "loss": 0.5626,
      "step": 7003
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7198296153213678,
      "learning_rate": 6.01819859237308e-08,
      "loss": 0.5298,
      "step": 7004
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9616764860856934,
      "learning_rate": 6.004106962295319e-08,
      "loss": 0.6104,
      "step": 7005
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9494099441791194,
      "learning_rate": 5.9900313387389e-08,
      "loss": 0.6133,
      "step": 7006
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0206743623902785,
      "learning_rate": 5.975971724100748e-08,
      "loss": 0.6901,
      "step": 7007
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8288342593858744,
      "learning_rate": 5.961928120775084e-08,
      "loss": 0.5236,
      "step": 7008
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6030666047651422,
      "learning_rate": 5.947900531153371e-08,
      "loss": 0.5813,
      "step": 7009
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.614061384405718,
      "learning_rate": 5.933888957624378e-08,
      "loss": 0.5164,
      "step": 7010
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7664054837683564,
      "learning_rate": 5.9198934025741164e-08,
      "loss": 0.6525,
      "step": 7011
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6988778252210806,
      "learning_rate": 5.905913868385892e-08,
      "loss": 0.5482,
      "step": 7012
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6528872784454941,
      "learning_rate": 5.8919503574402785e-08,
      "loss": 0.5366,
      "step": 7013
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7425451749060353,
      "learning_rate": 5.878002872115095e-08,
      "loss": 0.5561,
      "step": 7014
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7827920819464114,
      "learning_rate": 5.8640714147854766e-08,
      "loss": 0.5005,
      "step": 7015
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.806860379693222,
      "learning_rate": 5.8501559878237904e-08,
      "loss": 0.6517,
      "step": 7016
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6331917739821395,
      "learning_rate": 5.836256593599698e-08,
      "loss": 0.5347,
      "step": 7017
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8435690262669994,
      "learning_rate": 5.8223732344801046e-08,
      "loss": 0.5854,
      "step": 7018
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.67138523681544,
      "learning_rate": 5.80850591282922e-08,
      "loss": 0.5289,
      "step": 7019
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6205320590980302,
      "learning_rate": 5.794654631008511e-08,
      "loss": 0.4477,
      "step": 7020
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8250685349284711,
      "learning_rate": 5.780819391376679e-08,
      "loss": 0.642,
      "step": 7021
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8751999149495736,
      "learning_rate": 5.7670001962897396e-08,
      "loss": 0.5948,
      "step": 7022
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7425268216368743,
      "learning_rate": 5.753197048100955e-08,
      "loss": 0.5194,
      "step": 7023
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8262700752859442,
      "learning_rate": 5.739409949160867e-08,
      "loss": 0.6444,
      "step": 7024
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6358490444121793,
      "learning_rate": 5.725638901817254e-08,
      "loss": 0.5239,
      "step": 7025
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8395759610166903,
      "learning_rate": 5.711883908415194e-08,
      "loss": 0.5691,
      "step": 7026
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8892731931701809,
      "learning_rate": 5.698144971297003e-08,
      "loss": 0.6155,
      "step": 7027
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6750710652037517,
      "learning_rate": 5.68442209280231e-08,
      "loss": 0.506,
      "step": 7028
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6111345783371932,
      "learning_rate": 5.6707152752679567e-08,
      "loss": 0.4973,
      "step": 7029
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9993048765030773,
      "learning_rate": 5.6570245210280864e-08,
      "loss": 0.6385,
      "step": 7030
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7757039205668105,
      "learning_rate": 5.6433498324140795e-08,
      "loss": 0.5025,
      "step": 7031
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.85497661377225,
      "learning_rate": 5.629691211754595e-08,
      "loss": 0.6327,
      "step": 7032
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0247603448137435,
      "learning_rate": 5.61604866137555e-08,
      "loss": 0.5397,
      "step": 7033
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6443613605825536,
      "learning_rate": 5.602422183600131e-08,
      "loss": 0.5259,
      "step": 7034
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0876592870216841,
      "learning_rate": 5.588811780748792e-08,
      "loss": 0.6284,
      "step": 7035
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7177688025976733,
      "learning_rate": 5.575217455139247e-08,
      "loss": 0.5266,
      "step": 7036
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6234408428229516,
      "learning_rate": 5.561639209086444e-08,
      "loss": 0.4917,
      "step": 7037
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8060715388877318,
      "learning_rate": 5.548077044902644e-08,
      "loss": 0.6093,
      "step": 7038
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7742599930981701,
      "learning_rate": 5.534530964897322e-08,
      "loss": 0.571,
      "step": 7039
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7436829975707513,
      "learning_rate": 5.521000971377243e-08,
      "loss": 0.5458,
      "step": 7040
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.708445701556239,
      "learning_rate": 5.507487066646432e-08,
      "loss": 0.5171,
      "step": 7041
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7551362747512802,
      "learning_rate": 5.4939892530061485e-08,
      "loss": 0.5488,
      "step": 7042
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7865885025849543,
      "learning_rate": 5.480507532754941e-08,
      "loss": 0.5609,
      "step": 7043
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6160009906099309,
      "learning_rate": 5.4670419081886073e-08,
      "loss": 0.5023,
      "step": 7044
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7311021506957778,
      "learning_rate": 5.453592381600191e-08,
      "loss": 0.5586,
      "step": 7045
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6759879553861401,
      "learning_rate": 5.440158955280016e-08,
      "loss": 0.5435,
      "step": 7046
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8353157525552973,
      "learning_rate": 5.426741631515663e-08,
      "loss": 0.5944,
      "step": 7047
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7160200590554945,
      "learning_rate": 5.413340412591949e-08,
      "loss": 0.5461,
      "step": 7048
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9450780792153107,
      "learning_rate": 5.3999553007909594e-08,
      "loss": 0.6754,
      "step": 7049
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.760357082744667,
      "learning_rate": 5.3865862983920595e-08,
      "loss": 0.5559,
      "step": 7050
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6081104902986652,
      "learning_rate": 5.373233407671829e-08,
      "loss": 0.5177,
      "step": 7051
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6930087126034485,
      "learning_rate": 5.359896630904137e-08,
      "loss": 0.5835,
      "step": 7052
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8777596102280375,
      "learning_rate": 5.34657597036009e-08,
      "loss": 0.6064,
      "step": 7053
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5989904712053076,
      "learning_rate": 5.33327142830805e-08,
      "loss": 0.4793,
      "step": 7054
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9133858191939173,
      "learning_rate": 5.319983007013673e-08,
      "loss": 0.6029,
      "step": 7055
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9920762105671919,
      "learning_rate": 5.306710708739814e-08,
      "loss": 0.6254,
      "step": 7056
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7416429024815674,
      "learning_rate": 5.293454535746622e-08,
      "loss": 0.5798,
      "step": 7057
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7535890438574102,
      "learning_rate": 5.280214490291479e-08,
      "loss": 0.5535,
      "step": 7058
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8825413650833511,
      "learning_rate": 5.266990574629016e-08,
      "loss": 0.6337,
      "step": 7059
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7062372806110339,
      "learning_rate": 5.253782791011141e-08,
      "loss": 0.5501,
      "step": 7060
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7367564663337032,
      "learning_rate": 5.2405911416870006e-08,
      "loss": 0.5477,
      "step": 7061
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8149490217147053,
      "learning_rate": 5.227415628902987e-08,
      "loss": 0.5408,
      "step": 7062
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8822065742107744,
      "learning_rate": 5.2142562549027604e-08,
      "loss": 0.6295,
      "step": 7063
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6190621102572367,
      "learning_rate": 5.201113021927217e-08,
      "loss": 0.4959,
      "step": 7064
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8501209122122887,
      "learning_rate": 5.187985932214523e-08,
      "loss": 0.6135,
      "step": 7065
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8778404556831634,
      "learning_rate": 5.1748749880000884e-08,
      "loss": 0.634,
      "step": 7066
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9739838137615773,
      "learning_rate": 5.161780191516552e-08,
      "loss": 0.5767,
      "step": 7067
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8158645640468021,
      "learning_rate": 5.148701544993839e-08,
      "loss": 0.6124,
      "step": 7068
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7288472685719366,
      "learning_rate": 5.135639050659091e-08,
      "loss": 0.5519,
      "step": 7069
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6568639135983344,
      "learning_rate": 5.122592710736728e-08,
      "loss": 0.5571,
      "step": 7070
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6537953585821406,
      "learning_rate": 5.1095625274483964e-08,
      "loss": 0.5061,
      "step": 7071
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7381449002199798,
      "learning_rate": 5.096548503013009e-08,
      "loss": 0.5413,
      "step": 7072
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6862588099764385,
      "learning_rate": 5.0835506396467165e-08,
      "loss": 0.5268,
      "step": 7073
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7055649282177653,
      "learning_rate": 5.070568939562914e-08,
      "loss": 0.5105,
      "step": 7074
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7475369890384103,
      "learning_rate": 5.057603404972255e-08,
      "loss": 0.5104,
      "step": 7075
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7185419618830547,
      "learning_rate": 5.04465403808263e-08,
      "loss": 0.6057,
      "step": 7076
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7070674943234772,
      "learning_rate": 5.0317208410991854e-08,
      "loss": 0.5505,
      "step": 7077
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6663443623184074,
      "learning_rate": 5.018803816224315e-08,
      "loss": 0.5695,
      "step": 7078
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.882182931294623,
      "learning_rate": 5.005902965657638e-08,
      "loss": 0.6274,
      "step": 7079
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8666546667993995,
      "learning_rate": 4.993018291596041e-08,
      "loss": 0.6241,
      "step": 7080
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5910146516458711,
      "learning_rate": 4.980149796233657e-08,
      "loss": 0.4944,
      "step": 7081
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7272342859564757,
      "learning_rate": 4.967297481761856e-08,
      "loss": 0.618,
      "step": 7082
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8588081871627958,
      "learning_rate": 4.9544613503692546e-08,
      "loss": 0.6146,
      "step": 7083
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7477562946487415,
      "learning_rate": 4.9416414042417144e-08,
      "loss": 0.5459,
      "step": 7084
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6543054875937356,
      "learning_rate": 4.928837645562323e-08,
      "loss": 0.475,
      "step": 7085
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9231795196464616,
      "learning_rate": 4.916050076511435e-08,
      "loss": 0.657,
      "step": 7086
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8594975675436148,
      "learning_rate": 4.903278699266644e-08,
      "loss": 0.6339,
      "step": 7087
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8995687606157043,
      "learning_rate": 4.8905235160027644e-08,
      "loss": 0.6627,
      "step": 7088
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7371662714394093,
      "learning_rate": 4.877784528891904e-08,
      "loss": 0.5324,
      "step": 7089
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.833185405578994,
      "learning_rate": 4.86506174010336e-08,
      "loss": 0.6103,
      "step": 7090
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9626452603802091,
      "learning_rate": 4.8523551518036884e-08,
      "loss": 0.658,
      "step": 7091
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7900499206002348,
      "learning_rate": 4.839664766156703e-08,
      "loss": 0.6127,
      "step": 7092
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7760255512577368,
      "learning_rate": 4.8269905853234315e-08,
      "loss": 0.5384,
      "step": 7093
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9667553356722294,
      "learning_rate": 4.814332611462157e-08,
      "loss": 0.5627,
      "step": 7094
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8588471544107891,
      "learning_rate": 4.801690846728401e-08,
      "loss": 0.5258,
      "step": 7095
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8708853159121567,
      "learning_rate": 4.78906529327493e-08,
      "loss": 0.5456,
      "step": 7096
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9616157916829089,
      "learning_rate": 4.7764559532517345e-08,
      "loss": 0.6465,
      "step": 7097
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.817519178173823,
      "learning_rate": 4.7638628288060534e-08,
      "loss": 0.6502,
      "step": 7098
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8025424717523184,
      "learning_rate": 4.751285922082382e-08,
      "loss": 0.6346,
      "step": 7099
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7401134729771175,
      "learning_rate": 4.7387252352224186e-08,
      "loss": 0.5317,
      "step": 7100
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9694573693026775,
      "learning_rate": 4.726180770365118e-08,
      "loss": 0.6543,
      "step": 7101
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7251888551124789,
      "learning_rate": 4.713652529646684e-08,
      "loss": 0.5796,
      "step": 7102
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6199777705453603,
      "learning_rate": 4.70114051520053e-08,
      "loss": 0.4946,
      "step": 7103
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7119197393054936,
      "learning_rate": 4.688644729157332e-08,
      "loss": 0.5365,
      "step": 7104
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.021404825701395,
      "learning_rate": 4.676165173644997e-08,
      "loss": 0.6352,
      "step": 7105
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6912348957742143,
      "learning_rate": 4.663701850788648e-08,
      "loss": 0.596,
      "step": 7106
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5989807492777921,
      "learning_rate": 4.651254762710677e-08,
      "loss": 0.5125,
      "step": 7107
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7177950720290377,
      "learning_rate": 4.638823911530676e-08,
      "loss": 0.5482,
      "step": 7108
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7485929148265179,
      "learning_rate": 4.6264092993654965e-08,
      "loss": 0.5596,
      "step": 7109
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.748183996680338,
      "learning_rate": 4.614010928329226e-08,
      "loss": 0.5841,
      "step": 7110
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8976080435887559,
      "learning_rate": 4.601628800533164e-08,
      "loss": 0.6357,
      "step": 7111
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8661695501186198,
      "learning_rate": 4.5892629180858564e-08,
      "loss": 0.6162,
      "step": 7112
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.632409699117834,
      "learning_rate": 4.576913283093098e-08,
      "loss": 0.5431,
      "step": 7113
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7109373222620569,
      "learning_rate": 4.564579897657894e-08,
      "loss": 0.6196,
      "step": 7114
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6735811584042484,
      "learning_rate": 4.55226276388051e-08,
      "loss": 0.5906,
      "step": 7115
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9699331907104476,
      "learning_rate": 4.5399618838584006e-08,
      "loss": 0.6472,
      "step": 7116
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8931269347818763,
      "learning_rate": 4.52767725968628e-08,
      "loss": 0.6587,
      "step": 7117
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.81883981589009,
      "learning_rate": 4.5154088934561074e-08,
      "loss": 0.6021,
      "step": 7118
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7033671652305435,
      "learning_rate": 4.5031567872570455e-08,
      "loss": 0.5927,
      "step": 7119
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.857634813047258,
      "learning_rate": 4.490920943175513e-08,
      "loss": 0.6488,
      "step": 7120
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6349022435150049,
      "learning_rate": 4.478701363295123e-08,
      "loss": 0.516,
      "step": 7121
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7306331520582102,
      "learning_rate": 4.4664980496967654e-08,
      "loss": 0.5355,
      "step": 7122
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8457295956274435,
      "learning_rate": 4.454311004458533e-08,
      "loss": 0.6086,
      "step": 7123
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7958149532172307,
      "learning_rate": 4.4421402296557445e-08,
      "loss": 0.5688,
      "step": 7124
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9381254527256889,
      "learning_rate": 4.429985727360963e-08,
      "loss": 0.5554,
      "step": 7125
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6978996014654406,
      "learning_rate": 4.417847499643967e-08,
      "loss": 0.5839,
      "step": 7126
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8493705016348178,
      "learning_rate": 4.4057255485717703e-08,
      "loss": 0.5929,
      "step": 7127
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9178695732280715,
      "learning_rate": 4.3936198762086204e-08,
      "loss": 0.6366,
      "step": 7128
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6174457365629217,
      "learning_rate": 4.381530484615992e-08,
      "loss": 0.5038,
      "step": 7129
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6500445749860795,
      "learning_rate": 4.36945737585257e-08,
      "loss": 0.5456,
      "step": 7130
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9482588019428619,
      "learning_rate": 4.3574005519742886e-08,
      "loss": 0.6176,
      "step": 7131
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6950817042258106,
      "learning_rate": 4.345360015034294e-08,
      "loss": 0.5068,
      "step": 7132
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8381910895617419,
      "learning_rate": 4.3333357670829686e-08,
      "loss": 0.6411,
      "step": 7133
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6348147078184392,
      "learning_rate": 4.3213278101679185e-08,
      "loss": 0.5313,
      "step": 7134
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7145672213947187,
      "learning_rate": 4.309336146333953e-08,
      "loss": 0.5708,
      "step": 7135
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8620921036337574,
      "learning_rate": 4.29736077762316e-08,
      "loss": 0.632,
      "step": 7136
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.834679905708729,
      "learning_rate": 4.285401706074798e-08,
      "loss": 0.6311,
      "step": 7137
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0042730052916666,
      "learning_rate": 4.27345893372536e-08,
      "loss": 0.712,
      "step": 7138
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9379435061058117,
      "learning_rate": 4.261532462608608e-08,
      "loss": 0.5938,
      "step": 7139
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6124783480040105,
      "learning_rate": 4.249622294755484e-08,
      "loss": 0.5006,
      "step": 7140
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6456666245248436,
      "learning_rate": 4.237728432194165e-08,
      "loss": 0.5272,
      "step": 7141
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7507603412337811,
      "learning_rate": 4.2258508769500544e-08,
      "loss": 0.5754,
      "step": 7142
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6930989098774712,
      "learning_rate": 4.2139896310457664e-08,
      "loss": 0.5914,
      "step": 7143
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7154411481545742,
      "learning_rate": 4.202144696501142e-08,
      "loss": 0.6095,
      "step": 7144
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9725727757248239,
      "learning_rate": 4.190316075333278e-08,
      "loss": 0.5987,
      "step": 7145
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7052807949186115,
      "learning_rate": 4.1785037695564406e-08,
      "loss": 0.5113,
      "step": 7146
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7046130375006884,
      "learning_rate": 4.166707781182144e-08,
      "loss": 0.5749,
      "step": 7147
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.891805321843176,
      "learning_rate": 4.154928112219136e-08,
      "loss": 0.5487,
      "step": 7148
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8229178893233402,
      "learning_rate": 4.143164764673368e-08,
      "loss": 0.6539,
      "step": 7149
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8147697102789799,
      "learning_rate": 4.131417740548005e-08,
      "loss": 0.6011,
      "step": 7150
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8910696341610276,
      "learning_rate": 4.119687041843445e-08,
      "loss": 0.6408,
      "step": 7151
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.059194312791156,
      "learning_rate": 4.107972670557314e-08,
      "loss": 0.6251,
      "step": 7152
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7536534690818572,
      "learning_rate": 4.0962746286844484e-08,
      "loss": 0.4945,
      "step": 7153
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8423638996356531,
      "learning_rate": 4.084592918216878e-08,
      "loss": 0.6262,
      "step": 7154
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8618229194243596,
      "learning_rate": 4.072927541143911e-08,
      "loss": 0.6402,
      "step": 7155
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.714774034112328,
      "learning_rate": 4.061278499452014e-08,
      "loss": 0.5442,
      "step": 7156
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8441440093708332,
      "learning_rate": 4.0496457951249007e-08,
      "loss": 0.6261,
      "step": 7157
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8114637488635104,
      "learning_rate": 4.038029430143519e-08,
      "loss": 0.5562,
      "step": 7158
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.750838213921343,
      "learning_rate": 4.026429406485987e-08,
      "loss": 0.5349,
      "step": 7159
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7781925083901424,
      "learning_rate": 4.0148457261276915e-08,
      "loss": 0.5603,
      "step": 7160
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5867784781540418,
      "learning_rate": 4.003278391041198e-08,
      "loss": 0.524,
      "step": 7161
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7574011713934733,
      "learning_rate": 3.99172740319631e-08,
      "loss": 0.5695,
      "step": 7162
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8473887044771929,
      "learning_rate": 3.980192764560042e-08,
      "loss": 0.5997,
      "step": 7163
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9172164980396105,
      "learning_rate": 3.968674477096612e-08,
      "loss": 0.6173,
      "step": 7164
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6896181166092821,
      "learning_rate": 3.957172542767462e-08,
      "loss": 0.5707,
      "step": 7165
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.656604934678135,
      "learning_rate": 3.9456869635312696e-08,
      "loss": 0.555,
      "step": 7166
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7506505139070786,
      "learning_rate": 3.9342177413439034e-08,
      "loss": 0.5362,
      "step": 7167
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.853660700599245,
      "learning_rate": 3.922764878158458e-08,
      "loss": 0.5905,
      "step": 7168
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9314305138257395,
      "learning_rate": 3.911328375925216e-08,
      "loss": 0.6082,
      "step": 7169
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9067266282631755,
      "learning_rate": 3.8999082365917205e-08,
      "loss": 0.6039,
      "step": 7170
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9713194134395533,
      "learning_rate": 3.888504462102671e-08,
      "loss": 0.5715,
      "step": 7171
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7581396164467741,
      "learning_rate": 3.8771170544000476e-08,
      "loss": 0.5464,
      "step": 7172
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9723807679027652,
      "learning_rate": 3.865746015422977e-08,
      "loss": 0.6142,
      "step": 7173
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.653068481439923,
      "learning_rate": 3.854391347107844e-08,
      "loss": 0.5012,
      "step": 7174
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9595688151240492,
      "learning_rate": 3.8430530513882234e-08,
      "loss": 0.6621,
      "step": 7175
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7064121205284901,
      "learning_rate": 3.831731130194915e-08,
      "loss": 0.5366,
      "step": 7176
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.607973441060355,
      "learning_rate": 3.820425585455922e-08,
      "loss": 0.5553,
      "step": 7177
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7097374217166457,
      "learning_rate": 3.8091364190964594e-08,
      "loss": 0.5677,
      "step": 7178
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9128480656249681,
      "learning_rate": 3.797863633038956e-08,
      "loss": 0.6171,
      "step": 7179
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9570136409365261,
      "learning_rate": 3.7866072292030536e-08,
      "loss": 0.6547,
      "step": 7180
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8204851485444133,
      "learning_rate": 3.775367209505587e-08,
      "loss": 0.5881,
      "step": 7181
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8552027503442323,
      "learning_rate": 3.764143575860634e-08,
      "loss": 0.5816,
      "step": 7182
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8804620269774404,
      "learning_rate": 3.7529363301794456e-08,
      "loss": 0.595,
      "step": 7183
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9194297746200314,
      "learning_rate": 3.7417454743705055e-08,
      "loss": 0.6072,
      "step": 7184
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6583391070102755,
      "learning_rate": 3.7305710103395006e-08,
      "loss": 0.5236,
      "step": 7185
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9693601314804884,
      "learning_rate": 3.71941293998933e-08,
      "loss": 0.6088,
      "step": 7186
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.714101203682966,
      "learning_rate": 3.708271265220087e-08,
      "loss": 0.5191,
      "step": 7187
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.77974703696639,
      "learning_rate": 3.6971459879290845e-08,
      "loss": 0.5321,
      "step": 7188
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8239874487680002,
      "learning_rate": 3.6860371100108425e-08,
      "loss": 0.5488,
      "step": 7189
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8023767621208219,
      "learning_rate": 3.674944633357091e-08,
      "loss": 0.6049,
      "step": 7190
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7314401278371676,
      "learning_rate": 3.663868559856764e-08,
      "loss": 0.5922,
      "step": 7191
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7611048174198054,
      "learning_rate": 3.6528088913959975e-08,
      "loss": 0.4599,
      "step": 7192
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8167428607990136,
      "learning_rate": 3.64176562985814e-08,
      "loss": 0.5585,
      "step": 7193
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7146657342058549,
      "learning_rate": 3.630738777123743e-08,
      "loss": 0.5726,
      "step": 7194
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5956300473937473,
      "learning_rate": 3.61972833507056e-08,
      "loss": 0.534,
      "step": 7195
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7209344806653497,
      "learning_rate": 3.6087343055735685e-08,
      "loss": 0.5432,
      "step": 7196
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6448001863977473,
      "learning_rate": 3.597756690504916e-08,
      "loss": 0.5299,
      "step": 7197
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8827879915835674,
      "learning_rate": 3.586795491733996e-08,
      "loss": 0.6059,
      "step": 7198
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9292134261683238,
      "learning_rate": 3.575850711127371e-08,
      "loss": 0.6585,
      "step": 7199
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8468578807585195,
      "learning_rate": 3.564922350548838e-08,
      "loss": 0.6154,
      "step": 7200
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.780823537727861,
      "learning_rate": 3.5540104118593764e-08,
      "loss": 0.6104,
      "step": 7201
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8792149806695043,
      "learning_rate": 3.5431148969171765e-08,
      "loss": 0.6205,
      "step": 7202
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6579195156185577,
      "learning_rate": 3.5322358075776215e-08,
      "loss": 0.562,
      "step": 7203
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8699845775597669,
      "learning_rate": 3.5213731456933184e-08,
      "loss": 0.5848,
      "step": 7204
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.691209828793836,
      "learning_rate": 3.5105269131140644e-08,
      "loss": 0.5564,
      "step": 7205
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7016006248956878,
      "learning_rate": 3.4996971116868504e-08,
      "loss": 0.5791,
      "step": 7206
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7414726586664325,
      "learning_rate": 3.488883743255888e-08,
      "loss": 0.5418,
      "step": 7207
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8093075303033707,
      "learning_rate": 3.4780868096625836e-08,
      "loss": 0.5809,
      "step": 7208
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.774431031285429,
      "learning_rate": 3.467306312745533e-08,
      "loss": 0.6164,
      "step": 7209
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8157691724781662,
      "learning_rate": 3.4565422543405445e-08,
      "loss": 0.6628,
      "step": 7210
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8773993062084237,
      "learning_rate": 3.44579463628063e-08,
      "loss": 0.6411,
      "step": 7211
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8227400561991806,
      "learning_rate": 3.4350634603959926e-08,
      "loss": 0.6238,
      "step": 7212
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7074201918671139,
      "learning_rate": 3.4243487285140484e-08,
      "loss": 0.5413,
      "step": 7213
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6848832491727511,
      "learning_rate": 3.413650442459392e-08,
      "loss": 0.5555,
      "step": 7214
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7250389367073469,
      "learning_rate": 3.4029686040538355e-08,
      "loss": 0.5534,
      "step": 7215
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7594176957296942,
      "learning_rate": 3.392303215116388e-08,
      "loss": 0.6051,
      "step": 7216
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9641137288820174,
      "learning_rate": 3.3816542774632525e-08,
      "loss": 0.6634,
      "step": 7217
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5833329912636238,
      "learning_rate": 3.371021792907824e-08,
      "loss": 0.4933,
      "step": 7218
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7892558823960883,
      "learning_rate": 3.360405763260721e-08,
      "loss": 0.5776,
      "step": 7219
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9206306921635786,
      "learning_rate": 3.349806190329729e-08,
      "loss": 0.6446,
      "step": 7220
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.854440005895728,
      "learning_rate": 3.33922307591985e-08,
      "loss": 0.6386,
      "step": 7221
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7690595505962444,
      "learning_rate": 3.328656421833265e-08,
      "loss": 0.5581,
      "step": 7222
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7890436698791222,
      "learning_rate": 3.318106229869377e-08,
      "loss": 0.6469,
      "step": 7223
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7642773256452028,
      "learning_rate": 3.307572501824785e-08,
      "loss": 0.5162,
      "step": 7224
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6809035325164974,
      "learning_rate": 3.2970552394932514e-08,
      "loss": 0.5866,
      "step": 7225
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.915155489499076,
      "learning_rate": 3.286554444665779e-08,
      "loss": 0.6369,
      "step": 7226
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9001421047089216,
      "learning_rate": 3.276070119130525e-08,
      "loss": 0.5881,
      "step": 7227
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8949228771138282,
      "learning_rate": 3.265602264672862e-08,
      "loss": 0.5898,
      "step": 7228
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8707282010304314,
      "learning_rate": 3.255150883075364e-08,
      "loss": 0.6089,
      "step": 7229
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7934597702914037,
      "learning_rate": 3.244715976117795e-08,
      "loss": 0.5883,
      "step": 7230
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7319047635306315,
      "learning_rate": 3.234297545577103e-08,
      "loss": 0.4796,
      "step": 7231
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6470208084072265,
      "learning_rate": 3.2238955932274436e-08,
      "loss": 0.5395,
      "step": 7232
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8728715111073796,
      "learning_rate": 3.213510120840157e-08,
      "loss": 0.5963,
      "step": 7233
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6100379460119404,
      "learning_rate": 3.203141130183784e-08,
      "loss": 0.4884,
      "step": 7234
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.95594915873736,
      "learning_rate": 3.1927886230240565e-08,
      "loss": 0.6348,
      "step": 7235
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6372277462554713,
      "learning_rate": 3.182452601123886e-08,
      "loss": 0.5211,
      "step": 7236
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9355422071950238,
      "learning_rate": 3.1721330662434097e-08,
      "loss": 0.6102,
      "step": 7237
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6241549176875114,
      "learning_rate": 3.161830020139922e-08,
      "loss": 0.551,
      "step": 7238
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8099211821245752,
      "learning_rate": 3.151543464567929e-08,
      "loss": 0.6635,
      "step": 7239
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7190754261763039,
      "learning_rate": 3.141273401279121e-08,
      "loss": 0.5687,
      "step": 7240
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8301615452743462,
      "learning_rate": 3.1310198320223966e-08,
      "loss": 0.6394,
      "step": 7241
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9850435987507413,
      "learning_rate": 3.120782758543816e-08,
      "loss": 0.6667,
      "step": 7242
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9911629314843788,
      "learning_rate": 3.1105621825866404e-08,
      "loss": 0.6536,
      "step": 7243
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8571239668326772,
      "learning_rate": 3.100358105891354e-08,
      "loss": 0.6152,
      "step": 7244
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7551905552359298,
      "learning_rate": 3.0901705301955794e-08,
      "loss": 0.5224,
      "step": 7245
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0492874851314253,
      "learning_rate": 3.079999457234173e-08,
      "loss": 0.6088,
      "step": 7246
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8273582058139619,
      "learning_rate": 3.069844888739148e-08,
      "loss": 0.5396,
      "step": 7247
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8235138075880819,
      "learning_rate": 3.059706826439723e-08,
      "loss": 0.6047,
      "step": 7248
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8528860805385543,
      "learning_rate": 3.049585272062305e-08,
      "loss": 0.6051,
      "step": 7249
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.705706960238756,
      "learning_rate": 3.0394802273305045e-08,
      "loss": 0.5418,
      "step": 7250
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.596981970820245,
      "learning_rate": 3.0293916939650884e-08,
      "loss": 0.4724,
      "step": 7251
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8848243018426158,
      "learning_rate": 3.01931967368404e-08,
      "loss": 0.5155,
      "step": 7252
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8744242365468112,
      "learning_rate": 3.009264168202508e-08,
      "loss": 0.6706,
      "step": 7253
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7024447149278906,
      "learning_rate": 2.999225179232845e-08,
      "loss": 0.5425,
      "step": 7254
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8951284829159555,
      "learning_rate": 2.989202708484595e-08,
      "loss": 0.5822,
      "step": 7255
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7250759890585469,
      "learning_rate": 2.9791967576644594e-08,
      "loss": 0.6121,
      "step": 7256
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6234521575965822,
      "learning_rate": 2.969207328476375e-08,
      "loss": 0.5335,
      "step": 7257
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.749795267394905,
      "learning_rate": 2.959234422621415e-08,
      "loss": 0.5508,
      "step": 7258
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8636073533020747,
      "learning_rate": 2.9492780417978758e-08,
      "loss": 0.6294,
      "step": 7259
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.887876625054137,
      "learning_rate": 2.9393381877012123e-08,
      "loss": 0.6636,
      "step": 7260
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8324545958682875,
      "learning_rate": 2.9294148620240933e-08,
      "loss": 0.5188,
      "step": 7261
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7611168344149271,
      "learning_rate": 2.9195080664563442e-08,
      "loss": 0.5837,
      "step": 7262
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7661226206087224,
      "learning_rate": 2.9096178026849937e-08,
      "loss": 0.5137,
      "step": 7263
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8322838728884923,
      "learning_rate": 2.8997440723942503e-08,
      "loss": 0.6285,
      "step": 7264
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7744264146490837,
      "learning_rate": 2.8898868772655018e-08,
      "loss": 0.5542,
      "step": 7265
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7336767171354189,
      "learning_rate": 2.880046218977339e-08,
      "loss": 0.5468,
      "step": 7266
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7635367687773135,
      "learning_rate": 2.8702220992055214e-08,
      "loss": 0.5508,
      "step": 7267
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7487195728202541,
      "learning_rate": 2.860414519622989e-08,
      "loss": 0.5394,
      "step": 7268
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7158286138890791,
      "learning_rate": 2.8506234818998608e-08,
      "loss": 0.5556,
      "step": 7269
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6490072219759189,
      "learning_rate": 2.84084898770347e-08,
      "loss": 0.5256,
      "step": 7270
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8112796828001055,
      "learning_rate": 2.8310910386982966e-08,
      "loss": 0.5909,
      "step": 7271
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7572296856179627,
      "learning_rate": 2.821349636546011e-08,
      "loss": 0.5261,
      "step": 7272
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.816570987913261,
      "learning_rate": 2.8116247829054972e-08,
      "loss": 0.5648,
      "step": 7273
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8919148018207494,
      "learning_rate": 2.8019164794327755e-08,
      "loss": 0.6664,
      "step": 7274
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7051256950320341,
      "learning_rate": 2.7922247277810673e-08,
      "loss": 0.5406,
      "step": 7275
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6680055891827361,
      "learning_rate": 2.782549529600786e-08,
      "loss": 0.5274,
      "step": 7276
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8542932690428376,
      "learning_rate": 2.7728908865395252e-08,
      "loss": 0.6911,
      "step": 7277
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6551618724016277,
      "learning_rate": 2.7632488002420353e-08,
      "loss": 0.5093,
      "step": 7278
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8320909840614118,
      "learning_rate": 2.7536232723502695e-08,
      "loss": 0.608,
      "step": 7279
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.686917210990391,
      "learning_rate": 2.7440143045033614e-08,
      "loss": 0.5148,
      "step": 7280
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7453461359383318,
      "learning_rate": 2.734421898337613e-08,
      "loss": 0.5873,
      "step": 7281
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8993783897062503,
      "learning_rate": 2.724846055486507e-08,
      "loss": 0.5819,
      "step": 7282
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0555958199116768,
      "learning_rate": 2.7152867775807054e-08,
      "loss": 0.6366,
      "step": 7283
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8152790512325929,
      "learning_rate": 2.7057440662480724e-08,
      "loss": 0.6516,
      "step": 7284
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8240071192024564,
      "learning_rate": 2.6962179231136085e-08,
      "loss": 0.5548,
      "step": 7285
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6295839041812001,
      "learning_rate": 2.686708349799538e-08,
      "loss": 0.4942,
      "step": 7286
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9425429406384146,
      "learning_rate": 2.6772153479252322e-08,
      "loss": 0.6029,
      "step": 7287
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6957958570867699,
      "learning_rate": 2.667738919107243e-08,
      "loss": 0.5191,
      "step": 7288
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8516117539878484,
      "learning_rate": 2.6582790649593233e-08,
      "loss": 0.678,
      "step": 7289
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7935133376542478,
      "learning_rate": 2.648835787092385e-08,
      "loss": 0.5302,
      "step": 7290
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6482413083242479,
      "learning_rate": 2.6394090871144968e-08,
      "loss": 0.4846,
      "step": 7291
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5905827050214838,
      "learning_rate": 2.629998966630953e-08,
      "loss": 0.5223,
      "step": 7292
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6173549039822884,
      "learning_rate": 2.6206054272441934e-08,
      "loss": 0.4756,
      "step": 7293
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7609342824956519,
      "learning_rate": 2.6112284705538503e-08,
      "loss": 0.5816,
      "step": 7294
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7781689248715798,
      "learning_rate": 2.601868098156701e-08,
      "loss": 0.5823,
      "step": 7295
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9271994559550089,
      "learning_rate": 2.5925243116467265e-08,
      "loss": 0.6238,
      "step": 7296
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7379924298783651,
      "learning_rate": 2.5831971126150764e-08,
      "loss": 0.5769,
      "step": 7297
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5920473005317936,
      "learning_rate": 2.57388650265008e-08,
      "loss": 0.4842,
      "step": 7298
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9016036351096196,
      "learning_rate": 2.5645924833372356e-08,
      "loss": 0.6253,
      "step": 7299
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6805383060911628,
      "learning_rate": 2.5553150562592106e-08,
      "loss": 0.5866,
      "step": 7300
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8856669460180067,
      "learning_rate": 2.5460542229958746e-08,
      "loss": 0.5897,
      "step": 7301
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.649128930799852,
      "learning_rate": 2.536809985124222e-08,
      "loss": 0.48,
      "step": 7302
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.758725213366629,
      "learning_rate": 2.5275823442184707e-08,
      "loss": 0.5776,
      "step": 7303
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8264163011593813,
      "learning_rate": 2.5183713018499865e-08,
      "loss": 0.6253,
      "step": 7304
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7050846529919563,
      "learning_rate": 2.5091768595873143e-08,
      "loss": 0.5388,
      "step": 7305
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7560979029581031,
      "learning_rate": 2.4999990189961683e-08,
      "loss": 0.6035,
      "step": 7306
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7155094135096984,
      "learning_rate": 2.490837781639432e-08,
      "loss": 0.5835,
      "step": 7307
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8090863112793168,
      "learning_rate": 2.4816931490771797e-08,
      "loss": 0.6159,
      "step": 7308
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6639803448318626,
      "learning_rate": 2.4725651228666544e-08,
      "loss": 0.558,
      "step": 7309
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.634450185349951,
      "learning_rate": 2.463453704562257e-08,
      "loss": 0.56,
      "step": 7310
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9012663735561001,
      "learning_rate": 2.4543588957155693e-08,
      "loss": 0.6496,
      "step": 7311
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7179294875779447,
      "learning_rate": 2.4452806978753292e-08,
      "loss": 0.5553,
      "step": 7312
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7318889642436464,
      "learning_rate": 2.4362191125874676e-08,
      "loss": 0.5641,
      "step": 7313
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.629847707636561,
      "learning_rate": 2.427174141395083e-08,
      "loss": 0.4975,
      "step": 7314
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6811445202685522,
      "learning_rate": 2.4181457858384436e-08,
      "loss": 0.5409,
      "step": 7315
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8376259600602651,
      "learning_rate": 2.4091340474549636e-08,
      "loss": 0.5573,
      "step": 7316
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9537920904601983,
      "learning_rate": 2.4001389277792716e-08,
      "loss": 0.606,
      "step": 7317
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8517656877010927,
      "learning_rate": 2.3911604283431196e-08,
      "loss": 0.6274,
      "step": 7318
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8238777681106071,
      "learning_rate": 2.3821985506754737e-08,
      "loss": 0.5827,
      "step": 7319
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.042579964028282,
      "learning_rate": 2.3732532963024466e-08,
      "loss": 0.6567,
      "step": 7320
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8152585007025988,
      "learning_rate": 2.364324666747297e-08,
      "loss": 0.6257,
      "step": 7321
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9035497507915733,
      "learning_rate": 2.3554126635305095e-08,
      "loss": 0.725,
      "step": 7322
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6135358852854472,
      "learning_rate": 2.3465172881696914e-08,
      "loss": 0.5534,
      "step": 7323
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7256928190829117,
      "learning_rate": 2.337638542179632e-08,
      "loss": 0.4633,
      "step": 7324
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6861522571682235,
      "learning_rate": 2.3287764270722765e-08,
      "loss": 0.5559,
      "step": 7325
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7282070536821227,
      "learning_rate": 2.319930944356774e-08,
      "loss": 0.501,
      "step": 7326
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.917918311517255,
      "learning_rate": 2.3111020955394078e-08,
      "loss": 0.6442,
      "step": 7327
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7619179860137575,
      "learning_rate": 2.3022898821236424e-08,
      "loss": 0.5584,
      "step": 7328
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8653088917477231,
      "learning_rate": 2.2934943056101e-08,
      "loss": 0.6383,
      "step": 7329
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9468279249576645,
      "learning_rate": 2.2847153674965813e-08,
      "loss": 0.7024,
      "step": 7330
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7532869035839799,
      "learning_rate": 2.275953069278036e-08,
      "loss": 0.5766,
      "step": 7331
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7770475302388583,
      "learning_rate": 2.2672074124466145e-08,
      "loss": 0.5753,
      "step": 7332
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9552309094567915,
      "learning_rate": 2.2584783984915813e-08,
      "loss": 0.6889,
      "step": 7333
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9430881173078659,
      "learning_rate": 2.2497660288994137e-08,
      "loss": 0.681,
      "step": 7334
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7925380942108152,
      "learning_rate": 2.241070305153747e-08,
      "loss": 0.5865,
      "step": 7335
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8483727468173223,
      "learning_rate": 2.232391228735364e-08,
      "loss": 0.6069,
      "step": 7336
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9580201230614123,
      "learning_rate": 2.2237288011222156e-08,
      "loss": 0.6277,
      "step": 7337
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6540616035077086,
      "learning_rate": 2.2150830237894212e-08,
      "loss": 0.4598,
      "step": 7338
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5939019635070065,
      "learning_rate": 2.2064538982092818e-08,
      "loss": 0.5047,
      "step": 7339
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7803918132292064,
      "learning_rate": 2.1978414258512324e-08,
      "loss": 0.6045,
      "step": 7340
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8808798253134988,
      "learning_rate": 2.1892456081818888e-08,
      "loss": 0.6285,
      "step": 7341
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6342257557730554,
      "learning_rate": 2.180666446665036e-08,
      "loss": 0.478,
      "step": 7342
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9032675516702545,
      "learning_rate": 2.1721039427616162e-08,
      "loss": 0.6091,
      "step": 7343
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7420888302984697,
      "learning_rate": 2.1635580979297295e-08,
      "loss": 0.5302,
      "step": 7344
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7346921491147425,
      "learning_rate": 2.1550289136246458e-08,
      "loss": 0.5417,
      "step": 7345
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7045662810830867,
      "learning_rate": 2.1465163912988028e-08,
      "loss": 0.5171,
      "step": 7346
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7147259598666014,
      "learning_rate": 2.1380205324017852e-08,
      "loss": 0.5713,
      "step": 7347
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5889244485434171,
      "learning_rate": 2.1295413383803472e-08,
      "loss": 0.5493,
      "step": 7348
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7927584494338501,
      "learning_rate": 2.1210788106784116e-08,
      "loss": 0.5937,
      "step": 7349
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1904287972337695,
      "learning_rate": 2.1126329507370587e-08,
      "loss": 0.6513,
      "step": 7350
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7858485830915337,
      "learning_rate": 2.1042037599945382e-08,
      "loss": 0.5783,
      "step": 7351
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6998750538399491,
      "learning_rate": 2.0957912398862465e-08,
      "loss": 0.5638,
      "step": 7352
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7715934201734731,
      "learning_rate": 2.0873953918447374e-08,
      "loss": 0.5412,
      "step": 7353
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7332572984356304,
      "learning_rate": 2.079016217299756e-08,
      "loss": 0.5368,
      "step": 7354
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.906684047198429,
      "learning_rate": 2.0706537176781725e-08,
      "loss": 0.6709,
      "step": 7355
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6968944961918091,
      "learning_rate": 2.062307894404047e-08,
      "loss": 0.4994,
      "step": 7356
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7297407488979746,
      "learning_rate": 2.053978748898577e-08,
      "loss": 0.5915,
      "step": 7357
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6533872329803413,
      "learning_rate": 2.0456662825801385e-08,
      "loss": 0.5079,
      "step": 7358
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9324718133578759,
      "learning_rate": 2.0373704968642436e-08,
      "loss": 0.564,
      "step": 7359
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7564071407681818,
      "learning_rate": 2.0290913931635734e-08,
      "loss": 0.5515,
      "step": 7360
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.791927611658742,
      "learning_rate": 2.020828972888e-08,
      "loss": 0.5713,
      "step": 7361
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.688344054909427,
      "learning_rate": 2.0125832374445096e-08,
      "loss": 0.5076,
      "step": 7362
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0923643566937382,
      "learning_rate": 2.0043541882372783e-08,
      "loss": 0.6203,
      "step": 7363
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8687605679872175,
      "learning_rate": 1.9961418266676078e-08,
      "loss": 0.6389,
      "step": 7364
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7001244065773278,
      "learning_rate": 1.9879461541339905e-08,
      "loss": 0.5755,
      "step": 7365
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7328643788992382,
      "learning_rate": 1.979767172032054e-08,
      "loss": 0.5902,
      "step": 7366
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6380921976386079,
      "learning_rate": 1.971604881754607e-08,
      "loss": 0.5541,
      "step": 7367
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7210434911134327,
      "learning_rate": 1.963459284691593e-08,
      "loss": 0.5525,
      "step": 7368
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8062029944399082,
      "learning_rate": 1.955330382230136e-08,
      "loss": 0.6188,
      "step": 7369
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2224542636201894,
      "learning_rate": 1.9472181757544838e-08,
      "loss": 0.6359,
      "step": 7370
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8517990928845706,
      "learning_rate": 1.9391226666460757e-08,
      "loss": 0.6318,
      "step": 7371
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8870327592080854,
      "learning_rate": 1.9310438562834765e-08,
      "loss": 0.6024,
      "step": 7372
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7804076023913208,
      "learning_rate": 1.92298174604244e-08,
      "loss": 0.6133,
      "step": 7373
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7633874061136199,
      "learning_rate": 1.914936337295847e-08,
      "loss": 0.5507,
      "step": 7374
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6976314768152295,
      "learning_rate": 1.9069076314137678e-08,
      "loss": 0.5203,
      "step": 7375
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7132493863241013,
      "learning_rate": 1.898895629763375e-08,
      "loss": 0.5458,
      "step": 7376
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6431605460472087,
      "learning_rate": 1.8909003337090557e-08,
      "loss": 0.483,
      "step": 7377
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6924352692329612,
      "learning_rate": 1.8829217446123203e-08,
      "loss": 0.5264,
      "step": 7378
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6651086187715306,
      "learning_rate": 1.8749598638318265e-08,
      "loss": 0.5175,
      "step": 7379
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8293932579453627,
      "learning_rate": 1.867014692723412e-08,
      "loss": 0.5979,
      "step": 7380
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5332156596110267,
      "learning_rate": 1.8590862326400612e-08,
      "loss": 0.6547,
      "step": 7381
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8448468623784054,
      "learning_rate": 1.8511744849318833e-08,
      "loss": 0.6288,
      "step": 7382
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7525873697326189,
      "learning_rate": 1.8432794509462003e-08,
      "loss": 0.5914,
      "step": 7383
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6062885852603496,
      "learning_rate": 1.8354011320274255e-08,
      "loss": 0.5123,
      "step": 7384
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7319035825840275,
      "learning_rate": 1.827539529517175e-08,
      "loss": 0.5747,
      "step": 7385
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8080882088267415,
      "learning_rate": 1.8196946447541882e-08,
      "loss": 0.6239,
      "step": 7386
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6648203856734201,
      "learning_rate": 1.8118664790743532e-08,
      "loss": 0.4993,
      "step": 7387
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7230985894853645,
      "learning_rate": 1.8040550338107585e-08,
      "loss": 0.5071,
      "step": 7388
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8253701678454294,
      "learning_rate": 1.7962603102935847e-08,
      "loss": 0.5434,
      "step": 7389
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6550046141942629,
      "learning_rate": 1.7884823098502035e-08,
      "loss": 0.4814,
      "step": 7390
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9027399667574526,
      "learning_rate": 1.7807210338051216e-08,
      "loss": 0.5575,
      "step": 7391
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.697773544514756,
      "learning_rate": 1.7729764834800155e-08,
      "loss": 0.5795,
      "step": 7392
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7043387893402796,
      "learning_rate": 1.7652486601936745e-08,
      "loss": 0.5851,
      "step": 7393
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.774481556592844,
      "learning_rate": 1.7575375652621016e-08,
      "loss": 0.6229,
      "step": 7394
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7548891288143791,
      "learning_rate": 1.749843199998391e-08,
      "loss": 0.5257,
      "step": 7395
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9761776761402281,
      "learning_rate": 1.7421655657128274e-08,
      "loss": 0.5469,
      "step": 7396
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9522549230260436,
      "learning_rate": 1.7345046637128213e-08,
      "loss": 0.5645,
      "step": 7397
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6655055146099254,
      "learning_rate": 1.726860495302951e-08,
      "loss": 0.5063,
      "step": 7398
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.807614336920565,
      "learning_rate": 1.7192330617849415e-08,
      "loss": 0.5202,
      "step": 7399
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6579943421904548,
      "learning_rate": 1.7116223644576433e-08,
      "loss": 0.5387,
      "step": 7400
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8859530869317134,
      "learning_rate": 1.704028404617108e-08,
      "loss": 0.6233,
      "step": 7401
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7800016614422606,
      "learning_rate": 1.6964511835564908e-08,
      "loss": 0.5638,
      "step": 7402
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7031839129420778,
      "learning_rate": 1.688890702566126e-08,
      "loss": 0.5555,
      "step": 7403
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6573089507009723,
      "learning_rate": 1.681346962933472e-08,
      "loss": 0.5605,
      "step": 7404
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6060935506335959,
      "learning_rate": 1.6738199659431463e-08,
      "loss": 0.5208,
      "step": 7405
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9492646602240872,
      "learning_rate": 1.6663097128769343e-08,
      "loss": 0.6424,
      "step": 7406
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6295299022225685,
      "learning_rate": 1.6588162050137466e-08,
      "loss": 0.5073,
      "step": 7407
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.72031449700312,
      "learning_rate": 1.65133944362964e-08,
      "loss": 0.5363,
      "step": 7408
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5847103223098346,
      "learning_rate": 1.643879429997841e-08,
      "loss": 0.5116,
      "step": 7409
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9150892898250279,
      "learning_rate": 1.6364361653887104e-08,
      "loss": 0.6601,
      "step": 7410
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8738454243198037,
      "learning_rate": 1.629009651069757e-08,
      "loss": 0.6274,
      "step": 7411
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8284571349550164,
      "learning_rate": 1.621599888305636e-08,
      "loss": 0.6604,
      "step": 7412
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6714369343398056,
      "learning_rate": 1.61420687835816e-08,
      "loss": 0.5306,
      "step": 7413
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6938815098184956,
      "learning_rate": 1.606830622486266e-08,
      "loss": 0.5666,
      "step": 7414
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7671555204626684,
      "learning_rate": 1.599471121946072e-08,
      "loss": 0.5117,
      "step": 7415
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8545314562719327,
      "learning_rate": 1.5921283779908202e-08,
      "loss": 0.5999,
      "step": 7416
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8685526562210228,
      "learning_rate": 1.5848023918708986e-08,
      "loss": 0.6141,
      "step": 7417
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0092263866324755,
      "learning_rate": 1.577493164833843e-08,
      "loss": 0.6168,
      "step": 7418
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8591542960214308,
      "learning_rate": 1.5702006981243355e-08,
      "loss": 0.6605,
      "step": 7419
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8004052873341032,
      "learning_rate": 1.562924992984227e-08,
      "loss": 0.6,
      "step": 7420
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7848706576290132,
      "learning_rate": 1.555666050652471e-08,
      "loss": 0.6131,
      "step": 7421
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8597694772581106,
      "learning_rate": 1.548423872365212e-08,
      "loss": 0.5943,
      "step": 7422
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7626861860910683,
      "learning_rate": 1.541198459355708e-08,
      "loss": 0.5937,
      "step": 7423
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7279878451756645,
      "learning_rate": 1.5339898128543637e-08,
      "loss": 0.4908,
      "step": 7424
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8642458864623317,
      "learning_rate": 1.5267979340887415e-08,
      "loss": 0.6659,
      "step": 7425
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.822340669725819,
      "learning_rate": 1.5196228242835507e-08,
      "loss": 0.6116,
      "step": 7426
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8263607963159706,
      "learning_rate": 1.5124644846606248e-08,
      "loss": 0.6259,
      "step": 7427
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7760617981752432,
      "learning_rate": 1.505322916438967e-08,
      "loss": 0.5328,
      "step": 7428
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0307063617662702,
      "learning_rate": 1.498198120834704e-08,
      "loss": 0.6466,
      "step": 7429
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1231308044113038,
      "learning_rate": 1.4910900990611207e-08,
      "loss": 0.6534,
      "step": 7430
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7148018690501072,
      "learning_rate": 1.4839988523286384e-08,
      "loss": 0.56,
      "step": 7431
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9841373491236265,
      "learning_rate": 1.4769243818448129e-08,
      "loss": 0.6287,
      "step": 7432
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7792377440416043,
      "learning_rate": 1.4698666888143585e-08,
      "loss": 0.6155,
      "step": 7433
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.666459001113147,
      "learning_rate": 1.4628257744391248e-08,
      "loss": 0.5487,
      "step": 7434
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7087115673400899,
      "learning_rate": 1.4558016399181084e-08,
      "loss": 0.4958,
      "step": 7435
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6818492188497828,
      "learning_rate": 1.4487942864474522e-08,
      "loss": 0.5873,
      "step": 7436
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6286302695212599,
      "learning_rate": 1.4418037152204354e-08,
      "loss": 0.4887,
      "step": 7437
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8460944440847032,
      "learning_rate": 1.4348299274274612e-08,
      "loss": 0.5765,
      "step": 7438
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8292750788786889,
      "learning_rate": 1.4278729242561127e-08,
      "loss": 0.637,
      "step": 7439
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.678072001461441,
      "learning_rate": 1.420932706891087e-08,
      "loss": 0.5331,
      "step": 7440
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7772186134042712,
      "learning_rate": 1.4140092765142387e-08,
      "loss": 0.5564,
      "step": 7441
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6797351926525314,
      "learning_rate": 1.4071026343045355e-08,
      "loss": 0.5248,
      "step": 7442
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8103672034012288,
      "learning_rate": 1.400212781438126e-08,
      "loss": 0.6331,
      "step": 7443
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.936394948464033,
      "learning_rate": 1.3933397190882823e-08,
      "loss": 0.6616,
      "step": 7444
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7610103144312518,
      "learning_rate": 1.3864834484253906e-08,
      "loss": 0.508,
      "step": 7445
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8859683214603586,
      "learning_rate": 1.3796439706170282e-08,
      "loss": 0.6462,
      "step": 7446
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6440572387053101,
      "learning_rate": 1.3728212868278855e-08,
      "loss": 0.5618,
      "step": 7447
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.697272406276553,
      "learning_rate": 1.3660153982197775e-08,
      "loss": 0.5075,
      "step": 7448
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8296097730272598,
      "learning_rate": 1.3592263059516884e-08,
      "loss": 0.6706,
      "step": 7449
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6039206591322959,
      "learning_rate": 1.3524540111797378e-08,
      "loss": 0.5221,
      "step": 7450
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8865933372831674,
      "learning_rate": 1.3456985150571587e-08,
      "loss": 0.6587,
      "step": 7451
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7016480035513186,
      "learning_rate": 1.338959818734342e-08,
      "loss": 0.5517,
      "step": 7452
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9516630256351832,
      "learning_rate": 1.3322379233588365e-08,
      "loss": 0.58,
      "step": 7453
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9685653087456392,
      "learning_rate": 1.3255328300752932e-08,
      "loss": 0.6121,
      "step": 7454
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8724263003376227,
      "learning_rate": 1.318844540025521e-08,
      "loss": 0.5344,
      "step": 7455
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8628209379113889,
      "learning_rate": 1.3121730543484865e-08,
      "loss": 0.6213,
      "step": 7456
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8279215376848827,
      "learning_rate": 1.3055183741802478e-08,
      "loss": 0.6529,
      "step": 7457
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6928467558296452,
      "learning_rate": 1.2988805006540316e-08,
      "loss": 0.5445,
      "step": 7458
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8882408833985649,
      "learning_rate": 1.2922594349002114e-08,
      "loss": 0.6437,
      "step": 7459
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8568455764166591,
      "learning_rate": 1.2856551780462854e-08,
      "loss": 0.6094,
      "step": 7460
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8757168124385708,
      "learning_rate": 1.2790677312168763e-08,
      "loss": 0.5459,
      "step": 7461
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7134921079396285,
      "learning_rate": 1.2724970955337755e-08,
      "loss": 0.5288,
      "step": 7462
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.745076102036001,
      "learning_rate": 1.2659432721158769e-08,
      "loss": 0.5164,
      "step": 7463
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.316617603426175,
      "learning_rate": 1.2594062620792323e-08,
      "loss": 0.5783,
      "step": 7464
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6863625560447916,
      "learning_rate": 1.2528860665370288e-08,
      "loss": 0.5186,
      "step": 7465
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6393061656456104,
      "learning_rate": 1.2463826865995896e-08,
      "loss": 0.5161,
      "step": 7466
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0707017558819005,
      "learning_rate": 1.2398961233743733e-08,
      "loss": 0.6401,
      "step": 7467
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6396429205127209,
      "learning_rate": 1.2334263779659737e-08,
      "loss": 0.5309,
      "step": 7468
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9416409047348248,
      "learning_rate": 1.226973451476121e-08,
      "loss": 0.6542,
      "step": 7469
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8974829573647967,
      "learning_rate": 1.2205373450036694e-08,
      "loss": 0.6397,
      "step": 7470
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.811570936155315,
      "learning_rate": 1.214118059644631e-08,
      "loss": 0.5829,
      "step": 7471
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6586064858591256,
      "learning_rate": 1.2077155964921536e-08,
      "loss": 0.5774,
      "step": 7472
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7827020231054449,
      "learning_rate": 1.2013299566364988e-08,
      "loss": 0.6318,
      "step": 7473
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7747931346889434,
      "learning_rate": 1.1949611411650628e-08,
      "loss": 0.5929,
      "step": 7474
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0526926416869506,
      "learning_rate": 1.1886091511624118e-08,
      "loss": 0.6061,
      "step": 7475
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6893402279171401,
      "learning_rate": 1.182273987710214e-08,
      "loss": 0.5609,
      "step": 7476
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7755770870514901,
      "learning_rate": 1.1759556518872726e-08,
      "loss": 0.5239,
      "step": 7477
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.808182937599015,
      "learning_rate": 1.1696541447695495e-08,
      "loss": 0.5811,
      "step": 7478
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7554955483621599,
      "learning_rate": 1.1633694674301198e-08,
      "loss": 0.5583,
      "step": 7479
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6532446438972842,
      "learning_rate": 1.1571016209391938e-08,
      "loss": 0.5547,
      "step": 7480
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9858596140611197,
      "learning_rate": 1.1508506063641177e-08,
      "loss": 0.68,
      "step": 7481
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8526843507711517,
      "learning_rate": 1.1446164247693846e-08,
      "loss": 0.5612,
      "step": 7482
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8569036759493792,
      "learning_rate": 1.1383990772166118e-08,
      "loss": 0.6691,
      "step": 7483
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9169812059723932,
      "learning_rate": 1.1321985647645415e-08,
      "loss": 0.6925,
      "step": 7484
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6331171599888407,
      "learning_rate": 1.1260148884690512e-08,
      "loss": 0.5513,
      "step": 7485
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6837030869785222,
      "learning_rate": 1.1198480493831764e-08,
      "loss": 0.5164,
      "step": 7486
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7809302427240681,
      "learning_rate": 1.1136980485570436e-08,
      "loss": 0.5731,
      "step": 7487
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.138571079313505,
      "learning_rate": 1.1075648870379484e-08,
      "loss": 0.6818,
      "step": 7488
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9838949559959377,
      "learning_rate": 1.1014485658702888e-08,
      "loss": 0.6787,
      "step": 7489
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0092771007766697,
      "learning_rate": 1.0953490860956316e-08,
      "loss": 0.6073,
      "step": 7490
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8794828659376731,
      "learning_rate": 1.0892664487526349e-08,
      "loss": 0.6543,
      "step": 7491
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8545104096145523,
      "learning_rate": 1.0832006548771255e-08,
      "loss": 0.5683,
      "step": 7492
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7436780616702736,
      "learning_rate": 1.0771517055020329e-08,
      "loss": 0.4984,
      "step": 7493
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7381212442322622,
      "learning_rate": 1.071119601657433e-08,
      "loss": 0.5462,
      "step": 7494
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6773485257548485,
      "learning_rate": 1.0651043443705265e-08,
      "loss": 0.5475,
      "step": 7495
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7800486506403884,
      "learning_rate": 1.0591059346656606e-08,
      "loss": 0.5794,
      "step": 7496
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6908874198785283,
      "learning_rate": 1.0531243735642958e-08,
      "loss": 0.5335,
      "step": 7497
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7456726693453808,
      "learning_rate": 1.0471596620850287e-08,
      "loss": 0.5766,
      "step": 7498
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.609232105365008,
      "learning_rate": 1.0412118012435912e-08,
      "loss": 0.5124,
      "step": 7499
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.709611044411409,
      "learning_rate": 1.0352807920528284e-08,
      "loss": 0.6198,
      "step": 7500
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8431788329165055,
      "learning_rate": 1.0293666355227437e-08,
      "loss": 0.595,
      "step": 7501
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6091891525029052,
      "learning_rate": 1.0234693326604648e-08,
      "loss": 0.523,
      "step": 7502
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7246699663512606,
      "learning_rate": 1.0175888844702108e-08,
      "loss": 0.5218,
      "step": 7503
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8068193393415852,
      "learning_rate": 1.0117252919533802e-08,
      "loss": 0.5332,
      "step": 7504
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6373922915593233,
      "learning_rate": 1.005878556108486e-08,
      "loss": 0.5682,
      "step": 7505
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7515995577292833,
      "learning_rate": 1.0000486779311645e-08,
      "loss": 0.5191,
      "step": 7506
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6966555053826615,
      "learning_rate": 9.942356584141775e-09,
      "loss": 0.577,
      "step": 7507
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7129370865897096,
      "learning_rate": 9.88439498547422e-09,
      "loss": 0.5538,
      "step": 7508
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8789884911484607,
      "learning_rate": 9.826601993179196e-09,
      "loss": 0.6099,
      "step": 7509
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8510842384671977,
      "learning_rate": 9.768977617098383e-09,
      "loss": 0.6461,
      "step": 7510
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9996106063734769,
      "learning_rate": 9.711521867044492e-09,
      "loss": 0.6423,
      "step": 7511
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8052995363976938,
      "learning_rate": 9.654234752801693e-09,
      "loss": 0.5758,
      "step": 7512
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7165004307117471,
      "learning_rate": 9.597116284125294e-09,
      "loss": 0.567,
      "step": 7513
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.706665762132291,
      "learning_rate": 9.540166470742072e-09,
      "loss": 0.6398,
      "step": 7514
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8344278649040886,
      "learning_rate": 9.483385322350046e-09,
      "loss": 0.6218,
      "step": 7515
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7593176809706591,
      "learning_rate": 9.426772848618258e-09,
      "loss": 0.6078,
      "step": 7516
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9479314684841266,
      "learning_rate": 9.370329059187442e-09,
      "loss": 0.5904,
      "step": 7517
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.608550619826079,
      "learning_rate": 9.314053963669244e-09,
      "loss": 0.6117,
      "step": 7518
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8107647270814977,
      "learning_rate": 9.257947571646774e-09,
      "loss": 0.6328,
      "step": 7519
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6796869017596302,
      "learning_rate": 9.20200989267439e-09,
      "loss": 0.5194,
      "step": 7520
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8902480216979246,
      "learning_rate": 9.146240936277583e-09,
      "loss": 0.6832,
      "step": 7521
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7409395001127745,
      "learning_rate": 9.090640711953312e-09,
      "loss": 0.5787,
      "step": 7522
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7028859756724618,
      "learning_rate": 9.035209229169783e-09,
      "loss": 0.5208,
      "step": 7523
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7086890186586934,
      "learning_rate": 8.97994649736622e-09,
      "loss": 0.508,
      "step": 7524
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.708304875757451,
      "learning_rate": 8.92485252595343e-09,
      "loss": 0.5651,
      "step": 7525
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.238797266998316,
      "learning_rate": 8.869927324313354e-09,
      "loss": 0.6069,
      "step": 7526
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8580278984963516,
      "learning_rate": 8.815170901799174e-09,
      "loss": 0.6286,
      "step": 7527
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9809102282323726,
      "learning_rate": 8.76058326773521e-09,
      "loss": 0.6391,
      "step": 7528
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8829039812636805,
      "learning_rate": 8.706164431417251e-09,
      "loss": 0.6717,
      "step": 7529
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6128742736129354,
      "learning_rate": 8.651914402112326e-09,
      "loss": 0.5345,
      "step": 7530
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9039532102247477,
      "learning_rate": 8.59783318905849e-09,
      "loss": 0.6712,
      "step": 7531
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9371011545992333,
      "learning_rate": 8.543920801465488e-09,
      "loss": 0.6179,
      "step": 7532
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8420040601590139,
      "learning_rate": 8.490177248513753e-09,
      "loss": 0.6019,
      "step": 7533
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7675033169086196,
      "learning_rate": 8.436602539355297e-09,
      "loss": 0.587,
      "step": 7534
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7892858097666121,
      "learning_rate": 8.383196683113491e-09,
      "loss": 0.6465,
      "step": 7535
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8587179331982684,
      "learning_rate": 8.329959688882614e-09,
      "loss": 0.6084,
      "step": 7536
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6881603006339427,
      "learning_rate": 8.276891565728528e-09,
      "loss": 0.5315,
      "step": 7537
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8182522569649632,
      "learning_rate": 8.22399232268811e-09,
      "loss": 0.6257,
      "step": 7538
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8143999278603906,
      "learning_rate": 8.171261968769606e-09,
      "loss": 0.5258,
      "step": 7539
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7768436738322156,
      "learning_rate": 8.11870051295227e-09,
      "loss": 0.5145,
      "step": 7540
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6696638197929685,
      "learning_rate": 8.066307964186947e-09,
      "loss": 0.517,
      "step": 7541
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8719131752139349,
      "learning_rate": 8.014084331395498e-09,
      "loss": 0.5838,
      "step": 7542
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5781892332883033,
      "learning_rate": 7.962029623471033e-09,
      "loss": 0.5079,
      "step": 7543
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8020540480226601,
      "learning_rate": 7.910143849278017e-09,
      "loss": 0.5469,
      "step": 7544
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7054626042109186,
      "learning_rate": 7.858427017651825e-09,
      "loss": 0.5356,
      "step": 7545
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7413876744199586,
      "learning_rate": 7.806879137399636e-09,
      "loss": 0.5502,
      "step": 7546
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6756865411673445,
      "learning_rate": 7.755500217299205e-09,
      "loss": 0.5704,
      "step": 7547
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8095412022910256,
      "learning_rate": 7.704290266099977e-09,
      "loss": 0.6823,
      "step": 7548
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8373607604943534,
      "learning_rate": 7.653249292522535e-09,
      "loss": 0.5535,
      "step": 7549
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.935621437830354,
      "learning_rate": 7.602377305258478e-09,
      "loss": 0.6729,
      "step": 7550
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7419776401823711,
      "learning_rate": 7.55167431297088e-09,
      "loss": 0.5253,
      "step": 7551
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8696753973084435,
      "learning_rate": 7.501140324293943e-09,
      "loss": 0.5868,
      "step": 7552
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7322803708925193,
      "learning_rate": 7.450775347832893e-09,
      "loss": 0.5401,
      "step": 7553
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9341471986371082,
      "learning_rate": 7.400579392164541e-09,
      "loss": 0.6244,
      "step": 7554
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.685579468148357,
      "learning_rate": 7.350552465836712e-09,
      "loss": 0.5639,
      "step": 7555
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7995344196590488,
      "learning_rate": 7.300694577368482e-09,
      "loss": 0.5198,
      "step": 7556
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.86902026154038,
      "learning_rate": 7.2510057352500595e-09,
      "loss": 0.6428,
      "step": 7557
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8713701417272404,
      "learning_rate": 7.201485947943009e-09,
      "loss": 0.6575,
      "step": 7558
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7802405231514628,
      "learning_rate": 7.152135223879918e-09,
      "loss": 0.568,
      "step": 7559
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7373286111494213,
      "learning_rate": 7.102953571464953e-09,
      "loss": 0.5952,
      "step": 7560
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9750381066857418,
      "learning_rate": 7.053940999072971e-09,
      "loss": 0.6339,
      "step": 7561
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7696393262829982,
      "learning_rate": 7.005097515050296e-09,
      "loss": 0.5303,
      "step": 7562
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6953188160217638,
      "learning_rate": 6.956423127714717e-09,
      "loss": 0.5522,
      "step": 7563
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8061153277992281,
      "learning_rate": 6.907917845354827e-09,
      "loss": 0.6269,
      "step": 7564
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5999099740193771,
      "learning_rate": 6.8595816762304635e-09,
      "loss": 0.4523,
      "step": 7565
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7719315501949956,
      "learning_rate": 6.811414628572931e-09,
      "loss": 0.5961,
      "step": 7566
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6362304056432907,
      "learning_rate": 6.763416710584557e-09,
      "loss": 0.5383,
      "step": 7567
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7091126867296464,
      "learning_rate": 6.7155879304388045e-09,
      "loss": 0.571,
      "step": 7568
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.665901378440214,
      "learning_rate": 6.66792829628049e-09,
      "loss": 0.5673,
      "step": 7569
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6725077902039263,
      "learning_rate": 6.620437816225566e-09,
      "loss": 0.5197,
      "step": 7570
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7286549194993233,
      "learning_rate": 6.573116498361009e-09,
      "loss": 0.5076,
      "step": 7571
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6323778390854087,
      "learning_rate": 6.525964350745372e-09,
      "loss": 0.5159,
      "step": 7572
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8544654773466921,
      "learning_rate": 6.478981381408011e-09,
      "loss": 0.6284,
      "step": 7573
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7153054579615711,
      "learning_rate": 6.432167598349747e-09,
      "loss": 0.5424,
      "step": 7574
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6533476176169017,
      "learning_rate": 6.385523009542426e-09,
      "loss": 0.5575,
      "step": 7575
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7119882130740285,
      "learning_rate": 6.339047622929139e-09,
      "loss": 0.5327,
      "step": 7576
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6642632343696933,
      "learning_rate": 6.292741446424111e-09,
      "loss": 0.5612,
      "step": 7577
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6621859589178055,
      "learning_rate": 6.24660448791281e-09,
      "loss": 0.5462,
      "step": 7578
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8249447035337744,
      "learning_rate": 6.200636755251954e-09,
      "loss": 0.6099,
      "step": 7579
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8682403610990475,
      "learning_rate": 6.1548382562693905e-09,
      "loss": 0.6382,
      "step": 7580
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6928112996598041,
      "learning_rate": 6.109208998763993e-09,
      "loss": 0.5973,
      "step": 7581
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.693022665382722,
      "learning_rate": 6.06374899050599e-09,
      "loss": 0.5738,
      "step": 7582
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6516607857370383,
      "learning_rate": 6.018458239236857e-09,
      "loss": 0.5562,
      "step": 7583
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8842785760140708,
      "learning_rate": 5.973336752669089e-09,
      "loss": 0.4995,
      "step": 7584
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8075849493962141,
      "learning_rate": 5.9283845384863196e-09,
      "loss": 0.5766,
      "step": 7585
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6170562325150267,
      "learning_rate": 5.883601604343536e-09,
      "loss": 0.5061,
      "step": 7586
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7602226824331247,
      "learning_rate": 5.838987957866748e-09,
      "loss": 0.5354,
      "step": 7587
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9442038542114577,
      "learning_rate": 5.794543606653213e-09,
      "loss": 0.5684,
      "step": 7588
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7863367425134906,
      "learning_rate": 5.750268558271432e-09,
      "loss": 0.5711,
      "step": 7589
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6497559128150161,
      "learning_rate": 5.706162820260818e-09,
      "loss": 0.5574,
      "step": 7590
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8846944384559432,
      "learning_rate": 5.662226400132142e-09,
      "loss": 0.6085,
      "step": 7591
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9398363031465404,
      "learning_rate": 5.618459305367418e-09,
      "loss": 0.628,
      "step": 7592
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0153481351861122,
      "learning_rate": 5.574861543419685e-09,
      "loss": 0.6186,
      "step": 7593
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5649109335713043,
      "learning_rate": 5.531433121713225e-09,
      "loss": 0.4594,
      "step": 7594
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9364037205962281,
      "learning_rate": 5.488174047643346e-09,
      "loss": 0.5367,
      "step": 7595
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6014264577058027,
      "learning_rate": 5.44508432857671e-09,
      "loss": 0.55,
      "step": 7596
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9408193818242538,
      "learning_rate": 5.4021639718510036e-09,
      "loss": 0.6493,
      "step": 7597
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8171306413202737,
      "learning_rate": 5.359412984775158e-09,
      "loss": 0.6587,
      "step": 7598
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0144819978904633,
      "learning_rate": 5.316831374629127e-09,
      "loss": 0.5918,
      "step": 7599
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7954285966062634,
      "learning_rate": 5.274419148664222e-09,
      "loss": 0.5745,
      "step": 7600
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6878477292867142,
      "learning_rate": 5.232176314102776e-09,
      "loss": 0.5426,
      "step": 7601
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8673852009716464,
      "learning_rate": 5.1901028781382585e-09,
      "loss": 0.6206,
      "step": 7602
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9948198275492526,
      "learning_rate": 5.148198847935381e-09,
      "loss": 0.5714,
      "step": 7603
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9125179997256873,
      "learning_rate": 5.106464230629992e-09,
      "loss": 0.551,
      "step": 7604
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7157607717861895,
      "learning_rate": 5.064899033328962e-09,
      "loss": 0.5627,
      "step": 7605
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6682642272129844,
      "learning_rate": 5.023503263110407e-09,
      "loss": 0.4957,
      "step": 7606
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7953733617703158,
      "learning_rate": 4.982276927023799e-09,
      "loss": 0.5285,
      "step": 7607
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.718964951000397,
      "learning_rate": 4.9412200320893e-09,
      "loss": 0.5832,
      "step": 7608
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6123040979142189,
      "learning_rate": 4.90033258529865e-09,
      "loss": 0.5128,
      "step": 7609
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9467179917810105,
      "learning_rate": 4.859614593614392e-09,
      "loss": 0.6767,
      "step": 7610
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6992169519985888,
      "learning_rate": 4.819066063970534e-09,
      "loss": 0.5393,
      "step": 7611
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.131074199317081,
      "learning_rate": 4.778687003271997e-09,
      "loss": 0.6541,
      "step": 7612
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7338188373512529,
      "learning_rate": 4.738477418394948e-09,
      "loss": 0.5243,
      "step": 7613
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6583435644039286,
      "learning_rate": 4.698437316186577e-09,
      "loss": 0.5273,
      "step": 7614
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6466553645180572,
      "learning_rate": 4.6585667034654274e-09,
      "loss": 0.5048,
      "step": 7615
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5981060631320428,
      "learning_rate": 4.618865587020959e-09,
      "loss": 0.5504,
      "step": 7616
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6898142205021601,
      "learning_rate": 4.579333973613875e-09,
      "loss": 0.5743,
      "step": 7617
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6211126441222631,
      "learning_rate": 4.539971869976011e-09,
      "loss": 0.5334,
      "step": 7618
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6840162615366923,
      "learning_rate": 4.50077928281023e-09,
      "loss": 0.5295,
      "step": 7619
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7990232594480967,
      "learning_rate": 4.4617562187907465e-09,
      "loss": 0.528,
      "step": 7620
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.686368405213289,
      "learning_rate": 4.422902684562801e-09,
      "loss": 0.5203,
      "step": 7621
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6491699685270039,
      "learning_rate": 4.384218686742547e-09,
      "loss": 0.5418,
      "step": 7622
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9100667459930547,
      "learning_rate": 4.345704231917713e-09,
      "loss": 0.6337,
      "step": 7623
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1719803524622097,
      "learning_rate": 4.30735932664672e-09,
      "loss": 0.6221,
      "step": 7624
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8048379569254922,
      "learning_rate": 4.2691839774594565e-09,
      "loss": 0.5924,
      "step": 7625
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7802571740658417,
      "learning_rate": 4.2311781908567216e-09,
      "loss": 0.5241,
      "step": 7626
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.845511673519417,
      "learning_rate": 4.193341973310449e-09,
      "loss": 0.6649,
      "step": 7627
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6393972393716844,
      "learning_rate": 4.155675331263819e-09,
      "loss": 0.4927,
      "step": 7628
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6890336142953265,
      "learning_rate": 4.118178271131034e-09,
      "loss": 0.5183,
      "step": 7629
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8202212507536755,
      "learning_rate": 4.080850799297431e-09,
      "loss": 0.6735,
      "step": 7630
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.647954104831195,
      "learning_rate": 4.043692922119591e-09,
      "loss": 0.5316,
      "step": 7631
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8259723120291497,
      "learning_rate": 4.006704645925007e-09,
      "loss": 0.6294,
      "step": 7632
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7872624547178528,
      "learning_rate": 3.969885977012533e-09,
      "loss": 0.5916,
      "step": 7633
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6404077507063713,
      "learning_rate": 3.933236921651817e-09,
      "loss": 0.5218,
      "step": 7634
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5060169456152206,
      "learning_rate": 3.896757486083868e-09,
      "loss": 0.5951,
      "step": 7635
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7093728253530815,
      "learning_rate": 3.860447676520828e-09,
      "loss": 0.5539,
      "step": 7636
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8092384093324037,
      "learning_rate": 3.824307499145862e-09,
      "loss": 0.5684,
      "step": 7637
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7659066425808306,
      "learning_rate": 3.7883369601132694e-09,
      "loss": 0.5319,
      "step": 7638
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7761458426071753,
      "learning_rate": 3.752536065548373e-09,
      "loss": 0.5269,
      "step": 7639
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7568130168088266,
      "learning_rate": 3.7169048215478504e-09,
      "loss": 0.61,
      "step": 7640
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6244437764699241,
      "learning_rate": 3.6814432341790713e-09,
      "loss": 0.4834,
      "step": 7641
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8281084506953066,
      "learning_rate": 3.6461513094810936e-09,
      "loss": 0.5916,
      "step": 7642
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8286724921222312,
      "learning_rate": 3.6110290534635547e-09,
      "loss": 0.6562,
      "step": 7643
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8075634307883415,
      "learning_rate": 3.576076472107448e-09,
      "loss": 0.5911,
      "step": 7644
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.787500315394439,
      "learning_rate": 3.5412935713649004e-09,
      "loss": 0.5502,
      "step": 7645
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8787649689236644,
      "learning_rate": 3.506680357158953e-09,
      "loss": 0.6382,
      "step": 7646
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6937920361951332,
      "learning_rate": 3.4722368353840013e-09,
      "loss": 0.5517,
      "step": 7647
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6960367795578051,
      "learning_rate": 3.437963011905354e-09,
      "loss": 0.567,
      "step": 7648
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1229765833217076,
      "learning_rate": 3.4038588925594524e-09,
      "loss": 0.6191,
      "step": 7649
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8310550563839344,
      "learning_rate": 3.369924483153985e-09,
      "loss": 0.603,
      "step": 7650
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0452837904722785,
      "learning_rate": 3.336159789467552e-09,
      "loss": 0.6169,
      "step": 7651
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8207085435232901,
      "learning_rate": 3.3025648172498865e-09,
      "loss": 0.6807,
      "step": 7652
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7210016621085138,
      "learning_rate": 3.2691395722219685e-09,
      "loss": 0.6257,
      "step": 7653
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7735222422313226,
      "learning_rate": 3.2358840600756888e-09,
      "loss": 0.5507,
      "step": 7654
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9065330390227576,
      "learning_rate": 3.2027982864740733e-09,
      "loss": 0.6434,
      "step": 7655
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.770361162034577,
      "learning_rate": 3.169882257051504e-09,
      "loss": 0.6266,
      "step": 7656
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6561794731907594,
      "learning_rate": 3.1371359774129414e-09,
      "loss": 0.487,
      "step": 7657
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7370926196644199,
      "learning_rate": 3.104559453134925e-09,
      "loss": 0.5795,
      "step": 7658
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7709353637612014,
      "learning_rate": 3.0721526897649064e-09,
      "loss": 0.5098,
      "step": 7659
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.644998010024961,
      "learning_rate": 3.0399156928213596e-09,
      "loss": 0.5302,
      "step": 7660
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8908427809021644,
      "learning_rate": 3.007848467793894e-09,
      "loss": 0.5863,
      "step": 7661
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.905303774682902,
      "learning_rate": 2.9759510201432526e-09,
      "loss": 0.6324,
      "step": 7662
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9907247459803514,
      "learning_rate": 2.9442233553012006e-09,
      "loss": 0.6295,
      "step": 7663
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.99766259334281,
      "learning_rate": 2.9126654786706394e-09,
      "loss": 0.6072,
      "step": 7664
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8583573024255204,
      "learning_rate": 2.881277395625603e-09,
      "loss": 0.6481,
      "step": 7665
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7610332640630907,
      "learning_rate": 2.8500591115111493e-09,
      "loss": 0.5614,
      "step": 7666
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9271367726365909,
      "learning_rate": 2.8190106316432483e-09,
      "loss": 0.5837,
      "step": 7667
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6844169353815958,
      "learning_rate": 2.788131961309448e-09,
      "loss": 0.4852,
      "step": 7668
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8008408192047864,
      "learning_rate": 2.7574231057678753e-09,
      "loss": 0.5741,
      "step": 7669
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6076879576816461,
      "learning_rate": 2.7268840702479033e-09,
      "loss": 0.5035,
      "step": 7670
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8505519349028723,
      "learning_rate": 2.6965148599500387e-09,
      "loss": 0.6122,
      "step": 7671
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7036491110490439,
      "learning_rate": 2.666315480045922e-09,
      "loss": 0.5529,
      "step": 7672
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1616194954521126,
      "learning_rate": 2.6362859356781065e-09,
      "loss": 0.6597,
      "step": 7673
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6834273032154738,
      "learning_rate": 2.606426231960279e-09,
      "loss": 0.5729,
      "step": 7674
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9741503292856879,
      "learning_rate": 2.5767363739773728e-09,
      "loss": 0.6398,
      "step": 7675
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6662352293095339,
      "learning_rate": 2.547216366785121e-09,
      "loss": 0.5717,
      "step": 7676
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8953487963619526,
      "learning_rate": 2.5178662154106136e-09,
      "loss": 0.6119,
      "step": 7677
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8857828237363115,
      "learning_rate": 2.488685924851741e-09,
      "loss": 0.6418,
      "step": 7678
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7400300465617861,
      "learning_rate": 2.45967550007764e-09,
      "loss": 0.6139,
      "step": 7679
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6322052522698717,
      "learning_rate": 2.4308349460285817e-09,
      "loss": 0.5086,
      "step": 7680
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7535875327503674,
      "learning_rate": 2.4021642676156363e-09,
      "loss": 0.5437,
      "step": 7681
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9387460652472008,
      "learning_rate": 2.373663469721232e-09,
      "loss": 0.6471,
      "step": 7682
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8836071519759069,
      "learning_rate": 2.3453325571988203e-09,
      "loss": 0.6758,
      "step": 7683
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.751826562558573,
      "learning_rate": 2.317171534872764e-09,
      "loss": 0.6335,
      "step": 7684
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8693007266561128,
      "learning_rate": 2.28918040753856e-09,
      "loss": 0.5482,
      "step": 7685
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6629854904796727,
      "learning_rate": 2.2613591799628408e-09,
      "loss": 0.53,
      "step": 7686
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8247883006609313,
      "learning_rate": 2.2337078568833713e-09,
      "loss": 0.5913,
      "step": 7687
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9207997536665475,
      "learning_rate": 2.2062264430087185e-09,
      "loss": 0.6562,
      "step": 7688
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9820487790399751,
      "learning_rate": 2.178914943018806e-09,
      "loss": 0.6132,
      "step": 7689
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7787798399632668,
      "learning_rate": 2.1517733615644686e-09,
      "loss": 0.6295,
      "step": 7690
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8993581695610413,
      "learning_rate": 2.1248017032675645e-09,
      "loss": 0.6496,
      "step": 7691
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4198671075398537,
      "learning_rate": 2.0979999727211985e-09,
      "loss": 0.6371,
      "step": 7692
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9520256670361313,
      "learning_rate": 2.071368174489274e-09,
      "loss": 0.6709,
      "step": 7693
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6894919582458224,
      "learning_rate": 2.0449063131070534e-09,
      "loss": 0.5331,
      "step": 7694
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.906231865334116,
      "learning_rate": 2.0186143930807087e-09,
      "loss": 0.6682,
      "step": 7695
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8742463885006678,
      "learning_rate": 1.992492418887437e-09,
      "loss": 0.6608,
      "step": 7696
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6993170826483747,
      "learning_rate": 1.966540394975458e-09,
      "loss": 0.5378,
      "step": 7697
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6062064496637353,
      "learning_rate": 1.940758325764347e-09,
      "loss": 0.5204,
      "step": 7698
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1752873068564542,
      "learning_rate": 1.915146215644259e-09,
      "loss": 0.6338,
      "step": 7699
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6096760859842739,
      "learning_rate": 1.8897040689768164e-09,
      "loss": 0.5329,
      "step": 7700
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7857277761261512,
      "learning_rate": 1.8644318900945532e-09,
      "loss": 0.6288,
      "step": 7701
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.2029585915958085,
      "learning_rate": 1.8393296833011386e-09,
      "loss": 0.6395,
      "step": 7702
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6362993007555012,
      "learning_rate": 1.8143974528710414e-09,
      "loss": 0.4728,
      "step": 7703
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7797341961480369,
      "learning_rate": 1.7896352030499772e-09,
      "loss": 0.5878,
      "step": 7704
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8110437963890815,
      "learning_rate": 1.7650429380547948e-09,
      "loss": 0.5823,
      "step": 7705
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7500184973892016,
      "learning_rate": 1.7406206620732556e-09,
      "loss": 0.54,
      "step": 7706
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8058649993254625,
      "learning_rate": 1.7163683792642546e-09,
      "loss": 0.5941,
      "step": 7707
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8857687890081204,
      "learning_rate": 1.692286093757711e-09,
      "loss": 0.5895,
      "step": 7708
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8903378301260115,
      "learning_rate": 1.6683738096544553e-09,
      "loss": 0.6421,
      "step": 7709
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8627915731427043,
      "learning_rate": 1.6446315310266746e-09,
      "loss": 0.5657,
      "step": 7710
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6371493798251553,
      "learning_rate": 1.6210592619173569e-09,
      "loss": 0.5156,
      "step": 7711
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6422608474991933,
      "learning_rate": 1.5976570063405136e-09,
      "loss": 0.573,
      "step": 7712
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9502088844667549,
      "learning_rate": 1.5744247682815126e-09,
      "loss": 0.6595,
      "step": 7713
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7767363529129933,
      "learning_rate": 1.5513625516964113e-09,
      "loss": 0.5963,
      "step": 7714
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9082009653060671,
      "learning_rate": 1.5284703605125126e-09,
      "loss": 0.6714,
      "step": 7715
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7720588542362435,
      "learning_rate": 1.5057481986280318e-09,
      "loss": 0.5769,
      "step": 7716
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9967270142968973,
      "learning_rate": 1.48319606991254e-09,
      "loss": 0.5964,
      "step": 7717
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7945730618890671,
      "learning_rate": 1.460813978206188e-09,
      "loss": 0.5834,
      "step": 7718
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8832620734287215,
      "learning_rate": 1.4386019273204819e-09,
      "loss": 0.6154,
      "step": 7719
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.906122491245456,
      "learning_rate": 1.416559921037952e-09,
      "loss": 0.682,
      "step": 7720
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7870923371120171,
      "learning_rate": 1.3946879631121511e-09,
      "loss": 0.5618,
      "step": 7721
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9082631059173026,
      "learning_rate": 1.3729860572675446e-09,
      "loss": 0.5875,
      "step": 7722
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6666193874646723,
      "learning_rate": 1.3514542071998425e-09,
      "loss": 0.5145,
      "step": 7723
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7670320844392445,
      "learning_rate": 1.3300924165756678e-09,
      "loss": 0.6272,
      "step": 7724
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1991674558183119,
      "learning_rate": 1.3089006890325549e-09,
      "loss": 0.6411,
      "step": 7725
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.895322372351799,
      "learning_rate": 1.287879028179395e-09,
      "loss": 0.6169,
      "step": 7726
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.161419626608704,
      "learning_rate": 1.2670274375959911e-09,
      "loss": 0.5765,
      "step": 7727
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0821107551447402,
      "learning_rate": 1.2463459208331696e-09,
      "loss": 0.6794,
      "step": 7728
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7683768294915218,
      "learning_rate": 1.2258344814125577e-09,
      "loss": 0.5954,
      "step": 7729
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6318128180742039,
      "learning_rate": 1.2054931228272502e-09,
      "loss": 0.5592,
      "step": 7730
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7149259086210914,
      "learning_rate": 1.185321848541032e-09,
      "loss": 0.5495,
      "step": 7731
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7055943271727431,
      "learning_rate": 1.1653206619889334e-09,
      "loss": 0.5383,
      "step": 7732
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7022404021510396,
      "learning_rate": 1.145489566577007e-09,
      "loss": 0.5056,
      "step": 7733
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.125695385968074,
      "learning_rate": 1.1258285656822186e-09,
      "loss": 0.682,
      "step": 7734
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7435298571653468,
      "learning_rate": 1.1063376626525567e-09,
      "loss": 0.5697,
      "step": 7735
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6841377407229922,
      "learning_rate": 1.087016860807255e-09,
      "loss": 0.5257,
      "step": 7736
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6312959323075776,
      "learning_rate": 1.0678661634363484e-09,
      "loss": 0.5139,
      "step": 7737
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7454446328939776,
      "learning_rate": 1.048885573801117e-09,
      "loss": 0.5308,
      "step": 7738
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.930261987513385,
      "learning_rate": 1.0300750951335313e-09,
      "loss": 0.641,
      "step": 7739
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.651838990915629,
      "learning_rate": 1.0114347306370286e-09,
      "loss": 0.5046,
      "step": 7740
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6633323430717634,
      "learning_rate": 9.929644834858474e-10,
      "loss": 0.5125,
      "step": 7741
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.3962291920591583,
      "learning_rate": 9.7466435682525e-10,
      "loss": 0.6676,
      "step": 7742
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5905443934316996,
      "learning_rate": 9.565343537714098e-10,
      "loss": 0.4721,
      "step": 7743
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7652373649870421,
      "learning_rate": 9.385744774119687e-10,
      "loss": 0.5672,
      "step": 7744
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8355364862447022,
      "learning_rate": 9.207847308050354e-10,
      "loss": 0.5349,
      "step": 7745
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9812709719835333,
      "learning_rate": 9.031651169801868e-10,
      "loss": 0.6637,
      "step": 7746
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.640559651817363,
      "learning_rate": 8.857156389378007e-10,
      "loss": 0.5059,
      "step": 7747
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7622410352315059,
      "learning_rate": 8.684362996492778e-10,
      "loss": 0.6082,
      "step": 7748
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.759436239412125,
      "learning_rate": 8.513271020572644e-10,
      "loss": 0.5591,
      "step": 7749
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7087432165089936,
      "learning_rate": 8.343880490750965e-10,
      "loss": 0.5734,
      "step": 7750
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9115974543764354,
      "learning_rate": 8.176191435874668e-10,
      "loss": 0.684,
      "step": 7751
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8429322905559871,
      "learning_rate": 8.01020388449869e-10,
      "loss": 0.6412,
      "step": 7752
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7105458421511506,
      "learning_rate": 7.845917864889307e-10,
      "loss": 0.5514,
      "step": 7753
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7190693131106404,
      "learning_rate": 7.683333405023029e-10,
      "loss": 0.5358,
      "step": 7754
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6579933881946854,
      "learning_rate": 7.522450532585489e-10,
      "loss": 0.5174,
      "step": 7755
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8520475146054648,
      "learning_rate": 7.36326927497366e-10,
      "loss": 0.5876,
      "step": 7756
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8391052673828491,
      "learning_rate": 7.205789659294748e-10,
      "loss": 0.5632,
      "step": 7757
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.687771013003076,
      "learning_rate": 7.050011712365078e-10,
      "loss": 0.5678,
      "step": 7758
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7231398444955385,
      "learning_rate": 6.895935460713431e-10,
      "loss": 0.5073,
      "step": 7759
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0439955976693296,
      "learning_rate": 6.743560930576597e-10,
      "loss": 0.5563,
      "step": 7760
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7748237796749509,
      "learning_rate": 6.592888147901599e-10,
      "loss": 0.6211,
      "step": 7761
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6697530504609222,
      "learning_rate": 6.44391713834791e-10,
      "loss": 0.5906,
      "step": 7762
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8173992007651356,
      "learning_rate": 6.296647927283016e-10,
      "loss": 0.5727,
      "step": 7763
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6995655725056337,
      "learning_rate": 6.151080539784637e-10,
      "loss": 0.5837,
      "step": 7764
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8340377363415067,
      "learning_rate": 6.007215000641829e-10,
      "loss": 0.5955,
      "step": 7765
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1362039877575774,
      "learning_rate": 5.865051334354998e-10,
      "loss": 0.5827,
      "step": 7766
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9204452107089917,
      "learning_rate": 5.724589565130333e-10,
      "loss": 0.7043,
      "step": 7767
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8969727321660015,
      "learning_rate": 5.585829716889812e-10,
      "loss": 0.6007,
      "step": 7768
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7245955990855155,
      "learning_rate": 5.4487718132612e-10,
      "loss": 0.596,
      "step": 7769
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7821150569561376,
      "learning_rate": 5.313415877583605e-10,
      "loss": 0.5417,
      "step": 7770
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7902372250417986,
      "learning_rate": 5.179761932907478e-10,
      "loss": 0.5379,
      "step": 7771
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8452380216172978,
      "learning_rate": 5.047810001992392e-10,
      "loss": 0.5909,
      "step": 7772
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6879838393543596,
      "learning_rate": 4.91756010730926e-10,
      "loss": 0.5234,
      "step": 7773
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.811489517106738,
      "learning_rate": 4.789012271037008e-10,
      "loss": 0.631,
      "step": 7774
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8862776548730364,
      "learning_rate": 4.662166515067011e-10,
      "loss": 0.6016,
      "step": 7775
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8361010535139795,
      "learning_rate": 4.5370228609997727e-10,
      "loss": 0.6399,
      "step": 7776
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9146471098773087,
      "learning_rate": 4.413581330144911e-10,
      "loss": 0.6132,
      "step": 7777
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8103643952406688,
      "learning_rate": 4.291841943524499e-10,
      "loss": 0.5938,
      "step": 7778
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8131392206165737,
      "learning_rate": 4.1718047218686213e-10,
      "loss": 0.5584,
      "step": 7779
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6557486036260337,
      "learning_rate": 4.053469685617594e-10,
      "loss": 0.5386,
      "step": 7780
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8535844316312553,
      "learning_rate": 3.936836854925296e-10,
      "loss": 0.658,
      "step": 7781
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7168049015838625,
      "learning_rate": 3.821906249650286e-10,
      "loss": 0.5863,
      "step": 7782
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7211229890962785,
      "learning_rate": 3.7086778893657967e-10,
      "loss": 0.5311,
      "step": 7783
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6421714804378829,
      "learning_rate": 3.5971517933519604e-10,
      "loss": 0.494,
      "step": 7784
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6805517185959694,
      "learning_rate": 3.4873279806024725e-10,
      "loss": 0.5376,
      "step": 7785
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7969189570048373,
      "learning_rate": 3.37920646981793e-10,
      "loss": 0.607,
      "step": 7786
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8668643062043864,
      "learning_rate": 3.2727872794102716e-10,
      "loss": 0.6167,
      "step": 7787
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6634571073149956,
      "learning_rate": 3.168070427501668e-10,
      "loss": 0.5154,
      "step": 7788
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9179457487398366,
      "learning_rate": 3.0650559319245204e-10,
      "loss": 0.6571,
      "step": 7789
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9379168258139087,
      "learning_rate": 2.963743810221464e-10,
      "loss": 0.5888,
      "step": 7790
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8212681366255622,
      "learning_rate": 2.864134079644254e-10,
      "loss": 0.6102,
      "step": 7791
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7316386443460332,
      "learning_rate": 2.766226757154877e-10,
      "loss": 0.5935,
      "step": 7792
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6463410458828635,
      "learning_rate": 2.6700218594277737e-10,
      "loss": 0.5248,
      "step": 7793
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6446284100699337,
      "learning_rate": 2.575519402843174e-10,
      "loss": 0.578,
      "step": 7794
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6456447775135491,
      "learning_rate": 2.482719403495981e-10,
      "loss": 0.4896,
      "step": 7795
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9125556746929285,
      "learning_rate": 2.391621877189109e-10,
      "loss": 0.6629,
      "step": 7796
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8832089692283861,
      "learning_rate": 2.3022268394334855e-10,
      "loss": 0.6494,
      "step": 7797
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8066100811593482,
      "learning_rate": 2.214534305453597e-10,
      "loss": 0.5208,
      "step": 7798
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8096489835068553,
      "learning_rate": 2.1285442901830541e-10,
      "loss": 0.578,
      "step": 7799
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.006701301731855,
      "learning_rate": 2.0442568082634782e-10,
      "loss": 0.6289,
      "step": 7800
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.763797774964898,
      "learning_rate": 1.9616718740489425e-10,
      "loss": 0.5515,
      "step": 7801
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8841713862164309,
      "learning_rate": 1.8807895016037524e-10,
      "loss": 0.5832,
      "step": 7802
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.597487272678739,
      "learning_rate": 1.8016097047002243e-10,
      "loss": 0.4738,
      "step": 7803
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8886282154411391,
      "learning_rate": 1.724132496822017e-10,
      "loss": 0.6397,
      "step": 7804
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6359561392149431,
      "learning_rate": 1.648357891161911e-10,
      "loss": 0.4994,
      "step": 7805
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7587363409585012,
      "learning_rate": 1.5742859006251386e-10,
      "loss": 0.5851,
      "step": 7806
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8215674682659871,
      "learning_rate": 1.5019165378249432e-10,
      "loss": 0.5476,
      "step": 7807
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9217323252068718,
      "learning_rate": 1.4312498150848006e-10,
      "loss": 0.5978,
      "step": 7808
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.830744659980955,
      "learning_rate": 1.3622857444373082e-10,
      "loss": 0.5912,
      "step": 7809
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8300094806674186,
      "learning_rate": 1.295024337628625e-10,
      "loss": 0.5933,
      "step": 7810
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8814962220993272,
      "learning_rate": 1.2294656061107022e-10,
      "loss": 0.6512,
      "step": 7811
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6399356028053811,
      "learning_rate": 1.1656095610490523e-10,
      "loss": 0.5339,
      "step": 7812
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7287650404842321,
      "learning_rate": 1.1034562133171999e-10,
      "loss": 0.5567,
      "step": 7813
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8066223167633728,
      "learning_rate": 1.0430055734977904e-10,
      "loss": 0.6095,
      "step": 7814
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7740593184487227,
      "learning_rate": 9.842576518870327e-11,
      "loss": 0.5634,
      "step": 7815
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.707158995805899,
      "learning_rate": 9.272124584880358e-11,
      "loss": 0.5644,
      "step": 7816
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6842630828829599,
      "learning_rate": 8.718700030152515e-11,
      "loss": 0.5464,
      "step": 7817
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8533620146819175,
      "learning_rate": 8.182302948933629e-11,
      "loss": 0.6548,
      "step": 7818
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7664416011039629,
      "learning_rate": 7.662933432550644e-11,
      "loss": 0.5899,
      "step": 7819
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6175570252393078,
      "learning_rate": 7.160591569466134e-11,
      "loss": 0.5415,
      "step": 7820
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.11096689069913,
      "learning_rate": 6.67527744521168e-11,
      "loss": 0.6343,
      "step": 7821
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7960029206260274,
      "learning_rate": 6.206991142432283e-11,
      "loss": 0.6311,
      "step": 7822
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7914337769007089,
      "learning_rate": 5.7557327408863655e-11,
      "loss": 0.5674,
      "step": 7823
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7868433726722298,
      "learning_rate": 5.321502317390258e-11,
      "loss": 0.541,
      "step": 7824
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6164095389455341,
      "learning_rate": 4.9042999459181224e-11,
      "loss": 0.4973,
      "step": 7825
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8386565776760152,
      "learning_rate": 4.5041256974909236e-11,
      "loss": 0.6178,
      "step": 7826
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7972901883819165,
      "learning_rate": 4.120979640276356e-11,
      "loss": 0.5664,
      "step": 7827
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8440276444110991,
      "learning_rate": 3.754861839500023e-11,
      "loss": 0.555,
      "step": 7828
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.618396863229862,
      "learning_rate": 3.405772357523151e-11,
      "loss": 0.5868,
      "step": 7829
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0686622682608307,
      "learning_rate": 3.0737112537870816e-11,
      "loss": 0.6117,
      "step": 7830
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6916540299878454,
      "learning_rate": 2.758678584835472e-11,
      "loss": 0.526,
      "step": 7831
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8215045800812389,
      "learning_rate": 2.4606744043143e-11,
      "loss": 0.6567,
      "step": 7832
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.765299166724864,
      "learning_rate": 2.17969876297186e-11,
      "loss": 0.5321,
      "step": 7833
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6363264239762301,
      "learning_rate": 1.9157517086587638e-11,
      "loss": 0.5737,
      "step": 7834
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7716953401591824,
      "learning_rate": 1.668833286327942e-11,
      "loss": 0.536,
      "step": 7835
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7293217330316644,
      "learning_rate": 1.4389435380124382e-11,
      "loss": 0.5084,
      "step": 7836
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.820382705806208,
      "learning_rate": 1.2260825028698186e-11,
      "loss": 0.6494,
      "step": 7837
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6894220363458476,
      "learning_rate": 1.0302502171488648e-11,
      "loss": 0.5902,
      "step": 7838
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9321954346199218,
      "learning_rate": 8.514467141895743e-12,
      "loss": 0.616,
      "step": 7839
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6768627957877051,
      "learning_rate": 6.896720244453647e-12,
      "loss": 0.5578,
      "step": 7840
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7034212835050537,
      "learning_rate": 5.449261754608692e-12,
      "loss": 0.5526,
      "step": 7841
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7680212927368978,
      "learning_rate": 4.172091918941412e-12,
      "loss": 0.5579,
      "step": 7842
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7864126355901061,
      "learning_rate": 3.065210954944497e-12,
      "loss": 0.5619,
      "step": 7843
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7162419939094142,
      "learning_rate": 2.1286190510227952e-12,
      "loss": 0.5462,
      "step": 7844
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8687403737639042,
      "learning_rate": 1.3623163666043324e-12,
      "loss": 0.6186,
      "step": 7845
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8735836295831937,
      "learning_rate": 7.663030323623587e-13,
      "loss": 0.6025,
      "step": 7846
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.750286920792172,
      "learning_rate": 3.4057914966023613e-13,
      "loss": 0.5743,
      "step": 7847
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6733822382820582,
      "learning_rate": 8.514479099552829e-14,
      "loss": 0.5113,
      "step": 7848
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8236251711940762,
      "learning_rate": 0.0,
      "loss": 0.5915,
      "step": 7849
    },
    {
      "epoch": 1.0,
      "step": 7849,
      "total_flos": 4047659296358400.0,
      "train_loss": 0.5910339839694788,
      "train_runtime": 40587.4686,
      "train_samples_per_second": 24.755,
      "train_steps_per_second": 0.193
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 7849,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 60000,
  "total_flos": 4047659296358400.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}