{
  "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": 6.315755160919551,
      "learning_rate": 4.237288135593221e-08,
      "loss": 0.9479,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.100218490677534,
      "learning_rate": 8.474576271186442e-08,
      "loss": 1.2069,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.171566849825352,
      "learning_rate": 1.2711864406779662e-07,
      "loss": 1.1277,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.584630662833674,
      "learning_rate": 1.6949152542372883e-07,
      "loss": 0.9206,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.100846456380341,
      "learning_rate": 2.1186440677966102e-07,
      "loss": 0.9232,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.4705123107118965,
      "learning_rate": 2.5423728813559323e-07,
      "loss": 0.8992,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.295949465651534,
      "learning_rate": 2.966101694915255e-07,
      "loss": 1.0882,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.023526451128331,
      "learning_rate": 3.3898305084745766e-07,
      "loss": 0.9273,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.598934446265413,
      "learning_rate": 3.813559322033899e-07,
      "loss": 1.1617,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.255483401248126,
      "learning_rate": 4.2372881355932204e-07,
      "loss": 1.106,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.586857366208579,
      "learning_rate": 4.661016949152543e-07,
      "loss": 1.1648,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.903139126054202,
      "learning_rate": 5.084745762711865e-07,
      "loss": 0.883,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.202622651142046,
      "learning_rate": 5.508474576271187e-07,
      "loss": 1.1534,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.512018510847586,
      "learning_rate": 5.93220338983051e-07,
      "loss": 1.1664,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.499201174169701,
      "learning_rate": 6.355932203389831e-07,
      "loss": 0.8748,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.9090400452428966,
      "learning_rate": 6.779661016949153e-07,
      "loss": 0.8748,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.38152180767904,
      "learning_rate": 7.203389830508476e-07,
      "loss": 1.0645,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.053943119085202,
      "learning_rate": 7.627118644067798e-07,
      "loss": 1.1386,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.8066592158020813,
      "learning_rate": 8.050847457627118e-07,
      "loss": 0.8644,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.401110354097955,
      "learning_rate": 8.474576271186441e-07,
      "loss": 1.0844,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.559152082114432,
      "learning_rate": 8.898305084745763e-07,
      "loss": 1.0849,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8745037562094358,
      "learning_rate": 9.322033898305086e-07,
      "loss": 0.8012,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.88624059061097,
      "learning_rate": 9.745762711864408e-07,
      "loss": 1.0375,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.715535508413173,
      "learning_rate": 1.016949152542373e-06,
      "loss": 0.8254,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.6402729503537405,
      "learning_rate": 1.059322033898305e-06,
      "loss": 0.8766,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8482238838073206,
      "learning_rate": 1.1016949152542374e-06,
      "loss": 0.9553,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.0497968975133296,
      "learning_rate": 1.1440677966101696e-06,
      "loss": 0.9557,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2061093249472625,
      "learning_rate": 1.186440677966102e-06,
      "loss": 0.7813,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8067996733068905,
      "learning_rate": 1.228813559322034e-06,
      "loss": 0.7334,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.298191883020086,
      "learning_rate": 1.2711864406779662e-06,
      "loss": 0.9672,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5705838510255912,
      "learning_rate": 1.3135593220338985e-06,
      "loss": 0.712,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6088494092491419,
      "learning_rate": 1.3559322033898307e-06,
      "loss": 0.7545,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4744942694005478,
      "learning_rate": 1.3983050847457628e-06,
      "loss": 0.7158,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3407350211239946,
      "learning_rate": 1.4406779661016951e-06,
      "loss": 0.653,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3950625631306801,
      "learning_rate": 1.4830508474576273e-06,
      "loss": 0.6612,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.755463643787986,
      "learning_rate": 1.5254237288135596e-06,
      "loss": 0.9663,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6523762136186397,
      "learning_rate": 1.5677966101694915e-06,
      "loss": 0.8251,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2628654326647275,
      "learning_rate": 1.6101694915254237e-06,
      "loss": 0.6934,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5372606404011073,
      "learning_rate": 1.652542372881356e-06,
      "loss": 0.8577,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0473609509684376,
      "learning_rate": 1.6949152542372882e-06,
      "loss": 0.6877,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4049463579442723,
      "learning_rate": 1.7372881355932205e-06,
      "loss": 0.86,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0127261251945565,
      "learning_rate": 1.7796610169491526e-06,
      "loss": 0.745,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8838010865593211,
      "learning_rate": 1.8220338983050848e-06,
      "loss": 0.5871,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0609299135935981,
      "learning_rate": 1.8644067796610171e-06,
      "loss": 0.6314,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9159445100927852,
      "learning_rate": 1.9067796610169493e-06,
      "loss": 0.8977,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9773402968846524,
      "learning_rate": 1.9491525423728816e-06,
      "loss": 0.6562,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.11450266217315,
      "learning_rate": 1.9915254237288137e-06,
      "loss": 0.6779,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5612110238194168,
      "learning_rate": 2.033898305084746e-06,
      "loss": 0.8962,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4987795309713536,
      "learning_rate": 2.076271186440678e-06,
      "loss": 0.8172,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2072667175753797,
      "learning_rate": 2.11864406779661e-06,
      "loss": 0.6323,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1535750512983887,
      "learning_rate": 2.1610169491525427e-06,
      "loss": 0.8327,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0199047844541156,
      "learning_rate": 2.203389830508475e-06,
      "loss": 0.5918,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8864849172154937,
      "learning_rate": 2.245762711864407e-06,
      "loss": 0.6398,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1514523578372728,
      "learning_rate": 2.288135593220339e-06,
      "loss": 0.7384,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1239926372958742,
      "learning_rate": 2.3305084745762712e-06,
      "loss": 0.6809,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4453255002614818,
      "learning_rate": 2.372881355932204e-06,
      "loss": 0.7847,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1403389296713031,
      "learning_rate": 2.415254237288136e-06,
      "loss": 0.8264,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9067072880055732,
      "learning_rate": 2.457627118644068e-06,
      "loss": 0.6831,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8498698865320297,
      "learning_rate": 2.5e-06,
      "loss": 0.654,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9265346653911731,
      "learning_rate": 2.5423728813559323e-06,
      "loss": 0.6422,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1596522463743175,
      "learning_rate": 2.5847457627118645e-06,
      "loss": 0.8155,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9474405179772677,
      "learning_rate": 2.627118644067797e-06,
      "loss": 0.6162,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9095862376115516,
      "learning_rate": 2.669491525423729e-06,
      "loss": 0.7255,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8318515654495965,
      "learning_rate": 2.7118644067796613e-06,
      "loss": 0.5899,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1369539475488417,
      "learning_rate": 2.7542372881355934e-06,
      "loss": 0.731,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9723618370693417,
      "learning_rate": 2.7966101694915256e-06,
      "loss": 0.6611,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4000758610623976,
      "learning_rate": 2.838983050847458e-06,
      "loss": 0.8021,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8956344339444645,
      "learning_rate": 2.8813559322033903e-06,
      "loss": 0.6644,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8523212039125865,
      "learning_rate": 2.9237288135593224e-06,
      "loss": 0.6268,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0703861752332404,
      "learning_rate": 2.9661016949152545e-06,
      "loss": 0.7397,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9326192666797318,
      "learning_rate": 3.0084745762711862e-06,
      "loss": 0.6247,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8313311945247217,
      "learning_rate": 3.0508474576271192e-06,
      "loss": 0.5699,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0409291394971776,
      "learning_rate": 3.0932203389830514e-06,
      "loss": 0.8053,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.159631053983387,
      "learning_rate": 3.135593220338983e-06,
      "loss": 0.7713,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0142393261515645,
      "learning_rate": 3.1779661016949152e-06,
      "loss": 0.7452,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1177153115205338,
      "learning_rate": 3.2203389830508473e-06,
      "loss": 0.6191,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.934158324801368,
      "learning_rate": 3.26271186440678e-06,
      "loss": 0.5971,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8198664871974447,
      "learning_rate": 3.305084745762712e-06,
      "loss": 0.5894,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.12560097722721,
      "learning_rate": 3.347457627118644e-06,
      "loss": 0.7259,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1372264114353359,
      "learning_rate": 3.3898305084745763e-06,
      "loss": 0.7235,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.946308089908942,
      "learning_rate": 3.4322033898305084e-06,
      "loss": 0.7423,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.768706538717993,
      "learning_rate": 3.474576271186441e-06,
      "loss": 0.5814,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7882263274857597,
      "learning_rate": 3.516949152542373e-06,
      "loss": 0.6227,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0130484490917735,
      "learning_rate": 3.5593220338983053e-06,
      "loss": 0.6292,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0625999194751543,
      "learning_rate": 3.6016949152542374e-06,
      "loss": 0.756,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1806468563909829,
      "learning_rate": 3.6440677966101695e-06,
      "loss": 0.8181,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8915103932048728,
      "learning_rate": 3.686440677966102e-06,
      "loss": 0.7278,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0544449375949836,
      "learning_rate": 3.7288135593220342e-06,
      "loss": 0.6745,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.061924278121735,
      "learning_rate": 3.7711864406779664e-06,
      "loss": 0.762,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7838385105137881,
      "learning_rate": 3.8135593220338985e-06,
      "loss": 0.5319,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8940767055179175,
      "learning_rate": 3.8559322033898315e-06,
      "loss": 0.7681,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0434976840215688,
      "learning_rate": 3.898305084745763e-06,
      "loss": 0.7914,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9909272239591096,
      "learning_rate": 3.940677966101695e-06,
      "loss": 0.7253,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7948844412832544,
      "learning_rate": 3.9830508474576275e-06,
      "loss": 0.7036,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9165645687951253,
      "learning_rate": 4.025423728813559e-06,
      "loss": 0.6991,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9924627256847344,
      "learning_rate": 4.067796610169492e-06,
      "loss": 0.7001,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7887320484875681,
      "learning_rate": 4.110169491525424e-06,
      "loss": 0.5986,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9114726726862187,
      "learning_rate": 4.152542372881356e-06,
      "loss": 0.6901,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8754590567958539,
      "learning_rate": 4.1949152542372886e-06,
      "loss": 0.6272,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7911143593574747,
      "learning_rate": 4.23728813559322e-06,
      "loss": 0.6387,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1497787822708534,
      "learning_rate": 4.279661016949153e-06,
      "loss": 0.7806,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.961636545453514,
      "learning_rate": 4.322033898305085e-06,
      "loss": 0.6654,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8643465998348661,
      "learning_rate": 4.364406779661017e-06,
      "loss": 0.761,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0623573430856974,
      "learning_rate": 4.40677966101695e-06,
      "loss": 0.7737,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7924110178523148,
      "learning_rate": 4.449152542372881e-06,
      "loss": 0.5905,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0547457932502393,
      "learning_rate": 4.491525423728814e-06,
      "loss": 0.7155,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0021114670760642,
      "learning_rate": 4.5338983050847465e-06,
      "loss": 0.7308,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1147786425678303,
      "learning_rate": 4.576271186440678e-06,
      "loss": 0.7139,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9188378962317331,
      "learning_rate": 4.618644067796611e-06,
      "loss": 0.722,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8417003882693744,
      "learning_rate": 4.6610169491525425e-06,
      "loss": 0.6588,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9742289791982495,
      "learning_rate": 4.703389830508475e-06,
      "loss": 0.6528,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9680593640556735,
      "learning_rate": 4.745762711864408e-06,
      "loss": 0.6821,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.131304634234551,
      "learning_rate": 4.788135593220339e-06,
      "loss": 0.7428,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9588135285950953,
      "learning_rate": 4.830508474576272e-06,
      "loss": 0.6703,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9066996549067298,
      "learning_rate": 4.872881355932204e-06,
      "loss": 0.5918,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.833010228931143,
      "learning_rate": 4.915254237288136e-06,
      "loss": 0.6471,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0443031332541632,
      "learning_rate": 4.957627118644069e-06,
      "loss": 0.7252,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1057503996909768,
      "learning_rate": 5e-06,
      "loss": 0.7125,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.288078975820244,
      "learning_rate": 5.042372881355932e-06,
      "loss": 0.7089,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7787012931038324,
      "learning_rate": 5.084745762711865e-06,
      "loss": 0.614,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8671198361828785,
      "learning_rate": 5.127118644067796e-06,
      "loss": 0.6033,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0562366014378568,
      "learning_rate": 5.169491525423729e-06,
      "loss": 0.7258,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.825424722924917,
      "learning_rate": 5.211864406779662e-06,
      "loss": 0.6106,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9535747173058154,
      "learning_rate": 5.254237288135594e-06,
      "loss": 0.6698,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9964029349528971,
      "learning_rate": 5.296610169491526e-06,
      "loss": 0.6864,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8496061929889926,
      "learning_rate": 5.338983050847458e-06,
      "loss": 0.6085,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8169307113798493,
      "learning_rate": 5.38135593220339e-06,
      "loss": 0.577,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8828226478416715,
      "learning_rate": 5.423728813559323e-06,
      "loss": 0.6536,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8111566368468357,
      "learning_rate": 5.466101694915254e-06,
      "loss": 0.5861,
      "step": 129
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8629999756794673,
      "learning_rate": 5.508474576271187e-06,
      "loss": 0.6798,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0649694021172096,
      "learning_rate": 5.550847457627119e-06,
      "loss": 0.7477,
      "step": 131
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2379080105519054,
      "learning_rate": 5.593220338983051e-06,
      "loss": 0.8184,
      "step": 132
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9008620998842783,
      "learning_rate": 5.635593220338984e-06,
      "loss": 0.5844,
      "step": 133
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.057219847760786,
      "learning_rate": 5.677966101694916e-06,
      "loss": 0.6303,
      "step": 134
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8110716988102185,
      "learning_rate": 5.720338983050848e-06,
      "loss": 0.6138,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.007863377115401,
      "learning_rate": 5.7627118644067805e-06,
      "loss": 0.7345,
      "step": 136
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0012050419482001,
      "learning_rate": 5.805084745762712e-06,
      "loss": 0.7099,
      "step": 137
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8503972784915824,
      "learning_rate": 5.847457627118645e-06,
      "loss": 0.5918,
      "step": 138
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.203598531953053,
      "learning_rate": 5.8898305084745765e-06,
      "loss": 0.7367,
      "step": 139
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8131031453284981,
      "learning_rate": 5.932203389830509e-06,
      "loss": 0.5979,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1498567432766191,
      "learning_rate": 5.974576271186441e-06,
      "loss": 0.6825,
      "step": 141
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9748267037249195,
      "learning_rate": 6.0169491525423725e-06,
      "loss": 0.6309,
      "step": 142
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1147298161057293,
      "learning_rate": 6.059322033898306e-06,
      "loss": 0.7522,
      "step": 143
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7798848047498792,
      "learning_rate": 6.1016949152542385e-06,
      "loss": 0.5549,
      "step": 144
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9114141439174802,
      "learning_rate": 6.14406779661017e-06,
      "loss": 0.6766,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9897454653838555,
      "learning_rate": 6.186440677966103e-06,
      "loss": 0.7221,
      "step": 146
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9453356509228966,
      "learning_rate": 6.2288135593220344e-06,
      "loss": 0.6089,
      "step": 147
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1046986048460117,
      "learning_rate": 6.271186440677966e-06,
      "loss": 0.7237,
      "step": 148
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7679291507056515,
      "learning_rate": 6.313559322033899e-06,
      "loss": 0.5788,
      "step": 149
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8568576370735939,
      "learning_rate": 6.3559322033898304e-06,
      "loss": 0.6798,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8832945038337772,
      "learning_rate": 6.398305084745763e-06,
      "loss": 0.5891,
      "step": 151
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9761886742433304,
      "learning_rate": 6.440677966101695e-06,
      "loss": 0.719,
      "step": 152
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.898779892810249,
      "learning_rate": 6.483050847457628e-06,
      "loss": 0.625,
      "step": 153
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8946339479763311,
      "learning_rate": 6.52542372881356e-06,
      "loss": 0.7351,
      "step": 154
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7873558089294436,
      "learning_rate": 6.567796610169492e-06,
      "loss": 0.6204,
      "step": 155
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8160069510964497,
      "learning_rate": 6.610169491525424e-06,
      "loss": 0.5863,
      "step": 156
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9217008221003499,
      "learning_rate": 6.652542372881357e-06,
      "loss": 0.64,
      "step": 157
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8146830461284279,
      "learning_rate": 6.694915254237288e-06,
      "loss": 0.6413,
      "step": 158
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.119588204107743,
      "learning_rate": 6.737288135593221e-06,
      "loss": 0.7497,
      "step": 159
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8891169665144224,
      "learning_rate": 6.779661016949153e-06,
      "loss": 0.6159,
      "step": 160
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9544571651440298,
      "learning_rate": 6.822033898305085e-06,
      "loss": 0.6307,
      "step": 161
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9748541918685296,
      "learning_rate": 6.864406779661017e-06,
      "loss": 0.7042,
      "step": 162
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8928036088788414,
      "learning_rate": 6.90677966101695e-06,
      "loss": 0.5525,
      "step": 163
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9447750031112148,
      "learning_rate": 6.949152542372882e-06,
      "loss": 0.6309,
      "step": 164
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7293762637264555,
      "learning_rate": 6.9915254237288146e-06,
      "loss": 0.5824,
      "step": 165
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9935951614871427,
      "learning_rate": 7.033898305084746e-06,
      "loss": 0.7509,
      "step": 166
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2187072550538158,
      "learning_rate": 7.076271186440679e-06,
      "loss": 0.7195,
      "step": 167
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7743930446324943,
      "learning_rate": 7.1186440677966106e-06,
      "loss": 0.6039,
      "step": 168
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7757997144586963,
      "learning_rate": 7.161016949152543e-06,
      "loss": 0.5715,
      "step": 169
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9041618886185299,
      "learning_rate": 7.203389830508475e-06,
      "loss": 0.5617,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8781459413382111,
      "learning_rate": 7.2457627118644065e-06,
      "loss": 0.6267,
      "step": 171
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0213172173088207,
      "learning_rate": 7.288135593220339e-06,
      "loss": 0.7325,
      "step": 172
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.001334402469566,
      "learning_rate": 7.3305084745762725e-06,
      "loss": 0.7242,
      "step": 173
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9066076828879898,
      "learning_rate": 7.372881355932204e-06,
      "loss": 0.5799,
      "step": 174
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.011559773485519,
      "learning_rate": 7.415254237288137e-06,
      "loss": 0.6684,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1146404100163043,
      "learning_rate": 7.4576271186440685e-06,
      "loss": 0.7341,
      "step": 176
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0343761477564852,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.6357,
      "step": 177
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7725674905493598,
      "learning_rate": 7.542372881355933e-06,
      "loss": 0.575,
      "step": 178
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8621063821193666,
      "learning_rate": 7.5847457627118645e-06,
      "loss": 0.6182,
      "step": 179
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.844374090282496,
      "learning_rate": 7.627118644067797e-06,
      "loss": 0.6298,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9800783922527415,
      "learning_rate": 7.66949152542373e-06,
      "loss": 0.623,
      "step": 181
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1236727397729043,
      "learning_rate": 7.711864406779663e-06,
      "loss": 0.7356,
      "step": 182
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9659078982971598,
      "learning_rate": 7.754237288135595e-06,
      "loss": 0.7238,
      "step": 183
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.001387908180136,
      "learning_rate": 7.796610169491526e-06,
      "loss": 0.6177,
      "step": 184
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1512598543745955,
      "learning_rate": 7.838983050847458e-06,
      "loss": 0.7762,
      "step": 185
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9569207735160119,
      "learning_rate": 7.88135593220339e-06,
      "loss": 0.7724,
      "step": 186
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8822836126824283,
      "learning_rate": 7.923728813559323e-06,
      "loss": 0.586,
      "step": 187
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9410329153890431,
      "learning_rate": 7.966101694915255e-06,
      "loss": 0.6237,
      "step": 188
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8131163568514106,
      "learning_rate": 8.008474576271187e-06,
      "loss": 0.5999,
      "step": 189
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7878807001243879,
      "learning_rate": 8.050847457627118e-06,
      "loss": 0.6251,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3021336547354976,
      "learning_rate": 8.093220338983052e-06,
      "loss": 0.6943,
      "step": 191
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0357888044807833,
      "learning_rate": 8.135593220338983e-06,
      "loss": 0.7529,
      "step": 192
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9798088952833964,
      "learning_rate": 8.177966101694917e-06,
      "loss": 0.6445,
      "step": 193
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2025857531069202,
      "learning_rate": 8.220338983050849e-06,
      "loss": 0.7624,
      "step": 194
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2444329236365508,
      "learning_rate": 8.26271186440678e-06,
      "loss": 0.7065,
      "step": 195
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0099626060159643,
      "learning_rate": 8.305084745762712e-06,
      "loss": 0.6457,
      "step": 196
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8750853835711236,
      "learning_rate": 8.347457627118645e-06,
      "loss": 0.6198,
      "step": 197
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8943179057603692,
      "learning_rate": 8.389830508474577e-06,
      "loss": 0.5735,
      "step": 198
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.020139596778078,
      "learning_rate": 8.432203389830509e-06,
      "loss": 0.6681,
      "step": 199
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2038740689317602,
      "learning_rate": 8.47457627118644e-06,
      "loss": 0.6539,
      "step": 200
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0491948555402582,
      "learning_rate": 8.516949152542372e-06,
      "loss": 0.7375,
      "step": 201
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0492347673279663,
      "learning_rate": 8.559322033898306e-06,
      "loss": 0.6445,
      "step": 202
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0718128205114044,
      "learning_rate": 8.601694915254239e-06,
      "loss": 0.7002,
      "step": 203
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.734556758763377,
      "learning_rate": 8.64406779661017e-06,
      "loss": 0.5629,
      "step": 204
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7412973081240049,
      "learning_rate": 8.686440677966103e-06,
      "loss": 0.5947,
      "step": 205
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8180267124924756,
      "learning_rate": 8.728813559322034e-06,
      "loss": 0.5656,
      "step": 206
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8448009928921776,
      "learning_rate": 8.771186440677966e-06,
      "loss": 0.6197,
      "step": 207
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.892668338028599,
      "learning_rate": 8.8135593220339e-06,
      "loss": 0.6715,
      "step": 208
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0712961761495488,
      "learning_rate": 8.855932203389831e-06,
      "loss": 0.7181,
      "step": 209
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8682510234712424,
      "learning_rate": 8.898305084745763e-06,
      "loss": 0.577,
      "step": 210
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8059324335317097,
      "learning_rate": 8.940677966101694e-06,
      "loss": 0.6141,
      "step": 211
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2501186941750848,
      "learning_rate": 8.983050847457628e-06,
      "loss": 0.6886,
      "step": 212
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9097074020678482,
      "learning_rate": 9.02542372881356e-06,
      "loss": 0.7408,
      "step": 213
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.801724007745236,
      "learning_rate": 9.067796610169493e-06,
      "loss": 0.5711,
      "step": 214
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8516202040963996,
      "learning_rate": 9.110169491525425e-06,
      "loss": 0.6395,
      "step": 215
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9631488325325961,
      "learning_rate": 9.152542372881356e-06,
      "loss": 0.6496,
      "step": 216
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.084435462085635,
      "learning_rate": 9.194915254237288e-06,
      "loss": 0.6989,
      "step": 217
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8363119000814339,
      "learning_rate": 9.237288135593222e-06,
      "loss": 0.5734,
      "step": 218
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9511490838269513,
      "learning_rate": 9.279661016949153e-06,
      "loss": 0.6169,
      "step": 219
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8895238466035947,
      "learning_rate": 9.322033898305085e-06,
      "loss": 0.6267,
      "step": 220
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1514987460633765,
      "learning_rate": 9.364406779661017e-06,
      "loss": 0.6641,
      "step": 221
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9565801841708472,
      "learning_rate": 9.40677966101695e-06,
      "loss": 0.6133,
      "step": 222
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8669101354574144,
      "learning_rate": 9.449152542372882e-06,
      "loss": 0.6412,
      "step": 223
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7342602405274954,
      "learning_rate": 9.491525423728815e-06,
      "loss": 0.6167,
      "step": 224
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7382289769982713,
      "learning_rate": 9.533898305084747e-06,
      "loss": 0.6358,
      "step": 225
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6923445266648418,
      "learning_rate": 9.576271186440679e-06,
      "loss": 0.4961,
      "step": 226
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7434130124115883,
      "learning_rate": 9.61864406779661e-06,
      "loss": 0.5386,
      "step": 227
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9396672981659955,
      "learning_rate": 9.661016949152544e-06,
      "loss": 0.6244,
      "step": 228
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8403025796295461,
      "learning_rate": 9.703389830508475e-06,
      "loss": 0.6751,
      "step": 229
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.346216823414232,
      "learning_rate": 9.745762711864407e-06,
      "loss": 0.747,
      "step": 230
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0732661890075386,
      "learning_rate": 9.788135593220339e-06,
      "loss": 0.7309,
      "step": 231
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9194804417202522,
      "learning_rate": 9.830508474576272e-06,
      "loss": 0.5994,
      "step": 232
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8500279144644034,
      "learning_rate": 9.872881355932204e-06,
      "loss": 0.6503,
      "step": 233
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9479766715163666,
      "learning_rate": 9.915254237288137e-06,
      "loss": 0.7034,
      "step": 234
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.024106341163967,
      "learning_rate": 9.957627118644069e-06,
      "loss": 0.64,
      "step": 235
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8773894145226389,
      "learning_rate": 1e-05,
      "loss": 0.6003,
      "step": 236
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9473859245843204,
      "learning_rate": 9.999999574276046e-06,
      "loss": 0.7454,
      "step": 237
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9577208668886873,
      "learning_rate": 9.999998297104252e-06,
      "loss": 0.6419,
      "step": 238
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8602749658395881,
      "learning_rate": 9.99999616848484e-06,
      "loss": 0.603,
      "step": 239
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0045808750730014,
      "learning_rate": 9.999993188418168e-06,
      "loss": 0.6388,
      "step": 240
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0650052995930719,
      "learning_rate": 9.999989356904746e-06,
      "loss": 0.7113,
      "step": 241
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8045082422482904,
      "learning_rate": 9.999984673945226e-06,
      "loss": 0.6069,
      "step": 242
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8436870361222776,
      "learning_rate": 9.999979139540406e-06,
      "loss": 0.645,
      "step": 243
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9947510382185858,
      "learning_rate": 9.999972753691227e-06,
      "loss": 0.7262,
      "step": 244
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.742499938099767,
      "learning_rate": 9.999965516398778e-06,
      "loss": 0.7071,
      "step": 245
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9349758417642406,
      "learning_rate": 9.99995742766429e-06,
      "loss": 0.6056,
      "step": 246
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9643565158039966,
      "learning_rate": 9.999948487489144e-06,
      "loss": 0.6991,
      "step": 247
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.041471885230738,
      "learning_rate": 9.999938695874857e-06,
      "loss": 0.7836,
      "step": 248
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9026920203140789,
      "learning_rate": 9.9999280528231e-06,
      "loss": 0.6217,
      "step": 249
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7483237548345875,
      "learning_rate": 9.999916558335685e-06,
      "loss": 0.5728,
      "step": 250
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7523195268586251,
      "learning_rate": 9.999904212414569e-06,
      "loss": 0.5335,
      "step": 251
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8561396872975037,
      "learning_rate": 9.999891015061853e-06,
      "loss": 0.6514,
      "step": 252
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7684453975305775,
      "learning_rate": 9.999876966279784e-06,
      "loss": 0.6109,
      "step": 253
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0075623622613596,
      "learning_rate": 9.999862066070759e-06,
      "loss": 0.6978,
      "step": 254
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9330297959886233,
      "learning_rate": 9.999846314437312e-06,
      "loss": 0.6024,
      "step": 255
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8153131674765485,
      "learning_rate": 9.999829711382125e-06,
      "loss": 0.58,
      "step": 256
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0285588577508944,
      "learning_rate": 9.999812256908025e-06,
      "loss": 0.6716,
      "step": 257
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2039006035827808,
      "learning_rate": 9.999793951017986e-06,
      "loss": 0.6966,
      "step": 258
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8965745425797729,
      "learning_rate": 9.999774793715126e-06,
      "loss": 0.7421,
      "step": 259
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.917614617516794,
      "learning_rate": 9.999754785002705e-06,
      "loss": 0.679,
      "step": 260
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0362393643384793,
      "learning_rate": 9.999733924884132e-06,
      "loss": 0.7768,
      "step": 261
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0174413250119518,
      "learning_rate": 9.999712213362957e-06,
      "loss": 0.6778,
      "step": 262
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7670304828921283,
      "learning_rate": 9.999689650442878e-06,
      "loss": 0.5888,
      "step": 263
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0250762401374716,
      "learning_rate": 9.999666236127741e-06,
      "loss": 0.6437,
      "step": 264
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8724926110458359,
      "learning_rate": 9.999641970421528e-06,
      "loss": 0.6002,
      "step": 265
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7471948393811478,
      "learning_rate": 9.999616853328373e-06,
      "loss": 0.5884,
      "step": 266
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9431722393770324,
      "learning_rate": 9.999590884852554e-06,
      "loss": 0.6848,
      "step": 267
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8904762183461915,
      "learning_rate": 9.999564064998493e-06,
      "loss": 0.6573,
      "step": 268
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7787426366104326,
      "learning_rate": 9.999536393770757e-06,
      "loss": 0.6033,
      "step": 269
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7488704336463297,
      "learning_rate": 9.999507871174056e-06,
      "loss": 0.6111,
      "step": 270
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9063140732547219,
      "learning_rate": 9.999478497213252e-06,
      "loss": 0.7089,
      "step": 271
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1333482461657327,
      "learning_rate": 9.999448271893342e-06,
      "loss": 0.6964,
      "step": 272
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0304558799971852,
      "learning_rate": 9.999417195219476e-06,
      "loss": 0.6852,
      "step": 273
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7818934205810679,
      "learning_rate": 9.999385267196947e-06,
      "loss": 0.6542,
      "step": 274
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9892073103285758,
      "learning_rate": 9.999352487831187e-06,
      "loss": 0.626,
      "step": 275
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9616757237790202,
      "learning_rate": 9.999318857127783e-06,
      "loss": 0.6752,
      "step": 276
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9961448141781154,
      "learning_rate": 9.999284375092458e-06,
      "loss": 0.6269,
      "step": 277
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.795420515321003,
      "learning_rate": 9.999249041731089e-06,
      "loss": 0.5838,
      "step": 278
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8826509457705606,
      "learning_rate": 9.999212857049688e-06,
      "loss": 0.6113,
      "step": 279
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9583078016380316,
      "learning_rate": 9.99917582105442e-06,
      "loss": 0.61,
      "step": 280
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8359376697608909,
      "learning_rate": 9.999137933751591e-06,
      "loss": 0.5932,
      "step": 281
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7554090887881599,
      "learning_rate": 9.99909919514765e-06,
      "loss": 0.6292,
      "step": 282
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7969393806474914,
      "learning_rate": 9.9990596052492e-06,
      "loss": 0.5801,
      "step": 283
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8339723105361531,
      "learning_rate": 9.999019164062976e-06,
      "loss": 0.5922,
      "step": 284
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0184188001217274,
      "learning_rate": 9.99897787159587e-06,
      "loss": 0.7114,
      "step": 285
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9345821403070972,
      "learning_rate": 9.99893572785491e-06,
      "loss": 0.6693,
      "step": 286
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9489832490319505,
      "learning_rate": 9.998892732847275e-06,
      "loss": 0.6747,
      "step": 287
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9491822975095311,
      "learning_rate": 9.998848886580283e-06,
      "loss": 0.6135,
      "step": 288
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0307820768893219,
      "learning_rate": 9.998804189061407e-06,
      "loss": 0.6866,
      "step": 289
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7608263414528763,
      "learning_rate": 9.998758640298253e-06,
      "loss": 0.5903,
      "step": 290
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8599073151312877,
      "learning_rate": 9.998712240298579e-06,
      "loss": 0.69,
      "step": 291
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7205713598023247,
      "learning_rate": 9.998664989070287e-06,
      "loss": 0.5683,
      "step": 292
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8110085756398755,
      "learning_rate": 9.998616886621424e-06,
      "loss": 0.6288,
      "step": 293
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9154328323405975,
      "learning_rate": 9.99856793296018e-06,
      "loss": 0.6759,
      "step": 294
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8200883833415538,
      "learning_rate": 9.99851812809489e-06,
      "loss": 0.61,
      "step": 295
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9288553336473484,
      "learning_rate": 9.998467472034039e-06,
      "loss": 0.6731,
      "step": 296
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0248926938844098,
      "learning_rate": 9.99841596478625e-06,
      "loss": 0.675,
      "step": 297
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9662476530548694,
      "learning_rate": 9.998363606360295e-06,
      "loss": 0.7635,
      "step": 298
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8337459948893531,
      "learning_rate": 9.998310396765092e-06,
      "loss": 0.6224,
      "step": 299
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8930266443405209,
      "learning_rate": 9.9982563360097e-06,
      "loss": 0.6268,
      "step": 300
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8621128050741431,
      "learning_rate": 9.998201424103325e-06,
      "loss": 0.5759,
      "step": 301
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.218581113023825,
      "learning_rate": 9.998145661055318e-06,
      "loss": 0.6884,
      "step": 302
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8165414786659695,
      "learning_rate": 9.998089046875176e-06,
      "loss": 0.5984,
      "step": 303
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1533663305554112,
      "learning_rate": 9.998031581572538e-06,
      "loss": 0.6714,
      "step": 304
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8257686275582143,
      "learning_rate": 9.997973265157192e-06,
      "loss": 0.6049,
      "step": 305
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8418914276868043,
      "learning_rate": 9.997914097639066e-06,
      "loss": 0.6021,
      "step": 306
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.169835228410195,
      "learning_rate": 9.997854079028239e-06,
      "loss": 0.7795,
      "step": 307
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8813017380551073,
      "learning_rate": 9.997793209334928e-06,
      "loss": 0.6428,
      "step": 308
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.887480511846037,
      "learning_rate": 9.9977314885695e-06,
      "loss": 0.6193,
      "step": 309
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.02812925541616,
      "learning_rate": 9.997668916742466e-06,
      "loss": 0.761,
      "step": 310
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7943587048634801,
      "learning_rate": 9.997605493864482e-06,
      "loss": 0.5408,
      "step": 311
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8382077658643962,
      "learning_rate": 9.997541219946346e-06,
      "loss": 0.5964,
      "step": 312
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9590898612777877,
      "learning_rate": 9.997476094999005e-06,
      "loss": 0.7203,
      "step": 313
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7771017969668923,
      "learning_rate": 9.997410119033548e-06,
      "loss": 0.6217,
      "step": 314
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9717986029539818,
      "learning_rate": 9.997343292061208e-06,
      "loss": 0.7661,
      "step": 315
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8374225704031762,
      "learning_rate": 9.99727561409337e-06,
      "loss": 0.6195,
      "step": 316
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7693484657466653,
      "learning_rate": 9.997207085141555e-06,
      "loss": 0.6024,
      "step": 317
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5543932190442702,
      "learning_rate": 9.997137705217435e-06,
      "loss": 0.7393,
      "step": 318
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9669158468522525,
      "learning_rate": 9.997067474332824e-06,
      "loss": 0.7073,
      "step": 319
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9043492810600272,
      "learning_rate": 9.99699639249968e-06,
      "loss": 0.6805,
      "step": 320
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7212983946702716,
      "learning_rate": 9.996924459730108e-06,
      "loss": 0.5934,
      "step": 321
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9636016130383295,
      "learning_rate": 9.996851676036358e-06,
      "loss": 0.615,
      "step": 322
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1145403303057233,
      "learning_rate": 9.996778041430826e-06,
      "loss": 0.6685,
      "step": 323
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.647098903247357,
      "learning_rate": 9.99670355592605e-06,
      "loss": 0.7344,
      "step": 324
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8188894099687588,
      "learning_rate": 9.996628219534713e-06,
      "loss": 0.5963,
      "step": 325
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7971616385327054,
      "learning_rate": 9.996552032269643e-06,
      "loss": 0.5691,
      "step": 326
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7662516889475968,
      "learning_rate": 9.996474994143819e-06,
      "loss": 0.6288,
      "step": 327
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0796277220899329,
      "learning_rate": 9.996397105170353e-06,
      "loss": 0.6825,
      "step": 328
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8240757295568947,
      "learning_rate": 9.996318365362514e-06,
      "loss": 0.5904,
      "step": 329
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.805990665340262,
      "learning_rate": 9.996238774733708e-06,
      "loss": 0.5812,
      "step": 330
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8234970501145048,
      "learning_rate": 9.99615833329749e-06,
      "loss": 0.5773,
      "step": 331
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9420760857449071,
      "learning_rate": 9.996077041067555e-06,
      "loss": 0.5468,
      "step": 332
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9809175860162507,
      "learning_rate": 9.995994898057752e-06,
      "loss": 0.6663,
      "step": 333
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.942092634391354,
      "learning_rate": 9.995911904282063e-06,
      "loss": 0.6731,
      "step": 334
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1619584512173078,
      "learning_rate": 9.995828059754625e-06,
      "loss": 0.747,
      "step": 335
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9416861391157988,
      "learning_rate": 9.995743364489715e-06,
      "loss": 0.7163,
      "step": 336
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0041871482847478,
      "learning_rate": 9.995657818501754e-06,
      "loss": 0.7309,
      "step": 337
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8976708106014296,
      "learning_rate": 9.995571421805312e-06,
      "loss": 0.6954,
      "step": 338
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9970549295496145,
      "learning_rate": 9.9954841744151e-06,
      "loss": 0.7289,
      "step": 339
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1211833889866902,
      "learning_rate": 9.995396076345976e-06,
      "loss": 0.6742,
      "step": 340
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8830745823232112,
      "learning_rate": 9.995307127612942e-06,
      "loss": 0.612,
      "step": 341
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7209637307243821,
      "learning_rate": 9.995217328231144e-06,
      "loss": 0.614,
      "step": 342
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9340561683492581,
      "learning_rate": 9.995126678215874e-06,
      "loss": 0.7104,
      "step": 343
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9738537795595587,
      "learning_rate": 9.995035177582572e-06,
      "loss": 0.6642,
      "step": 344
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9440417239739113,
      "learning_rate": 9.994942826346817e-06,
      "loss": 0.6892,
      "step": 345
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7497514014731164,
      "learning_rate": 9.994849624524333e-06,
      "loss": 0.5784,
      "step": 346
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9106193950516023,
      "learning_rate": 9.994755572130996e-06,
      "loss": 0.7091,
      "step": 347
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.053805118303543,
      "learning_rate": 9.994660669182818e-06,
      "loss": 0.62,
      "step": 348
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8778116627318417,
      "learning_rate": 9.994564915695965e-06,
      "loss": 0.6443,
      "step": 349
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0489939813220084,
      "learning_rate": 9.994468311686738e-06,
      "loss": 0.713,
      "step": 350
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.938005710011667,
      "learning_rate": 9.99437085717159e-06,
      "loss": 0.7186,
      "step": 351
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9125421530777931,
      "learning_rate": 9.994272552167116e-06,
      "loss": 0.6405,
      "step": 352
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0192868039618457,
      "learning_rate": 9.994173396690055e-06,
      "loss": 0.7072,
      "step": 353
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.017436335831891,
      "learning_rate": 9.994073390757296e-06,
      "loss": 0.6565,
      "step": 354
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9485778133461242,
      "learning_rate": 9.993972534385866e-06,
      "loss": 0.5994,
      "step": 355
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0398677839945885,
      "learning_rate": 9.993870827592937e-06,
      "loss": 0.7275,
      "step": 356
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8255473602037663,
      "learning_rate": 9.993768270395836e-06,
      "loss": 0.5959,
      "step": 357
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8258622309155117,
      "learning_rate": 9.99366486281202e-06,
      "loss": 0.5621,
      "step": 358
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8949061455906837,
      "learning_rate": 9.993560604859103e-06,
      "loss": 0.5521,
      "step": 359
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8527556019739144,
      "learning_rate": 9.993455496554838e-06,
      "loss": 0.58,
      "step": 360
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8789357795080626,
      "learning_rate": 9.993349537917123e-06,
      "loss": 0.6461,
      "step": 361
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8246177657611419,
      "learning_rate": 9.993242728964001e-06,
      "loss": 0.6181,
      "step": 362
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0375105332126155,
      "learning_rate": 9.993135069713663e-06,
      "loss": 0.6614,
      "step": 363
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8104965800012588,
      "learning_rate": 9.99302656018444e-06,
      "loss": 0.6015,
      "step": 364
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7596096101521556,
      "learning_rate": 9.992917200394812e-06,
      "loss": 0.5891,
      "step": 365
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1644348170500525,
      "learning_rate": 9.992806990363399e-06,
      "loss": 0.6752,
      "step": 366
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1436977846077838,
      "learning_rate": 9.99269593010897e-06,
      "loss": 0.7497,
      "step": 367
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7724466008975204,
      "learning_rate": 9.992584019650437e-06,
      "loss": 0.5779,
      "step": 368
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0039725391305003,
      "learning_rate": 9.992471259006861e-06,
      "loss": 0.6915,
      "step": 369
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8643750102532078,
      "learning_rate": 9.992357648197438e-06,
      "loss": 0.6565,
      "step": 370
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8769587422934415,
      "learning_rate": 9.992243187241519e-06,
      "loss": 0.6275,
      "step": 371
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.999357823422199,
      "learning_rate": 9.992127876158593e-06,
      "loss": 0.588,
      "step": 372
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9815912258356045,
      "learning_rate": 9.992011714968299e-06,
      "loss": 0.6181,
      "step": 373
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9379284359788672,
      "learning_rate": 9.991894703690414e-06,
      "loss": 0.6335,
      "step": 374
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8802045143811273,
      "learning_rate": 9.991776842344866e-06,
      "loss": 0.6009,
      "step": 375
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7951951084578475,
      "learning_rate": 9.991658130951729e-06,
      "loss": 0.6046,
      "step": 376
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0391377129467918,
      "learning_rate": 9.991538569531211e-06,
      "loss": 0.6763,
      "step": 377
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3049012800130029,
      "learning_rate": 9.99141815810368e-06,
      "loss": 0.6774,
      "step": 378
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.945194996853557,
      "learning_rate": 9.991296896689635e-06,
      "loss": 0.6464,
      "step": 379
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7312511061611014,
      "learning_rate": 9.991174785309727e-06,
      "loss": 0.5764,
      "step": 380
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2765573937218397,
      "learning_rate": 9.991051823984751e-06,
      "loss": 0.8081,
      "step": 381
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9244902302904036,
      "learning_rate": 9.990928012735646e-06,
      "loss": 0.605,
      "step": 382
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7325463852480184,
      "learning_rate": 9.990803351583496e-06,
      "loss": 0.5762,
      "step": 383
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.994216536194795,
      "learning_rate": 9.990677840549528e-06,
      "loss": 0.7333,
      "step": 384
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8182606885795133,
      "learning_rate": 9.990551479655117e-06,
      "loss": 0.6297,
      "step": 385
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.840992663713055,
      "learning_rate": 9.990424268921779e-06,
      "loss": 0.68,
      "step": 386
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0141039755293508,
      "learning_rate": 9.99029620837118e-06,
      "loss": 0.7713,
      "step": 387
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7619333670671702,
      "learning_rate": 9.990167298025123e-06,
      "loss": 0.5444,
      "step": 388
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.766071906719017,
      "learning_rate": 9.990037537905563e-06,
      "loss": 0.5622,
      "step": 389
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9833012585342708,
      "learning_rate": 9.989906928034597e-06,
      "loss": 0.6846,
      "step": 390
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8688067861021305,
      "learning_rate": 9.989775468434467e-06,
      "loss": 0.6543,
      "step": 391
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9549303508672472,
      "learning_rate": 9.989643159127554e-06,
      "loss": 0.6923,
      "step": 392
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.826282494354979,
      "learning_rate": 9.989510000136395e-06,
      "loss": 0.6737,
      "step": 393
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7552895110973618,
      "learning_rate": 9.989375991483664e-06,
      "loss": 0.6333,
      "step": 394
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8315473541600976,
      "learning_rate": 9.989241133192178e-06,
      "loss": 0.6379,
      "step": 395
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8378720067306025,
      "learning_rate": 9.989105425284907e-06,
      "loss": 0.6259,
      "step": 396
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0566130118550656,
      "learning_rate": 9.988968867784958e-06,
      "loss": 0.6804,
      "step": 397
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8419946463946731,
      "learning_rate": 9.988831460715584e-06,
      "loss": 0.6163,
      "step": 398
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.834007810611835,
      "learning_rate": 9.988693204100185e-06,
      "loss": 0.6101,
      "step": 399
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8746465329802927,
      "learning_rate": 9.988554097962308e-06,
      "loss": 0.6724,
      "step": 400
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8639812892611826,
      "learning_rate": 9.988414142325637e-06,
      "loss": 0.6684,
      "step": 401
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9759578689345811,
      "learning_rate": 9.988273337214008e-06,
      "loss": 0.6557,
      "step": 402
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7408369216955119,
      "learning_rate": 9.988131682651394e-06,
      "loss": 0.6094,
      "step": 403
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0295894842371458,
      "learning_rate": 9.987989178661923e-06,
      "loss": 0.7053,
      "step": 404
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8860845771509531,
      "learning_rate": 9.987845825269858e-06,
      "loss": 0.6481,
      "step": 405
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7513098847924715,
      "learning_rate": 9.987701622499612e-06,
      "loss": 0.5924,
      "step": 406
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0426045003443716,
      "learning_rate": 9.987556570375743e-06,
      "loss": 0.6915,
      "step": 407
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8521945757805716,
      "learning_rate": 9.987410668922948e-06,
      "loss": 0.5732,
      "step": 408
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9585014676721031,
      "learning_rate": 9.987263918166075e-06,
      "loss": 0.6817,
      "step": 409
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0589047866327423,
      "learning_rate": 9.987116318130115e-06,
      "loss": 0.664,
      "step": 410
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7451349527570841,
      "learning_rate": 9.9869678688402e-06,
      "loss": 0.5642,
      "step": 411
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9152208995938579,
      "learning_rate": 9.98681857032161e-06,
      "loss": 0.5812,
      "step": 412
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9758349805011866,
      "learning_rate": 9.986668422599772e-06,
      "loss": 0.6745,
      "step": 413
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0349726333391913,
      "learning_rate": 9.986517425700251e-06,
      "loss": 0.7296,
      "step": 414
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.841148207353582,
      "learning_rate": 9.986365579648762e-06,
      "loss": 0.5802,
      "step": 415
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8046401674836037,
      "learning_rate": 9.986212884471162e-06,
      "loss": 0.5903,
      "step": 416
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8075564935762903,
      "learning_rate": 9.986059340193453e-06,
      "loss": 0.6427,
      "step": 417
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8285220615452736,
      "learning_rate": 9.985904946841785e-06,
      "loss": 0.59,
      "step": 418
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8361566593322827,
      "learning_rate": 9.985749704442446e-06,
      "loss": 0.5477,
      "step": 419
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.894200543261818,
      "learning_rate": 9.985593613021873e-06,
      "loss": 0.6825,
      "step": 420
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9402822126531339,
      "learning_rate": 9.985436672606647e-06,
      "loss": 0.7059,
      "step": 421
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9927059480094714,
      "learning_rate": 9.985278883223494e-06,
      "loss": 0.7266,
      "step": 422
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6917812724522772,
      "learning_rate": 9.985120244899285e-06,
      "loss": 0.5574,
      "step": 423
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7832286582978668,
      "learning_rate": 9.984960757661031e-06,
      "loss": 0.5573,
      "step": 424
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7987835239334181,
      "learning_rate": 9.984800421535893e-06,
      "loss": 0.6408,
      "step": 425
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7297111616191886,
      "learning_rate": 9.984639236551177e-06,
      "loss": 0.5983,
      "step": 426
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8390909979092573,
      "learning_rate": 9.984477202734327e-06,
      "loss": 0.5989,
      "step": 427
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9470731744767806,
      "learning_rate": 9.984314320112936e-06,
      "loss": 0.6401,
      "step": 428
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.906598707603198,
      "learning_rate": 9.984150588714742e-06,
      "loss": 0.6463,
      "step": 429
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8337135665759178,
      "learning_rate": 9.98398600856763e-06,
      "loss": 0.6216,
      "step": 430
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0073563290493255,
      "learning_rate": 9.983820579699622e-06,
      "loss": 0.6583,
      "step": 431
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7707538345782596,
      "learning_rate": 9.983654302138892e-06,
      "loss": 0.5752,
      "step": 432
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9674923073232958,
      "learning_rate": 9.98348717591375e-06,
      "loss": 0.6957,
      "step": 433
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.905475220676663,
      "learning_rate": 9.983319201052664e-06,
      "loss": 0.5946,
      "step": 434
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9246782522026172,
      "learning_rate": 9.983150377584231e-06,
      "loss": 0.6663,
      "step": 435
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8247841052316118,
      "learning_rate": 9.982980705537204e-06,
      "loss": 0.5997,
      "step": 436
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7650610923549515,
      "learning_rate": 9.982810184940475e-06,
      "loss": 0.557,
      "step": 437
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8039381465215515,
      "learning_rate": 9.98263881582308e-06,
      "loss": 0.6065,
      "step": 438
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0640664396586468,
      "learning_rate": 9.982466598214207e-06,
      "loss": 0.7222,
      "step": 439
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7784472993334844,
      "learning_rate": 9.982293532143177e-06,
      "loss": 0.6365,
      "step": 440
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2232123825374588,
      "learning_rate": 9.982119617639463e-06,
      "loss": 0.6252,
      "step": 441
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9452941580986269,
      "learning_rate": 9.981944854732684e-06,
      "loss": 0.6777,
      "step": 442
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8886308700699889,
      "learning_rate": 9.981769243452595e-06,
      "loss": 0.6608,
      "step": 443
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7735213545941916,
      "learning_rate": 9.981592783829106e-06,
      "loss": 0.6055,
      "step": 444
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0651443112878105,
      "learning_rate": 9.981415475892261e-06,
      "loss": 0.6116,
      "step": 445
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9515438486310432,
      "learning_rate": 9.981237319672258e-06,
      "loss": 0.6564,
      "step": 446
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8446593836206666,
      "learning_rate": 9.981058315199435e-06,
      "loss": 0.6036,
      "step": 447
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7648027821877263,
      "learning_rate": 9.980878462504271e-06,
      "loss": 0.5905,
      "step": 448
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9424532217216225,
      "learning_rate": 9.980697761617397e-06,
      "loss": 0.6954,
      "step": 449
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7914859399082917,
      "learning_rate": 9.980516212569582e-06,
      "loss": 0.5867,
      "step": 450
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7296700315535216,
      "learning_rate": 9.980333815391742e-06,
      "loss": 0.5423,
      "step": 451
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8157784098786468,
      "learning_rate": 9.980150570114939e-06,
      "loss": 0.5467,
      "step": 452
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9873444507617037,
      "learning_rate": 9.979966476770375e-06,
      "loss": 0.7114,
      "step": 453
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9925090296729879,
      "learning_rate": 9.979781535389404e-06,
      "loss": 0.6591,
      "step": 454
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7257964395187778,
      "learning_rate": 9.979595746003514e-06,
      "loss": 0.6221,
      "step": 455
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9116833056992497,
      "learning_rate": 9.979409108644345e-06,
      "loss": 0.6573,
      "step": 456
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8609637143918085,
      "learning_rate": 9.979221623343682e-06,
      "loss": 0.5552,
      "step": 457
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8324267584272339,
      "learning_rate": 9.979033290133449e-06,
      "loss": 0.6656,
      "step": 458
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8641787671699229,
      "learning_rate": 9.978844109045718e-06,
      "loss": 0.6022,
      "step": 459
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9705356247339784,
      "learning_rate": 9.978654080112704e-06,
      "loss": 0.6856,
      "step": 460
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0588435046428086,
      "learning_rate": 9.978463203366768e-06,
      "loss": 0.6653,
      "step": 461
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8165401999178358,
      "learning_rate": 9.978271478840412e-06,
      "loss": 0.6163,
      "step": 462
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.895253816049949,
      "learning_rate": 9.978078906566289e-06,
      "loss": 0.6314,
      "step": 463
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3960561444274502,
      "learning_rate": 9.977885486577188e-06,
      "loss": 0.6623,
      "step": 464
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1365427772340655,
      "learning_rate": 9.977691218906048e-06,
      "loss": 0.6384,
      "step": 465
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7533101727198697,
      "learning_rate": 9.977496103585949e-06,
      "loss": 0.5776,
      "step": 466
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9827800301025504,
      "learning_rate": 9.977300140650121e-06,
      "loss": 0.7097,
      "step": 467
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8250401177590654,
      "learning_rate": 9.977103330131933e-06,
      "loss": 0.5915,
      "step": 468
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9309698276185471,
      "learning_rate": 9.976905672064897e-06,
      "loss": 0.678,
      "step": 469
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9348244603567505,
      "learning_rate": 9.976707166482674e-06,
      "loss": 0.6381,
      "step": 470
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7649254438122951,
      "learning_rate": 9.976507813419067e-06,
      "loss": 0.604,
      "step": 471
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1565357102438112,
      "learning_rate": 9.976307612908027e-06,
      "loss": 0.6867,
      "step": 472
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.879665977717225,
      "learning_rate": 9.976106564983641e-06,
      "loss": 0.6238,
      "step": 473
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8888154006042165,
      "learning_rate": 9.975904669680148e-06,
      "loss": 0.6245,
      "step": 474
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8553969990167989,
      "learning_rate": 9.975701927031929e-06,
      "loss": 0.585,
      "step": 475
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1480627460703725,
      "learning_rate": 9.975498337073508e-06,
      "loss": 0.7234,
      "step": 476
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8510315847624589,
      "learning_rate": 9.975293899839555e-06,
      "loss": 0.5932,
      "step": 477
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.959623216416685,
      "learning_rate": 9.975088615364883e-06,
      "loss": 0.6432,
      "step": 478
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9991654847289766,
      "learning_rate": 9.97488248368445e-06,
      "loss": 0.6237,
      "step": 479
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9434243787216019,
      "learning_rate": 9.974675504833356e-06,
      "loss": 0.7033,
      "step": 480
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8816702660460409,
      "learning_rate": 9.974467678846851e-06,
      "loss": 0.6907,
      "step": 481
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.698224354471496,
      "learning_rate": 9.974259005760324e-06,
      "loss": 0.5637,
      "step": 482
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9431702468716794,
      "learning_rate": 9.97404948560931e-06,
      "loss": 0.6639,
      "step": 483
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0089455605187896,
      "learning_rate": 9.973839118429486e-06,
      "loss": 0.6859,
      "step": 484
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7341477743553783,
      "learning_rate": 9.97362790425668e-06,
      "loss": 0.6228,
      "step": 485
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.788455171090866,
      "learning_rate": 9.973415843126854e-06,
      "loss": 0.5872,
      "step": 486
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.989000580701188,
      "learning_rate": 9.973202935076125e-06,
      "loss": 0.7115,
      "step": 487
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8341745173777312,
      "learning_rate": 9.972989180140746e-06,
      "loss": 0.6161,
      "step": 488
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8408517299637173,
      "learning_rate": 9.972774578357118e-06,
      "loss": 0.603,
      "step": 489
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7641029908200004,
      "learning_rate": 9.972559129761784e-06,
      "loss": 0.597,
      "step": 490
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.049208228809266,
      "learning_rate": 9.972342834391435e-06,
      "loss": 0.6909,
      "step": 491
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9352269185055221,
      "learning_rate": 9.972125692282902e-06,
      "loss": 0.6827,
      "step": 492
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.027543683357878,
      "learning_rate": 9.971907703473164e-06,
      "loss": 0.6594,
      "step": 493
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8274700480961436,
      "learning_rate": 9.97168886799934e-06,
      "loss": 0.6168,
      "step": 494
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7663034374898712,
      "learning_rate": 9.971469185898697e-06,
      "loss": 0.5452,
      "step": 495
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7609980260288177,
      "learning_rate": 9.971248657208644e-06,
      "loss": 0.5837,
      "step": 496
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8611117285561701,
      "learning_rate": 9.971027281966735e-06,
      "loss": 0.6567,
      "step": 497
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0093881198913883,
      "learning_rate": 9.970805060210666e-06,
      "loss": 0.7164,
      "step": 498
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7038884292490618,
      "learning_rate": 9.970581991978283e-06,
      "loss": 0.5605,
      "step": 499
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8478959947775854,
      "learning_rate": 9.97035807730757e-06,
      "loss": 0.5768,
      "step": 500
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8660748011522885,
      "learning_rate": 9.970133316236656e-06,
      "loss": 0.6082,
      "step": 501
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.970012780534241,
      "learning_rate": 9.969907708803816e-06,
      "loss": 0.6396,
      "step": 502
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7214940225898968,
      "learning_rate": 9.96968125504747e-06,
      "loss": 0.6538,
      "step": 503
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9216898815650768,
      "learning_rate": 9.969453955006182e-06,
      "loss": 0.6315,
      "step": 504
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9806306613287539,
      "learning_rate": 9.969225808718655e-06,
      "loss": 0.7167,
      "step": 505
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0923750163480554,
      "learning_rate": 9.968996816223741e-06,
      "loss": 0.7147,
      "step": 506
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8489171584194607,
      "learning_rate": 9.968766977560437e-06,
      "loss": 0.6071,
      "step": 507
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8024965268568347,
      "learning_rate": 9.968536292767881e-06,
      "loss": 0.5921,
      "step": 508
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9430179010466601,
      "learning_rate": 9.968304761885355e-06,
      "loss": 0.5994,
      "step": 509
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0330657022410388,
      "learning_rate": 9.96807238495229e-06,
      "loss": 0.7171,
      "step": 510
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7778810841555546,
      "learning_rate": 9.967839162008253e-06,
      "loss": 0.5984,
      "step": 511
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9431657191058711,
      "learning_rate": 9.96760509309296e-06,
      "loss": 0.6662,
      "step": 512
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.756202180613133,
      "learning_rate": 9.967370178246274e-06,
      "loss": 0.5899,
      "step": 513
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.802061654331361,
      "learning_rate": 9.967134417508196e-06,
      "loss": 0.5971,
      "step": 514
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9856204798795912,
      "learning_rate": 9.966897810918874e-06,
      "loss": 0.6816,
      "step": 515
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8534539607791247,
      "learning_rate": 9.966660358518599e-06,
      "loss": 0.5815,
      "step": 516
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9286042081053918,
      "learning_rate": 9.966422060347806e-06,
      "loss": 0.6834,
      "step": 517
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9700822499918169,
      "learning_rate": 9.966182916447077e-06,
      "loss": 0.7254,
      "step": 518
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9878118402578369,
      "learning_rate": 9.965942926857136e-06,
      "loss": 0.5882,
      "step": 519
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7579724665624524,
      "learning_rate": 9.965702091618848e-06,
      "loss": 0.6427,
      "step": 520
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7994295267037167,
      "learning_rate": 9.965460410773227e-06,
      "loss": 0.5647,
      "step": 521
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7749684827307922,
      "learning_rate": 9.965217884361427e-06,
      "loss": 0.6228,
      "step": 522
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7561752724398992,
      "learning_rate": 9.964974512424749e-06,
      "loss": 0.5752,
      "step": 523
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.858260704256212,
      "learning_rate": 9.964730295004637e-06,
      "loss": 0.6834,
      "step": 524
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8504676037936518,
      "learning_rate": 9.964485232142676e-06,
      "loss": 0.611,
      "step": 525
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9507580212115982,
      "learning_rate": 9.964239323880602e-06,
      "loss": 0.6919,
      "step": 526
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.926602669416903,
      "learning_rate": 9.963992570260286e-06,
      "loss": 0.7052,
      "step": 527
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7900489724282319,
      "learning_rate": 9.96374497132375e-06,
      "loss": 0.592,
      "step": 528
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8126066341296388,
      "learning_rate": 9.96349652711316e-06,
      "loss": 0.6392,
      "step": 529
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8378157404905768,
      "learning_rate": 9.963247237670816e-06,
      "loss": 0.629,
      "step": 530
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0248692233377634,
      "learning_rate": 9.962997103039177e-06,
      "loss": 0.618,
      "step": 531
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8909205812077798,
      "learning_rate": 9.962746123260835e-06,
      "loss": 0.6763,
      "step": 532
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8930382251743185,
      "learning_rate": 9.962494298378532e-06,
      "loss": 0.7251,
      "step": 533
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9299771533511041,
      "learning_rate": 9.962241628435146e-06,
      "loss": 0.6301,
      "step": 534
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7208837913193927,
      "learning_rate": 9.961988113473708e-06,
      "loss": 0.5523,
      "step": 535
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9322336634144031,
      "learning_rate": 9.961733753537388e-06,
      "loss": 0.6001,
      "step": 536
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9973221223155072,
      "learning_rate": 9.9614785486695e-06,
      "loss": 0.6837,
      "step": 537
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9298660693205547,
      "learning_rate": 9.961222498913506e-06,
      "loss": 0.715,
      "step": 538
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0199642534082014,
      "learning_rate": 9.960965604313003e-06,
      "loss": 0.5889,
      "step": 539
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9184903210157217,
      "learning_rate": 9.960707864911741e-06,
      "loss": 0.5917,
      "step": 540
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.75524194232963,
      "learning_rate": 9.960449280753612e-06,
      "loss": 0.5755,
      "step": 541
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.490969590131637,
      "learning_rate": 9.960189851882646e-06,
      "loss": 0.6529,
      "step": 542
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7892066654765252,
      "learning_rate": 9.959929578343024e-06,
      "loss": 0.5954,
      "step": 543
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7625967089780826,
      "learning_rate": 9.959668460179066e-06,
      "loss": 0.5567,
      "step": 544
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9337933156009779,
      "learning_rate": 9.95940649743524e-06,
      "loss": 0.6746,
      "step": 545
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8018923757576564,
      "learning_rate": 9.959143690156152e-06,
      "loss": 0.5953,
      "step": 546
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.772375140157595,
      "learning_rate": 9.95888003838656e-06,
      "loss": 0.551,
      "step": 547
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7270472389387665,
      "learning_rate": 9.958615542171359e-06,
      "loss": 0.566,
      "step": 548
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7037495319898727,
      "learning_rate": 9.958350201555587e-06,
      "loss": 0.5796,
      "step": 549
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8365102880514173,
      "learning_rate": 9.958084016584432e-06,
      "loss": 0.6753,
      "step": 550
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8268714161478222,
      "learning_rate": 9.957816987303223e-06,
      "loss": 0.6888,
      "step": 551
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9462676729925515,
      "learning_rate": 9.957549113757433e-06,
      "loss": 0.6411,
      "step": 552
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7633213086191619,
      "learning_rate": 9.957280395992674e-06,
      "loss": 0.5946,
      "step": 553
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8291074079135722,
      "learning_rate": 9.957010834054707e-06,
      "loss": 0.6093,
      "step": 554
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7439977973378293,
      "learning_rate": 9.95674042798944e-06,
      "loss": 0.592,
      "step": 555
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.031637133465603,
      "learning_rate": 9.956469177842914e-06,
      "loss": 0.7773,
      "step": 556
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7283800531091312,
      "learning_rate": 9.956197083661325e-06,
      "loss": 0.6187,
      "step": 557
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8848565692866931,
      "learning_rate": 9.955924145491005e-06,
      "loss": 0.6969,
      "step": 558
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8844647691938945,
      "learning_rate": 9.955650363378434e-06,
      "loss": 0.6591,
      "step": 559
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8726042709977126,
      "learning_rate": 9.955375737370234e-06,
      "loss": 0.5797,
      "step": 560
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7313265588369671,
      "learning_rate": 9.95510026751317e-06,
      "loss": 0.5666,
      "step": 561
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9607523313588162,
      "learning_rate": 9.954823953854152e-06,
      "loss": 0.6443,
      "step": 562
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8490946250137187,
      "learning_rate": 9.954546796440234e-06,
      "loss": 0.6299,
      "step": 563
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.750610669112727,
      "learning_rate": 9.954268795318614e-06,
      "loss": 0.5952,
      "step": 564
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0445261825511276,
      "learning_rate": 9.95398995053663e-06,
      "loss": 0.7069,
      "step": 565
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2254757431723826,
      "learning_rate": 9.953710262141767e-06,
      "loss": 0.7476,
      "step": 566
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8103706528243231,
      "learning_rate": 9.953429730181653e-06,
      "loss": 0.6309,
      "step": 567
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8165130976313479,
      "learning_rate": 9.953148354704063e-06,
      "loss": 0.6171,
      "step": 568
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8806847613412369,
      "learning_rate": 9.95286613575691e-06,
      "loss": 0.5946,
      "step": 569
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.908668295729378,
      "learning_rate": 9.952583073388251e-06,
      "loss": 0.7032,
      "step": 570
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8053027613304646,
      "learning_rate": 9.95229916764629e-06,
      "loss": 0.584,
      "step": 571
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.93489348644883,
      "learning_rate": 9.952014418579375e-06,
      "loss": 0.6011,
      "step": 572
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0654744724998089,
      "learning_rate": 9.951728826235993e-06,
      "loss": 0.7225,
      "step": 573
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9128936705681024,
      "learning_rate": 9.951442390664778e-06,
      "loss": 0.6607,
      "step": 574
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7638159978482243,
      "learning_rate": 9.951155111914508e-06,
      "loss": 0.6214,
      "step": 575
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0592710862221595,
      "learning_rate": 9.950866990034105e-06,
      "loss": 0.7441,
      "step": 576
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7626376664673241,
      "learning_rate": 9.95057802507263e-06,
      "loss": 0.5457,
      "step": 577
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7101151695410787,
      "learning_rate": 9.950288217079292e-06,
      "loss": 0.5885,
      "step": 578
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8943339463215348,
      "learning_rate": 9.949997566103442e-06,
      "loss": 0.6849,
      "step": 579
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8531538235528026,
      "learning_rate": 9.949706072194575e-06,
      "loss": 0.6846,
      "step": 580
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8623475976212491,
      "learning_rate": 9.949413735402332e-06,
      "loss": 0.7354,
      "step": 581
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8640465576649092,
      "learning_rate": 9.94912055577649e-06,
      "loss": 0.6962,
      "step": 582
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9768915207729412,
      "learning_rate": 9.948826533366977e-06,
      "loss": 0.7436,
      "step": 583
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8315723975788107,
      "learning_rate": 9.948531668223865e-06,
      "loss": 0.5185,
      "step": 584
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3308345548950664,
      "learning_rate": 9.948235960397358e-06,
      "loss": 0.6698,
      "step": 585
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9270428971647223,
      "learning_rate": 9.94793940993782e-06,
      "loss": 0.5367,
      "step": 586
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7989050894598666,
      "learning_rate": 9.947642016895749e-06,
      "loss": 0.6386,
      "step": 587
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9777221564182457,
      "learning_rate": 9.947343781321786e-06,
      "loss": 0.6644,
      "step": 588
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.76484889469589,
      "learning_rate": 9.947044703266718e-06,
      "loss": 0.6045,
      "step": 589
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.93063026304871,
      "learning_rate": 9.946744782781474e-06,
      "loss": 0.6963,
      "step": 590
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8551049796434639,
      "learning_rate": 9.94644401991713e-06,
      "loss": 0.6274,
      "step": 591
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8134392412997804,
      "learning_rate": 9.9461424147249e-06,
      "loss": 0.5948,
      "step": 592
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.896278160288477,
      "learning_rate": 9.945839967256146e-06,
      "loss": 0.7359,
      "step": 593
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9812497988712363,
      "learning_rate": 9.945536677562368e-06,
      "loss": 0.6704,
      "step": 594
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8209110223875445,
      "learning_rate": 9.94523254569522e-06,
      "loss": 0.6057,
      "step": 595
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6947022442188788,
      "learning_rate": 9.944927571706486e-06,
      "loss": 0.5691,
      "step": 596
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0114013030643787,
      "learning_rate": 9.944621755648105e-06,
      "loss": 0.6591,
      "step": 597
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7373920211522685,
      "learning_rate": 9.944315097572148e-06,
      "loss": 0.5652,
      "step": 598
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8274232703460835,
      "learning_rate": 9.944007597530842e-06,
      "loss": 0.6304,
      "step": 599
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.922224773657226,
      "learning_rate": 9.943699255576549e-06,
      "loss": 0.6006,
      "step": 600
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8164923669342885,
      "learning_rate": 9.943390071761774e-06,
      "loss": 0.5697,
      "step": 601
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7756446463136811,
      "learning_rate": 9.94308004613917e-06,
      "loss": 0.5573,
      "step": 602
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9780185005685222,
      "learning_rate": 9.94276917876153e-06,
      "loss": 0.6268,
      "step": 603
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.750615458466306,
      "learning_rate": 9.942457469681794e-06,
      "loss": 0.5706,
      "step": 604
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7668287276372201,
      "learning_rate": 9.94214491895304e-06,
      "loss": 0.6163,
      "step": 605
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7516588483787702,
      "learning_rate": 9.941831526628495e-06,
      "loss": 0.5768,
      "step": 606
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7837604787075109,
      "learning_rate": 9.941517292761523e-06,
      "loss": 0.577,
      "step": 607
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8484282516972161,
      "learning_rate": 9.941202217405637e-06,
      "loss": 0.6096,
      "step": 608
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7569221912628195,
      "learning_rate": 9.94088630061449e-06,
      "loss": 0.6125,
      "step": 609
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7038152962167636,
      "learning_rate": 9.940569542441881e-06,
      "loss": 0.5778,
      "step": 610
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9820651063216177,
      "learning_rate": 9.940251942941747e-06,
      "loss": 0.7132,
      "step": 611
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8446938223008503,
      "learning_rate": 9.939933502168175e-06,
      "loss": 0.5826,
      "step": 612
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9444598688114014,
      "learning_rate": 9.939614220175393e-06,
      "loss": 0.6982,
      "step": 613
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1620093172968176,
      "learning_rate": 9.939294097017768e-06,
      "loss": 0.6894,
      "step": 614
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9226729624115031,
      "learning_rate": 9.938973132749818e-06,
      "loss": 0.6751,
      "step": 615
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8861209902999666,
      "learning_rate": 9.938651327426196e-06,
      "loss": 0.7189,
      "step": 616
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.730152321451675,
      "learning_rate": 9.938328681101703e-06,
      "loss": 0.5676,
      "step": 617
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.18406401003547,
      "learning_rate": 9.938005193831282e-06,
      "loss": 0.6978,
      "step": 618
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7328855186267786,
      "learning_rate": 9.93768086567002e-06,
      "loss": 0.5415,
      "step": 619
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7631245160779453,
      "learning_rate": 9.93735569667315e-06,
      "loss": 0.6004,
      "step": 620
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1313154696333698,
      "learning_rate": 9.93702968689604e-06,
      "loss": 0.6476,
      "step": 621
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7750062025343325,
      "learning_rate": 9.936702836394208e-06,
      "loss": 0.6006,
      "step": 622
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.074813819328632,
      "learning_rate": 9.936375145223313e-06,
      "loss": 0.6401,
      "step": 623
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6498973923795885,
      "learning_rate": 9.936046613439157e-06,
      "loss": 0.5269,
      "step": 624
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9313633736715803,
      "learning_rate": 9.935717241097686e-06,
      "loss": 0.698,
      "step": 625
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9218926344512484,
      "learning_rate": 9.93538702825499e-06,
      "loss": 0.6667,
      "step": 626
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7298388773146208,
      "learning_rate": 9.935055974967299e-06,
      "loss": 0.618,
      "step": 627
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.845635745197338,
      "learning_rate": 9.93472408129099e-06,
      "loss": 0.6254,
      "step": 628
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6506275909397085,
      "learning_rate": 9.934391347282576e-06,
      "loss": 0.5231,
      "step": 629
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.994721563230209,
      "learning_rate": 9.934057772998725e-06,
      "loss": 0.7173,
      "step": 630
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8938602928885192,
      "learning_rate": 9.933723358496236e-06,
      "loss": 0.6704,
      "step": 631
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8833262463756573,
      "learning_rate": 9.93338810383206e-06,
      "loss": 0.609,
      "step": 632
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9583228384857131,
      "learning_rate": 9.933052009063285e-06,
      "loss": 0.7423,
      "step": 633
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9040067970946756,
      "learning_rate": 9.932715074247144e-06,
      "loss": 0.6348,
      "step": 634
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7926578120685713,
      "learning_rate": 9.932377299441014e-06,
      "loss": 0.5922,
      "step": 635
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8436771748734967,
      "learning_rate": 9.932038684702417e-06,
      "loss": 0.5635,
      "step": 636
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8237469260465538,
      "learning_rate": 9.931699230089011e-06,
      "loss": 0.6019,
      "step": 637
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8090287198120906,
      "learning_rate": 9.931358935658606e-06,
      "loss": 0.5803,
      "step": 638
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9953598802689478,
      "learning_rate": 9.93101780146915e-06,
      "loss": 0.6923,
      "step": 639
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7630701341604019,
      "learning_rate": 9.930675827578731e-06,
      "loss": 0.5498,
      "step": 640
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7716374554508619,
      "learning_rate": 9.930333014045587e-06,
      "loss": 0.607,
      "step": 641
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8305878537378821,
      "learning_rate": 9.929989360928094e-06,
      "loss": 0.5926,
      "step": 642
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3823498431942076,
      "learning_rate": 9.929644868284773e-06,
      "loss": 0.6768,
      "step": 643
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7682614029356418,
      "learning_rate": 9.929299536174289e-06,
      "loss": 0.5367,
      "step": 644
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7677354413042392,
      "learning_rate": 9.928953364655447e-06,
      "loss": 0.5381,
      "step": 645
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7548625077982958,
      "learning_rate": 9.928606353787195e-06,
      "loss": 0.5819,
      "step": 646
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8086537365085523,
      "learning_rate": 9.928258503628627e-06,
      "loss": 0.6115,
      "step": 647
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7668718255159341,
      "learning_rate": 9.927909814238979e-06,
      "loss": 0.5757,
      "step": 648
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9101451122571584,
      "learning_rate": 9.927560285677628e-06,
      "loss": 0.676,
      "step": 649
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8663523860531017,
      "learning_rate": 9.927209918004095e-06,
      "loss": 0.6648,
      "step": 650
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9804079749783077,
      "learning_rate": 9.926858711278045e-06,
      "loss": 0.7192,
      "step": 651
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7713409436857132,
      "learning_rate": 9.926506665559283e-06,
      "loss": 0.612,
      "step": 652
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8105425218434699,
      "learning_rate": 9.926153780907761e-06,
      "loss": 0.6037,
      "step": 653
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9602434806882082,
      "learning_rate": 9.92580005738357e-06,
      "loss": 0.6451,
      "step": 654
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7355057116304701,
      "learning_rate": 9.925445495046945e-06,
      "loss": 0.5907,
      "step": 655
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.922586656832762,
      "learning_rate": 9.925090093958265e-06,
      "loss": 0.6592,
      "step": 656
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8218299017388514,
      "learning_rate": 9.924733854178052e-06,
      "loss": 0.5916,
      "step": 657
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8246134875640933,
      "learning_rate": 9.92437677576697e-06,
      "loss": 0.5924,
      "step": 658
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8761614197107747,
      "learning_rate": 9.924018858785824e-06,
      "loss": 0.7165,
      "step": 659
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.843966322355839,
      "learning_rate": 9.923660103295563e-06,
      "loss": 0.6611,
      "step": 660
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7858714903646885,
      "learning_rate": 9.923300509357283e-06,
      "loss": 0.6406,
      "step": 661
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7209605043597681,
      "learning_rate": 9.922940077032214e-06,
      "loss": 0.6448,
      "step": 662
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9623215583902911,
      "learning_rate": 9.92257880638174e-06,
      "loss": 0.7363,
      "step": 663
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9502695108766952,
      "learning_rate": 9.922216697467378e-06,
      "loss": 0.6628,
      "step": 664
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7806007967785518,
      "learning_rate": 9.92185375035079e-06,
      "loss": 0.54,
      "step": 665
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8806724514659502,
      "learning_rate": 9.921489965093785e-06,
      "loss": 0.6124,
      "step": 666
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1387755783997202,
      "learning_rate": 9.921125341758308e-06,
      "loss": 0.7379,
      "step": 667
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9299146349283249,
      "learning_rate": 9.920759880406456e-06,
      "loss": 0.6754,
      "step": 668
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8669405325971468,
      "learning_rate": 9.92039358110046e-06,
      "loss": 0.6673,
      "step": 669
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.717965757974735,
      "learning_rate": 9.920026443902698e-06,
      "loss": 0.6088,
      "step": 670
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0134517646772696,
      "learning_rate": 9.919658468875689e-06,
      "loss": 0.7304,
      "step": 671
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8253834412205865,
      "learning_rate": 9.919289656082093e-06,
      "loss": 0.6032,
      "step": 672
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8908513498578015,
      "learning_rate": 9.91892000558472e-06,
      "loss": 0.6556,
      "step": 673
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.692491090448771,
      "learning_rate": 9.918549517446513e-06,
      "loss": 0.5595,
      "step": 674
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7525718663539719,
      "learning_rate": 9.918178191730565e-06,
      "loss": 0.585,
      "step": 675
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.708981498945706,
      "learning_rate": 9.917806028500109e-06,
      "loss": 0.5522,
      "step": 676
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8877344425437592,
      "learning_rate": 9.91743302781852e-06,
      "loss": 0.5998,
      "step": 677
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9016956231450495,
      "learning_rate": 9.917059189749314e-06,
      "loss": 0.6565,
      "step": 678
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6946495146054391,
      "learning_rate": 9.916684514356153e-06,
      "loss": 0.5896,
      "step": 679
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7027121668087554,
      "learning_rate": 9.916309001702844e-06,
      "loss": 0.5723,
      "step": 680
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9192736379936258,
      "learning_rate": 9.915932651853327e-06,
      "loss": 0.6864,
      "step": 681
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7098074011381186,
      "learning_rate": 9.915555464871694e-06,
      "loss": 0.5533,
      "step": 682
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8120449469522245,
      "learning_rate": 9.915177440822177e-06,
      "loss": 0.61,
      "step": 683
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7409337740136418,
      "learning_rate": 9.914798579769144e-06,
      "loss": 0.6112,
      "step": 684
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0142076692103514,
      "learning_rate": 9.914418881777118e-06,
      "loss": 0.6815,
      "step": 685
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8327428479309273,
      "learning_rate": 9.914038346910754e-06,
      "loss": 0.6142,
      "step": 686
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7391678451146036,
      "learning_rate": 9.913656975234853e-06,
      "loss": 0.5659,
      "step": 687
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6909004767866131,
      "learning_rate": 9.91327476681436e-06,
      "loss": 0.5616,
      "step": 688
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7291099800620955,
      "learning_rate": 9.91289172171436e-06,
      "loss": 0.5801,
      "step": 689
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8525452639489284,
      "learning_rate": 9.912507840000081e-06,
      "loss": 0.6405,
      "step": 690
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8416155895502007,
      "learning_rate": 9.912123121736896e-06,
      "loss": 0.6471,
      "step": 691
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9151314287094644,
      "learning_rate": 9.911737566990316e-06,
      "loss": 0.696,
      "step": 692
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7510500670706307,
      "learning_rate": 9.911351175826e-06,
      "loss": 0.579,
      "step": 693
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9476143372389542,
      "learning_rate": 9.910963948309744e-06,
      "loss": 0.6292,
      "step": 694
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7858968747578426,
      "learning_rate": 9.91057588450749e-06,
      "loss": 0.5726,
      "step": 695
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7324210337557129,
      "learning_rate": 9.910186984485321e-06,
      "loss": 0.5577,
      "step": 696
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7029992476164638,
      "learning_rate": 9.909797248309463e-06,
      "loss": 0.6156,
      "step": 697
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9636352807473783,
      "learning_rate": 9.909406676046282e-06,
      "loss": 0.6438,
      "step": 698
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8213434682319596,
      "learning_rate": 9.909015267762292e-06,
      "loss": 0.6516,
      "step": 699
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7998111431369197,
      "learning_rate": 9.908623023524143e-06,
      "loss": 0.5824,
      "step": 700
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8959014133293823,
      "learning_rate": 9.908229943398629e-06,
      "loss": 0.7201,
      "step": 701
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8667368634098876,
      "learning_rate": 9.90783602745269e-06,
      "loss": 0.5919,
      "step": 702
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9801267598627209,
      "learning_rate": 9.907441275753407e-06,
      "loss": 0.7418,
      "step": 703
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9703574617065216,
      "learning_rate": 9.907045688367998e-06,
      "loss": 0.6869,
      "step": 704
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8406415430843424,
      "learning_rate": 9.90664926536383e-06,
      "loss": 0.6407,
      "step": 705
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7629023886992453,
      "learning_rate": 9.906252006808411e-06,
      "loss": 0.5756,
      "step": 706
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0298108255803995,
      "learning_rate": 9.905853912769385e-06,
      "loss": 0.6129,
      "step": 707
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9282339551154486,
      "learning_rate": 9.90545498331455e-06,
      "loss": 0.6178,
      "step": 708
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7636587035815693,
      "learning_rate": 9.905055218511832e-06,
      "loss": 0.5936,
      "step": 709
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9119017883934157,
      "learning_rate": 9.904654618429312e-06,
      "loss": 0.6629,
      "step": 710
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8810501453237173,
      "learning_rate": 9.904253183135209e-06,
      "loss": 0.6775,
      "step": 711
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.917049858842309,
      "learning_rate": 9.903850912697879e-06,
      "loss": 0.6797,
      "step": 712
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1023512708693004,
      "learning_rate": 9.903447807185827e-06,
      "loss": 0.6732,
      "step": 713
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8893156064284122,
      "learning_rate": 9.903043866667697e-06,
      "loss": 0.6614,
      "step": 714
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7901393067993758,
      "learning_rate": 9.902639091212276e-06,
      "loss": 0.6138,
      "step": 715
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7873768019751562,
      "learning_rate": 9.902233480888495e-06,
      "loss": 0.5492,
      "step": 716
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9237841453100953,
      "learning_rate": 9.901827035765422e-06,
      "loss": 0.685,
      "step": 717
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7998758371783121,
      "learning_rate": 9.90141975591227e-06,
      "loss": 0.5834,
      "step": 718
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6702074175253155,
      "learning_rate": 9.901011641398398e-06,
      "loss": 0.524,
      "step": 719
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8423372947256208,
      "learning_rate": 9.900602692293303e-06,
      "loss": 0.6597,
      "step": 720
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8275800935623158,
      "learning_rate": 9.90019290866662e-06,
      "loss": 0.5679,
      "step": 721
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7686967913973015,
      "learning_rate": 9.899782290588138e-06,
      "loss": 0.604,
      "step": 722
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.857706491871605,
      "learning_rate": 9.899370838127775e-06,
      "loss": 0.6264,
      "step": 723
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7333631846418711,
      "learning_rate": 9.8989585513556e-06,
      "loss": 0.5393,
      "step": 724
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9838373877664329,
      "learning_rate": 9.898545430341821e-06,
      "loss": 0.7017,
      "step": 725
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.974486654890674,
      "learning_rate": 9.898131475156788e-06,
      "loss": 0.739,
      "step": 726
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.812168267768625,
      "learning_rate": 9.897716685870994e-06,
      "loss": 0.5399,
      "step": 727
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7579017390354547,
      "learning_rate": 9.89730106255507e-06,
      "loss": 0.5417,
      "step": 728
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9102695778461423,
      "learning_rate": 9.896884605279798e-06,
      "loss": 0.6113,
      "step": 729
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8304422211649709,
      "learning_rate": 9.89646731411609e-06,
      "loss": 0.6445,
      "step": 730
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9882610697091398,
      "learning_rate": 9.896049189135012e-06,
      "loss": 0.7118,
      "step": 731
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8567763171882996,
      "learning_rate": 9.895630230407764e-06,
      "loss": 0.5456,
      "step": 732
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9809733207959377,
      "learning_rate": 9.89521043800569e-06,
      "loss": 0.6782,
      "step": 733
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9884907501295046,
      "learning_rate": 9.894789812000273e-06,
      "loss": 0.6151,
      "step": 734
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7460169974638555,
      "learning_rate": 9.89436835246315e-06,
      "loss": 0.5616,
      "step": 735
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.721660580584569,
      "learning_rate": 9.89394605946608e-06,
      "loss": 0.566,
      "step": 736
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7413767170113644,
      "learning_rate": 9.893522933080984e-06,
      "loss": 0.5614,
      "step": 737
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9417369873345561,
      "learning_rate": 9.893098973379912e-06,
      "loss": 0.6768,
      "step": 738
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8578953015537314,
      "learning_rate": 9.89267418043506e-06,
      "loss": 0.6425,
      "step": 739
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7480721188181566,
      "learning_rate": 9.892248554318769e-06,
      "loss": 0.5589,
      "step": 740
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8452017810861409,
      "learning_rate": 9.891822095103513e-06,
      "loss": 0.5668,
      "step": 741
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9039908276260701,
      "learning_rate": 9.89139480286192e-06,
      "loss": 0.573,
      "step": 742
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6779024497797658,
      "learning_rate": 9.890966677666749e-06,
      "loss": 0.5731,
      "step": 743
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6946726295782174,
      "learning_rate": 9.890537719590907e-06,
      "loss": 0.5142,
      "step": 744
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.808765705975854,
      "learning_rate": 9.89010792870744e-06,
      "loss": 0.6342,
      "step": 745
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6628179380064558,
      "learning_rate": 9.889677305089536e-06,
      "loss": 0.4992,
      "step": 746
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0453292672839902,
      "learning_rate": 9.889245848810529e-06,
      "loss": 0.6636,
      "step": 747
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7621998662806297,
      "learning_rate": 9.888813559943891e-06,
      "loss": 0.612,
      "step": 748
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9565742612518761,
      "learning_rate": 9.888380438563232e-06,
      "loss": 0.6449,
      "step": 749
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8487784993806144,
      "learning_rate": 9.887946484742314e-06,
      "loss": 0.6162,
      "step": 750
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8597309250550237,
      "learning_rate": 9.88751169855503e-06,
      "loss": 0.6689,
      "step": 751
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6981979908490888,
      "learning_rate": 9.887076080075422e-06,
      "loss": 0.5867,
      "step": 752
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6548144431248555,
      "learning_rate": 9.88663962937767e-06,
      "loss": 0.5657,
      "step": 753
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8881334549929241,
      "learning_rate": 9.8862023465361e-06,
      "loss": 0.628,
      "step": 754
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9692547090827943,
      "learning_rate": 9.885764231625172e-06,
      "loss": 0.6624,
      "step": 755
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8393730405810584,
      "learning_rate": 9.885325284719495e-06,
      "loss": 0.6269,
      "step": 756
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7104667944001845,
      "learning_rate": 9.884885505893819e-06,
      "loss": 0.6165,
      "step": 757
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.872908550944843,
      "learning_rate": 9.88444489522303e-06,
      "loss": 0.6199,
      "step": 758
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7429284831172942,
      "learning_rate": 9.884003452782162e-06,
      "loss": 0.5557,
      "step": 759
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8818667407877973,
      "learning_rate": 9.883561178646387e-06,
      "loss": 0.6332,
      "step": 760
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.74189317405622,
      "learning_rate": 9.88311807289102e-06,
      "loss": 0.5621,
      "step": 761
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8038857385308639,
      "learning_rate": 9.882674135591516e-06,
      "loss": 0.6475,
      "step": 762
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.694763821643562,
      "learning_rate": 9.882229366823475e-06,
      "loss": 0.5596,
      "step": 763
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8216853227179713,
      "learning_rate": 9.881783766662635e-06,
      "loss": 0.5651,
      "step": 764
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7505648721723389,
      "learning_rate": 9.881337335184879e-06,
      "loss": 0.5246,
      "step": 765
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8171642955631234,
      "learning_rate": 9.880890072466227e-06,
      "loss": 0.6126,
      "step": 766
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8696487145327768,
      "learning_rate": 9.880441978582844e-06,
      "loss": 0.687,
      "step": 767
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7238153115144139,
      "learning_rate": 9.879993053611037e-06,
      "loss": 0.5398,
      "step": 768
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9478120655667572,
      "learning_rate": 9.879543297627252e-06,
      "loss": 0.6493,
      "step": 769
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9027328102335653,
      "learning_rate": 9.879092710708078e-06,
      "loss": 0.599,
      "step": 770
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9284578286283937,
      "learning_rate": 9.878641292930247e-06,
      "loss": 0.6689,
      "step": 771
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8548342980449497,
      "learning_rate": 9.878189044370628e-06,
      "loss": 0.6352,
      "step": 772
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6962359676542369,
      "learning_rate": 9.877735965106234e-06,
      "loss": 0.5981,
      "step": 773
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7441704767941549,
      "learning_rate": 9.877282055214222e-06,
      "loss": 0.5656,
      "step": 774
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9518230764048553,
      "learning_rate": 9.876827314771889e-06,
      "loss": 0.6684,
      "step": 775
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8658989510790782,
      "learning_rate": 9.876371743856667e-06,
      "loss": 0.631,
      "step": 776
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8626920094777617,
      "learning_rate": 9.87591534254614e-06,
      "loss": 0.678,
      "step": 777
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7569179234065797,
      "learning_rate": 9.87545811091803e-06,
      "loss": 0.5777,
      "step": 778
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7667165817688688,
      "learning_rate": 9.875000049050193e-06,
      "loss": 0.6184,
      "step": 779
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0001633898335596,
      "learning_rate": 9.874541157020636e-06,
      "loss": 0.7321,
      "step": 780
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8045789331930552,
      "learning_rate": 9.8740814349075e-06,
      "loss": 0.6109,
      "step": 781
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7588845064764372,
      "learning_rate": 9.873620882789077e-06,
      "loss": 0.5983,
      "step": 782
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7544801666102019,
      "learning_rate": 9.87315950074379e-06,
      "loss": 0.5767,
      "step": 783
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9743045985283472,
      "learning_rate": 9.872697288850208e-06,
      "loss": 0.6516,
      "step": 784
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8479404444370521,
      "learning_rate": 9.87223424718704e-06,
      "loss": 0.5399,
      "step": 785
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9695049182869461,
      "learning_rate": 9.871770375833139e-06,
      "loss": 0.6951,
      "step": 786
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6634064398127952,
      "learning_rate": 9.871305674867497e-06,
      "loss": 0.5388,
      "step": 787
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8137494509652877,
      "learning_rate": 9.870840144369247e-06,
      "loss": 0.5487,
      "step": 788
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7223817338627702,
      "learning_rate": 9.870373784417664e-06,
      "loss": 0.6069,
      "step": 789
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9629317971760376,
      "learning_rate": 9.869906595092165e-06,
      "loss": 0.703,
      "step": 790
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0079896733283227,
      "learning_rate": 9.869438576472307e-06,
      "loss": 0.6507,
      "step": 791
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9078559674663124,
      "learning_rate": 9.86896972863779e-06,
      "loss": 0.6823,
      "step": 792
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7656748314841163,
      "learning_rate": 9.868500051668453e-06,
      "loss": 0.5989,
      "step": 793
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9071940990542837,
      "learning_rate": 9.868029545644275e-06,
      "loss": 0.6498,
      "step": 794
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9636092570882405,
      "learning_rate": 9.867558210645383e-06,
      "loss": 0.77,
      "step": 795
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8433778188273502,
      "learning_rate": 9.867086046752035e-06,
      "loss": 0.6403,
      "step": 796
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9317821873088836,
      "learning_rate": 9.866613054044638e-06,
      "loss": 0.6834,
      "step": 797
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7757055786101275,
      "learning_rate": 9.86613923260374e-06,
      "loss": 0.5763,
      "step": 798
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7242323135700258,
      "learning_rate": 9.865664582510024e-06,
      "loss": 0.5487,
      "step": 799
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7321232632336565,
      "learning_rate": 9.86518910384432e-06,
      "loss": 0.5557,
      "step": 800
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9355060804299221,
      "learning_rate": 9.864712796687598e-06,
      "loss": 0.656,
      "step": 801
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7276862239843759,
      "learning_rate": 9.864235661120966e-06,
      "loss": 0.5042,
      "step": 802
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9988728639970739,
      "learning_rate": 9.863757697225675e-06,
      "loss": 0.6934,
      "step": 803
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0564490011879095,
      "learning_rate": 9.86327890508312e-06,
      "loss": 0.697,
      "step": 804
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7895425899944992,
      "learning_rate": 9.862799284774834e-06,
      "loss": 0.5954,
      "step": 805
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0199665631266697,
      "learning_rate": 9.862318836382487e-06,
      "loss": 0.632,
      "step": 806
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8763614495723234,
      "learning_rate": 9.861837559987898e-06,
      "loss": 0.6926,
      "step": 807
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7360426015650645,
      "learning_rate": 9.861355455673024e-06,
      "loss": 0.586,
      "step": 808
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7586022371468344,
      "learning_rate": 9.860872523519962e-06,
      "loss": 0.5994,
      "step": 809
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0711323291671404,
      "learning_rate": 9.86038876361095e-06,
      "loss": 0.7022,
      "step": 810
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.631180684540505,
      "learning_rate": 9.859904176028364e-06,
      "loss": 0.5343,
      "step": 811
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1013725126350422,
      "learning_rate": 9.859418760854726e-06,
      "loss": 0.687,
      "step": 812
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2576703117967731,
      "learning_rate": 9.858932518172701e-06,
      "loss": 0.7069,
      "step": 813
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.933101646011243,
      "learning_rate": 9.858445448065086e-06,
      "loss": 0.7437,
      "step": 814
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7101302992080752,
      "learning_rate": 9.857957550614828e-06,
      "loss": 0.5372,
      "step": 815
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1048112345458718,
      "learning_rate": 9.857468825905007e-06,
      "loss": 0.6348,
      "step": 816
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7125169819096057,
      "learning_rate": 9.856979274018851e-06,
      "loss": 0.5432,
      "step": 817
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7520784771603117,
      "learning_rate": 9.856488895039725e-06,
      "loss": 0.5822,
      "step": 818
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7589155967095328,
      "learning_rate": 9.855997689051133e-06,
      "loss": 0.5628,
      "step": 819
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7577179625755758,
      "learning_rate": 9.855505656136726e-06,
      "loss": 0.5711,
      "step": 820
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1601702201883908,
      "learning_rate": 9.855012796380288e-06,
      "loss": 0.6733,
      "step": 821
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7271300461503057,
      "learning_rate": 9.854519109865752e-06,
      "loss": 0.5451,
      "step": 822
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8171650892152054,
      "learning_rate": 9.854024596677184e-06,
      "loss": 0.6004,
      "step": 823
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9927864718958673,
      "learning_rate": 9.853529256898797e-06,
      "loss": 0.6765,
      "step": 824
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8764586363158297,
      "learning_rate": 9.85303309061494e-06,
      "loss": 0.6576,
      "step": 825
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8268840784096786,
      "learning_rate": 9.852536097910106e-06,
      "loss": 0.5165,
      "step": 826
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9348262430461438,
      "learning_rate": 9.852038278868929e-06,
      "loss": 0.6729,
      "step": 827
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.806380040487769,
      "learning_rate": 9.851539633576181e-06,
      "loss": 0.5675,
      "step": 828
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8787725180343682,
      "learning_rate": 9.851040162116779e-06,
      "loss": 0.5907,
      "step": 829
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8522407660506429,
      "learning_rate": 9.85053986457577e-06,
      "loss": 0.584,
      "step": 830
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8465694669618473,
      "learning_rate": 9.850038741038359e-06,
      "loss": 0.5721,
      "step": 831
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8621850695821015,
      "learning_rate": 9.849536791589875e-06,
      "loss": 0.6153,
      "step": 832
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7578180186210306,
      "learning_rate": 9.849034016315799e-06,
      "loss": 0.6304,
      "step": 833
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8566597011108543,
      "learning_rate": 9.848530415301748e-06,
      "loss": 0.5615,
      "step": 834
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7449314252925586,
      "learning_rate": 9.848025988633476e-06,
      "loss": 0.5538,
      "step": 835
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7898047349879547,
      "learning_rate": 9.847520736396885e-06,
      "loss": 0.641,
      "step": 836
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.701933988336018,
      "learning_rate": 9.847014658678016e-06,
      "loss": 0.5408,
      "step": 837
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9429442665495136,
      "learning_rate": 9.846507755563044e-06,
      "loss": 0.6466,
      "step": 838
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9355697954917016,
      "learning_rate": 9.846000027138292e-06,
      "loss": 0.634,
      "step": 839
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0728623110408484,
      "learning_rate": 9.845491473490221e-06,
      "loss": 0.7414,
      "step": 840
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8166910649109188,
      "learning_rate": 9.844982094705433e-06,
      "loss": 0.5779,
      "step": 841
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8737318423935748,
      "learning_rate": 9.844471890870668e-06,
      "loss": 0.6269,
      "step": 842
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7338586418238993,
      "learning_rate": 9.84396086207281e-06,
      "loss": 0.5636,
      "step": 843
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7501059775882307,
      "learning_rate": 9.843449008398883e-06,
      "loss": 0.5934,
      "step": 844
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9017164094919561,
      "learning_rate": 9.842936329936045e-06,
      "loss": 0.5763,
      "step": 845
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6615896203996664,
      "learning_rate": 9.842422826771605e-06,
      "loss": 0.52,
      "step": 846
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9864792548753551,
      "learning_rate": 9.841908498993005e-06,
      "loss": 0.6916,
      "step": 847
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8356849710899119,
      "learning_rate": 9.84139334668783e-06,
      "loss": 0.626,
      "step": 848
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8603619971110945,
      "learning_rate": 9.840877369943806e-06,
      "loss": 0.5858,
      "step": 849
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0552369639591088,
      "learning_rate": 9.8403605688488e-06,
      "loss": 0.7282,
      "step": 850
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8398213414730651,
      "learning_rate": 9.839842943490811e-06,
      "loss": 0.5538,
      "step": 851
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9752097830125197,
      "learning_rate": 9.839324493957993e-06,
      "loss": 0.6833,
      "step": 852
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8183433100358727,
      "learning_rate": 9.838805220338628e-06,
      "loss": 0.615,
      "step": 853
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7559456200466549,
      "learning_rate": 9.838285122721145e-06,
      "loss": 0.5609,
      "step": 854
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9765798333547278,
      "learning_rate": 9.83776420119411e-06,
      "loss": 0.6724,
      "step": 855
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1344909109455168,
      "learning_rate": 9.837242455846232e-06,
      "loss": 0.7112,
      "step": 856
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7375559015233665,
      "learning_rate": 9.836719886766357e-06,
      "loss": 0.5447,
      "step": 857
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.966985778646837,
      "learning_rate": 9.836196494043476e-06,
      "loss": 0.6375,
      "step": 858
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6824484636669631,
      "learning_rate": 9.835672277766713e-06,
      "loss": 0.5211,
      "step": 859
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9439544582968047,
      "learning_rate": 9.835147238025337e-06,
      "loss": 0.6153,
      "step": 860
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.819608376113469,
      "learning_rate": 9.834621374908761e-06,
      "loss": 0.5687,
      "step": 861
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6926186871755378,
      "learning_rate": 9.834094688506533e-06,
      "loss": 0.5735,
      "step": 862
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9745182323192518,
      "learning_rate": 9.833567178908337e-06,
      "loss": 0.6365,
      "step": 863
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.989031005255609,
      "learning_rate": 9.833038846204008e-06,
      "loss": 0.6627,
      "step": 864
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7080838838706122,
      "learning_rate": 9.832509690483516e-06,
      "loss": 0.5545,
      "step": 865
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.107339068251972,
      "learning_rate": 9.831979711836966e-06,
      "loss": 0.6803,
      "step": 866
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3062900015879466,
      "learning_rate": 9.83144891035461e-06,
      "loss": 0.7197,
      "step": 867
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8497236986328994,
      "learning_rate": 9.830917286126838e-06,
      "loss": 0.7121,
      "step": 868
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7757920744231476,
      "learning_rate": 9.83038483924418e-06,
      "loss": 0.5507,
      "step": 869
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0615132639191454,
      "learning_rate": 9.829851569797308e-06,
      "loss": 0.6602,
      "step": 870
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.956227306772144,
      "learning_rate": 9.829317477877031e-06,
      "loss": 0.5922,
      "step": 871
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9965856925151164,
      "learning_rate": 9.828782563574299e-06,
      "loss": 0.7016,
      "step": 872
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.950627188504573,
      "learning_rate": 9.8282468269802e-06,
      "loss": 0.6849,
      "step": 873
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9460454498869647,
      "learning_rate": 9.827710268185969e-06,
      "loss": 0.7197,
      "step": 874
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.019841546323306,
      "learning_rate": 9.827172887282974e-06,
      "loss": 0.7185,
      "step": 875
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.889620093786492,
      "learning_rate": 9.826634684362724e-06,
      "loss": 0.6965,
      "step": 876
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7465778324800886,
      "learning_rate": 9.826095659516872e-06,
      "loss": 0.5581,
      "step": 877
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0637192538791145,
      "learning_rate": 9.825555812837207e-06,
      "loss": 0.6928,
      "step": 878
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7646975750362563,
      "learning_rate": 9.825015144415658e-06,
      "loss": 0.5791,
      "step": 879
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9466364333058634,
      "learning_rate": 9.824473654344297e-06,
      "loss": 0.7225,
      "step": 880
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0179728010770526,
      "learning_rate": 9.823931342715335e-06,
      "loss": 0.6604,
      "step": 881
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.894929363812397,
      "learning_rate": 9.82338820962112e-06,
      "loss": 0.5946,
      "step": 882
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9527469721754865,
      "learning_rate": 9.822844255154143e-06,
      "loss": 0.6836,
      "step": 883
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7842014338538598,
      "learning_rate": 9.822299479407031e-06,
      "loss": 0.6258,
      "step": 884
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8438387077422554,
      "learning_rate": 9.82175388247256e-06,
      "loss": 0.5754,
      "step": 885
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.995184189225101,
      "learning_rate": 9.821207464443631e-06,
      "loss": 0.7388,
      "step": 886
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.818907499545714,
      "learning_rate": 9.8206602254133e-06,
      "loss": 0.6103,
      "step": 887
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7209778471552986,
      "learning_rate": 9.820112165474756e-06,
      "loss": 0.6015,
      "step": 888
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7540231073394064,
      "learning_rate": 9.819563284721323e-06,
      "loss": 0.5681,
      "step": 889
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9391713549848318,
      "learning_rate": 9.819013583246473e-06,
      "loss": 0.7238,
      "step": 890
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9053162761539533,
      "learning_rate": 9.818463061143813e-06,
      "loss": 0.6896,
      "step": 891
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0561921375057277,
      "learning_rate": 9.817911718507093e-06,
      "loss": 0.6872,
      "step": 892
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6846470415644579,
      "learning_rate": 9.817359555430201e-06,
      "loss": 0.5836,
      "step": 893
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7982779920786313,
      "learning_rate": 9.816806572007162e-06,
      "loss": 0.57,
      "step": 894
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9620989650618625,
      "learning_rate": 9.816252768332146e-06,
      "loss": 0.6669,
      "step": 895
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.056540597627523,
      "learning_rate": 9.81569814449946e-06,
      "loss": 0.7202,
      "step": 896
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7390798962525975,
      "learning_rate": 9.815142700603546e-06,
      "loss": 0.5573,
      "step": 897
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1473763586635846,
      "learning_rate": 9.814586436738998e-06,
      "loss": 0.7548,
      "step": 898
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7025789120657356,
      "learning_rate": 9.814029353000536e-06,
      "loss": 0.5642,
      "step": 899
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9986804575352153,
      "learning_rate": 9.813471449483027e-06,
      "loss": 0.6569,
      "step": 900
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9814569058152091,
      "learning_rate": 9.812912726281475e-06,
      "loss": 0.6935,
      "step": 901
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1753092195010622,
      "learning_rate": 9.812353183491029e-06,
      "loss": 0.7441,
      "step": 902
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8935409273789665,
      "learning_rate": 9.81179282120697e-06,
      "loss": 0.6555,
      "step": 903
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7452936131510355,
      "learning_rate": 9.81123163952472e-06,
      "loss": 0.5895,
      "step": 904
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9321491826631151,
      "learning_rate": 9.810669638539849e-06,
      "loss": 0.6648,
      "step": 905
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8000600526917718,
      "learning_rate": 9.810106818348054e-06,
      "loss": 0.6396,
      "step": 906
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7797029623000101,
      "learning_rate": 9.809543179045178e-06,
      "loss": 0.5644,
      "step": 907
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8896708866855105,
      "learning_rate": 9.808978720727204e-06,
      "loss": 0.6514,
      "step": 908
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.154650611722061,
      "learning_rate": 9.808413443490254e-06,
      "loss": 0.6867,
      "step": 909
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6915883733884495,
      "learning_rate": 9.80784734743059e-06,
      "loss": 0.5851,
      "step": 910
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9574323984000062,
      "learning_rate": 9.80728043264461e-06,
      "loss": 0.7087,
      "step": 911
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8863031933018034,
      "learning_rate": 9.806712699228852e-06,
      "loss": 0.7166,
      "step": 912
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.996658414550817,
      "learning_rate": 9.80614414728e-06,
      "loss": 0.6715,
      "step": 913
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8143489316746981,
      "learning_rate": 9.805574776894867e-06,
      "loss": 0.6165,
      "step": 914
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0770707197919507,
      "learning_rate": 9.805004588170415e-06,
      "loss": 0.7589,
      "step": 915
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7684996000512531,
      "learning_rate": 9.804433581203741e-06,
      "loss": 0.5971,
      "step": 916
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8456081358683584,
      "learning_rate": 9.803861756092078e-06,
      "loss": 0.559,
      "step": 917
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0837457215584743,
      "learning_rate": 9.803289112932805e-06,
      "loss": 0.706,
      "step": 918
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9236713359625933,
      "learning_rate": 9.802715651823438e-06,
      "loss": 0.6298,
      "step": 919
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7318138955841155,
      "learning_rate": 9.802141372861628e-06,
      "loss": 0.5385,
      "step": 920
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0370023513206552,
      "learning_rate": 9.801566276145172e-06,
      "loss": 0.6184,
      "step": 921
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9136035452552146,
      "learning_rate": 9.800990361772e-06,
      "loss": 0.6461,
      "step": 922
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6927797318964658,
      "learning_rate": 9.800413629840187e-06,
      "loss": 0.555,
      "step": 923
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9253387844988584,
      "learning_rate": 9.799836080447943e-06,
      "loss": 0.6749,
      "step": 924
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7927433710973957,
      "learning_rate": 9.799257713693617e-06,
      "loss": 0.6099,
      "step": 925
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9807196104183729,
      "learning_rate": 9.798678529675702e-06,
      "loss": 0.6982,
      "step": 926
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0085773156376008,
      "learning_rate": 9.798098528492826e-06,
      "loss": 0.7441,
      "step": 927
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7927967848273345,
      "learning_rate": 9.797517710243756e-06,
      "loss": 0.5853,
      "step": 928
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.069726899127923,
      "learning_rate": 9.796936075027401e-06,
      "loss": 0.6674,
      "step": 929
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.783194424160951,
      "learning_rate": 9.796353622942807e-06,
      "loss": 0.5844,
      "step": 930
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6975517719419201,
      "learning_rate": 9.795770354089157e-06,
      "loss": 0.5501,
      "step": 931
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8645177028322643,
      "learning_rate": 9.79518626856578e-06,
      "loss": 0.6335,
      "step": 932
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0269379995644559,
      "learning_rate": 9.794601366472136e-06,
      "loss": 0.6961,
      "step": 933
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7705554334294366,
      "learning_rate": 9.794015647907828e-06,
      "loss": 0.6035,
      "step": 934
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7594167003974919,
      "learning_rate": 9.7934291129726e-06,
      "loss": 0.6144,
      "step": 935
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7408399246519117,
      "learning_rate": 9.792841761766333e-06,
      "loss": 0.5881,
      "step": 936
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7355307957508304,
      "learning_rate": 9.792253594389044e-06,
      "loss": 0.5657,
      "step": 937
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7755297683590929,
      "learning_rate": 9.791664610940893e-06,
      "loss": 0.6467,
      "step": 938
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7664237885590591,
      "learning_rate": 9.79107481152218e-06,
      "loss": 0.533,
      "step": 939
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7168882984652725,
      "learning_rate": 9.790484196233336e-06,
      "loss": 0.584,
      "step": 940
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7241817065915962,
      "learning_rate": 9.789892765174944e-06,
      "loss": 0.6039,
      "step": 941
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8749790333098371,
      "learning_rate": 9.789300518447713e-06,
      "loss": 0.664,
      "step": 942
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7044735322261202,
      "learning_rate": 9.788707456152498e-06,
      "loss": 0.5767,
      "step": 943
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.970296065815631,
      "learning_rate": 9.788113578390293e-06,
      "loss": 0.6558,
      "step": 944
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7398011612547214,
      "learning_rate": 9.787518885262226e-06,
      "loss": 0.5542,
      "step": 945
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8183708747049194,
      "learning_rate": 9.78692337686957e-06,
      "loss": 0.6625,
      "step": 946
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7834175913686896,
      "learning_rate": 9.786327053313733e-06,
      "loss": 0.5959,
      "step": 947
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0816430807169304,
      "learning_rate": 9.785729914696261e-06,
      "loss": 0.6339,
      "step": 948
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0556911246687808,
      "learning_rate": 9.785131961118843e-06,
      "loss": 0.6434,
      "step": 949
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9440275994707563,
      "learning_rate": 9.784533192683303e-06,
      "loss": 0.6687,
      "step": 950
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6913959567580054,
      "learning_rate": 9.783933609491605e-06,
      "loss": 0.5648,
      "step": 951
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9497479875551993,
      "learning_rate": 9.783333211645853e-06,
      "loss": 0.6779,
      "step": 952
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4999423740102389,
      "learning_rate": 9.782731999248287e-06,
      "loss": 0.7535,
      "step": 953
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7762702970572679,
      "learning_rate": 9.782129972401286e-06,
      "loss": 0.5445,
      "step": 954
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7175732985264717,
      "learning_rate": 9.781527131207372e-06,
      "loss": 0.5621,
      "step": 955
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8913573472220098,
      "learning_rate": 9.780923475769202e-06,
      "loss": 0.6791,
      "step": 956
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9284652814579771,
      "learning_rate": 9.780319006189568e-06,
      "loss": 0.6466,
      "step": 957
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9185713718739321,
      "learning_rate": 9.779713722571413e-06,
      "loss": 0.6671,
      "step": 958
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8139619101491286,
      "learning_rate": 9.779107625017802e-06,
      "loss": 0.5951,
      "step": 959
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8664524718030011,
      "learning_rate": 9.778500713631953e-06,
      "loss": 0.6892,
      "step": 960
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8937139509912793,
      "learning_rate": 9.777892988517214e-06,
      "loss": 0.6415,
      "step": 961
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.991570211257772,
      "learning_rate": 9.777284449777073e-06,
      "loss": 0.7085,
      "step": 962
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8168177573579105,
      "learning_rate": 9.776675097515162e-06,
      "loss": 0.6058,
      "step": 963
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6871005300940596,
      "learning_rate": 9.776064931835246e-06,
      "loss": 0.5371,
      "step": 964
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8165778920391509,
      "learning_rate": 9.775453952841225e-06,
      "loss": 0.5662,
      "step": 965
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8464232421113248,
      "learning_rate": 9.774842160637149e-06,
      "loss": 0.6342,
      "step": 966
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8658763667272781,
      "learning_rate": 9.774229555327195e-06,
      "loss": 0.6706,
      "step": 967
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7637882509715292,
      "learning_rate": 9.773616137015687e-06,
      "loss": 0.5882,
      "step": 968
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.734113029645848,
      "learning_rate": 9.77300190580708e-06,
      "loss": 0.5914,
      "step": 969
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4586489261626505,
      "learning_rate": 9.772386861805976e-06,
      "loss": 0.6932,
      "step": 970
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6788801684356507,
      "learning_rate": 9.771771005117105e-06,
      "loss": 0.638,
      "step": 971
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9238597025655824,
      "learning_rate": 9.771154335845345e-06,
      "loss": 0.6903,
      "step": 972
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9639763461980686,
      "learning_rate": 9.770536854095709e-06,
      "loss": 0.6939,
      "step": 973
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.731914543768533,
      "learning_rate": 9.769918559973345e-06,
      "loss": 0.521,
      "step": 974
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6916666229147426,
      "learning_rate": 9.76929945358354e-06,
      "loss": 0.5847,
      "step": 975
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9747492556513749,
      "learning_rate": 9.768679535031727e-06,
      "loss": 0.7139,
      "step": 976
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9273153089722869,
      "learning_rate": 9.768058804423469e-06,
      "loss": 0.6703,
      "step": 977
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.072224791794499,
      "learning_rate": 9.767437261864468e-06,
      "loss": 0.6733,
      "step": 978
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7342018129204232,
      "learning_rate": 9.766814907460568e-06,
      "loss": 0.5607,
      "step": 979
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7804168063653115,
      "learning_rate": 9.766191741317752e-06,
      "loss": 0.6068,
      "step": 980
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9542109348711877,
      "learning_rate": 9.765567763542134e-06,
      "loss": 0.6467,
      "step": 981
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9321754241138277,
      "learning_rate": 9.764942974239974e-06,
      "loss": 0.6776,
      "step": 982
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7488462299163969,
      "learning_rate": 9.764317373517666e-06,
      "loss": 0.5895,
      "step": 983
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6984439554320233,
      "learning_rate": 9.763690961481744e-06,
      "loss": 0.553,
      "step": 984
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.770024500901702,
      "learning_rate": 9.76306373823888e-06,
      "loss": 0.5555,
      "step": 985
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9338611833990433,
      "learning_rate": 9.76243570389588e-06,
      "loss": 0.6835,
      "step": 986
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7757083981569457,
      "learning_rate": 9.761806858559699e-06,
      "loss": 0.5928,
      "step": 987
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9439698032040154,
      "learning_rate": 9.761177202337414e-06,
      "loss": 0.7229,
      "step": 988
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8339692414501271,
      "learning_rate": 9.760546735336254e-06,
      "loss": 0.6523,
      "step": 989
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9450676285213673,
      "learning_rate": 9.759915457663581e-06,
      "loss": 0.6762,
      "step": 990
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8757514882704517,
      "learning_rate": 9.759283369426893e-06,
      "loss": 0.6528,
      "step": 991
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9245344168940118,
      "learning_rate": 9.758650470733829e-06,
      "loss": 0.6634,
      "step": 992
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9880739165135637,
      "learning_rate": 9.758016761692166e-06,
      "loss": 0.7216,
      "step": 993
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7660910515782134,
      "learning_rate": 9.757382242409816e-06,
      "loss": 0.583,
      "step": 994
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0359208438444523,
      "learning_rate": 9.756746912994832e-06,
      "loss": 0.7472,
      "step": 995
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1092944594479166,
      "learning_rate": 9.756110773555404e-06,
      "loss": 0.718,
      "step": 996
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.94668456975757,
      "learning_rate": 9.75547382419986e-06,
      "loss": 0.6427,
      "step": 997
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8017944595849557,
      "learning_rate": 9.754836065036669e-06,
      "loss": 0.5633,
      "step": 998
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9561908986980844,
      "learning_rate": 9.754197496174429e-06,
      "loss": 0.6899,
      "step": 999
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7211650157357844,
      "learning_rate": 9.753558117721885e-06,
      "loss": 0.5579,
      "step": 1000
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7897919614351709,
      "learning_rate": 9.752917929787915e-06,
      "loss": 0.5281,
      "step": 1001
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7201909321133236,
      "learning_rate": 9.752276932481539e-06,
      "loss": 0.5745,
      "step": 1002
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9483731542265567,
      "learning_rate": 9.751635125911907e-06,
      "loss": 0.6339,
      "step": 1003
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9626046220155712,
      "learning_rate": 9.750992510188317e-06,
      "loss": 0.6568,
      "step": 1004
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7637031277927749,
      "learning_rate": 9.7503490854202e-06,
      "loss": 0.5791,
      "step": 1005
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.976385085270203,
      "learning_rate": 9.74970485171712e-06,
      "loss": 0.6398,
      "step": 1006
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8391172780535088,
      "learning_rate": 9.749059809188785e-06,
      "loss": 0.5802,
      "step": 1007
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9220323658444436,
      "learning_rate": 9.748413957945043e-06,
      "loss": 0.6955,
      "step": 1008
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9578984673934916,
      "learning_rate": 9.74776729809587e-06,
      "loss": 0.6702,
      "step": 1009
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7191793711644885,
      "learning_rate": 9.747119829751389e-06,
      "loss": 0.5447,
      "step": 1010
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7480485905936854,
      "learning_rate": 9.746471553021856e-06,
      "loss": 0.573,
      "step": 1011
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8264376770169287,
      "learning_rate": 9.745822468017665e-06,
      "loss": 0.6534,
      "step": 1012
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9759465547374736,
      "learning_rate": 9.745172574849351e-06,
      "loss": 0.6313,
      "step": 1013
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1134486251567228,
      "learning_rate": 9.744521873627582e-06,
      "loss": 0.6856,
      "step": 1014
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1389548492991892,
      "learning_rate": 9.743870364463166e-06,
      "loss": 0.6529,
      "step": 1015
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.882265415250145,
      "learning_rate": 9.743218047467046e-06,
      "loss": 0.6679,
      "step": 1016
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8827573084123569,
      "learning_rate": 9.742564922750309e-06,
      "loss": 0.6371,
      "step": 1017
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6788938172861039,
      "learning_rate": 9.741910990424173e-06,
      "loss": 0.5494,
      "step": 1018
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0172654869506512,
      "learning_rate": 9.741256250599997e-06,
      "loss": 0.6905,
      "step": 1019
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9582774285962188,
      "learning_rate": 9.740600703389274e-06,
      "loss": 0.7014,
      "step": 1020
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1862352499822466,
      "learning_rate": 9.73994434890364e-06,
      "loss": 0.7549,
      "step": 1021
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6761142935246048,
      "learning_rate": 9.739287187254864e-06,
      "loss": 0.5652,
      "step": 1022
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9980643274991466,
      "learning_rate": 9.73862921855485e-06,
      "loss": 0.6802,
      "step": 1023
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7314237102873798,
      "learning_rate": 9.737970442915653e-06,
      "loss": 0.6126,
      "step": 1024
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7040482478083115,
      "learning_rate": 9.737310860449444e-06,
      "loss": 0.5493,
      "step": 1025
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8643812791877247,
      "learning_rate": 9.73665047126855e-06,
      "loss": 0.6013,
      "step": 1026
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9441418949555234,
      "learning_rate": 9.735989275485428e-06,
      "loss": 0.6912,
      "step": 1027
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0338277339784687,
      "learning_rate": 9.73532727321267e-06,
      "loss": 0.6561,
      "step": 1028
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9062938706512511,
      "learning_rate": 9.734664464563011e-06,
      "loss": 0.6404,
      "step": 1029
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7982633464234161,
      "learning_rate": 9.734000849649317e-06,
      "loss": 0.5534,
      "step": 1030
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8492831011762463,
      "learning_rate": 9.7333364285846e-06,
      "loss": 0.6233,
      "step": 1031
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1000241608977386,
      "learning_rate": 9.732671201481997e-06,
      "loss": 0.6886,
      "step": 1032
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7073212107498936,
      "learning_rate": 9.732005168454795e-06,
      "loss": 0.5611,
      "step": 1033
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0494888540451057,
      "learning_rate": 9.73133832961641e-06,
      "loss": 0.6855,
      "step": 1034
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.781161144930962,
      "learning_rate": 9.7306706850804e-06,
      "loss": 0.5918,
      "step": 1035
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9902122493389596,
      "learning_rate": 9.730002234960453e-06,
      "loss": 0.6479,
      "step": 1036
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.737029365030554,
      "learning_rate": 9.729332979370405e-06,
      "loss": 0.6123,
      "step": 1037
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8339284811106819,
      "learning_rate": 9.728662918424218e-06,
      "loss": 0.672,
      "step": 1038
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8840243708359882,
      "learning_rate": 9.727992052235999e-06,
      "loss": 0.6238,
      "step": 1039
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7438545594674684,
      "learning_rate": 9.727320380919992e-06,
      "loss": 0.5543,
      "step": 1040
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7579648842249721,
      "learning_rate": 9.726647904590572e-06,
      "loss": 0.5899,
      "step": 1041
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8268839444898627,
      "learning_rate": 9.725974623362254e-06,
      "loss": 0.6403,
      "step": 1042
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8628467212515322,
      "learning_rate": 9.725300537349694e-06,
      "loss": 0.6215,
      "step": 1043
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9261329941212265,
      "learning_rate": 9.72462564666768e-06,
      "loss": 0.7053,
      "step": 1044
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7392891572898306,
      "learning_rate": 9.723949951431138e-06,
      "loss": 0.6042,
      "step": 1045
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9499640161631635,
      "learning_rate": 9.723273451755136e-06,
      "loss": 0.7164,
      "step": 1046
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0904090502528248,
      "learning_rate": 9.722596147754869e-06,
      "loss": 0.6382,
      "step": 1047
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7117735979327462,
      "learning_rate": 9.72191803954568e-06,
      "loss": 0.5743,
      "step": 1048
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0810130601145984,
      "learning_rate": 9.721239127243039e-06,
      "loss": 0.7218,
      "step": 1049
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0109747313607806,
      "learning_rate": 9.720559410962562e-06,
      "loss": 0.5746,
      "step": 1050
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7235026809880478,
      "learning_rate": 9.719878890819993e-06,
      "loss": 0.5635,
      "step": 1051
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6637332522030321,
      "learning_rate": 9.719197566931224e-06,
      "loss": 0.5365,
      "step": 1052
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7273680135982696,
      "learning_rate": 9.718515439412271e-06,
      "loss": 0.5412,
      "step": 1053
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9994433539312929,
      "learning_rate": 9.717832508379296e-06,
      "loss": 0.7267,
      "step": 1054
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7587576190483052,
      "learning_rate": 9.717148773948597e-06,
      "loss": 0.6008,
      "step": 1055
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6657828765042793,
      "learning_rate": 9.716464236236602e-06,
      "loss": 0.561,
      "step": 1056
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9822613120832019,
      "learning_rate": 9.715778895359886e-06,
      "loss": 0.721,
      "step": 1057
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.913418529960109,
      "learning_rate": 9.715092751435151e-06,
      "loss": 0.6781,
      "step": 1058
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.197345944173032,
      "learning_rate": 9.714405804579241e-06,
      "loss": 0.6978,
      "step": 1059
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8180931306214322,
      "learning_rate": 9.713718054909139e-06,
      "loss": 0.5931,
      "step": 1060
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8300969100386351,
      "learning_rate": 9.713029502541957e-06,
      "loss": 0.5673,
      "step": 1061
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8834312789834103,
      "learning_rate": 9.712340147594954e-06,
      "loss": 0.6409,
      "step": 1062
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8853772685576812,
      "learning_rate": 9.711649990185514e-06,
      "loss": 0.667,
      "step": 1063
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7104042442121519,
      "learning_rate": 9.710959030431167e-06,
      "loss": 0.533,
      "step": 1064
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7145865737843454,
      "learning_rate": 9.710267268449577e-06,
      "loss": 0.5199,
      "step": 1065
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.776888570192149,
      "learning_rate": 9.70957470435854e-06,
      "loss": 0.5413,
      "step": 1066
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7498830641191208,
      "learning_rate": 9.708881338275995e-06,
      "loss": 0.617,
      "step": 1067
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8598604092731292,
      "learning_rate": 9.708187170320015e-06,
      "loss": 0.6306,
      "step": 1068
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.956591081146534,
      "learning_rate": 9.707492200608811e-06,
      "loss": 0.68,
      "step": 1069
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7254653266397849,
      "learning_rate": 9.706796429260728e-06,
      "loss": 0.599,
      "step": 1070
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7433384368002686,
      "learning_rate": 9.706099856394246e-06,
      "loss": 0.5751,
      "step": 1071
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0059953111056341,
      "learning_rate": 9.705402482127986e-06,
      "loss": 0.5667,
      "step": 1072
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7208744575145551,
      "learning_rate": 9.704704306580705e-06,
      "loss": 0.4906,
      "step": 1073
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8653827368334114,
      "learning_rate": 9.704005329871295e-06,
      "loss": 0.6629,
      "step": 1074
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7497916614217609,
      "learning_rate": 9.703305552118782e-06,
      "loss": 0.5831,
      "step": 1075
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.88776582778491,
      "learning_rate": 9.702604973442332e-06,
      "loss": 0.6538,
      "step": 1076
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9699490968295508,
      "learning_rate": 9.701903593961245e-06,
      "loss": 0.6528,
      "step": 1077
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7545594653859838,
      "learning_rate": 9.701201413794964e-06,
      "loss": 0.5888,
      "step": 1078
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7820740378328453,
      "learning_rate": 9.700498433063056e-06,
      "loss": 0.5955,
      "step": 1079
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8793005282937691,
      "learning_rate": 9.699794651885235e-06,
      "loss": 0.6599,
      "step": 1080
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.924265123805297,
      "learning_rate": 9.699090070381348e-06,
      "loss": 0.5911,
      "step": 1081
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0259195837887995,
      "learning_rate": 9.698384688671374e-06,
      "loss": 0.685,
      "step": 1082
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8459164126899845,
      "learning_rate": 9.697678506875436e-06,
      "loss": 0.585,
      "step": 1083
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7689786376811757,
      "learning_rate": 9.696971525113789e-06,
      "loss": 0.5344,
      "step": 1084
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4437406001675164,
      "learning_rate": 9.696263743506825e-06,
      "loss": 0.6077,
      "step": 1085
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8605546054943277,
      "learning_rate": 9.695555162175069e-06,
      "loss": 0.5968,
      "step": 1086
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7742404794889682,
      "learning_rate": 9.694845781239188e-06,
      "loss": 0.6106,
      "step": 1087
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8577012986494054,
      "learning_rate": 9.69413560081998e-06,
      "loss": 0.5776,
      "step": 1088
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0295973009575785,
      "learning_rate": 9.693424621038384e-06,
      "loss": 0.7274,
      "step": 1089
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6782095924667602,
      "learning_rate": 9.692712842015469e-06,
      "loss": 0.517,
      "step": 1090
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7516306029792222,
      "learning_rate": 9.692000263872445e-06,
      "loss": 0.5644,
      "step": 1091
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9706041776408542,
      "learning_rate": 9.691286886730657e-06,
      "loss": 0.7394,
      "step": 1092
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.06954227014014,
      "learning_rate": 9.690572710711588e-06,
      "loss": 0.7047,
      "step": 1093
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9821778866755354,
      "learning_rate": 9.689857735936849e-06,
      "loss": 0.628,
      "step": 1094
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6886707091719464,
      "learning_rate": 9.6891419625282e-06,
      "loss": 0.5254,
      "step": 1095
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7397853229300382,
      "learning_rate": 9.688425390607523e-06,
      "loss": 0.5499,
      "step": 1096
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8823471648877111,
      "learning_rate": 9.687708020296846e-06,
      "loss": 0.5849,
      "step": 1097
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9308471678744554,
      "learning_rate": 9.68698985171833e-06,
      "loss": 0.6343,
      "step": 1098
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0676986761338185,
      "learning_rate": 9.686270884994272e-06,
      "loss": 0.6127,
      "step": 1099
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.969104487868543,
      "learning_rate": 9.685551120247103e-06,
      "loss": 0.6349,
      "step": 1100
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1454085753369472,
      "learning_rate": 9.68483055759939e-06,
      "loss": 0.6521,
      "step": 1101
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.953063094752734,
      "learning_rate": 9.684109197173841e-06,
      "loss": 0.6181,
      "step": 1102
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8904523114796692,
      "learning_rate": 9.683387039093295e-06,
      "loss": 0.6284,
      "step": 1103
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8860311215614417,
      "learning_rate": 9.682664083480726e-06,
      "loss": 0.6905,
      "step": 1104
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7844583441293042,
      "learning_rate": 9.681940330459248e-06,
      "loss": 0.5937,
      "step": 1105
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8099289491219696,
      "learning_rate": 9.681215780152107e-06,
      "loss": 0.6305,
      "step": 1106
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6478141978785894,
      "learning_rate": 9.680490432682687e-06,
      "loss": 0.5326,
      "step": 1107
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9145285861436038,
      "learning_rate": 9.679764288174508e-06,
      "loss": 0.6394,
      "step": 1108
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6784913547196753,
      "learning_rate": 9.679037346751226e-06,
      "loss": 0.5186,
      "step": 1109
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8815448592817704,
      "learning_rate": 9.678309608536626e-06,
      "loss": 0.5811,
      "step": 1110
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7106822261694445,
      "learning_rate": 9.677581073654641e-06,
      "loss": 0.5921,
      "step": 1111
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9254064954950437,
      "learning_rate": 9.676851742229328e-06,
      "loss": 0.656,
      "step": 1112
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7159298742841554,
      "learning_rate": 9.676121614384889e-06,
      "loss": 0.562,
      "step": 1113
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9774645473746579,
      "learning_rate": 9.675390690245651e-06,
      "loss": 0.6846,
      "step": 1114
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7475468147852461,
      "learning_rate": 9.674658969936089e-06,
      "loss": 0.6124,
      "step": 1115
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0075918023127493,
      "learning_rate": 9.673926453580802e-06,
      "loss": 0.6877,
      "step": 1116
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7062696716317111,
      "learning_rate": 9.673193141304535e-06,
      "loss": 0.5099,
      "step": 1117
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0169237664621988,
      "learning_rate": 9.67245903323216e-06,
      "loss": 0.6672,
      "step": 1118
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6701822605912964,
      "learning_rate": 9.67172412948869e-06,
      "loss": 0.576,
      "step": 1119
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8420563298436072,
      "learning_rate": 9.670988430199269e-06,
      "loss": 0.6324,
      "step": 1120
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7481025134684111,
      "learning_rate": 9.67025193548918e-06,
      "loss": 0.5686,
      "step": 1121
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.673254264784351,
      "learning_rate": 9.669514645483842e-06,
      "loss": 0.5544,
      "step": 1122
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0703485066815706,
      "learning_rate": 9.668776560308806e-06,
      "loss": 0.6644,
      "step": 1123
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6549672369818021,
      "learning_rate": 9.66803768008976e-06,
      "loss": 0.5379,
      "step": 1124
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6630891429727088,
      "learning_rate": 9.66729800495253e-06,
      "loss": 0.5412,
      "step": 1125
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9211641810654386,
      "learning_rate": 9.666557535023073e-06,
      "loss": 0.6177,
      "step": 1126
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2112562188855034,
      "learning_rate": 9.665816270427483e-06,
      "loss": 0.6545,
      "step": 1127
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.996255440801724,
      "learning_rate": 9.66507421129199e-06,
      "loss": 0.667,
      "step": 1128
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7445179415592746,
      "learning_rate": 9.66433135774296e-06,
      "loss": 0.5892,
      "step": 1129
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8526749575390422,
      "learning_rate": 9.663587709906891e-06,
      "loss": 0.5533,
      "step": 1130
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7790559722425667,
      "learning_rate": 9.662843267910422e-06,
      "loss": 0.6401,
      "step": 1131
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0261150291855647,
      "learning_rate": 9.662098031880321e-06,
      "loss": 0.703,
      "step": 1132
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7651288290761387,
      "learning_rate": 9.661352001943494e-06,
      "loss": 0.5477,
      "step": 1133
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9821048154097823,
      "learning_rate": 9.660605178226983e-06,
      "loss": 0.7015,
      "step": 1134
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6986033135320731,
      "learning_rate": 9.659857560857963e-06,
      "loss": 0.5451,
      "step": 1135
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8463580272517893,
      "learning_rate": 9.659109149963748e-06,
      "loss": 0.5994,
      "step": 1136
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8270293495072578,
      "learning_rate": 9.658359945671782e-06,
      "loss": 0.6094,
      "step": 1137
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6989255274229249,
      "learning_rate": 9.657609948109648e-06,
      "loss": 0.5407,
      "step": 1138
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9548477427367056,
      "learning_rate": 9.656859157405063e-06,
      "loss": 0.6926,
      "step": 1139
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.054333808008999,
      "learning_rate": 9.656107573685877e-06,
      "loss": 0.6763,
      "step": 1140
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9114501941058868,
      "learning_rate": 9.655355197080078e-06,
      "loss": 0.6377,
      "step": 1141
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9065384295139649,
      "learning_rate": 9.65460202771579e-06,
      "loss": 0.6585,
      "step": 1142
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8744252898688392,
      "learning_rate": 9.653848065721268e-06,
      "loss": 0.5465,
      "step": 1143
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8317512666969036,
      "learning_rate": 9.6530933112249e-06,
      "loss": 0.5598,
      "step": 1144
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7442850105164438,
      "learning_rate": 9.65233776435522e-06,
      "loss": 0.5808,
      "step": 1145
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8213509926674526,
      "learning_rate": 9.651581425240888e-06,
      "loss": 0.6625,
      "step": 1146
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8414158128669929,
      "learning_rate": 9.650824294010697e-06,
      "loss": 0.618,
      "step": 1147
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.999969920526632,
      "learning_rate": 9.65006637079358e-06,
      "loss": 0.5781,
      "step": 1148
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.842766238559155,
      "learning_rate": 9.649307655718605e-06,
      "loss": 0.5731,
      "step": 1149
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9281001688366916,
      "learning_rate": 9.64854814891497e-06,
      "loss": 0.619,
      "step": 1150
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6950859861173151,
      "learning_rate": 9.647787850512017e-06,
      "loss": 0.5583,
      "step": 1151
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.035776547415945,
      "learning_rate": 9.647026760639212e-06,
      "loss": 0.7074,
      "step": 1152
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6491145986864243,
      "learning_rate": 9.646264879426161e-06,
      "loss": 0.5103,
      "step": 1153
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.125381227607009,
      "learning_rate": 9.645502207002607e-06,
      "loss": 0.5963,
      "step": 1154
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9732219064604967,
      "learning_rate": 9.644738743498423e-06,
      "loss": 0.7124,
      "step": 1155
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8424016143749036,
      "learning_rate": 9.64397448904362e-06,
      "loss": 0.5485,
      "step": 1156
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8702349203298858,
      "learning_rate": 9.643209443768343e-06,
      "loss": 0.6251,
      "step": 1157
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0510521108240087,
      "learning_rate": 9.642443607802867e-06,
      "loss": 0.6843,
      "step": 1158
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8156081835652739,
      "learning_rate": 9.641676981277612e-06,
      "loss": 0.5847,
      "step": 1159
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8363072598497219,
      "learning_rate": 9.640909564323123e-06,
      "loss": 0.5835,
      "step": 1160
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7463112659389821,
      "learning_rate": 9.640141357070085e-06,
      "loss": 0.5928,
      "step": 1161
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8886616963084485,
      "learning_rate": 9.639372359649313e-06,
      "loss": 0.6857,
      "step": 1162
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8355239815009524,
      "learning_rate": 9.63860257219176e-06,
      "loss": 0.6026,
      "step": 1163
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6925469113905485,
      "learning_rate": 9.637831994828515e-06,
      "loss": 0.5464,
      "step": 1164
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7175132947005278,
      "learning_rate": 9.637060627690798e-06,
      "loss": 0.593,
      "step": 1165
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8518135673261236,
      "learning_rate": 9.636288470909964e-06,
      "loss": 0.6057,
      "step": 1166
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7355591355569097,
      "learning_rate": 9.635515524617505e-06,
      "loss": 0.5536,
      "step": 1167
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1557573645824277,
      "learning_rate": 9.634741788945044e-06,
      "loss": 0.6778,
      "step": 1168
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9569740068844284,
      "learning_rate": 9.63396726402434e-06,
      "loss": 0.6761,
      "step": 1169
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7362876019275366,
      "learning_rate": 9.633191949987288e-06,
      "loss": 0.5451,
      "step": 1170
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7998783587460482,
      "learning_rate": 9.632415846965915e-06,
      "loss": 0.5286,
      "step": 1171
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0378395289191231,
      "learning_rate": 9.631638955092383e-06,
      "loss": 0.7157,
      "step": 1172
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8621606932984904,
      "learning_rate": 9.630861274498988e-06,
      "loss": 0.64,
      "step": 1173
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.934780351828164,
      "learning_rate": 9.630082805318164e-06,
      "loss": 0.6658,
      "step": 1174
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7837739866398066,
      "learning_rate": 9.629303547682474e-06,
      "loss": 0.5764,
      "step": 1175
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7628641187468533,
      "learning_rate": 9.628523501724615e-06,
      "loss": 0.5772,
      "step": 1176
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.884087789020235,
      "learning_rate": 9.627742667577426e-06,
      "loss": 0.6771,
      "step": 1177
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9218733121321974,
      "learning_rate": 9.626961045373872e-06,
      "loss": 0.6625,
      "step": 1178
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7472406617081606,
      "learning_rate": 9.626178635247054e-06,
      "loss": 0.5881,
      "step": 1179
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8232236242235816,
      "learning_rate": 9.625395437330211e-06,
      "loss": 0.5835,
      "step": 1180
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6886164365483731,
      "learning_rate": 9.62461145175671e-06,
      "loss": 0.5571,
      "step": 1181
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7064570403578206,
      "learning_rate": 9.62382667866006e-06,
      "loss": 0.5474,
      "step": 1182
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.869138090757438,
      "learning_rate": 9.623041118173896e-06,
      "loss": 0.6563,
      "step": 1183
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6795762825107204,
      "learning_rate": 9.622254770431994e-06,
      "loss": 0.5333,
      "step": 1184
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7212851449776427,
      "learning_rate": 9.621467635568255e-06,
      "loss": 0.6618,
      "step": 1185
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7538214840226601,
      "learning_rate": 9.620679713716728e-06,
      "loss": 0.5698,
      "step": 1186
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6892806362148062,
      "learning_rate": 9.619891005011582e-06,
      "loss": 0.6479,
      "step": 1187
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9848723775223823,
      "learning_rate": 9.619101509587125e-06,
      "loss": 0.7322,
      "step": 1188
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9732895235932517,
      "learning_rate": 9.618311227577806e-06,
      "loss": 0.6589,
      "step": 1189
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0724050384550887,
      "learning_rate": 9.617520159118197e-06,
      "loss": 0.6994,
      "step": 1190
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0181995790334541,
      "learning_rate": 9.616728304343008e-06,
      "loss": 0.7214,
      "step": 1191
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7228228092266846,
      "learning_rate": 9.615935663387088e-06,
      "loss": 0.5514,
      "step": 1192
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7450269701663594,
      "learning_rate": 9.615142236385412e-06,
      "loss": 0.5309,
      "step": 1193
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3238983007013911,
      "learning_rate": 9.614348023473094e-06,
      "loss": 0.6923,
      "step": 1194
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9288096349008675,
      "learning_rate": 9.613553024785379e-06,
      "loss": 0.6487,
      "step": 1195
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.208575480176333,
      "learning_rate": 9.612757240457646e-06,
      "loss": 0.6699,
      "step": 1196
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8757025919147509,
      "learning_rate": 9.611960670625413e-06,
      "loss": 0.6494,
      "step": 1197
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8056622177242443,
      "learning_rate": 9.611163315424323e-06,
      "loss": 0.5831,
      "step": 1198
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8641291452915447,
      "learning_rate": 9.610365174990159e-06,
      "loss": 0.6328,
      "step": 1199
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7917733455120687,
      "learning_rate": 9.609566249458837e-06,
      "loss": 0.5737,
      "step": 1200
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9591997879783773,
      "learning_rate": 9.608766538966406e-06,
      "loss": 0.7323,
      "step": 1201
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.750861223364628,
      "learning_rate": 9.607966043649047e-06,
      "loss": 0.5482,
      "step": 1202
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7101822686666331,
      "learning_rate": 9.607164763643074e-06,
      "loss": 0.5283,
      "step": 1203
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6666417525300337,
      "learning_rate": 9.606362699084939e-06,
      "loss": 0.5455,
      "step": 1204
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7032687074743561,
      "learning_rate": 9.605559850111227e-06,
      "loss": 0.551,
      "step": 1205
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7607037512055422,
      "learning_rate": 9.604756216858652e-06,
      "loss": 0.6034,
      "step": 1206
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7806540494720198,
      "learning_rate": 9.603951799464066e-06,
      "loss": 0.6233,
      "step": 1207
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0171141749292825,
      "learning_rate": 9.60314659806445e-06,
      "loss": 0.7097,
      "step": 1208
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6747970315407621,
      "learning_rate": 9.602340612796925e-06,
      "loss": 0.5566,
      "step": 1209
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.758521470820783,
      "learning_rate": 9.60153384379874e-06,
      "loss": 0.6382,
      "step": 1210
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7822713992315741,
      "learning_rate": 9.60072629120728e-06,
      "loss": 0.5669,
      "step": 1211
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7808871287274356,
      "learning_rate": 9.599917955160065e-06,
      "loss": 0.6584,
      "step": 1212
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8656125058638835,
      "learning_rate": 9.599108835794742e-06,
      "loss": 0.6772,
      "step": 1213
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9191812405082243,
      "learning_rate": 9.598298933249096e-06,
      "loss": 0.6139,
      "step": 1214
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8571340533160847,
      "learning_rate": 9.597488247661048e-06,
      "loss": 0.6567,
      "step": 1215
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8026308537690293,
      "learning_rate": 9.596676779168647e-06,
      "loss": 0.6198,
      "step": 1216
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8233099851703983,
      "learning_rate": 9.59586452791008e-06,
      "loss": 0.6195,
      "step": 1217
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9362317565985335,
      "learning_rate": 9.59505149402366e-06,
      "loss": 0.7417,
      "step": 1218
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7583445333257788,
      "learning_rate": 9.594237677647844e-06,
      "loss": 0.6168,
      "step": 1219
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1582105541248802,
      "learning_rate": 9.593423078921214e-06,
      "loss": 0.6568,
      "step": 1220
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7023939603418193,
      "learning_rate": 9.592607697982485e-06,
      "loss": 0.5315,
      "step": 1221
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6786928111652569,
      "learning_rate": 9.591791534970514e-06,
      "loss": 0.5904,
      "step": 1222
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9867821033906398,
      "learning_rate": 9.59097459002428e-06,
      "loss": 0.6369,
      "step": 1223
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8624348907891688,
      "learning_rate": 9.590156863282901e-06,
      "loss": 0.6836,
      "step": 1224
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.641528820287354,
      "learning_rate": 9.58933835488563e-06,
      "loss": 0.552,
      "step": 1225
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7724136707887936,
      "learning_rate": 9.588519064971847e-06,
      "loss": 0.6094,
      "step": 1226
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7523022956947315,
      "learning_rate": 9.58769899368107e-06,
      "loss": 0.5516,
      "step": 1227
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7636949449716391,
      "learning_rate": 9.58687814115295e-06,
      "loss": 0.5556,
      "step": 1228
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.747497773473688,
      "learning_rate": 9.586056507527266e-06,
      "loss": 0.5844,
      "step": 1229
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8793282485212105,
      "learning_rate": 9.585234092943938e-06,
      "loss": 0.6807,
      "step": 1230
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7666083980324859,
      "learning_rate": 9.584410897543011e-06,
      "loss": 0.5529,
      "step": 1231
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8713840967095233,
      "learning_rate": 9.583586921464669e-06,
      "loss": 0.6026,
      "step": 1232
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9035009561916743,
      "learning_rate": 9.582762164849227e-06,
      "loss": 0.6771,
      "step": 1233
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7725721359950758,
      "learning_rate": 9.58193662783713e-06,
      "loss": 0.5566,
      "step": 1234
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7362983873862653,
      "learning_rate": 9.58111031056896e-06,
      "loss": 0.5788,
      "step": 1235
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8966396035795065,
      "learning_rate": 9.580283213185429e-06,
      "loss": 0.6745,
      "step": 1236
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9777643020445494,
      "learning_rate": 9.579455335827383e-06,
      "loss": 0.6537,
      "step": 1237
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.770767692946826,
      "learning_rate": 9.578626678635802e-06,
      "loss": 0.564,
      "step": 1238
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8181996530156633,
      "learning_rate": 9.577797241751799e-06,
      "loss": 0.5622,
      "step": 1239
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8244602581987415,
      "learning_rate": 9.576967025316617e-06,
      "loss": 0.6193,
      "step": 1240
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8935014812877475,
      "learning_rate": 9.576136029471632e-06,
      "loss": 0.6623,
      "step": 1241
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8503335464566736,
      "learning_rate": 9.575304254358355e-06,
      "loss": 0.5882,
      "step": 1242
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.783896665510255,
      "learning_rate": 9.57447170011843e-06,
      "loss": 0.5873,
      "step": 1243
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0204014653181868,
      "learning_rate": 9.573638366893629e-06,
      "loss": 0.7158,
      "step": 1244
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8902354987488832,
      "learning_rate": 9.572804254825863e-06,
      "loss": 0.6518,
      "step": 1245
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2378426385612038,
      "learning_rate": 9.571969364057172e-06,
      "loss": 0.7121,
      "step": 1246
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.702352719161442,
      "learning_rate": 9.57113369472973e-06,
      "loss": 0.5464,
      "step": 1247
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7896264029247491,
      "learning_rate": 9.570297246985838e-06,
      "loss": 0.5968,
      "step": 1248
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8472804287380558,
      "learning_rate": 9.569460020967939e-06,
      "loss": 0.6041,
      "step": 1249
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.98904717772101,
      "learning_rate": 9.568622016818604e-06,
      "loss": 0.6474,
      "step": 1250
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1273179779897353,
      "learning_rate": 9.567783234680534e-06,
      "loss": 0.5979,
      "step": 1251
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9400886228454184,
      "learning_rate": 9.566943674696567e-06,
      "loss": 0.6036,
      "step": 1252
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6921289897057276,
      "learning_rate": 9.56610333700967e-06,
      "loss": 0.5782,
      "step": 1253
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.932953292240861,
      "learning_rate": 9.565262221762942e-06,
      "loss": 0.6507,
      "step": 1254
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7559671494446775,
      "learning_rate": 9.56442032909962e-06,
      "loss": 0.5761,
      "step": 1255
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6580896751907516,
      "learning_rate": 9.563577659163069e-06,
      "loss": 0.5418,
      "step": 1256
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.095064655241494,
      "learning_rate": 9.562734212096783e-06,
      "loss": 0.733,
      "step": 1257
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.061601469498517,
      "learning_rate": 9.561889988044396e-06,
      "loss": 0.6513,
      "step": 1258
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6710504504104681,
      "learning_rate": 9.56104498714967e-06,
      "loss": 0.5667,
      "step": 1259
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8084330586767373,
      "learning_rate": 9.560199209556498e-06,
      "loss": 0.5473,
      "step": 1260
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7853431439067693,
      "learning_rate": 9.55935265540891e-06,
      "loss": 0.5789,
      "step": 1261
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9051643593523143,
      "learning_rate": 9.558505324851063e-06,
      "loss": 0.6987,
      "step": 1262
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7178026246597083,
      "learning_rate": 9.55765721802725e-06,
      "loss": 0.6186,
      "step": 1263
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.855279285925669,
      "learning_rate": 9.556808335081894e-06,
      "loss": 0.554,
      "step": 1264
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.721062302262685,
      "learning_rate": 9.555958676159549e-06,
      "loss": 0.6089,
      "step": 1265
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0955402584854657,
      "learning_rate": 9.555108241404906e-06,
      "loss": 0.624,
      "step": 1266
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9105813138786871,
      "learning_rate": 9.554257030962786e-06,
      "loss": 0.6591,
      "step": 1267
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1865448796817364,
      "learning_rate": 9.553405044978138e-06,
      "loss": 0.6059,
      "step": 1268
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7758621117933575,
      "learning_rate": 9.552552283596049e-06,
      "loss": 0.5479,
      "step": 1269
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.656123552463779,
      "learning_rate": 9.551698746961732e-06,
      "loss": 0.525,
      "step": 1270
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0307526875301047,
      "learning_rate": 9.55084443522054e-06,
      "loss": 0.6545,
      "step": 1271
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6740851988235872,
      "learning_rate": 9.549989348517949e-06,
      "loss": 0.5257,
      "step": 1272
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6725997419652654,
      "learning_rate": 9.549133486999575e-06,
      "loss": 0.5246,
      "step": 1273
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7537544187323888,
      "learning_rate": 9.54827685081116e-06,
      "loss": 0.5513,
      "step": 1274
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6948168622802673,
      "learning_rate": 9.547419440098582e-06,
      "loss": 0.5217,
      "step": 1275
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6915872697364935,
      "learning_rate": 9.546561255007847e-06,
      "loss": 0.5671,
      "step": 1276
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9430528685843382,
      "learning_rate": 9.545702295685097e-06,
      "loss": 0.6592,
      "step": 1277
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1604352641866587,
      "learning_rate": 9.544842562276602e-06,
      "loss": 0.6832,
      "step": 1278
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9445783621589375,
      "learning_rate": 9.543982054928766e-06,
      "loss": 0.627,
      "step": 1279
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.863848836315885,
      "learning_rate": 9.543120773788126e-06,
      "loss": 0.666,
      "step": 1280
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8108027169148443,
      "learning_rate": 9.542258719001348e-06,
      "loss": 0.5592,
      "step": 1281
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9487478444979899,
      "learning_rate": 9.54139589071523e-06,
      "loss": 0.7096,
      "step": 1282
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9399022546125204,
      "learning_rate": 9.540532289076706e-06,
      "loss": 0.63,
      "step": 1283
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0395451375644704,
      "learning_rate": 9.539667914232832e-06,
      "loss": 0.6457,
      "step": 1284
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8584812975254499,
      "learning_rate": 9.538802766330809e-06,
      "loss": 0.6111,
      "step": 1285
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8812697105514897,
      "learning_rate": 9.53793684551796e-06,
      "loss": 0.6983,
      "step": 1286
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7534564534921276,
      "learning_rate": 9.537070151941741e-06,
      "loss": 0.5386,
      "step": 1287
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8743037647883589,
      "learning_rate": 9.536202685749743e-06,
      "loss": 0.6558,
      "step": 1288
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7103679472254497,
      "learning_rate": 9.535334447089685e-06,
      "loss": 0.5814,
      "step": 1289
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8371217587202098,
      "learning_rate": 9.534465436109418e-06,
      "loss": 0.5793,
      "step": 1290
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.740545149772196,
      "learning_rate": 9.533595652956929e-06,
      "loss": 0.5435,
      "step": 1291
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8682122680849824,
      "learning_rate": 9.532725097780331e-06,
      "loss": 0.6506,
      "step": 1292
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9426054733014517,
      "learning_rate": 9.53185377072787e-06,
      "loss": 0.6164,
      "step": 1293
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0308009824249327,
      "learning_rate": 9.530981671947924e-06,
      "loss": 0.6619,
      "step": 1294
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7947818632864568,
      "learning_rate": 9.530108801589002e-06,
      "loss": 0.6304,
      "step": 1295
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.764682880171863,
      "learning_rate": 9.529235159799746e-06,
      "loss": 0.6653,
      "step": 1296
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1214070766818558,
      "learning_rate": 9.528360746728928e-06,
      "loss": 0.6993,
      "step": 1297
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7562199877765657,
      "learning_rate": 9.52748556252545e-06,
      "loss": 0.6169,
      "step": 1298
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8306518017777018,
      "learning_rate": 9.526609607338348e-06,
      "loss": 0.5494,
      "step": 1299
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0674848113812165,
      "learning_rate": 9.525732881316789e-06,
      "loss": 0.7042,
      "step": 1300
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8407257477850494,
      "learning_rate": 9.524855384610068e-06,
      "loss": 0.6177,
      "step": 1301
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8408582982523104,
      "learning_rate": 9.523977117367614e-06,
      "loss": 0.6294,
      "step": 1302
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.704748987946337,
      "learning_rate": 9.523098079738989e-06,
      "loss": 0.4795,
      "step": 1303
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2722567543723173,
      "learning_rate": 9.52221827187388e-06,
      "loss": 0.6661,
      "step": 1304
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8704658826792098,
      "learning_rate": 9.521337693922114e-06,
      "loss": 0.6194,
      "step": 1305
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7226552441015462,
      "learning_rate": 9.52045634603364e-06,
      "loss": 0.5709,
      "step": 1306
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6664619971419585,
      "learning_rate": 9.519574228358544e-06,
      "loss": 0.5347,
      "step": 1307
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9934761130127118,
      "learning_rate": 9.518691341047043e-06,
      "loss": 0.648,
      "step": 1308
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8571797360041568,
      "learning_rate": 9.517807684249481e-06,
      "loss": 0.6536,
      "step": 1309
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0939841004661959,
      "learning_rate": 9.516923258116336e-06,
      "loss": 0.681,
      "step": 1310
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7984360821842511,
      "learning_rate": 9.516038062798219e-06,
      "loss": 0.7158,
      "step": 1311
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8572021021466015,
      "learning_rate": 9.515152098445866e-06,
      "loss": 0.6399,
      "step": 1312
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9215105632257056,
      "learning_rate": 9.51426536521015e-06,
      "loss": 0.6357,
      "step": 1313
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8020917486611238,
      "learning_rate": 9.513377863242071e-06,
      "loss": 0.6523,
      "step": 1314
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9130889751057123,
      "learning_rate": 9.51248959269276e-06,
      "loss": 0.6568,
      "step": 1315
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8971737636326172,
      "learning_rate": 9.511600553713485e-06,
      "loss": 0.6094,
      "step": 1316
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.035346530355026,
      "learning_rate": 9.510710746455636e-06,
      "loss": 0.6682,
      "step": 1317
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6889771668440428,
      "learning_rate": 9.509820171070741e-06,
      "loss": 0.5252,
      "step": 1318
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.766535392284087,
      "learning_rate": 9.508928827710452e-06,
      "loss": 0.5695,
      "step": 1319
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0384726436607847,
      "learning_rate": 9.508036716526556e-06,
      "loss": 0.6701,
      "step": 1320
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9105623989806091,
      "learning_rate": 9.507143837670973e-06,
      "loss": 0.6165,
      "step": 1321
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7046746961580885,
      "learning_rate": 9.506250191295751e-06,
      "loss": 0.5403,
      "step": 1322
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8150641543105733,
      "learning_rate": 9.505355777553065e-06,
      "loss": 0.5389,
      "step": 1323
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7873057040910879,
      "learning_rate": 9.504460596595228e-06,
      "loss": 0.612,
      "step": 1324
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8925400447272772,
      "learning_rate": 9.503564648574678e-06,
      "loss": 0.6575,
      "step": 1325
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2570727146303182,
      "learning_rate": 9.502667933643985e-06,
      "loss": 0.6012,
      "step": 1326
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8297815118344147,
      "learning_rate": 9.501770451955855e-06,
      "loss": 0.626,
      "step": 1327
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0230441873015468,
      "learning_rate": 9.500872203663112e-06,
      "loss": 0.6949,
      "step": 1328
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7587205312937768,
      "learning_rate": 9.499973188918724e-06,
      "loss": 0.5581,
      "step": 1329
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0792922226017754,
      "learning_rate": 9.499073407875783e-06,
      "loss": 0.67,
      "step": 1330
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.682004069953878,
      "learning_rate": 9.498172860687511e-06,
      "loss": 0.5813,
      "step": 1331
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8220654565512721,
      "learning_rate": 9.497271547507263e-06,
      "loss": 0.572,
      "step": 1332
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7103120261987019,
      "learning_rate": 9.496369468488522e-06,
      "loss": 0.5478,
      "step": 1333
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8726501575090189,
      "learning_rate": 9.495466623784903e-06,
      "loss": 0.6616,
      "step": 1334
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6891245386464124,
      "learning_rate": 9.494563013550151e-06,
      "loss": 0.5409,
      "step": 1335
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9539126902889865,
      "learning_rate": 9.493658637938143e-06,
      "loss": 0.6693,
      "step": 1336
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8299801617799516,
      "learning_rate": 9.492753497102884e-06,
      "loss": 0.6006,
      "step": 1337
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7813806787708597,
      "learning_rate": 9.491847591198509e-06,
      "loss": 0.5414,
      "step": 1338
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9879701067871018,
      "learning_rate": 9.490940920379286e-06,
      "loss": 0.6847,
      "step": 1339
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.818481226282068,
      "learning_rate": 9.490033484799608e-06,
      "loss": 0.6516,
      "step": 1340
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8626262788636114,
      "learning_rate": 9.489125284614006e-06,
      "loss": 0.6466,
      "step": 1341
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2256309471064777,
      "learning_rate": 9.488216319977135e-06,
      "loss": 0.6807,
      "step": 1342
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7882644073183519,
      "learning_rate": 9.487306591043783e-06,
      "loss": 0.5856,
      "step": 1343
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6529156386967632,
      "learning_rate": 9.486396097968866e-06,
      "loss": 0.5762,
      "step": 1344
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6751147266659726,
      "learning_rate": 9.485484840907434e-06,
      "loss": 0.5548,
      "step": 1345
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7995661715136811,
      "learning_rate": 9.484572820014662e-06,
      "loss": 0.6517,
      "step": 1346
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6993121254898311,
      "learning_rate": 9.483660035445858e-06,
      "loss": 0.5523,
      "step": 1347
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7295088926293203,
      "learning_rate": 9.482746487356462e-06,
      "loss": 0.5845,
      "step": 1348
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1168786182956545,
      "learning_rate": 9.48183217590204e-06,
      "loss": 0.6238,
      "step": 1349
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0466771172496752,
      "learning_rate": 9.48091710123829e-06,
      "loss": 0.6458,
      "step": 1350
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8604326667107473,
      "learning_rate": 9.480001263521039e-06,
      "loss": 0.6148,
      "step": 1351
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8604001846611475,
      "learning_rate": 9.479084662906243e-06,
      "loss": 0.6519,
      "step": 1352
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9135423547478635,
      "learning_rate": 9.478167299549995e-06,
      "loss": 0.6636,
      "step": 1353
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8622045769363359,
      "learning_rate": 9.477249173608507e-06,
      "loss": 0.6096,
      "step": 1354
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8074485690598203,
      "learning_rate": 9.47633028523813e-06,
      "loss": 0.56,
      "step": 1355
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8264406834688497,
      "learning_rate": 9.475410634595339e-06,
      "loss": 0.6559,
      "step": 1356
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6960446781572884,
      "learning_rate": 9.474490221836741e-06,
      "loss": 0.611,
      "step": 1357
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8979341051489285,
      "learning_rate": 9.473569047119073e-06,
      "loss": 0.6103,
      "step": 1358
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7018476691732929,
      "learning_rate": 9.472647110599201e-06,
      "loss": 0.6043,
      "step": 1359
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.716661844343984,
      "learning_rate": 9.471724412434122e-06,
      "loss": 0.5832,
      "step": 1360
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7050578556926759,
      "learning_rate": 9.470800952780963e-06,
      "loss": 0.5809,
      "step": 1361
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9556523897582249,
      "learning_rate": 9.469876731796976e-06,
      "loss": 0.7005,
      "step": 1362
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8066380348719958,
      "learning_rate": 9.468951749639552e-06,
      "loss": 0.5963,
      "step": 1363
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.175657773107267,
      "learning_rate": 9.4680260064662e-06,
      "loss": 0.7185,
      "step": 1364
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0799387829964824,
      "learning_rate": 9.467099502434566e-06,
      "loss": 0.6695,
      "step": 1365
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4124809392199231,
      "learning_rate": 9.466172237702425e-06,
      "loss": 0.6987,
      "step": 1366
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8914422200422153,
      "learning_rate": 9.46524421242768e-06,
      "loss": 0.6439,
      "step": 1367
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7152675167335548,
      "learning_rate": 9.464315426768365e-06,
      "loss": 0.5595,
      "step": 1368
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9298567589824199,
      "learning_rate": 9.463385880882641e-06,
      "loss": 0.6938,
      "step": 1369
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1389939999792575,
      "learning_rate": 9.462455574928802e-06,
      "loss": 0.728,
      "step": 1370
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8292786689476357,
      "learning_rate": 9.461524509065266e-06,
      "loss": 0.6418,
      "step": 1371
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7443732277340811,
      "learning_rate": 9.460592683450587e-06,
      "loss": 0.5763,
      "step": 1372
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.935505321703371,
      "learning_rate": 9.459660098243444e-06,
      "loss": 0.6357,
      "step": 1373
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8762314456229187,
      "learning_rate": 9.458726753602648e-06,
      "loss": 0.6572,
      "step": 1374
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8708782293166181,
      "learning_rate": 9.457792649687135e-06,
      "loss": 0.6781,
      "step": 1375
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8148322706339097,
      "learning_rate": 9.456857786655975e-06,
      "loss": 0.6513,
      "step": 1376
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7563675711411383,
      "learning_rate": 9.455922164668366e-06,
      "loss": 0.5445,
      "step": 1377
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7395653501631607,
      "learning_rate": 9.454985783883633e-06,
      "loss": 0.5728,
      "step": 1378
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6643466566999617,
      "learning_rate": 9.454048644461233e-06,
      "loss": 0.5562,
      "step": 1379
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8586519316683581,
      "learning_rate": 9.45311074656075e-06,
      "loss": 0.6678,
      "step": 1380
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7106282287523121,
      "learning_rate": 9.4521720903419e-06,
      "loss": 0.5403,
      "step": 1381
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7872496072655488,
      "learning_rate": 9.451232675964526e-06,
      "loss": 0.6083,
      "step": 1382
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6381917221280797,
      "learning_rate": 9.450292503588599e-06,
      "loss": 0.5332,
      "step": 1383
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9896206723666053,
      "learning_rate": 9.449351573374221e-06,
      "loss": 0.6607,
      "step": 1384
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.487989563253944,
      "learning_rate": 9.448409885481623e-06,
      "loss": 0.6584,
      "step": 1385
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7021826800844788,
      "learning_rate": 9.447467440071165e-06,
      "loss": 0.5486,
      "step": 1386
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8628202573182997,
      "learning_rate": 9.446524237303337e-06,
      "loss": 0.565,
      "step": 1387
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6472861922924706,
      "learning_rate": 9.445580277338753e-06,
      "loss": 0.5045,
      "step": 1388
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9832946693572294,
      "learning_rate": 9.444635560338161e-06,
      "loss": 0.7067,
      "step": 1389
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.870460099038199,
      "learning_rate": 9.443690086462438e-06,
      "loss": 0.6368,
      "step": 1390
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9355886593765449,
      "learning_rate": 9.442743855872588e-06,
      "loss": 0.7125,
      "step": 1391
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8267630267433251,
      "learning_rate": 9.441796868729743e-06,
      "loss": 0.6197,
      "step": 1392
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3247391721524735,
      "learning_rate": 9.440849125195165e-06,
      "loss": 0.6262,
      "step": 1393
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7697499940467353,
      "learning_rate": 9.439900625430245e-06,
      "loss": 0.564,
      "step": 1394
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.731071347295179,
      "learning_rate": 9.438951369596502e-06,
      "loss": 0.5628,
      "step": 1395
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9867238236472955,
      "learning_rate": 9.438001357855589e-06,
      "loss": 0.6645,
      "step": 1396
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6913602809795198,
      "learning_rate": 9.437050590369277e-06,
      "loss": 0.5717,
      "step": 1397
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7644375256545279,
      "learning_rate": 9.436099067299476e-06,
      "loss": 0.5158,
      "step": 1398
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0514777782654239,
      "learning_rate": 9.435146788808217e-06,
      "loss": 0.7343,
      "step": 1399
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6504492789035554,
      "learning_rate": 9.434193755057666e-06,
      "loss": 0.5177,
      "step": 1400
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8861569283497982,
      "learning_rate": 9.433239966210113e-06,
      "loss": 0.6767,
      "step": 1401
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7366680027550628,
      "learning_rate": 9.432285422427979e-06,
      "loss": 0.5461,
      "step": 1402
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8129918274310444,
      "learning_rate": 9.431330123873813e-06,
      "loss": 0.5718,
      "step": 1403
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7986296907262868,
      "learning_rate": 9.430374070710293e-06,
      "loss": 0.6201,
      "step": 1404
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9412767660410114,
      "learning_rate": 9.429417263100222e-06,
      "loss": 0.7098,
      "step": 1405
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7980304664743195,
      "learning_rate": 9.428459701206538e-06,
      "loss": 0.5718,
      "step": 1406
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7384508780938681,
      "learning_rate": 9.427501385192302e-06,
      "loss": 0.509,
      "step": 1407
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7535811350858405,
      "learning_rate": 9.426542315220706e-06,
      "loss": 0.5938,
      "step": 1408
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.880508787461307,
      "learning_rate": 9.425582491455068e-06,
      "loss": 0.6789,
      "step": 1409
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6830295549212533,
      "learning_rate": 9.424621914058837e-06,
      "loss": 0.5382,
      "step": 1410
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7009254136090947,
      "learning_rate": 9.423660583195591e-06,
      "loss": 0.5395,
      "step": 1411
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9452855425942293,
      "learning_rate": 9.422698499029032e-06,
      "loss": 0.6383,
      "step": 1412
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0505196889008737,
      "learning_rate": 9.421735661722993e-06,
      "loss": 0.7178,
      "step": 1413
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0373377940917945,
      "learning_rate": 9.420772071441437e-06,
      "loss": 0.6746,
      "step": 1414
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7321410015833552,
      "learning_rate": 9.419807728348454e-06,
      "loss": 0.5227,
      "step": 1415
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9335362591314534,
      "learning_rate": 9.418842632608258e-06,
      "loss": 0.7,
      "step": 1416
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8204885066714896,
      "learning_rate": 9.417876784385197e-06,
      "loss": 0.5958,
      "step": 1417
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8012454116399874,
      "learning_rate": 9.416910183843746e-06,
      "loss": 0.6025,
      "step": 1418
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7928185226323354,
      "learning_rate": 9.415942831148503e-06,
      "loss": 0.5419,
      "step": 1419
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5483265539463231,
      "learning_rate": 9.414974726464203e-06,
      "loss": 0.6869,
      "step": 1420
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.900369485492935,
      "learning_rate": 9.414005869955702e-06,
      "loss": 0.6095,
      "step": 1421
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9433803283459624,
      "learning_rate": 9.413036261787987e-06,
      "loss": 0.6284,
      "step": 1422
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7701997617779117,
      "learning_rate": 9.41206590212617e-06,
      "loss": 0.5691,
      "step": 1423
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8970706584427929,
      "learning_rate": 9.411094791135494e-06,
      "loss": 0.6914,
      "step": 1424
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6680764281951888,
      "learning_rate": 9.410122928981332e-06,
      "loss": 0.5313,
      "step": 1425
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9388747901596838,
      "learning_rate": 9.409150315829177e-06,
      "loss": 0.6981,
      "step": 1426
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7466058230171558,
      "learning_rate": 9.408176951844657e-06,
      "loss": 0.553,
      "step": 1427
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.867009821561423,
      "learning_rate": 9.407202837193527e-06,
      "loss": 0.6286,
      "step": 1428
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9830153456580065,
      "learning_rate": 9.406227972041668e-06,
      "loss": 0.7027,
      "step": 1429
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9238394132248928,
      "learning_rate": 9.405252356555089e-06,
      "loss": 0.6758,
      "step": 1430
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9919844551003053,
      "learning_rate": 9.404275990899927e-06,
      "loss": 0.6995,
      "step": 1431
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.00166222259337,
      "learning_rate": 9.403298875242448e-06,
      "loss": 0.7228,
      "step": 1432
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9935346146699167,
      "learning_rate": 9.402321009749043e-06,
      "loss": 0.6908,
      "step": 1433
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9314062931808536,
      "learning_rate": 9.401342394586234e-06,
      "loss": 0.6446,
      "step": 1434
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8467541347200406,
      "learning_rate": 9.400363029920667e-06,
      "loss": 0.5836,
      "step": 1435
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8297993023382004,
      "learning_rate": 9.399382915919119e-06,
      "loss": 0.61,
      "step": 1436
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7607304868667946,
      "learning_rate": 9.398402052748492e-06,
      "loss": 0.5807,
      "step": 1437
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0092052694956086,
      "learning_rate": 9.39742044057582e-06,
      "loss": 0.6518,
      "step": 1438
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.800278530033686,
      "learning_rate": 9.396438079568257e-06,
      "loss": 0.5804,
      "step": 1439
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.724408868218778,
      "learning_rate": 9.395454969893092e-06,
      "loss": 0.5308,
      "step": 1440
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6459185893983147,
      "learning_rate": 9.394471111717735e-06,
      "loss": 0.5805,
      "step": 1441
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2806999126351153,
      "learning_rate": 9.393486505209731e-06,
      "loss": 0.7055,
      "step": 1442
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8090690580736293,
      "learning_rate": 9.392501150536747e-06,
      "loss": 0.6058,
      "step": 1443
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8876635050002758,
      "learning_rate": 9.391515047866575e-06,
      "loss": 0.6519,
      "step": 1444
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6573797049659426,
      "learning_rate": 9.390528197367143e-06,
      "loss": 0.5477,
      "step": 1445
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7708956511975157,
      "learning_rate": 9.389540599206497e-06,
      "loss": 0.5361,
      "step": 1446
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9360230370143557,
      "learning_rate": 9.388552253552818e-06,
      "loss": 0.6693,
      "step": 1447
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.970158527075951,
      "learning_rate": 9.387563160574411e-06,
      "loss": 0.6898,
      "step": 1448
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8756441277447434,
      "learning_rate": 9.386573320439706e-06,
      "loss": 0.6457,
      "step": 1449
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8956500569817887,
      "learning_rate": 9.385582733317264e-06,
      "loss": 0.7085,
      "step": 1450
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9998782305202425,
      "learning_rate": 9.38459139937577e-06,
      "loss": 0.7146,
      "step": 1451
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7230468564212061,
      "learning_rate": 9.38359931878404e-06,
      "loss": 0.5747,
      "step": 1452
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.713890040657229,
      "learning_rate": 9.382606491711015e-06,
      "loss": 0.5823,
      "step": 1453
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.867719149009511,
      "learning_rate": 9.38161291832576e-06,
      "loss": 0.6558,
      "step": 1454
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8203425675832203,
      "learning_rate": 9.380618598797473e-06,
      "loss": 0.5727,
      "step": 1455
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7142825890240405,
      "learning_rate": 9.379623533295476e-06,
      "loss": 0.5146,
      "step": 1456
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6592433207121559,
      "learning_rate": 9.378627721989218e-06,
      "loss": 0.5358,
      "step": 1457
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9731278707589606,
      "learning_rate": 9.377631165048276e-06,
      "loss": 0.6394,
      "step": 1458
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8385952273910086,
      "learning_rate": 9.37663386264235e-06,
      "loss": 0.5822,
      "step": 1459
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7254346473306581,
      "learning_rate": 9.375635814941276e-06,
      "loss": 0.553,
      "step": 1460
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7584583063743576,
      "learning_rate": 9.374637022115005e-06,
      "loss": 0.5965,
      "step": 1461
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7261967805197976,
      "learning_rate": 9.373637484333625e-06,
      "loss": 0.6309,
      "step": 1462
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.03802615713192,
      "learning_rate": 9.372637201767345e-06,
      "loss": 0.6953,
      "step": 1463
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6553971967453723,
      "learning_rate": 9.371636174586503e-06,
      "loss": 0.5668,
      "step": 1464
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6941242968879597,
      "learning_rate": 9.370634402961565e-06,
      "loss": 0.5796,
      "step": 1465
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8691198834770186,
      "learning_rate": 9.36963188706312e-06,
      "loss": 0.602,
      "step": 1466
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.951613735756575,
      "learning_rate": 9.368628627061886e-06,
      "loss": 0.6696,
      "step": 1467
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8382284889157989,
      "learning_rate": 9.367624623128711e-06,
      "loss": 0.6489,
      "step": 1468
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7626337112380835,
      "learning_rate": 9.366619875434563e-06,
      "loss": 0.6022,
      "step": 1469
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8098200177881252,
      "learning_rate": 9.365614384150543e-06,
      "loss": 0.6116,
      "step": 1470
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7226955687707713,
      "learning_rate": 9.364608149447873e-06,
      "loss": 0.587,
      "step": 1471
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9743341693176809,
      "learning_rate": 9.363601171497906e-06,
      "loss": 0.6565,
      "step": 1472
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9084918718709991,
      "learning_rate": 9.362593450472118e-06,
      "loss": 0.7229,
      "step": 1473
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9738504066676951,
      "learning_rate": 9.361584986542116e-06,
      "loss": 0.7066,
      "step": 1474
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3025922797655187,
      "learning_rate": 9.360575779879628e-06,
      "loss": 0.6047,
      "step": 1475
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8598603129016218,
      "learning_rate": 9.359565830656514e-06,
      "loss": 0.6249,
      "step": 1476
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.717159892697812,
      "learning_rate": 9.358555139044755e-06,
      "loss": 0.4972,
      "step": 1477
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.812296559753277,
      "learning_rate": 9.357543705216465e-06,
      "loss": 0.5865,
      "step": 1478
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.91581960766695,
      "learning_rate": 9.356531529343878e-06,
      "loss": 0.6833,
      "step": 1479
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8470751419432861,
      "learning_rate": 9.355518611599357e-06,
      "loss": 0.6275,
      "step": 1480
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7002160091450838,
      "learning_rate": 9.354504952155392e-06,
      "loss": 0.5048,
      "step": 1481
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7655283760417843,
      "learning_rate": 9.353490551184597e-06,
      "loss": 0.595,
      "step": 1482
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8797984194132432,
      "learning_rate": 9.352475408859718e-06,
      "loss": 0.6278,
      "step": 1483
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9037980559426629,
      "learning_rate": 9.351459525353618e-06,
      "loss": 0.6017,
      "step": 1484
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7717744164747941,
      "learning_rate": 9.350442900839297e-06,
      "loss": 0.5593,
      "step": 1485
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0467950755337083,
      "learning_rate": 9.349425535489871e-06,
      "loss": 0.658,
      "step": 1486
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9252749318662788,
      "learning_rate": 9.348407429478589e-06,
      "loss": 0.6179,
      "step": 1487
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7601673282213051,
      "learning_rate": 9.347388582978822e-06,
      "loss": 0.5189,
      "step": 1488
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8738311803154118,
      "learning_rate": 9.346368996164072e-06,
      "loss": 0.652,
      "step": 1489
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0064289054746205,
      "learning_rate": 9.345348669207961e-06,
      "loss": 0.6466,
      "step": 1490
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8673882636723294,
      "learning_rate": 9.34432760228424e-06,
      "loss": 0.6283,
      "step": 1491
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8999332216547168,
      "learning_rate": 9.343305795566788e-06,
      "loss": 0.7124,
      "step": 1492
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.797703822431825,
      "learning_rate": 9.342283249229608e-06,
      "loss": 0.5313,
      "step": 1493
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7253846052120263,
      "learning_rate": 9.341259963446827e-06,
      "loss": 0.6194,
      "step": 1494
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7344136739618077,
      "learning_rate": 9.340235938392703e-06,
      "loss": 0.6112,
      "step": 1495
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7522858647091321,
      "learning_rate": 9.339211174241613e-06,
      "loss": 0.6099,
      "step": 1496
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8406951207448761,
      "learning_rate": 9.338185671168067e-06,
      "loss": 0.7125,
      "step": 1497
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8489361295417559,
      "learning_rate": 9.337159429346697e-06,
      "loss": 0.6684,
      "step": 1498
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6619646370906785,
      "learning_rate": 9.336132448952258e-06,
      "loss": 0.583,
      "step": 1499
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.942165444345931,
      "learning_rate": 9.335104730159637e-06,
      "loss": 0.6481,
      "step": 1500
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7299514233441786,
      "learning_rate": 9.334076273143843e-06,
      "loss": 0.5574,
      "step": 1501
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9408891725261744,
      "learning_rate": 9.333047078080013e-06,
      "loss": 0.6766,
      "step": 1502
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7670756329385489,
      "learning_rate": 9.332017145143407e-06,
      "loss": 0.5757,
      "step": 1503
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8372802244388492,
      "learning_rate": 9.33098647450941e-06,
      "loss": 0.6382,
      "step": 1504
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8219203876378051,
      "learning_rate": 9.329955066353538e-06,
      "loss": 0.6065,
      "step": 1505
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7913249540779064,
      "learning_rate": 9.328922920851426e-06,
      "loss": 0.6177,
      "step": 1506
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0051561912811227,
      "learning_rate": 9.32789003817884e-06,
      "loss": 0.6596,
      "step": 1507
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7105315374674107,
      "learning_rate": 9.326856418511668e-06,
      "loss": 0.6082,
      "step": 1508
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6807468876100213,
      "learning_rate": 9.325822062025924e-06,
      "loss": 0.5192,
      "step": 1509
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8061319371190606,
      "learning_rate": 9.32478696889775e-06,
      "loss": 0.6076,
      "step": 1510
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2858866682190981,
      "learning_rate": 9.32375113930341e-06,
      "loss": 0.7408,
      "step": 1511
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7252193360405388,
      "learning_rate": 9.322714573419296e-06,
      "loss": 0.6354,
      "step": 1512
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.79000626868656,
      "learning_rate": 9.321677271421923e-06,
      "loss": 0.6011,
      "step": 1513
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.861675515987892,
      "learning_rate": 9.320639233487934e-06,
      "loss": 0.6566,
      "step": 1514
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9193408449032802,
      "learning_rate": 9.319600459794097e-06,
      "loss": 0.6488,
      "step": 1515
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9942805917249699,
      "learning_rate": 9.318560950517298e-06,
      "loss": 0.6469,
      "step": 1516
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8442956214935163,
      "learning_rate": 9.317520705834566e-06,
      "loss": 0.5933,
      "step": 1517
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7672275061678321,
      "learning_rate": 9.316479725923034e-06,
      "loss": 0.5912,
      "step": 1518
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7771469973618885,
      "learning_rate": 9.315438010959974e-06,
      "loss": 0.591,
      "step": 1519
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9186584532423744,
      "learning_rate": 9.314395561122778e-06,
      "loss": 0.704,
      "step": 1520
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.922574073865881,
      "learning_rate": 9.313352376588966e-06,
      "loss": 0.7353,
      "step": 1521
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9866940849240626,
      "learning_rate": 9.31230845753618e-06,
      "loss": 0.739,
      "step": 1522
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7401484272040525,
      "learning_rate": 9.31126380414219e-06,
      "loss": 0.583,
      "step": 1523
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7426936246810004,
      "learning_rate": 9.310218416584887e-06,
      "loss": 0.6274,
      "step": 1524
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7480885517503246,
      "learning_rate": 9.309172295042291e-06,
      "loss": 0.6283,
      "step": 1525
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7414999595938194,
      "learning_rate": 9.308125439692546e-06,
      "loss": 0.5747,
      "step": 1526
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8200381512798903,
      "learning_rate": 9.307077850713922e-06,
      "loss": 0.5237,
      "step": 1527
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.696438969130969,
      "learning_rate": 9.30602952828481e-06,
      "loss": 0.5271,
      "step": 1528
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8900412487177588,
      "learning_rate": 9.304980472583729e-06,
      "loss": 0.601,
      "step": 1529
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9933262422355958,
      "learning_rate": 9.303930683789322e-06,
      "loss": 0.6783,
      "step": 1530
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8104814466678987,
      "learning_rate": 9.302880162080358e-06,
      "loss": 0.6113,
      "step": 1531
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.105396965825193,
      "learning_rate": 9.30182890763573e-06,
      "loss": 0.6562,
      "step": 1532
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.872229444170377,
      "learning_rate": 9.300776920634454e-06,
      "loss": 0.5773,
      "step": 1533
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.818441395272117,
      "learning_rate": 9.299724201255676e-06,
      "loss": 0.5879,
      "step": 1534
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7809104368939722,
      "learning_rate": 9.298670749678657e-06,
      "loss": 0.5796,
      "step": 1535
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7604625908175191,
      "learning_rate": 9.297616566082796e-06,
      "loss": 0.604,
      "step": 1536
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7545217657442671,
      "learning_rate": 9.296561650647605e-06,
      "loss": 0.6624,
      "step": 1537
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8019431217621308,
      "learning_rate": 9.295506003552725e-06,
      "loss": 0.6174,
      "step": 1538
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.718438041337846,
      "learning_rate": 9.294449624977923e-06,
      "loss": 0.5909,
      "step": 1539
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7164170654957996,
      "learning_rate": 9.29339251510309e-06,
      "loss": 0.5872,
      "step": 1540
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.028751660001763,
      "learning_rate": 9.292334674108239e-06,
      "loss": 0.6968,
      "step": 1541
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8817943770725851,
      "learning_rate": 9.291276102173512e-06,
      "loss": 0.6208,
      "step": 1542
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.775162344489918,
      "learning_rate": 9.290216799479169e-06,
      "loss": 0.5801,
      "step": 1543
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1747453556725396,
      "learning_rate": 9.289156766205601e-06,
      "loss": 0.6815,
      "step": 1544
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9637029153848131,
      "learning_rate": 9.28809600253332e-06,
      "loss": 0.6674,
      "step": 1545
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.843118713672653,
      "learning_rate": 9.287034508642962e-06,
      "loss": 0.6359,
      "step": 1546
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7437236261442519,
      "learning_rate": 9.285972284715291e-06,
      "loss": 0.6044,
      "step": 1547
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6883973872906871,
      "learning_rate": 9.28490933093119e-06,
      "loss": 0.5307,
      "step": 1548
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7291908889033247,
      "learning_rate": 9.28384564747167e-06,
      "loss": 0.5675,
      "step": 1549
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0589610166955687,
      "learning_rate": 9.282781234517863e-06,
      "loss": 0.6647,
      "step": 1550
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8798381364021589,
      "learning_rate": 9.281716092251033e-06,
      "loss": 0.6854,
      "step": 1551
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0137580050664556,
      "learning_rate": 9.280650220852556e-06,
      "loss": 0.5418,
      "step": 1552
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8788401563020618,
      "learning_rate": 9.279583620503942e-06,
      "loss": 0.6015,
      "step": 1553
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7614720491735161,
      "learning_rate": 9.278516291386823e-06,
      "loss": 0.6122,
      "step": 1554
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1049815457556047,
      "learning_rate": 9.277448233682953e-06,
      "loss": 0.6349,
      "step": 1555
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7226597521231889,
      "learning_rate": 9.27637944757421e-06,
      "loss": 0.6001,
      "step": 1556
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7922304003103005,
      "learning_rate": 9.2753099332426e-06,
      "loss": 0.6968,
      "step": 1557
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6425191668751903,
      "learning_rate": 9.274239690870247e-06,
      "loss": 0.6281,
      "step": 1558
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6448625054268583,
      "learning_rate": 9.273168720639403e-06,
      "loss": 0.5568,
      "step": 1559
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.738643651642791,
      "learning_rate": 9.272097022732444e-06,
      "loss": 0.566,
      "step": 1560
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8697136168268381,
      "learning_rate": 9.271024597331868e-06,
      "loss": 0.6514,
      "step": 1561
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8536718043641645,
      "learning_rate": 9.269951444620298e-06,
      "loss": 0.6756,
      "step": 1562
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.773309672526406,
      "learning_rate": 9.26887756478048e-06,
      "loss": 0.5816,
      "step": 1563
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.739379961373571,
      "learning_rate": 9.267802957995288e-06,
      "loss": 0.5694,
      "step": 1564
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.759744457689315,
      "learning_rate": 9.266727624447713e-06,
      "loss": 0.5789,
      "step": 1565
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8303819792975942,
      "learning_rate": 9.265651564320874e-06,
      "loss": 0.6046,
      "step": 1566
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9527546143783323,
      "learning_rate": 9.264574777798012e-06,
      "loss": 0.6698,
      "step": 1567
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.732659884433006,
      "learning_rate": 9.263497265062495e-06,
      "loss": 0.6422,
      "step": 1568
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0661966462269492,
      "learning_rate": 9.262419026297808e-06,
      "loss": 0.6183,
      "step": 1569
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8858284874157666,
      "learning_rate": 9.26134006168757e-06,
      "loss": 0.6316,
      "step": 1570
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9242246793291119,
      "learning_rate": 9.26026037141551e-06,
      "loss": 0.6513,
      "step": 1571
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9055532390060402,
      "learning_rate": 9.259179955665494e-06,
      "loss": 0.5904,
      "step": 1572
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.841539947337721,
      "learning_rate": 9.258098814621504e-06,
      "loss": 0.6753,
      "step": 1573
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.00011037120198,
      "learning_rate": 9.257016948467645e-06,
      "loss": 0.5826,
      "step": 1574
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0750876747016627,
      "learning_rate": 9.25593435738815e-06,
      "loss": 0.6633,
      "step": 1575
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7224467352543059,
      "learning_rate": 9.25485104156737e-06,
      "loss": 0.5668,
      "step": 1576
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7844720425299436,
      "learning_rate": 9.253767001189786e-06,
      "loss": 0.5961,
      "step": 1577
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8788938058359038,
      "learning_rate": 9.252682236439997e-06,
      "loss": 0.5804,
      "step": 1578
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9190541094053893,
      "learning_rate": 9.251596747502727e-06,
      "loss": 0.6116,
      "step": 1579
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9083903462274635,
      "learning_rate": 9.250510534562825e-06,
      "loss": 0.73,
      "step": 1580
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8906140248633336,
      "learning_rate": 9.249423597805259e-06,
      "loss": 0.6472,
      "step": 1581
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8347708668401195,
      "learning_rate": 9.248335937415123e-06,
      "loss": 0.6227,
      "step": 1582
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9770089252885229,
      "learning_rate": 9.247247553577638e-06,
      "loss": 0.679,
      "step": 1583
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8986923003889647,
      "learning_rate": 9.246158446478142e-06,
      "loss": 0.6786,
      "step": 1584
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6988598417055969,
      "learning_rate": 9.245068616302097e-06,
      "loss": 0.5696,
      "step": 1585
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7718127107832289,
      "learning_rate": 9.243978063235094e-06,
      "loss": 0.5774,
      "step": 1586
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6919991601864117,
      "learning_rate": 9.242886787462839e-06,
      "loss": 0.5694,
      "step": 1587
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.72027523637707,
      "learning_rate": 9.241794789171165e-06,
      "loss": 0.5833,
      "step": 1588
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9107096864147478,
      "learning_rate": 9.240702068546031e-06,
      "loss": 0.6604,
      "step": 1589
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7746850364965003,
      "learning_rate": 9.239608625773514e-06,
      "loss": 0.5806,
      "step": 1590
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9129621240403727,
      "learning_rate": 9.238514461039814e-06,
      "loss": 0.6693,
      "step": 1591
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9149265552176682,
      "learning_rate": 9.237419574531258e-06,
      "loss": 0.6662,
      "step": 1592
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.758431309483103,
      "learning_rate": 9.236323966434296e-06,
      "loss": 0.587,
      "step": 1593
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7210470304216425,
      "learning_rate": 9.235227636935493e-06,
      "loss": 0.579,
      "step": 1594
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8891628584429903,
      "learning_rate": 9.234130586221547e-06,
      "loss": 0.6658,
      "step": 1595
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7837551821738634,
      "learning_rate": 9.233032814479274e-06,
      "loss": 0.586,
      "step": 1596
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8054776919415938,
      "learning_rate": 9.231934321895612e-06,
      "loss": 0.6173,
      "step": 1597
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7998421405052893,
      "learning_rate": 9.230835108657622e-06,
      "loss": 0.6281,
      "step": 1598
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.834651104824456,
      "learning_rate": 9.229735174952492e-06,
      "loss": 0.6568,
      "step": 1599
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8726927003821472,
      "learning_rate": 9.228634520967524e-06,
      "loss": 0.6238,
      "step": 1600
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6707846465045022,
      "learning_rate": 9.227533146890151e-06,
      "loss": 0.5032,
      "step": 1601
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9350381129189336,
      "learning_rate": 9.226431052907928e-06,
      "loss": 0.7071,
      "step": 1602
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8920358796328383,
      "learning_rate": 9.225328239208524e-06,
      "loss": 0.6769,
      "step": 1603
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8886833812133887,
      "learning_rate": 9.22422470597974e-06,
      "loss": 0.6678,
      "step": 1604
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9323716529954269,
      "learning_rate": 9.223120453409497e-06,
      "loss": 0.6494,
      "step": 1605
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.928100267784525,
      "learning_rate": 9.222015481685838e-06,
      "loss": 0.6367,
      "step": 1606
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5399558746976043,
      "learning_rate": 9.220909790996927e-06,
      "loss": 0.6123,
      "step": 1607
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6509323967942054,
      "learning_rate": 9.219803381531052e-06,
      "loss": 0.585,
      "step": 1608
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9086364129417434,
      "learning_rate": 9.218696253476621e-06,
      "loss": 0.6239,
      "step": 1609
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9596573688039909,
      "learning_rate": 9.217588407022169e-06,
      "loss": 0.6381,
      "step": 1610
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.232707657005479,
      "learning_rate": 9.21647984235635e-06,
      "loss": 0.636,
      "step": 1611
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6594714692610681,
      "learning_rate": 9.21537055966794e-06,
      "loss": 0.5623,
      "step": 1612
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.737598094075262,
      "learning_rate": 9.21426055914584e-06,
      "loss": 0.5699,
      "step": 1613
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7689082783417887,
      "learning_rate": 9.21314984097907e-06,
      "loss": 0.5382,
      "step": 1614
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8047690448365722,
      "learning_rate": 9.212038405356775e-06,
      "loss": 0.6147,
      "step": 1615
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.796643699226512,
      "learning_rate": 9.21092625246822e-06,
      "loss": 0.5642,
      "step": 1616
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8466573833017538,
      "learning_rate": 9.209813382502793e-06,
      "loss": 0.6775,
      "step": 1617
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7409986778646053,
      "learning_rate": 9.208699795650005e-06,
      "loss": 0.5849,
      "step": 1618
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9326774107754888,
      "learning_rate": 9.207585492099487e-06,
      "loss": 0.6519,
      "step": 1619
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7115929101705958,
      "learning_rate": 9.206470472040994e-06,
      "loss": 0.596,
      "step": 1620
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7739898060274144,
      "learning_rate": 9.205354735664402e-06,
      "loss": 0.5937,
      "step": 1621
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.672379987028906,
      "learning_rate": 9.20423828315971e-06,
      "loss": 0.5775,
      "step": 1622
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9324137991561445,
      "learning_rate": 9.203121114717039e-06,
      "loss": 0.6453,
      "step": 1623
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7412879055921451,
      "learning_rate": 9.202003230526628e-06,
      "loss": 0.5937,
      "step": 1624
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8740019331205825,
      "learning_rate": 9.200884630778846e-06,
      "loss": 0.5968,
      "step": 1625
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.970542072675399,
      "learning_rate": 9.199765315664174e-06,
      "loss": 0.6984,
      "step": 1626
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8887041774449264,
      "learning_rate": 9.19864528537322e-06,
      "loss": 0.705,
      "step": 1627
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6978549815955,
      "learning_rate": 9.197524540096717e-06,
      "loss": 0.545,
      "step": 1628
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7511206735398799,
      "learning_rate": 9.196403080025513e-06,
      "loss": 0.5219,
      "step": 1629
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7507031381515049,
      "learning_rate": 9.195280905350585e-06,
      "loss": 0.5712,
      "step": 1630
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0106114340176788,
      "learning_rate": 9.194158016263022e-06,
      "loss": 0.5972,
      "step": 1631
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0284939054733948,
      "learning_rate": 9.193034412954045e-06,
      "loss": 0.7125,
      "step": 1632
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3734633990939342,
      "learning_rate": 9.191910095614988e-06,
      "loss": 0.6419,
      "step": 1633
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7741141269209555,
      "learning_rate": 9.190785064437316e-06,
      "loss": 0.6414,
      "step": 1634
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6746760975869747,
      "learning_rate": 9.189659319612605e-06,
      "loss": 0.5929,
      "step": 1635
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7608116551985806,
      "learning_rate": 9.18853286133256e-06,
      "loss": 0.5565,
      "step": 1636
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7455855580757017,
      "learning_rate": 9.187405689789004e-06,
      "loss": 0.5991,
      "step": 1637
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.972677599332281,
      "learning_rate": 9.186277805173883e-06,
      "loss": 0.65,
      "step": 1638
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0057386471776937,
      "learning_rate": 9.185149207679263e-06,
      "loss": 0.6243,
      "step": 1639
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.064579559959113,
      "learning_rate": 9.184019897497336e-06,
      "loss": 0.6059,
      "step": 1640
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7814696682955395,
      "learning_rate": 9.182889874820407e-06,
      "loss": 0.559,
      "step": 1641
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9138967319158429,
      "learning_rate": 9.181759139840911e-06,
      "loss": 0.6777,
      "step": 1642
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.730233423577774,
      "learning_rate": 9.180627692751399e-06,
      "loss": 0.6225,
      "step": 1643
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7141906934945305,
      "learning_rate": 9.179495533744543e-06,
      "loss": 0.5305,
      "step": 1644
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.946653839003957,
      "learning_rate": 9.178362663013138e-06,
      "loss": 0.6481,
      "step": 1645
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9400709446865988,
      "learning_rate": 9.177229080750104e-06,
      "loss": 0.6749,
      "step": 1646
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7072012908197184,
      "learning_rate": 9.176094787148476e-06,
      "loss": 0.5231,
      "step": 1647
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9977195913128668,
      "learning_rate": 9.174959782401408e-06,
      "loss": 0.6293,
      "step": 1648
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7188404747338388,
      "learning_rate": 9.173824066702189e-06,
      "loss": 0.5823,
      "step": 1649
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7602720468912013,
      "learning_rate": 9.17268764024421e-06,
      "loss": 0.559,
      "step": 1650
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8279132093175969,
      "learning_rate": 9.171550503220999e-06,
      "loss": 0.6642,
      "step": 1651
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6879512022624757,
      "learning_rate": 9.170412655826194e-06,
      "loss": 0.5689,
      "step": 1652
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9531262221201743,
      "learning_rate": 9.169274098253563e-06,
      "loss": 0.6646,
      "step": 1653
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0268354778495947,
      "learning_rate": 9.168134830696986e-06,
      "loss": 0.6945,
      "step": 1654
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8427359424288263,
      "learning_rate": 9.16699485335047e-06,
      "loss": 0.6899,
      "step": 1655
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6398606680437324,
      "learning_rate": 9.165854166408144e-06,
      "loss": 0.5371,
      "step": 1656
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8065401688150998,
      "learning_rate": 9.164712770064254e-06,
      "loss": 0.6532,
      "step": 1657
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8182513753790843,
      "learning_rate": 9.163570664513166e-06,
      "loss": 0.7089,
      "step": 1658
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8535790071663967,
      "learning_rate": 9.162427849949367e-06,
      "loss": 0.6781,
      "step": 1659
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8844564779764534,
      "learning_rate": 9.161284326567472e-06,
      "loss": 0.6756,
      "step": 1660
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8595583674542,
      "learning_rate": 9.160140094562208e-06,
      "loss": 0.6069,
      "step": 1661
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6414844032512167,
      "learning_rate": 9.158995154128425e-06,
      "loss": 0.5205,
      "step": 1662
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8286912256625293,
      "learning_rate": 9.157849505461095e-06,
      "loss": 0.5599,
      "step": 1663
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6441001181128698,
      "learning_rate": 9.156703148755311e-06,
      "loss": 0.5708,
      "step": 1664
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8654583278911271,
      "learning_rate": 9.155556084206285e-06,
      "loss": 0.6391,
      "step": 1665
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6150761946345886,
      "learning_rate": 9.15440831200935e-06,
      "loss": 0.5328,
      "step": 1666
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0571055738759445,
      "learning_rate": 9.15325983235996e-06,
      "loss": 0.6307,
      "step": 1667
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9494184518754794,
      "learning_rate": 9.152110645453689e-06,
      "loss": 0.64,
      "step": 1668
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6707018124037974,
      "learning_rate": 9.150960751486231e-06,
      "loss": 0.5204,
      "step": 1669
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9007994494048571,
      "learning_rate": 9.149810150653401e-06,
      "loss": 0.707,
      "step": 1670
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6784877630337013,
      "learning_rate": 9.148658843151135e-06,
      "loss": 0.5799,
      "step": 1671
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9809801270704389,
      "learning_rate": 9.14750682917549e-06,
      "loss": 0.6248,
      "step": 1672
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7054687903373112,
      "learning_rate": 9.14635410892264e-06,
      "loss": 0.5414,
      "step": 1673
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0399084742103415,
      "learning_rate": 9.14520068258888e-06,
      "loss": 0.6032,
      "step": 1674
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8591677486606304,
      "learning_rate": 9.144046550370628e-06,
      "loss": 0.6036,
      "step": 1675
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7084094942979969,
      "learning_rate": 9.142891712464422e-06,
      "loss": 0.5596,
      "step": 1676
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7105250801983417,
      "learning_rate": 9.141736169066917e-06,
      "loss": 0.5632,
      "step": 1677
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6914877695214648,
      "learning_rate": 9.140579920374892e-06,
      "loss": 0.5886,
      "step": 1678
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2709845408500606,
      "learning_rate": 9.139422966585242e-06,
      "loss": 0.6392,
      "step": 1679
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7384326267337248,
      "learning_rate": 9.138265307894985e-06,
      "loss": 0.5864,
      "step": 1680
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6790450115939309,
      "learning_rate": 9.137106944501258e-06,
      "loss": 0.5476,
      "step": 1681
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7258679594378835,
      "learning_rate": 9.13594787660132e-06,
      "loss": 0.5925,
      "step": 1682
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8256179175482496,
      "learning_rate": 9.134788104392545e-06,
      "loss": 0.5882,
      "step": 1683
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.730446393041975,
      "learning_rate": 9.13362762807243e-06,
      "loss": 0.5312,
      "step": 1684
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8232794625624785,
      "learning_rate": 9.132466447838598e-06,
      "loss": 0.6386,
      "step": 1685
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7983654532685514,
      "learning_rate": 9.131304563888779e-06,
      "loss": 0.5708,
      "step": 1686
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.732272239705348,
      "learning_rate": 9.130141976420833e-06,
      "loss": 0.5916,
      "step": 1687
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7224169664842395,
      "learning_rate": 9.128978685632735e-06,
      "loss": 0.5226,
      "step": 1688
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.235211970309404,
      "learning_rate": 9.127814691722583e-06,
      "loss": 0.6462,
      "step": 1689
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7541626413456836,
      "learning_rate": 9.126649994888592e-06,
      "loss": 0.6395,
      "step": 1690
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6406834135485577,
      "learning_rate": 9.125484595329098e-06,
      "loss": 0.5465,
      "step": 1691
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.726974134855615,
      "learning_rate": 9.124318493242556e-06,
      "loss": 0.6037,
      "step": 1692
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.704239500325948,
      "learning_rate": 9.123151688827542e-06,
      "loss": 0.5871,
      "step": 1693
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7566892783724908,
      "learning_rate": 9.12198418228275e-06,
      "loss": 0.56,
      "step": 1694
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9994032720300379,
      "learning_rate": 9.120815973806996e-06,
      "loss": 0.6398,
      "step": 1695
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8762560320159125,
      "learning_rate": 9.11964706359921e-06,
      "loss": 0.7044,
      "step": 1696
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7967810767753024,
      "learning_rate": 9.118477451858448e-06,
      "loss": 0.5665,
      "step": 1697
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7243554820040298,
      "learning_rate": 9.117307138783881e-06,
      "loss": 0.5448,
      "step": 1698
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.275838034418713,
      "learning_rate": 9.116136124574803e-06,
      "loss": 0.6344,
      "step": 1699
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8811931568828208,
      "learning_rate": 9.114964409430624e-06,
      "loss": 0.6334,
      "step": 1700
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0282201244574742,
      "learning_rate": 9.113791993550876e-06,
      "loss": 0.6758,
      "step": 1701
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.845582938833086,
      "learning_rate": 9.112618877135208e-06,
      "loss": 0.6672,
      "step": 1702
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7532367719484728,
      "learning_rate": 9.11144506038339e-06,
      "loss": 0.615,
      "step": 1703
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6218901306944322,
      "learning_rate": 9.110270543495312e-06,
      "loss": 0.5041,
      "step": 1704
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8728065675683571,
      "learning_rate": 9.10909532667098e-06,
      "loss": 0.6425,
      "step": 1705
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.630190568494268,
      "learning_rate": 9.107919410110522e-06,
      "loss": 0.5451,
      "step": 1706
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9244379329995707,
      "learning_rate": 9.106742794014186e-06,
      "loss": 0.7089,
      "step": 1707
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7055678750403788,
      "learning_rate": 9.105565478582335e-06,
      "loss": 0.5625,
      "step": 1708
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7722422835886537,
      "learning_rate": 9.104387464015453e-06,
      "loss": 0.5857,
      "step": 1709
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.805357366465741,
      "learning_rate": 9.103208750514147e-06,
      "loss": 0.6148,
      "step": 1710
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9550107373176977,
      "learning_rate": 9.102029338279138e-06,
      "loss": 0.6898,
      "step": 1711
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0599975987628432,
      "learning_rate": 9.100849227511266e-06,
      "loss": 0.6153,
      "step": 1712
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0974351460383835,
      "learning_rate": 9.099668418411494e-06,
      "loss": 0.6296,
      "step": 1713
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6993288675937175,
      "learning_rate": 9.0984869111809e-06,
      "loss": 0.5713,
      "step": 1714
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9196388164774751,
      "learning_rate": 9.097304706020682e-06,
      "loss": 0.6892,
      "step": 1715
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7456097217629195,
      "learning_rate": 9.096121803132158e-06,
      "loss": 0.5695,
      "step": 1716
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6867136241361311,
      "learning_rate": 9.094938202716764e-06,
      "loss": 0.553,
      "step": 1717
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7778179548021698,
      "learning_rate": 9.093753904976056e-06,
      "loss": 0.5442,
      "step": 1718
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.81808374340776,
      "learning_rate": 9.092568910111705e-06,
      "loss": 0.619,
      "step": 1719
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9386670540754333,
      "learning_rate": 9.091383218325506e-06,
      "loss": 0.6255,
      "step": 1720
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8158233269671321,
      "learning_rate": 9.090196829819368e-06,
      "loss": 0.548,
      "step": 1721
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8769296571815728,
      "learning_rate": 9.08900974479532e-06,
      "loss": 0.6549,
      "step": 1722
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7363202803849038,
      "learning_rate": 9.087821963455514e-06,
      "loss": 0.6063,
      "step": 1723
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7527555082525598,
      "learning_rate": 9.086633486002213e-06,
      "loss": 0.608,
      "step": 1724
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7091510522164963,
      "learning_rate": 9.085444312637804e-06,
      "loss": 0.5448,
      "step": 1725
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6896993880892248,
      "learning_rate": 9.08425444356479e-06,
      "loss": 0.6033,
      "step": 1726
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9764830673685948,
      "learning_rate": 9.083063878985793e-06,
      "loss": 0.6537,
      "step": 1727
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.785031296757777,
      "learning_rate": 9.081872619103556e-06,
      "loss": 0.6464,
      "step": 1728
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8211053894022589,
      "learning_rate": 9.080680664120935e-06,
      "loss": 0.6085,
      "step": 1729
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6683698084612498,
      "learning_rate": 9.07948801424091e-06,
      "loss": 0.5627,
      "step": 1730
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7623640773008702,
      "learning_rate": 9.078294669666577e-06,
      "loss": 0.5943,
      "step": 1731
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8322766288219557,
      "learning_rate": 9.077100630601147e-06,
      "loss": 0.5815,
      "step": 1732
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7214948388412612,
      "learning_rate": 9.075905897247955e-06,
      "loss": 0.5807,
      "step": 1733
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.809341957344064,
      "learning_rate": 9.074710469810453e-06,
      "loss": 0.5881,
      "step": 1734
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8104278580302012,
      "learning_rate": 9.073514348492204e-06,
      "loss": 0.6027,
      "step": 1735
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7238055896556168,
      "learning_rate": 9.072317533496905e-06,
      "loss": 0.5938,
      "step": 1736
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7911169611676274,
      "learning_rate": 9.07112002502835e-06,
      "loss": 0.6824,
      "step": 1737
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7938304964870317,
      "learning_rate": 9.06992182329047e-06,
      "loss": 0.5913,
      "step": 1738
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7636631258455324,
      "learning_rate": 9.068722928487302e-06,
      "loss": 0.5965,
      "step": 1739
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6691466688006485,
      "learning_rate": 9.067523340823007e-06,
      "loss": 0.5975,
      "step": 1740
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0070475167764037,
      "learning_rate": 9.066323060501865e-06,
      "loss": 0.6221,
      "step": 1741
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9511724200450111,
      "learning_rate": 9.065122087728267e-06,
      "loss": 0.7474,
      "step": 1742
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7252248964130821,
      "learning_rate": 9.063920422706727e-06,
      "loss": 0.551,
      "step": 1743
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1211443025301424,
      "learning_rate": 9.062718065641875e-06,
      "loss": 0.6351,
      "step": 1744
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7876488197922967,
      "learning_rate": 9.061515016738464e-06,
      "loss": 0.6354,
      "step": 1745
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7531567610715258,
      "learning_rate": 9.060311276201356e-06,
      "loss": 0.571,
      "step": 1746
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8503792231372225,
      "learning_rate": 9.059106844235539e-06,
      "loss": 0.6851,
      "step": 1747
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8001217877679286,
      "learning_rate": 9.057901721046113e-06,
      "loss": 0.5749,
      "step": 1748
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0012384941941663,
      "learning_rate": 9.056695906838302e-06,
      "loss": 0.6708,
      "step": 1749
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8541094011939935,
      "learning_rate": 9.055489401817438e-06,
      "loss": 0.656,
      "step": 1750
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8357890172107632,
      "learning_rate": 9.054282206188978e-06,
      "loss": 0.6269,
      "step": 1751
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6866886328457728,
      "learning_rate": 9.053074320158497e-06,
      "loss": 0.5281,
      "step": 1752
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7532696921676845,
      "learning_rate": 9.051865743931683e-06,
      "loss": 0.5627,
      "step": 1753
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9603772805197949,
      "learning_rate": 9.050656477714345e-06,
      "loss": 0.728,
      "step": 1754
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7283825730725442,
      "learning_rate": 9.049446521712408e-06,
      "loss": 0.511,
      "step": 1755
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8273130154245885,
      "learning_rate": 9.048235876131917e-06,
      "loss": 0.5645,
      "step": 1756
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9765117143848513,
      "learning_rate": 9.047024541179029e-06,
      "loss": 0.662,
      "step": 1757
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9010999202200949,
      "learning_rate": 9.045812517060023e-06,
      "loss": 0.6461,
      "step": 1758
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8374348905093684,
      "learning_rate": 9.044599803981294e-06,
      "loss": 0.6681,
      "step": 1759
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9809866375278119,
      "learning_rate": 9.043386402149355e-06,
      "loss": 0.6289,
      "step": 1760
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8456193304173618,
      "learning_rate": 9.042172311770835e-06,
      "loss": 0.6256,
      "step": 1761
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7710525567112971,
      "learning_rate": 9.040957533052483e-06,
      "loss": 0.6517,
      "step": 1762
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7273969252066462,
      "learning_rate": 9.03974206620116e-06,
      "loss": 0.564,
      "step": 1763
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6965816596919657,
      "learning_rate": 9.03852591142385e-06,
      "loss": 0.5681,
      "step": 1764
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7580748481208671,
      "learning_rate": 9.03730906892765e-06,
      "loss": 0.5378,
      "step": 1765
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7221588641200527,
      "learning_rate": 9.036091538919776e-06,
      "loss": 0.5671,
      "step": 1766
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9371739324533304,
      "learning_rate": 9.03487332160756e-06,
      "loss": 0.6793,
      "step": 1767
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6651522889103836,
      "learning_rate": 9.033654417198452e-06,
      "loss": 0.5774,
      "step": 1768
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7734725954649293,
      "learning_rate": 9.03243482590002e-06,
      "loss": 0.5849,
      "step": 1769
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7610915536999594,
      "learning_rate": 9.031214547919946e-06,
      "loss": 0.5784,
      "step": 1770
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.803952299480627,
      "learning_rate": 9.029993583466033e-06,
      "loss": 0.546,
      "step": 1771
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8000546888386627,
      "learning_rate": 9.028771932746195e-06,
      "loss": 0.5937,
      "step": 1772
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8291186462310567,
      "learning_rate": 9.02754959596847e-06,
      "loss": 0.6171,
      "step": 1773
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0146855901019032,
      "learning_rate": 9.026326573341006e-06,
      "loss": 0.6356,
      "step": 1774
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.702740851301099,
      "learning_rate": 9.025102865072072e-06,
      "loss": 0.6019,
      "step": 1775
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9512756159800014,
      "learning_rate": 9.023878471370056e-06,
      "loss": 0.6472,
      "step": 1776
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9773728277386027,
      "learning_rate": 9.022653392443455e-06,
      "loss": 0.6328,
      "step": 1777
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.878590570708946,
      "learning_rate": 9.02142762850089e-06,
      "loss": 0.6452,
      "step": 1778
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8192464553656723,
      "learning_rate": 9.020201179751094e-06,
      "loss": 0.6292,
      "step": 1779
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7253603418913679,
      "learning_rate": 9.018974046402918e-06,
      "loss": 0.5598,
      "step": 1780
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7347599244829282,
      "learning_rate": 9.017746228665332e-06,
      "loss": 0.5605,
      "step": 1781
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6342052185363972,
      "learning_rate": 9.01651772674742e-06,
      "loss": 0.5442,
      "step": 1782
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3426891750251602,
      "learning_rate": 9.015288540858384e-06,
      "loss": 0.6333,
      "step": 1783
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.753102034929604,
      "learning_rate": 9.01405867120754e-06,
      "loss": 0.5563,
      "step": 1784
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.842818485641212,
      "learning_rate": 9.012828118004322e-06,
      "loss": 0.6807,
      "step": 1785
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8012205485276382,
      "learning_rate": 9.011596881458282e-06,
      "loss": 0.5676,
      "step": 1786
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7302957077071275,
      "learning_rate": 9.010364961779084e-06,
      "loss": 0.6211,
      "step": 1787
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6700622046011556,
      "learning_rate": 9.009132359176514e-06,
      "loss": 0.5806,
      "step": 1788
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8229276054442922,
      "learning_rate": 9.00789907386047e-06,
      "loss": 0.6697,
      "step": 1789
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6765304334818132,
      "learning_rate": 9.006665106040967e-06,
      "loss": 0.5546,
      "step": 1790
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8205021457066586,
      "learning_rate": 9.005430455928137e-06,
      "loss": 0.5767,
      "step": 1791
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8226513884576943,
      "learning_rate": 9.004195123732229e-06,
      "loss": 0.6665,
      "step": 1792
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5984631525126797,
      "learning_rate": 9.002959109663607e-06,
      "loss": 0.4969,
      "step": 1793
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7540334707310535,
      "learning_rate": 9.001722413932749e-06,
      "loss": 0.5493,
      "step": 1794
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7725749445519982,
      "learning_rate": 9.000485036750258e-06,
      "loss": 0.5964,
      "step": 1795
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7516895027875931,
      "learning_rate": 8.99924697832684e-06,
      "loss": 0.5661,
      "step": 1796
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6852269325010806,
      "learning_rate": 8.998008238873323e-06,
      "loss": 0.5695,
      "step": 1797
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7228251665479638,
      "learning_rate": 8.996768818600657e-06,
      "loss": 0.5431,
      "step": 1798
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9622573445433384,
      "learning_rate": 8.995528717719899e-06,
      "loss": 0.6439,
      "step": 1799
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9910610099206466,
      "learning_rate": 8.994287936442226e-06,
      "loss": 0.6857,
      "step": 1800
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7672007423183251,
      "learning_rate": 8.993046474978927e-06,
      "loss": 0.6335,
      "step": 1801
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6598043353365445,
      "learning_rate": 8.991804333541414e-06,
      "loss": 0.5456,
      "step": 1802
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.486939874109273,
      "learning_rate": 8.990561512341209e-06,
      "loss": 0.6917,
      "step": 1803
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6741282238255342,
      "learning_rate": 8.989318011589953e-06,
      "loss": 0.5372,
      "step": 1804
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9392212109917977,
      "learning_rate": 8.988073831499399e-06,
      "loss": 0.6905,
      "step": 1805
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8339190627582693,
      "learning_rate": 8.986828972281419e-06,
      "loss": 0.5738,
      "step": 1806
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7192273804772419,
      "learning_rate": 8.985583434148e-06,
      "loss": 0.5996,
      "step": 1807
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9992670727070457,
      "learning_rate": 8.984337217311243e-06,
      "loss": 0.6253,
      "step": 1808
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7980599567684804,
      "learning_rate": 8.983090321983366e-06,
      "loss": 0.6055,
      "step": 1809
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8102957500803778,
      "learning_rate": 8.981842748376703e-06,
      "loss": 0.641,
      "step": 1810
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.159963884646134,
      "learning_rate": 8.980594496703704e-06,
      "loss": 0.7301,
      "step": 1811
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8683511509988018,
      "learning_rate": 8.97934556717693e-06,
      "loss": 0.6774,
      "step": 1812
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6996519858164988,
      "learning_rate": 8.978095960009063e-06,
      "loss": 0.5469,
      "step": 1813
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7021957454306138,
      "learning_rate": 8.976845675412898e-06,
      "loss": 0.5955,
      "step": 1814
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8300163218034267,
      "learning_rate": 8.975594713601344e-06,
      "loss": 0.6219,
      "step": 1815
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8599336208706447,
      "learning_rate": 8.974343074787428e-06,
      "loss": 0.6434,
      "step": 1816
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7045302061021801,
      "learning_rate": 8.973090759184292e-06,
      "loss": 0.5583,
      "step": 1817
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8920187062229696,
      "learning_rate": 8.971837767005189e-06,
      "loss": 0.6269,
      "step": 1818
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.776925122013056,
      "learning_rate": 8.970584098463495e-06,
      "loss": 0.5931,
      "step": 1819
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7999708058542344,
      "learning_rate": 8.969329753772694e-06,
      "loss": 0.5413,
      "step": 1820
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9039186862186455,
      "learning_rate": 8.968074733146386e-06,
      "loss": 0.6863,
      "step": 1821
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8842326024923454,
      "learning_rate": 8.966819036798292e-06,
      "loss": 0.6523,
      "step": 1822
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8641127840864393,
      "learning_rate": 8.96556266494224e-06,
      "loss": 0.6172,
      "step": 1823
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7052621432635187,
      "learning_rate": 8.96430561779218e-06,
      "loss": 0.5831,
      "step": 1824
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.803996299174675,
      "learning_rate": 8.963047895562174e-06,
      "loss": 0.6715,
      "step": 1825
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.82726587738202,
      "learning_rate": 8.961789498466398e-06,
      "loss": 0.634,
      "step": 1826
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7321807215769772,
      "learning_rate": 8.960530426719143e-06,
      "loss": 0.55,
      "step": 1827
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8171572146780561,
      "learning_rate": 8.959270680534817e-06,
      "loss": 0.6361,
      "step": 1828
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7719496052869266,
      "learning_rate": 8.958010260127943e-06,
      "loss": 0.5752,
      "step": 1829
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6335785748131367,
      "learning_rate": 8.956749165713154e-06,
      "loss": 0.5056,
      "step": 1830
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7282550491907417,
      "learning_rate": 8.955487397505203e-06,
      "loss": 0.5757,
      "step": 1831
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7236594269106582,
      "learning_rate": 8.954224955718957e-06,
      "loss": 0.5752,
      "step": 1832
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.784011160239139,
      "learning_rate": 8.952961840569397e-06,
      "loss": 0.635,
      "step": 1833
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8652267571371319,
      "learning_rate": 8.951698052271617e-06,
      "loss": 0.6014,
      "step": 1834
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7044958461739276,
      "learning_rate": 8.950433591040826e-06,
      "loss": 0.5222,
      "step": 1835
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7727336242658484,
      "learning_rate": 8.94916845709235e-06,
      "loss": 0.5573,
      "step": 1836
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8327714610213779,
      "learning_rate": 8.94790265064163e-06,
      "loss": 0.6279,
      "step": 1837
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8263280956362629,
      "learning_rate": 8.946636171904214e-06,
      "loss": 0.6368,
      "step": 1838
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7042401870432945,
      "learning_rate": 8.945369021095775e-06,
      "loss": 0.5422,
      "step": 1839
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7334952707005559,
      "learning_rate": 8.944101198432095e-06,
      "loss": 0.553,
      "step": 1840
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8826243897435279,
      "learning_rate": 8.94283270412907e-06,
      "loss": 0.6735,
      "step": 1841
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.651128765798537,
      "learning_rate": 8.941563538402711e-06,
      "loss": 0.525,
      "step": 1842
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6543518611513667,
      "learning_rate": 8.940293701469145e-06,
      "loss": 0.4693,
      "step": 1843
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7884774038528798,
      "learning_rate": 8.939023193544611e-06,
      "loss": 0.6058,
      "step": 1844
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7757565788220621,
      "learning_rate": 8.937752014845465e-06,
      "loss": 0.5616,
      "step": 1845
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9395952151613353,
      "learning_rate": 8.936480165588174e-06,
      "loss": 0.6875,
      "step": 1846
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.830597098331696,
      "learning_rate": 8.935207645989318e-06,
      "loss": 0.6265,
      "step": 1847
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7788980540091662,
      "learning_rate": 8.933934456265599e-06,
      "loss": 0.5913,
      "step": 1848
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8883618040930412,
      "learning_rate": 8.932660596633824e-06,
      "loss": 0.7046,
      "step": 1849
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7456816563417507,
      "learning_rate": 8.931386067310919e-06,
      "loss": 0.5248,
      "step": 1850
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8936548297518376,
      "learning_rate": 8.930110868513927e-06,
      "loss": 0.6754,
      "step": 1851
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7254390140125193,
      "learning_rate": 8.928835000459994e-06,
      "loss": 0.556,
      "step": 1852
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8332523834069967,
      "learning_rate": 8.927558463366393e-06,
      "loss": 0.621,
      "step": 1853
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7278124760817184,
      "learning_rate": 8.9262812574505e-06,
      "loss": 0.5773,
      "step": 1854
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8343111125701007,
      "learning_rate": 8.925003382929816e-06,
      "loss": 0.6167,
      "step": 1855
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7610900922685739,
      "learning_rate": 8.923724840021945e-06,
      "loss": 0.4841,
      "step": 1856
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8924361275608047,
      "learning_rate": 8.92244562894461e-06,
      "loss": 0.6835,
      "step": 1857
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9062770758503514,
      "learning_rate": 8.921165749915647e-06,
      "loss": 0.7286,
      "step": 1858
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9391657111707795,
      "learning_rate": 8.919885203153012e-06,
      "loss": 0.7292,
      "step": 1859
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.691653049080268,
      "learning_rate": 8.918603988874758e-06,
      "loss": 0.5435,
      "step": 1860
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6736304768430776,
      "learning_rate": 8.917322107299073e-06,
      "loss": 0.5695,
      "step": 1861
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.079970504137458,
      "learning_rate": 8.916039558644244e-06,
      "loss": 0.7181,
      "step": 1862
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7031316216922165,
      "learning_rate": 8.91475634312867e-06,
      "loss": 0.5656,
      "step": 1863
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.970306401055872,
      "learning_rate": 8.913472460970878e-06,
      "loss": 0.6201,
      "step": 1864
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7128104168366728,
      "learning_rate": 8.912187912389496e-06,
      "loss": 0.5623,
      "step": 1865
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7749996497785987,
      "learning_rate": 8.91090269760327e-06,
      "loss": 0.6453,
      "step": 1866
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.830955404676633,
      "learning_rate": 8.909616816831056e-06,
      "loss": 0.6385,
      "step": 1867
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9974152088693675,
      "learning_rate": 8.90833027029183e-06,
      "loss": 0.6511,
      "step": 1868
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7839959446154744,
      "learning_rate": 8.907043058204674e-06,
      "loss": 0.5792,
      "step": 1869
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6862329878214277,
      "learning_rate": 8.905755180788788e-06,
      "loss": 0.5807,
      "step": 1870
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.30580512635051,
      "learning_rate": 8.904466638263488e-06,
      "loss": 0.6334,
      "step": 1871
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8340907851346976,
      "learning_rate": 8.903177430848193e-06,
      "loss": 0.6501,
      "step": 1872
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.72020565012702,
      "learning_rate": 8.901887558762446e-06,
      "loss": 0.5641,
      "step": 1873
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8439023503821602,
      "learning_rate": 8.900597022225896e-06,
      "loss": 0.5621,
      "step": 1874
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.770583617352573,
      "learning_rate": 8.89930582145831e-06,
      "loss": 0.5813,
      "step": 1875
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8259131544552497,
      "learning_rate": 8.898013956679566e-06,
      "loss": 0.5932,
      "step": 1876
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7522258289376651,
      "learning_rate": 8.896721428109654e-06,
      "loss": 0.5499,
      "step": 1877
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6484770788920831,
      "learning_rate": 8.895428235968677e-06,
      "loss": 0.5207,
      "step": 1878
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9698414328940885,
      "learning_rate": 8.894134380476856e-06,
      "loss": 0.6768,
      "step": 1879
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7966621324568566,
      "learning_rate": 8.892839861854517e-06,
      "loss": 0.6589,
      "step": 1880
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7419699525830253,
      "learning_rate": 8.891544680322104e-06,
      "loss": 0.5834,
      "step": 1881
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.902361146062035,
      "learning_rate": 8.890248836100177e-06,
      "loss": 0.6265,
      "step": 1882
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.749160813780667,
      "learning_rate": 8.8889523294094e-06,
      "loss": 0.598,
      "step": 1883
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0104706830256829,
      "learning_rate": 8.887655160470556e-06,
      "loss": 0.7211,
      "step": 1884
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8059422574537072,
      "learning_rate": 8.886357329504538e-06,
      "loss": 0.5739,
      "step": 1885
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9080984138864188,
      "learning_rate": 8.885058836732357e-06,
      "loss": 0.6633,
      "step": 1886
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8006087752635738,
      "learning_rate": 8.883759682375128e-06,
      "loss": 0.6584,
      "step": 1887
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8747630444494514,
      "learning_rate": 8.882459866654087e-06,
      "loss": 0.6383,
      "step": 1888
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7102603370532257,
      "learning_rate": 8.88115938979058e-06,
      "loss": 0.5734,
      "step": 1889
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6597827751743804,
      "learning_rate": 8.879858252006059e-06,
      "loss": 0.5551,
      "step": 1890
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8325595763530709,
      "learning_rate": 8.8785564535221e-06,
      "loss": 0.6503,
      "step": 1891
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7679174860187776,
      "learning_rate": 8.877253994560381e-06,
      "loss": 0.5856,
      "step": 1892
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0062543563982924,
      "learning_rate": 8.875950875342702e-06,
      "loss": 0.705,
      "step": 1893
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8122397657471703,
      "learning_rate": 8.87464709609097e-06,
      "loss": 0.6204,
      "step": 1894
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9111254883782234,
      "learning_rate": 8.8733426570272e-06,
      "loss": 0.6752,
      "step": 1895
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7652815044790452,
      "learning_rate": 8.872037558373528e-06,
      "loss": 0.5467,
      "step": 1896
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6484927033413499,
      "learning_rate": 8.870731800352199e-06,
      "loss": 0.5948,
      "step": 1897
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9539929020240348,
      "learning_rate": 8.869425383185571e-06,
      "loss": 0.7258,
      "step": 1898
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.756187742469807,
      "learning_rate": 8.86811830709611e-06,
      "loss": 0.5285,
      "step": 1899
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7696346106774211,
      "learning_rate": 8.866810572306398e-06,
      "loss": 0.5931,
      "step": 1900
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6536423669380937,
      "learning_rate": 8.865502179039132e-06,
      "loss": 0.506,
      "step": 1901
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.706241097957143,
      "learning_rate": 8.864193127517114e-06,
      "loss": 0.5268,
      "step": 1902
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6289214904022691,
      "learning_rate": 8.862883417963264e-06,
      "loss": 0.5271,
      "step": 1903
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6635196321111894,
      "learning_rate": 8.861573050600611e-06,
      "loss": 0.5283,
      "step": 1904
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7934363408931292,
      "learning_rate": 8.860262025652296e-06,
      "loss": 0.6488,
      "step": 1905
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0827572303624151,
      "learning_rate": 8.858950343341574e-06,
      "loss": 0.61,
      "step": 1906
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8718578971625196,
      "learning_rate": 8.857638003891812e-06,
      "loss": 0.6515,
      "step": 1907
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7607392679233425,
      "learning_rate": 8.856325007526486e-06,
      "loss": 0.594,
      "step": 1908
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6656428915012981,
      "learning_rate": 8.855011354469186e-06,
      "loss": 0.5829,
      "step": 1909
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8437637996622123,
      "learning_rate": 8.853697044943614e-06,
      "loss": 0.6929,
      "step": 1910
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6464399577651874,
      "learning_rate": 8.852382079173583e-06,
      "loss": 0.5059,
      "step": 1911
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0195887543999693,
      "learning_rate": 8.851066457383017e-06,
      "loss": 0.672,
      "step": 1912
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8240469623607751,
      "learning_rate": 8.849750179795953e-06,
      "loss": 0.6511,
      "step": 1913
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8413518626973495,
      "learning_rate": 8.848433246636541e-06,
      "loss": 0.5762,
      "step": 1914
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9879940685040388,
      "learning_rate": 8.84711565812904e-06,
      "loss": 0.6609,
      "step": 1915
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0832942220529356,
      "learning_rate": 8.84579741449782e-06,
      "loss": 0.6648,
      "step": 1916
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9730298180017065,
      "learning_rate": 8.844478515967367e-06,
      "loss": 0.6809,
      "step": 1917
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8061031482338935,
      "learning_rate": 8.843158962762273e-06,
      "loss": 0.6742,
      "step": 1918
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9104671072644314,
      "learning_rate": 8.841838755107246e-06,
      "loss": 0.6817,
      "step": 1919
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7060884077682582,
      "learning_rate": 8.840517893227101e-06,
      "loss": 0.5717,
      "step": 1920
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7810269809370137,
      "learning_rate": 8.839196377346772e-06,
      "loss": 0.5964,
      "step": 1921
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9504857012050918,
      "learning_rate": 8.837874207691296e-06,
      "loss": 0.6704,
      "step": 1922
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7884607914060926,
      "learning_rate": 8.836551384485823e-06,
      "loss": 0.6114,
      "step": 1923
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8739752002779261,
      "learning_rate": 8.83522790795562e-06,
      "loss": 0.6485,
      "step": 1924
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.867517280808017,
      "learning_rate": 8.833903778326058e-06,
      "loss": 0.6386,
      "step": 1925
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8877451867182851,
      "learning_rate": 8.832578995822625e-06,
      "loss": 0.5696,
      "step": 1926
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7559521946462631,
      "learning_rate": 8.831253560670916e-06,
      "loss": 0.6256,
      "step": 1927
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6354190981571656,
      "learning_rate": 8.82992747309664e-06,
      "loss": 0.5828,
      "step": 1928
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7795912661459384,
      "learning_rate": 8.828600733325614e-06,
      "loss": 0.5932,
      "step": 1929
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7028386677922696,
      "learning_rate": 8.827273341583772e-06,
      "loss": 0.5791,
      "step": 1930
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6564266314487655,
      "learning_rate": 8.825945298097148e-06,
      "loss": 0.5675,
      "step": 1931
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7149073789063699,
      "learning_rate": 8.8246166030919e-06,
      "loss": 0.4908,
      "step": 1932
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8483383277559516,
      "learning_rate": 8.823287256794291e-06,
      "loss": 0.5936,
      "step": 1933
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9041161346428437,
      "learning_rate": 8.82195725943069e-06,
      "loss": 0.653,
      "step": 1934
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7650128872232571,
      "learning_rate": 8.820626611227585e-06,
      "loss": 0.5333,
      "step": 1935
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.89738252844998,
      "learning_rate": 8.819295312411574e-06,
      "loss": 0.6814,
      "step": 1936
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9622346150592043,
      "learning_rate": 8.817963363209357e-06,
      "loss": 0.6191,
      "step": 1937
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7158658712391445,
      "learning_rate": 8.816630763847756e-06,
      "loss": 0.5994,
      "step": 1938
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9612611211488961,
      "learning_rate": 8.815297514553698e-06,
      "loss": 0.6603,
      "step": 1939
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8063463112974308,
      "learning_rate": 8.81396361555422e-06,
      "loss": 0.6003,
      "step": 1940
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6659642173391825,
      "learning_rate": 8.81262906707647e-06,
      "loss": 0.5648,
      "step": 1941
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7847860001456343,
      "learning_rate": 8.811293869347712e-06,
      "loss": 0.5353,
      "step": 1942
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8425879345472,
      "learning_rate": 8.809958022595312e-06,
      "loss": 0.6594,
      "step": 1943
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8463724987506398,
      "learning_rate": 8.808621527046754e-06,
      "loss": 0.6396,
      "step": 1944
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6834593236146296,
      "learning_rate": 8.807284382929629e-06,
      "loss": 0.5694,
      "step": 1945
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8389541406839688,
      "learning_rate": 8.805946590471635e-06,
      "loss": 0.5769,
      "step": 1946
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6989983001587766,
      "learning_rate": 8.804608149900588e-06,
      "loss": 0.6015,
      "step": 1947
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7093666612422437,
      "learning_rate": 8.80326906144441e-06,
      "loss": 0.5425,
      "step": 1948
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6224127019888014,
      "learning_rate": 8.80192932533113e-06,
      "loss": 0.5313,
      "step": 1949
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6918084500758087,
      "learning_rate": 8.800588941788897e-06,
      "loss": 0.5208,
      "step": 1950
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7703729385959307,
      "learning_rate": 8.79924791104596e-06,
      "loss": 0.5499,
      "step": 1951
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7450610941849798,
      "learning_rate": 8.797906233330686e-06,
      "loss": 0.6096,
      "step": 1952
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.928819808935517,
      "learning_rate": 8.796563908871545e-06,
      "loss": 0.6055,
      "step": 1953
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6956657056066134,
      "learning_rate": 8.795220937897122e-06,
      "loss": 0.5585,
      "step": 1954
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6905355248613142,
      "learning_rate": 8.793877320636114e-06,
      "loss": 0.6038,
      "step": 1955
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7962590954490573,
      "learning_rate": 8.792533057317321e-06,
      "loss": 0.6896,
      "step": 1956
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0255821883870986,
      "learning_rate": 8.79118814816966e-06,
      "loss": 0.6259,
      "step": 1957
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1322157432206734,
      "learning_rate": 8.789842593422152e-06,
      "loss": 0.6198,
      "step": 1958
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7255377115872316,
      "learning_rate": 8.788496393303933e-06,
      "loss": 0.6005,
      "step": 1959
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7339380270816392,
      "learning_rate": 8.787149548044247e-06,
      "loss": 0.5793,
      "step": 1960
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7054192692519413,
      "learning_rate": 8.785802057872447e-06,
      "loss": 0.5719,
      "step": 1961
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3124111360469808,
      "learning_rate": 8.784453923017996e-06,
      "loss": 0.6837,
      "step": 1962
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9634097673238025,
      "learning_rate": 8.783105143710469e-06,
      "loss": 0.6715,
      "step": 1963
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8045861810672543,
      "learning_rate": 8.781755720179548e-06,
      "loss": 0.5531,
      "step": 1964
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7543958285659339,
      "learning_rate": 8.780405652655027e-06,
      "loss": 0.5886,
      "step": 1965
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9289003264134142,
      "learning_rate": 8.779054941366805e-06,
      "loss": 0.6553,
      "step": 1966
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.827454154354535,
      "learning_rate": 8.777703586544897e-06,
      "loss": 0.6391,
      "step": 1967
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9853509429499995,
      "learning_rate": 8.776351588419425e-06,
      "loss": 0.6532,
      "step": 1968
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6937018570118536,
      "learning_rate": 8.774998947220619e-06,
      "loss": 0.5313,
      "step": 1969
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9342839057222853,
      "learning_rate": 8.77364566317882e-06,
      "loss": 0.6544,
      "step": 1970
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6729214109393628,
      "learning_rate": 8.772291736524478e-06,
      "loss": 0.5335,
      "step": 1971
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8305344686297439,
      "learning_rate": 8.770937167488154e-06,
      "loss": 0.598,
      "step": 1972
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6595802411010885,
      "learning_rate": 8.769581956300512e-06,
      "loss": 0.4895,
      "step": 1973
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8545399367197936,
      "learning_rate": 8.768226103192336e-06,
      "loss": 0.6485,
      "step": 1974
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7915246781934338,
      "learning_rate": 8.766869608394512e-06,
      "loss": 0.5833,
      "step": 1975
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.93530170660416,
      "learning_rate": 8.765512472138037e-06,
      "loss": 0.6544,
      "step": 1976
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8310656400882053,
      "learning_rate": 8.764154694654016e-06,
      "loss": 0.6418,
      "step": 1977
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7014259729144102,
      "learning_rate": 8.762796276173667e-06,
      "loss": 0.5694,
      "step": 1978
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0598943972449064,
      "learning_rate": 8.76143721692831e-06,
      "loss": 0.6319,
      "step": 1979
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6898015713549526,
      "learning_rate": 8.760077517149384e-06,
      "loss": 0.5441,
      "step": 1980
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8634276901857205,
      "learning_rate": 8.758717177068427e-06,
      "loss": 0.62,
      "step": 1981
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6935936183248117,
      "learning_rate": 8.757356196917095e-06,
      "loss": 0.5618,
      "step": 1982
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9987592371884148,
      "learning_rate": 8.755994576927145e-06,
      "loss": 0.6239,
      "step": 1983
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7415630301483688,
      "learning_rate": 8.754632317330448e-06,
      "loss": 0.582,
      "step": 1984
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8938029349833673,
      "learning_rate": 8.753269418358983e-06,
      "loss": 0.6617,
      "step": 1985
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8725319532925707,
      "learning_rate": 8.751905880244837e-06,
      "loss": 0.5492,
      "step": 1986
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7396266314568015,
      "learning_rate": 8.750541703220208e-06,
      "loss": 0.5339,
      "step": 1987
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.809179527782143,
      "learning_rate": 8.749176887517399e-06,
      "loss": 0.6409,
      "step": 1988
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0185237417094761,
      "learning_rate": 8.747811433368825e-06,
      "loss": 0.6568,
      "step": 1989
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8025513635725298,
      "learning_rate": 8.746445341007009e-06,
      "loss": 0.6102,
      "step": 1990
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.071151409477144,
      "learning_rate": 8.745078610664578e-06,
      "loss": 0.7174,
      "step": 1991
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7697855789527388,
      "learning_rate": 8.74371124257428e-06,
      "loss": 0.5793,
      "step": 1992
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6946729035521078,
      "learning_rate": 8.742343236968956e-06,
      "loss": 0.5093,
      "step": 1993
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1491730457924645,
      "learning_rate": 8.740974594081569e-06,
      "loss": 0.6336,
      "step": 1994
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7804596962556815,
      "learning_rate": 8.73960531414518e-06,
      "loss": 0.6033,
      "step": 1995
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7455237343297116,
      "learning_rate": 8.738235397392964e-06,
      "loss": 0.563,
      "step": 1996
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6483947551739727,
      "learning_rate": 8.736864844058206e-06,
      "loss": 0.5164,
      "step": 1997
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9117178337068256,
      "learning_rate": 8.735493654374296e-06,
      "loss": 0.6814,
      "step": 1998
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8589054560981413,
      "learning_rate": 8.734121828574731e-06,
      "loss": 0.5597,
      "step": 1999
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8987904440844824,
      "learning_rate": 8.73274936689312e-06,
      "loss": 0.7028,
      "step": 2000
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1207140563730151,
      "learning_rate": 8.73137626956318e-06,
      "loss": 0.6755,
      "step": 2001
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6509964449240633,
      "learning_rate": 8.730002536818736e-06,
      "loss": 0.5016,
      "step": 2002
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7609586453415649,
      "learning_rate": 8.728628168893717e-06,
      "loss": 0.5753,
      "step": 2003
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0883921733033437,
      "learning_rate": 8.727253166022168e-06,
      "loss": 0.6415,
      "step": 2004
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8460363062499621,
      "learning_rate": 8.725877528438234e-06,
      "loss": 0.6809,
      "step": 2005
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8701782027194154,
      "learning_rate": 8.72450125637617e-06,
      "loss": 0.6428,
      "step": 2006
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7733324420419228,
      "learning_rate": 8.723124350070347e-06,
      "loss": 0.5572,
      "step": 2007
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7843190467587263,
      "learning_rate": 8.721746809755235e-06,
      "loss": 0.5808,
      "step": 2008
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6977752975021547,
      "learning_rate": 8.72036863566541e-06,
      "loss": 0.5221,
      "step": 2009
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6500606511990463,
      "learning_rate": 8.71898982803557e-06,
      "loss": 0.5569,
      "step": 2010
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8051197501810271,
      "learning_rate": 8.717610387100501e-06,
      "loss": 0.5888,
      "step": 2011
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1629330578255817,
      "learning_rate": 8.716230313095116e-06,
      "loss": 0.64,
      "step": 2012
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0711681351535367,
      "learning_rate": 8.714849606254424e-06,
      "loss": 0.6834,
      "step": 2013
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6166175941373604,
      "learning_rate": 8.713468266813545e-06,
      "loss": 0.5019,
      "step": 2014
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9143238490482865,
      "learning_rate": 8.712086295007707e-06,
      "loss": 0.6894,
      "step": 2015
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7229066480285856,
      "learning_rate": 8.710703691072245e-06,
      "loss": 0.5706,
      "step": 2016
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9226781049553117,
      "learning_rate": 8.7093204552426e-06,
      "loss": 0.6209,
      "step": 2017
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.765747774540585,
      "learning_rate": 8.707936587754328e-06,
      "loss": 0.6107,
      "step": 2018
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0083131241533017,
      "learning_rate": 8.706552088843083e-06,
      "loss": 0.6685,
      "step": 2019
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7511748652686437,
      "learning_rate": 8.70516695874463e-06,
      "loss": 0.5829,
      "step": 2020
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7984665351737803,
      "learning_rate": 8.703781197694846e-06,
      "loss": 0.5806,
      "step": 2021
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.795803786220064,
      "learning_rate": 8.70239480592971e-06,
      "loss": 0.6181,
      "step": 2022
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8257924125457154,
      "learning_rate": 8.70100778368531e-06,
      "loss": 0.6598,
      "step": 2023
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9263239734883454,
      "learning_rate": 8.699620131197841e-06,
      "loss": 0.717,
      "step": 2024
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.876756335859248,
      "learning_rate": 8.698231848703607e-06,
      "loss": 0.6539,
      "step": 2025
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6942243270989207,
      "learning_rate": 8.696842936439014e-06,
      "loss": 0.5172,
      "step": 2026
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8696010987635775,
      "learning_rate": 8.695453394640585e-06,
      "loss": 0.5492,
      "step": 2027
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9299807553823802,
      "learning_rate": 8.69406322354494e-06,
      "loss": 0.5929,
      "step": 2028
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6878120117457226,
      "learning_rate": 8.692672423388815e-06,
      "loss": 0.5435,
      "step": 2029
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9066210113148992,
      "learning_rate": 8.691280994409044e-06,
      "loss": 0.7152,
      "step": 2030
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6301839089769938,
      "learning_rate": 8.689888936842577e-06,
      "loss": 0.536,
      "step": 2031
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7807011748418788,
      "learning_rate": 8.688496250926462e-06,
      "loss": 0.5718,
      "step": 2032
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8825215192963675,
      "learning_rate": 8.687102936897865e-06,
      "loss": 0.6461,
      "step": 2033
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8887339265334426,
      "learning_rate": 8.685708994994049e-06,
      "loss": 0.6902,
      "step": 2034
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8185013271988214,
      "learning_rate": 8.684314425452387e-06,
      "loss": 0.606,
      "step": 2035
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6625731532622935,
      "learning_rate": 8.682919228510363e-06,
      "loss": 0.5619,
      "step": 2036
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8317143721022817,
      "learning_rate": 8.681523404405563e-06,
      "loss": 0.5546,
      "step": 2037
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8095247090684597,
      "learning_rate": 8.68012695337568e-06,
      "loss": 0.6133,
      "step": 2038
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8171315384294896,
      "learning_rate": 8.678729875658515e-06,
      "loss": 0.6712,
      "step": 2039
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7197046499482802,
      "learning_rate": 8.67733217149198e-06,
      "loss": 0.5778,
      "step": 2040
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7384342032029247,
      "learning_rate": 8.675933841114084e-06,
      "loss": 0.527,
      "step": 2041
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7967624935348813,
      "learning_rate": 8.674534884762952e-06,
      "loss": 0.6029,
      "step": 2042
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.832874454259647,
      "learning_rate": 8.67313530267681e-06,
      "loss": 0.5684,
      "step": 2043
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7255863271746238,
      "learning_rate": 8.671735095093991e-06,
      "loss": 0.5276,
      "step": 2044
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6575818971835546,
      "learning_rate": 8.670334262252937e-06,
      "loss": 0.5409,
      "step": 2045
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8841140027179337,
      "learning_rate": 8.668932804392198e-06,
      "loss": 0.6146,
      "step": 2046
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9824928661209361,
      "learning_rate": 8.667530721750423e-06,
      "loss": 0.6324,
      "step": 2047
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9411398889212877,
      "learning_rate": 8.666128014566375e-06,
      "loss": 0.6287,
      "step": 2048
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7561122228860881,
      "learning_rate": 8.664724683078918e-06,
      "loss": 0.5325,
      "step": 2049
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.859191991494865,
      "learning_rate": 8.663320727527027e-06,
      "loss": 0.6567,
      "step": 2050
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.728821202243004,
      "learning_rate": 8.661916148149782e-06,
      "loss": 0.57,
      "step": 2051
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6767639622734398,
      "learning_rate": 8.660510945186365e-06,
      "loss": 0.5133,
      "step": 2052
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9279525090801648,
      "learning_rate": 8.659105118876068e-06,
      "loss": 0.6907,
      "step": 2053
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2235500374261368,
      "learning_rate": 8.657698669458292e-06,
      "loss": 0.6988,
      "step": 2054
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.902774933098236,
      "learning_rate": 8.656291597172537e-06,
      "loss": 0.6603,
      "step": 2055
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7326495475483635,
      "learning_rate": 8.654883902258414e-06,
      "loss": 0.5314,
      "step": 2056
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9244288847282367,
      "learning_rate": 8.653475584955638e-06,
      "loss": 0.6417,
      "step": 2057
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0393911334648622,
      "learning_rate": 8.652066645504033e-06,
      "loss": 0.7027,
      "step": 2058
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7509214752343563,
      "learning_rate": 8.650657084143525e-06,
      "loss": 0.5611,
      "step": 2059
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6756456299191473,
      "learning_rate": 8.649246901114145e-06,
      "loss": 0.5491,
      "step": 2060
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8305792834618921,
      "learning_rate": 8.64783609665604e-06,
      "loss": 0.603,
      "step": 2061
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7344523217111673,
      "learning_rate": 8.646424671009448e-06,
      "loss": 0.6001,
      "step": 2062
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5961418564054419,
      "learning_rate": 8.645012624414722e-06,
      "loss": 0.4929,
      "step": 2063
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8494504540561819,
      "learning_rate": 8.643599957112321e-06,
      "loss": 0.6588,
      "step": 2064
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6721975809201963,
      "learning_rate": 8.642186669342804e-06,
      "loss": 0.5578,
      "step": 2065
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0692361988277472,
      "learning_rate": 8.640772761346844e-06,
      "loss": 0.6489,
      "step": 2066
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6710279541943218,
      "learning_rate": 8.639358233365211e-06,
      "loss": 0.5862,
      "step": 2067
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.740578187892062,
      "learning_rate": 8.637943085638785e-06,
      "loss": 0.6169,
      "step": 2068
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7688921713149206,
      "learning_rate": 8.636527318408552e-06,
      "loss": 0.6221,
      "step": 2069
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.639425782952498,
      "learning_rate": 8.635110931915602e-06,
      "loss": 0.4929,
      "step": 2070
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7070039780932834,
      "learning_rate": 8.63369392640113e-06,
      "loss": 0.5593,
      "step": 2071
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7868259360319192,
      "learning_rate": 8.632276302106437e-06,
      "loss": 0.5261,
      "step": 2072
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6942834473668794,
      "learning_rate": 8.63085805927293e-06,
      "loss": 0.5338,
      "step": 2073
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7524201881278602,
      "learning_rate": 8.629439198142124e-06,
      "loss": 0.591,
      "step": 2074
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8374464490401248,
      "learning_rate": 8.628019718955633e-06,
      "loss": 0.566,
      "step": 2075
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6701416033371944,
      "learning_rate": 8.626599621955179e-06,
      "loss": 0.5474,
      "step": 2076
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7375947444405496,
      "learning_rate": 8.625178907382593e-06,
      "loss": 0.5714,
      "step": 2077
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8588690128028347,
      "learning_rate": 8.623757575479803e-06,
      "loss": 0.6954,
      "step": 2078
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7770740039038444,
      "learning_rate": 8.622335626488852e-06,
      "loss": 0.5761,
      "step": 2079
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7126757195708782,
      "learning_rate": 8.62091306065188e-06,
      "loss": 0.5358,
      "step": 2080
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6357528170915606,
      "learning_rate": 8.619489878211138e-06,
      "loss": 0.5538,
      "step": 2081
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.773575886183575,
      "learning_rate": 8.618066079408976e-06,
      "loss": 0.5655,
      "step": 2082
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.103543041083038,
      "learning_rate": 8.616641664487856e-06,
      "loss": 0.621,
      "step": 2083
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6786409522469372,
      "learning_rate": 8.615216633690336e-06,
      "loss": 0.5249,
      "step": 2084
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.709366941217034,
      "learning_rate": 8.613790987259085e-06,
      "loss": 0.5657,
      "step": 2085
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9936592780503476,
      "learning_rate": 8.612364725436879e-06,
      "loss": 0.6814,
      "step": 2086
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8070679307666362,
      "learning_rate": 8.610937848466593e-06,
      "loss": 0.5997,
      "step": 2087
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7741105837673501,
      "learning_rate": 8.60951035659121e-06,
      "loss": 0.6954,
      "step": 2088
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8598441994349721,
      "learning_rate": 8.608082250053817e-06,
      "loss": 0.5753,
      "step": 2089
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6871093253471288,
      "learning_rate": 8.606653529097606e-06,
      "loss": 0.5978,
      "step": 2090
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8402978610523177,
      "learning_rate": 8.60522419396587e-06,
      "loss": 0.6622,
      "step": 2091
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8025997635649196,
      "learning_rate": 8.603794244902017e-06,
      "loss": 0.5746,
      "step": 2092
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8079648866775898,
      "learning_rate": 8.602363682149546e-06,
      "loss": 0.5526,
      "step": 2093
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7848806863731498,
      "learning_rate": 8.600932505952068e-06,
      "loss": 0.6087,
      "step": 2094
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9165143762221796,
      "learning_rate": 8.599500716553301e-06,
      "loss": 0.6672,
      "step": 2095
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7098452875125559,
      "learning_rate": 8.598068314197059e-06,
      "loss": 0.5533,
      "step": 2096
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7106573150898514,
      "learning_rate": 8.596635299127269e-06,
      "loss": 0.5544,
      "step": 2097
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6275054786560426,
      "learning_rate": 8.595201671587956e-06,
      "loss": 0.5001,
      "step": 2098
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9032591792337562,
      "learning_rate": 8.593767431823255e-06,
      "loss": 0.6973,
      "step": 2099
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9832893791400691,
      "learning_rate": 8.592332580077398e-06,
      "loss": 0.6853,
      "step": 2100
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7983767797769051,
      "learning_rate": 8.590897116594727e-06,
      "loss": 0.618,
      "step": 2101
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8611346675740268,
      "learning_rate": 8.589461041619687e-06,
      "loss": 0.5636,
      "step": 2102
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8881373293479193,
      "learning_rate": 8.588024355396827e-06,
      "loss": 0.6137,
      "step": 2103
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9065098589149884,
      "learning_rate": 8.586587058170799e-06,
      "loss": 0.634,
      "step": 2104
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9678113091015726,
      "learning_rate": 8.585149150186359e-06,
      "loss": 0.6795,
      "step": 2105
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6585278408268294,
      "learning_rate": 8.583710631688369e-06,
      "loss": 0.4703,
      "step": 2106
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.666075991773359,
      "learning_rate": 8.582271502921794e-06,
      "loss": 0.5236,
      "step": 2107
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8308239244690797,
      "learning_rate": 8.5808317641317e-06,
      "loss": 0.696,
      "step": 2108
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6109566217595789,
      "learning_rate": 8.579391415563263e-06,
      "loss": 0.4739,
      "step": 2109
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6665019209388863,
      "learning_rate": 8.57795045746176e-06,
      "loss": 0.552,
      "step": 2110
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8794364432256047,
      "learning_rate": 8.576508890072565e-06,
      "loss": 0.6399,
      "step": 2111
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7373689901029139,
      "learning_rate": 8.575066713641168e-06,
      "loss": 0.5586,
      "step": 2112
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1756731041277932,
      "learning_rate": 8.573623928413152e-06,
      "loss": 0.6884,
      "step": 2113
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7926003558780015,
      "learning_rate": 8.572180534634214e-06,
      "loss": 0.5712,
      "step": 2114
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8698030506537611,
      "learning_rate": 8.570736532550142e-06,
      "loss": 0.5424,
      "step": 2115
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8811133398906865,
      "learning_rate": 8.569291922406841e-06,
      "loss": 0.688,
      "step": 2116
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8866401031040106,
      "learning_rate": 8.56784670445031e-06,
      "loss": 0.65,
      "step": 2117
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6260009011465755,
      "learning_rate": 8.566400878926653e-06,
      "loss": 0.5443,
      "step": 2118
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8172623392123525,
      "learning_rate": 8.564954446082083e-06,
      "loss": 0.5753,
      "step": 2119
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8622881918556228,
      "learning_rate": 8.56350740616291e-06,
      "loss": 0.625,
      "step": 2120
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9047210862684324,
      "learning_rate": 8.56205975941555e-06,
      "loss": 0.6794,
      "step": 2121
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7415236097481217,
      "learning_rate": 8.560611506086518e-06,
      "loss": 0.576,
      "step": 2122
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8138831101181325,
      "learning_rate": 8.559162646422445e-06,
      "loss": 0.6395,
      "step": 2123
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8191648361935001,
      "learning_rate": 8.557713180670052e-06,
      "loss": 0.5721,
      "step": 2124
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7113374776305473,
      "learning_rate": 8.556263109076168e-06,
      "loss": 0.6063,
      "step": 2125
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.070745353392222,
      "learning_rate": 8.554812431887725e-06,
      "loss": 0.6671,
      "step": 2126
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9631473166536872,
      "learning_rate": 8.55336114935176e-06,
      "loss": 0.5968,
      "step": 2127
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8063882176780597,
      "learning_rate": 8.551909261715408e-06,
      "loss": 0.6017,
      "step": 2128
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3713162586413101,
      "learning_rate": 8.550456769225914e-06,
      "loss": 0.6019,
      "step": 2129
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.895627671462431,
      "learning_rate": 8.549003672130619e-06,
      "loss": 0.7221,
      "step": 2130
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.115825569332588,
      "learning_rate": 8.547549970676972e-06,
      "loss": 0.7126,
      "step": 2131
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6744296830114518,
      "learning_rate": 8.546095665112524e-06,
      "loss": 0.5879,
      "step": 2132
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0542291389317007,
      "learning_rate": 8.544640755684927e-06,
      "loss": 0.6625,
      "step": 2133
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7067083329411854,
      "learning_rate": 8.543185242641938e-06,
      "loss": 0.5762,
      "step": 2134
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8899550250092108,
      "learning_rate": 8.541729126231414e-06,
      "loss": 0.6177,
      "step": 2135
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9358455833625566,
      "learning_rate": 8.540272406701316e-06,
      "loss": 0.6917,
      "step": 2136
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8019144856821593,
      "learning_rate": 8.53881508429971e-06,
      "loss": 0.5475,
      "step": 2137
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7309143066553038,
      "learning_rate": 8.537357159274764e-06,
      "loss": 0.5758,
      "step": 2138
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8548437747811373,
      "learning_rate": 8.535898631874744e-06,
      "loss": 0.6729,
      "step": 2139
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7278671687332745,
      "learning_rate": 8.534439502348025e-06,
      "loss": 0.5547,
      "step": 2140
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8136868999320196,
      "learning_rate": 8.53297977094308e-06,
      "loss": 0.6425,
      "step": 2141
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7242295356994636,
      "learning_rate": 8.531519437908485e-06,
      "loss": 0.5474,
      "step": 2142
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6885692419793408,
      "learning_rate": 8.530058503492921e-06,
      "loss": 0.539,
      "step": 2143
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7968254199191971,
      "learning_rate": 8.528596967945172e-06,
      "loss": 0.6923,
      "step": 2144
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8958611999211049,
      "learning_rate": 8.527134831514116e-06,
      "loss": 0.6885,
      "step": 2145
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9463412415335809,
      "learning_rate": 8.525672094448746e-06,
      "loss": 0.5787,
      "step": 2146
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0415556795944747,
      "learning_rate": 8.524208756998148e-06,
      "loss": 0.5712,
      "step": 2147
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9288608356670949,
      "learning_rate": 8.522744819411515e-06,
      "loss": 0.6482,
      "step": 2148
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8879779258522612,
      "learning_rate": 8.52128028193814e-06,
      "loss": 0.6896,
      "step": 2149
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7540301814467247,
      "learning_rate": 8.519815144827415e-06,
      "loss": 0.5766,
      "step": 2150
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8111217497985278,
      "learning_rate": 8.51834940832884e-06,
      "loss": 0.612,
      "step": 2151
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7946179201424982,
      "learning_rate": 8.516883072692015e-06,
      "loss": 0.5573,
      "step": 2152
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0863149604567819,
      "learning_rate": 8.51541613816664e-06,
      "loss": 0.6631,
      "step": 2153
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.006336809524803,
      "learning_rate": 8.513948605002523e-06,
      "loss": 0.6602,
      "step": 2154
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.779702080676902,
      "learning_rate": 8.512480473449564e-06,
      "loss": 0.6007,
      "step": 2155
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7109553376450497,
      "learning_rate": 8.511011743757776e-06,
      "loss": 0.5873,
      "step": 2156
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9299938710168193,
      "learning_rate": 8.509542416177264e-06,
      "loss": 0.6237,
      "step": 2157
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7855393275143485,
      "learning_rate": 8.508072490958239e-06,
      "loss": 0.5287,
      "step": 2158
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7615302768685427,
      "learning_rate": 8.506601968351018e-06,
      "loss": 0.5529,
      "step": 2159
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2983792695241654,
      "learning_rate": 8.505130848606012e-06,
      "loss": 0.6771,
      "step": 2160
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0047322485772034,
      "learning_rate": 8.50365913197374e-06,
      "loss": 0.6538,
      "step": 2161
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.814560739020013,
      "learning_rate": 8.502186818704818e-06,
      "loss": 0.641,
      "step": 2162
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8877892445975503,
      "learning_rate": 8.500713909049964e-06,
      "loss": 0.6746,
      "step": 2163
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1054896961186658,
      "learning_rate": 8.499240403260005e-06,
      "loss": 0.6849,
      "step": 2164
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.757828476419521,
      "learning_rate": 8.49776630158586e-06,
      "loss": 0.573,
      "step": 2165
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8272076603238785,
      "learning_rate": 8.49629160427855e-06,
      "loss": 0.6607,
      "step": 2166
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0791597327946632,
      "learning_rate": 8.494816311589207e-06,
      "loss": 0.6825,
      "step": 2167
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7144652410692978,
      "learning_rate": 8.493340423769054e-06,
      "loss": 0.5706,
      "step": 2168
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9188422557919373,
      "learning_rate": 8.491863941069418e-06,
      "loss": 0.6679,
      "step": 2169
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7159994207143344,
      "learning_rate": 8.490386863741734e-06,
      "loss": 0.5178,
      "step": 2170
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.858370155632041,
      "learning_rate": 8.488909192037528e-06,
      "loss": 0.526,
      "step": 2171
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9027915665559302,
      "learning_rate": 8.487430926208434e-06,
      "loss": 0.6456,
      "step": 2172
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6498804864319441,
      "learning_rate": 8.485952066506184e-06,
      "loss": 0.5328,
      "step": 2173
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8150303077014934,
      "learning_rate": 8.484472613182616e-06,
      "loss": 0.6359,
      "step": 2174
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.830216916423939,
      "learning_rate": 8.482992566489662e-06,
      "loss": 0.632,
      "step": 2175
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6720845793888305,
      "learning_rate": 8.481511926679359e-06,
      "loss": 0.5683,
      "step": 2176
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8320690779942326,
      "learning_rate": 8.480030694003845e-06,
      "loss": 0.6374,
      "step": 2177
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7880540618495274,
      "learning_rate": 8.478548868715357e-06,
      "loss": 0.6154,
      "step": 2178
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8951410434168142,
      "learning_rate": 8.47706645106624e-06,
      "loss": 0.7151,
      "step": 2179
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8925758755852837,
      "learning_rate": 8.475583441308928e-06,
      "loss": 0.6223,
      "step": 2180
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6324310257727276,
      "learning_rate": 8.474099839695964e-06,
      "loss": 0.5392,
      "step": 2181
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7507939165391605,
      "learning_rate": 8.47261564647999e-06,
      "loss": 0.5969,
      "step": 2182
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.787797717060103,
      "learning_rate": 8.471130861913749e-06,
      "loss": 0.5623,
      "step": 2183
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8845384114704159,
      "learning_rate": 8.469645486250084e-06,
      "loss": 0.6375,
      "step": 2184
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7383754094764808,
      "learning_rate": 8.468159519741938e-06,
      "loss": 0.5635,
      "step": 2185
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8098315815047168,
      "learning_rate": 8.46667296264236e-06,
      "loss": 0.6853,
      "step": 2186
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9490789806280994,
      "learning_rate": 8.46518581520449e-06,
      "loss": 0.6579,
      "step": 2187
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7422561054979356,
      "learning_rate": 8.463698077681577e-06,
      "loss": 0.5551,
      "step": 2188
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7858395228715994,
      "learning_rate": 8.462209750326963e-06,
      "loss": 0.6412,
      "step": 2189
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9364773322415074,
      "learning_rate": 8.4607208333941e-06,
      "loss": 0.6497,
      "step": 2190
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7143508810113824,
      "learning_rate": 8.459231327136532e-06,
      "loss": 0.5127,
      "step": 2191
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2191503626928628,
      "learning_rate": 8.457741231807907e-06,
      "loss": 0.7118,
      "step": 2192
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7736365530016299,
      "learning_rate": 8.456250547661973e-06,
      "loss": 0.608,
      "step": 2193
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8706011162297675,
      "learning_rate": 8.454759274952577e-06,
      "loss": 0.5982,
      "step": 2194
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7150820882839987,
      "learning_rate": 8.453267413933669e-06,
      "loss": 0.5272,
      "step": 2195
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6566668156139139,
      "learning_rate": 8.451774964859296e-06,
      "loss": 0.5797,
      "step": 2196
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7185485218865221,
      "learning_rate": 8.450281927983607e-06,
      "loss": 0.5722,
      "step": 2197
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7063778632695681,
      "learning_rate": 8.44878830356085e-06,
      "loss": 0.5434,
      "step": 2198
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8246634333512559,
      "learning_rate": 8.447294091845373e-06,
      "loss": 0.65,
      "step": 2199
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6656324357775885,
      "learning_rate": 8.445799293091628e-06,
      "loss": 0.5146,
      "step": 2200
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7478585890620928,
      "learning_rate": 8.44430390755416e-06,
      "loss": 0.532,
      "step": 2201
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7583049251737343,
      "learning_rate": 8.44280793548762e-06,
      "loss": 0.5526,
      "step": 2202
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6890454817297638,
      "learning_rate": 8.441311377146754e-06,
      "loss": 0.5538,
      "step": 2203
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7254490833623586,
      "learning_rate": 8.439814232786413e-06,
      "loss": 0.553,
      "step": 2204
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7564763081740711,
      "learning_rate": 8.438316502661544e-06,
      "loss": 0.5945,
      "step": 2205
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7738617156740418,
      "learning_rate": 8.436818187027193e-06,
      "loss": 0.5651,
      "step": 2206
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9692124796333613,
      "learning_rate": 8.435319286138511e-06,
      "loss": 0.6351,
      "step": 2207
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.86735897283853,
      "learning_rate": 8.433819800250744e-06,
      "loss": 0.667,
      "step": 2208
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2284460624112894,
      "learning_rate": 8.432319729619237e-06,
      "loss": 0.6638,
      "step": 2209
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6791999343574601,
      "learning_rate": 8.430819074499437e-06,
      "loss": 0.5608,
      "step": 2210
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7325113456201207,
      "learning_rate": 8.429317835146894e-06,
      "loss": 0.5556,
      "step": 2211
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6633786406712496,
      "learning_rate": 8.427816011817248e-06,
      "loss": 0.606,
      "step": 2212
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.777425623423544,
      "learning_rate": 8.426313604766244e-06,
      "loss": 0.6582,
      "step": 2213
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8435551586757459,
      "learning_rate": 8.42481061424973e-06,
      "loss": 0.6145,
      "step": 2214
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.697428677223934,
      "learning_rate": 8.423307040523647e-06,
      "loss": 0.5103,
      "step": 2215
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7217355041361375,
      "learning_rate": 8.421802883844039e-06,
      "loss": 0.584,
      "step": 2216
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7151265216951087,
      "learning_rate": 8.42029814446705e-06,
      "loss": 0.5571,
      "step": 2217
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0484622953667317,
      "learning_rate": 8.418792822648916e-06,
      "loss": 0.6418,
      "step": 2218
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.722841475289307,
      "learning_rate": 8.417286918645983e-06,
      "loss": 0.5698,
      "step": 2219
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8564957188519151,
      "learning_rate": 8.41578043271469e-06,
      "loss": 0.7363,
      "step": 2220
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9328001249275764,
      "learning_rate": 8.414273365111573e-06,
      "loss": 0.6782,
      "step": 2221
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7156457025596012,
      "learning_rate": 8.412765716093273e-06,
      "loss": 0.5452,
      "step": 2222
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8842615329605965,
      "learning_rate": 8.411257485916521e-06,
      "loss": 0.6599,
      "step": 2223
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7209672383047154,
      "learning_rate": 8.409748674838162e-06,
      "loss": 0.5264,
      "step": 2224
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2792306948369405,
      "learning_rate": 8.408239283115125e-06,
      "loss": 0.649,
      "step": 2225
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7170581910545252,
      "learning_rate": 8.406729311004445e-06,
      "loss": 0.5572,
      "step": 2226
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1869669307617312,
      "learning_rate": 8.405218758763254e-06,
      "loss": 0.5861,
      "step": 2227
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9366676348224308,
      "learning_rate": 8.403707626648782e-06,
      "loss": 0.6616,
      "step": 2228
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9093745631113076,
      "learning_rate": 8.402195914918364e-06,
      "loss": 0.6157,
      "step": 2229
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9514606703634301,
      "learning_rate": 8.400683623829424e-06,
      "loss": 0.7011,
      "step": 2230
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.849618490958349,
      "learning_rate": 8.399170753639491e-06,
      "loss": 0.6317,
      "step": 2231
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8293474873591513,
      "learning_rate": 8.39765730460619e-06,
      "loss": 0.6104,
      "step": 2232
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7281890059835165,
      "learning_rate": 8.396143276987248e-06,
      "loss": 0.5497,
      "step": 2233
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6626105277162515,
      "learning_rate": 8.394628671040486e-06,
      "loss": 0.5255,
      "step": 2234
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7165942482190107,
      "learning_rate": 8.393113487023828e-06,
      "loss": 0.5581,
      "step": 2235
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7369761403757858,
      "learning_rate": 8.39159772519529e-06,
      "loss": 0.5464,
      "step": 2236
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8380325066830904,
      "learning_rate": 8.390081385812993e-06,
      "loss": 0.6802,
      "step": 2237
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7614642265833558,
      "learning_rate": 8.388564469135155e-06,
      "loss": 0.5304,
      "step": 2238
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.811008603365971,
      "learning_rate": 8.38704697542009e-06,
      "loss": 0.6395,
      "step": 2239
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8465238608175868,
      "learning_rate": 8.385528904926209e-06,
      "loss": 0.5597,
      "step": 2240
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7472055324454421,
      "learning_rate": 8.384010257912026e-06,
      "loss": 0.5564,
      "step": 2241
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7080499019280277,
      "learning_rate": 8.382491034636151e-06,
      "loss": 0.5721,
      "step": 2242
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6533341600262631,
      "learning_rate": 8.380971235357292e-06,
      "loss": 0.5327,
      "step": 2243
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.939988898220049,
      "learning_rate": 8.379450860334253e-06,
      "loss": 0.6453,
      "step": 2244
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.80635138992417,
      "learning_rate": 8.377929909825942e-06,
      "loss": 0.5354,
      "step": 2245
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9563518196570222,
      "learning_rate": 8.376408384091354e-06,
      "loss": 0.6526,
      "step": 2246
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9389607629729652,
      "learning_rate": 8.374886283389597e-06,
      "loss": 0.6261,
      "step": 2247
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8393685822603515,
      "learning_rate": 8.373363607979865e-06,
      "loss": 0.5596,
      "step": 2248
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7701859541160611,
      "learning_rate": 8.371840358121452e-06,
      "loss": 0.5583,
      "step": 2249
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6863315195615804,
      "learning_rate": 8.370316534073754e-06,
      "loss": 0.5918,
      "step": 2250
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7404503845856089,
      "learning_rate": 8.368792136096265e-06,
      "loss": 0.5826,
      "step": 2251
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1365211872584071,
      "learning_rate": 8.367267164448569e-06,
      "loss": 0.6445,
      "step": 2252
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6686324976583061,
      "learning_rate": 8.365741619390353e-06,
      "loss": 0.573,
      "step": 2253
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6669577025316342,
      "learning_rate": 8.364215501181405e-06,
      "loss": 0.56,
      "step": 2254
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7743203676057054,
      "learning_rate": 8.362688810081605e-06,
      "loss": 0.5571,
      "step": 2255
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6463048592606979,
      "learning_rate": 8.361161546350934e-06,
      "loss": 0.4956,
      "step": 2256
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6923095161337977,
      "learning_rate": 8.359633710249467e-06,
      "loss": 0.5083,
      "step": 2257
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.70794842662976,
      "learning_rate": 8.35810530203738e-06,
      "loss": 0.6492,
      "step": 2258
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9378970275223311,
      "learning_rate": 8.356576321974943e-06,
      "loss": 0.5743,
      "step": 2259
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9333887998052925,
      "learning_rate": 8.355046770322528e-06,
      "loss": 0.6174,
      "step": 2260
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.944990552343333,
      "learning_rate": 8.353516647340601e-06,
      "loss": 0.6071,
      "step": 2261
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7379723372631333,
      "learning_rate": 8.351985953289725e-06,
      "loss": 0.5888,
      "step": 2262
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0192604728391195,
      "learning_rate": 8.350454688430563e-06,
      "loss": 0.6379,
      "step": 2263
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8257305988462342,
      "learning_rate": 8.348922853023872e-06,
      "loss": 0.6037,
      "step": 2264
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.650025901664581,
      "learning_rate": 8.347390447330507e-06,
      "loss": 0.5053,
      "step": 2265
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8250901334042201,
      "learning_rate": 8.345857471611423e-06,
      "loss": 0.5287,
      "step": 2266
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7989820808565055,
      "learning_rate": 8.344323926127666e-06,
      "loss": 0.5452,
      "step": 2267
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8819463352133603,
      "learning_rate": 8.342789811140388e-06,
      "loss": 0.5485,
      "step": 2268
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7577632775211289,
      "learning_rate": 8.341255126910827e-06,
      "loss": 0.5688,
      "step": 2269
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9011162908581771,
      "learning_rate": 8.33971987370033e-06,
      "loss": 0.655,
      "step": 2270
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9716433692526733,
      "learning_rate": 8.33818405177033e-06,
      "loss": 0.6465,
      "step": 2271
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8084445685435369,
      "learning_rate": 8.336647661382364e-06,
      "loss": 0.6368,
      "step": 2272
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7236691898606522,
      "learning_rate": 8.335110702798061e-06,
      "loss": 0.5251,
      "step": 2273
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1478440821584148,
      "learning_rate": 8.333573176279151e-06,
      "loss": 0.5994,
      "step": 2274
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7554411600406318,
      "learning_rate": 8.332035082087457e-06,
      "loss": 0.5746,
      "step": 2275
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9094963471762483,
      "learning_rate": 8.330496420484904e-06,
      "loss": 0.6002,
      "step": 2276
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8301530304303779,
      "learning_rate": 8.328957191733504e-06,
      "loss": 0.6467,
      "step": 2277
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8254058430689047,
      "learning_rate": 8.327417396095377e-06,
      "loss": 0.647,
      "step": 2278
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8778719242595915,
      "learning_rate": 8.325877033832732e-06,
      "loss": 0.6989,
      "step": 2279
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0124746902448545,
      "learning_rate": 8.324336105207878e-06,
      "loss": 0.6239,
      "step": 2280
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8055130580747742,
      "learning_rate": 8.322794610483216e-06,
      "loss": 0.6248,
      "step": 2281
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1166269802304396,
      "learning_rate": 8.321252549921248e-06,
      "loss": 0.6788,
      "step": 2282
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7585865734303134,
      "learning_rate": 8.319709923784573e-06,
      "loss": 0.5829,
      "step": 2283
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8856231181748737,
      "learning_rate": 8.318166732335884e-06,
      "loss": 0.6253,
      "step": 2284
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8972869238426218,
      "learning_rate": 8.316622975837966e-06,
      "loss": 0.694,
      "step": 2285
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9705902192043009,
      "learning_rate": 8.315078654553708e-06,
      "loss": 0.6195,
      "step": 2286
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7017308734996244,
      "learning_rate": 8.313533768746091e-06,
      "loss": 0.5384,
      "step": 2287
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0124319702531217,
      "learning_rate": 8.311988318678194e-06,
      "loss": 0.6089,
      "step": 2288
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7586416027830107,
      "learning_rate": 8.310442304613191e-06,
      "loss": 0.5994,
      "step": 2289
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8293707163254753,
      "learning_rate": 8.308895726814351e-06,
      "loss": 0.6004,
      "step": 2290
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7028920180702602,
      "learning_rate": 8.30734858554504e-06,
      "loss": 0.5458,
      "step": 2291
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7471238512523348,
      "learning_rate": 8.305800881068723e-06,
      "loss": 0.6993,
      "step": 2292
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8885433476333644,
      "learning_rate": 8.304252613648951e-06,
      "loss": 0.6521,
      "step": 2293
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9279617129950012,
      "learning_rate": 8.302703783549386e-06,
      "loss": 0.7125,
      "step": 2294
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8296039701038359,
      "learning_rate": 8.301154391033772e-06,
      "loss": 0.6176,
      "step": 2295
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.965010297378682,
      "learning_rate": 8.299604436365959e-06,
      "loss": 0.5779,
      "step": 2296
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.693672460330611,
      "learning_rate": 8.298053919809883e-06,
      "loss": 0.5199,
      "step": 2297
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.833706684496561,
      "learning_rate": 8.296502841629583e-06,
      "loss": 0.5847,
      "step": 2298
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8793859250764897,
      "learning_rate": 8.294951202089193e-06,
      "loss": 0.6586,
      "step": 2299
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.754715350764544,
      "learning_rate": 8.293399001452941e-06,
      "loss": 0.5095,
      "step": 2300
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8577411749327027,
      "learning_rate": 8.291846239985149e-06,
      "loss": 0.5888,
      "step": 2301
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0214445655733677,
      "learning_rate": 8.290292917950236e-06,
      "loss": 0.7071,
      "step": 2302
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7875576995868492,
      "learning_rate": 8.288739035612717e-06,
      "loss": 0.6436,
      "step": 2303
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6641594132293381,
      "learning_rate": 8.287184593237201e-06,
      "loss": 0.5595,
      "step": 2304
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7055590548372608,
      "learning_rate": 8.285629591088395e-06,
      "loss": 0.5394,
      "step": 2305
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8464940831198519,
      "learning_rate": 8.2840740294311e-06,
      "loss": 0.5961,
      "step": 2306
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.642164613007127,
      "learning_rate": 8.28251790853021e-06,
      "loss": 0.4882,
      "step": 2307
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.788710077963095,
      "learning_rate": 8.280961228650717e-06,
      "loss": 0.5749,
      "step": 2308
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6686451690887508,
      "learning_rate": 8.279403990057709e-06,
      "loss": 0.5973,
      "step": 2309
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8393608167533704,
      "learning_rate": 8.277846193016365e-06,
      "loss": 0.617,
      "step": 2310
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8723931358823335,
      "learning_rate": 8.276287837791964e-06,
      "loss": 0.5834,
      "step": 2311
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7358226441137731,
      "learning_rate": 8.274728924649875e-06,
      "loss": 0.5282,
      "step": 2312
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7070045408010385,
      "learning_rate": 8.273169453855567e-06,
      "loss": 0.5765,
      "step": 2313
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6809552001776816,
      "learning_rate": 8.271609425674599e-06,
      "loss": 0.5126,
      "step": 2314
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.673713676762806,
      "learning_rate": 8.27004884037263e-06,
      "loss": 0.5998,
      "step": 2315
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7309875676978831,
      "learning_rate": 8.268487698215411e-06,
      "loss": 0.5706,
      "step": 2316
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7287597480529917,
      "learning_rate": 8.266925999468788e-06,
      "loss": 0.5583,
      "step": 2317
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6670359682652354,
      "learning_rate": 8.265363744398701e-06,
      "loss": 0.5383,
      "step": 2318
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6799576113854798,
      "learning_rate": 8.263800933271186e-06,
      "loss": 0.51,
      "step": 2319
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.662289019802724,
      "learning_rate": 8.262237566352376e-06,
      "loss": 0.5019,
      "step": 2320
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8462811630433921,
      "learning_rate": 8.260673643908495e-06,
      "loss": 0.6381,
      "step": 2321
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6189393677821143,
      "learning_rate": 8.25910916620586e-06,
      "loss": 0.5138,
      "step": 2322
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7925747174293012,
      "learning_rate": 8.257544133510887e-06,
      "loss": 0.5927,
      "step": 2323
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0187331849280385,
      "learning_rate": 8.255978546090085e-06,
      "loss": 0.6674,
      "step": 2324
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7066273633781068,
      "learning_rate": 8.254412404210056e-06,
      "loss": 0.5002,
      "step": 2325
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6916764410112194,
      "learning_rate": 8.252845708137502e-06,
      "loss": 0.5348,
      "step": 2326
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8730779608962944,
      "learning_rate": 8.25127845813921e-06,
      "loss": 0.6412,
      "step": 2327
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8922911818776277,
      "learning_rate": 8.249710654482067e-06,
      "loss": 0.6298,
      "step": 2328
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2167202797266854,
      "learning_rate": 8.248142297433058e-06,
      "loss": 0.6156,
      "step": 2329
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.764427728420986,
      "learning_rate": 8.246573387259253e-06,
      "loss": 0.581,
      "step": 2330
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.795355208072967,
      "learning_rate": 8.245003924227821e-06,
      "loss": 0.593,
      "step": 2331
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6713290959598207,
      "learning_rate": 8.243433908606029e-06,
      "loss": 0.5503,
      "step": 2332
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7684664533688154,
      "learning_rate": 8.24186334066123e-06,
      "loss": 0.6085,
      "step": 2333
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6375726576552677,
      "learning_rate": 8.240292220660881e-06,
      "loss": 0.5144,
      "step": 2334
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9105157625967314,
      "learning_rate": 8.238720548872523e-06,
      "loss": 0.6389,
      "step": 2335
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8387875591422956,
      "learning_rate": 8.237148325563795e-06,
      "loss": 0.6228,
      "step": 2336
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6983564368234124,
      "learning_rate": 8.23557555100243e-06,
      "loss": 0.56,
      "step": 2337
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7232134892748823,
      "learning_rate": 8.234002225456259e-06,
      "loss": 0.6053,
      "step": 2338
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7311475337497146,
      "learning_rate": 8.232428349193201e-06,
      "loss": 0.5067,
      "step": 2339
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8301871923589819,
      "learning_rate": 8.23085392248127e-06,
      "loss": 0.6399,
      "step": 2340
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1605622126682777,
      "learning_rate": 8.229278945588574e-06,
      "loss": 0.6874,
      "step": 2341
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7515228924479411,
      "learning_rate": 8.227703418783315e-06,
      "loss": 0.5694,
      "step": 2342
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7367554868434087,
      "learning_rate": 8.226127342333791e-06,
      "loss": 0.5751,
      "step": 2343
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1195425747080676,
      "learning_rate": 8.224550716508391e-06,
      "loss": 0.7148,
      "step": 2344
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9707320690293301,
      "learning_rate": 8.222973541575596e-06,
      "loss": 0.6403,
      "step": 2345
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7795332197889939,
      "learning_rate": 8.221395817803983e-06,
      "loss": 0.5996,
      "step": 2346
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.799105812375165,
      "learning_rate": 8.219817545462223e-06,
      "loss": 0.6068,
      "step": 2347
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0960130367747722,
      "learning_rate": 8.218238724819079e-06,
      "loss": 0.6576,
      "step": 2348
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.758546164841495,
      "learning_rate": 8.216659356143409e-06,
      "loss": 0.6209,
      "step": 2349
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9558347850765019,
      "learning_rate": 8.215079439704159e-06,
      "loss": 0.6516,
      "step": 2350
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8539978001923682,
      "learning_rate": 8.213498975770376e-06,
      "loss": 0.5873,
      "step": 2351
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8760203002805484,
      "learning_rate": 8.211917964611197e-06,
      "loss": 0.7309,
      "step": 2352
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9270815012110396,
      "learning_rate": 8.210336406495846e-06,
      "loss": 0.5724,
      "step": 2353
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7070308310026626,
      "learning_rate": 8.208754301693653e-06,
      "loss": 0.5629,
      "step": 2354
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.912194533535536,
      "learning_rate": 8.207171650474031e-06,
      "loss": 0.6649,
      "step": 2355
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8121180500589926,
      "learning_rate": 8.20558845310649e-06,
      "loss": 0.559,
      "step": 2356
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9841246901195542,
      "learning_rate": 8.20400470986063e-06,
      "loss": 0.6554,
      "step": 2357
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9320436045071452,
      "learning_rate": 8.202420421006145e-06,
      "loss": 0.6312,
      "step": 2358
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6420075735054956,
      "learning_rate": 8.200835586812827e-06,
      "loss": 0.5281,
      "step": 2359
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.882644943412692,
      "learning_rate": 8.199250207550553e-06,
      "loss": 0.6597,
      "step": 2360
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8829214326658736,
      "learning_rate": 8.1976642834893e-06,
      "loss": 0.6419,
      "step": 2361
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7359545705589885,
      "learning_rate": 8.196077814899133e-06,
      "loss": 0.6393,
      "step": 2362
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8197324154581176,
      "learning_rate": 8.194490802050207e-06,
      "loss": 0.7147,
      "step": 2363
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7461906058265823,
      "learning_rate": 8.192903245212783e-06,
      "loss": 0.5423,
      "step": 2364
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7084348334767367,
      "learning_rate": 8.191315144657195e-06,
      "loss": 0.5623,
      "step": 2365
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7559982367656779,
      "learning_rate": 8.189726500653885e-06,
      "loss": 0.6456,
      "step": 2366
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8946669145049603,
      "learning_rate": 8.188137313473385e-06,
      "loss": 0.7122,
      "step": 2367
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1765562511966037,
      "learning_rate": 8.186547583386313e-06,
      "loss": 0.6626,
      "step": 2368
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7738727324336859,
      "learning_rate": 8.184957310663383e-06,
      "loss": 0.6152,
      "step": 2369
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.93837278706296,
      "learning_rate": 8.183366495575407e-06,
      "loss": 0.6255,
      "step": 2370
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7344891325672279,
      "learning_rate": 8.18177513839328e-06,
      "loss": 0.6352,
      "step": 2371
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8200370488893685,
      "learning_rate": 8.180183239387993e-06,
      "loss": 0.6523,
      "step": 2372
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8367676183361057,
      "learning_rate": 8.178590798830632e-06,
      "loss": 0.6063,
      "step": 2373
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.679871202151405,
      "learning_rate": 8.176997816992371e-06,
      "loss": 0.594,
      "step": 2374
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8633656632553506,
      "learning_rate": 8.175404294144482e-06,
      "loss": 0.6361,
      "step": 2375
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7516707451458304,
      "learning_rate": 8.173810230558321e-06,
      "loss": 0.5853,
      "step": 2376
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6695076153640627,
      "learning_rate": 8.172215626505342e-06,
      "loss": 0.5495,
      "step": 2377
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7809538479366334,
      "learning_rate": 8.170620482257091e-06,
      "loss": 0.5839,
      "step": 2378
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6874530502127046,
      "learning_rate": 8.169024798085203e-06,
      "loss": 0.5773,
      "step": 2379
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7855193159842875,
      "learning_rate": 8.167428574261405e-06,
      "loss": 0.6299,
      "step": 2380
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8657778906017942,
      "learning_rate": 8.16583181105752e-06,
      "loss": 0.6432,
      "step": 2381
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6966905917563424,
      "learning_rate": 8.16423450874546e-06,
      "loss": 0.5877,
      "step": 2382
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6483608972281321,
      "learning_rate": 8.162636667597227e-06,
      "loss": 0.5026,
      "step": 2383
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8628910982855723,
      "learning_rate": 8.161038287884917e-06,
      "loss": 0.6635,
      "step": 2384
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.784810443655044,
      "learning_rate": 8.15943936988072e-06,
      "loss": 0.5554,
      "step": 2385
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8347137420278369,
      "learning_rate": 8.15783991385691e-06,
      "loss": 0.6538,
      "step": 2386
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7984138701956977,
      "learning_rate": 8.156239920085863e-06,
      "loss": 0.5875,
      "step": 2387
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7000855657456476,
      "learning_rate": 8.154639388840037e-06,
      "loss": 0.5287,
      "step": 2388
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9287400096514483,
      "learning_rate": 8.15303832039199e-06,
      "loss": 0.6628,
      "step": 2389
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7305778435529804,
      "learning_rate": 8.151436715014365e-06,
      "loss": 0.5841,
      "step": 2390
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.87603672255926,
      "learning_rate": 8.149834572979898e-06,
      "loss": 0.5822,
      "step": 2391
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7445847748529795,
      "learning_rate": 8.148231894561419e-06,
      "loss": 0.5502,
      "step": 2392
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.709503236081038,
      "learning_rate": 8.146628680031846e-06,
      "loss": 0.5739,
      "step": 2393
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.691261548060571,
      "learning_rate": 8.145024929664189e-06,
      "loss": 0.5187,
      "step": 2394
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6685156980087661,
      "learning_rate": 8.143420643731552e-06,
      "loss": 0.6101,
      "step": 2395
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7061501973708891,
      "learning_rate": 8.141815822507127e-06,
      "loss": 0.5506,
      "step": 2396
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.797025716621591,
      "learning_rate": 8.1402104662642e-06,
      "loss": 0.6619,
      "step": 2397
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8455375934747812,
      "learning_rate": 8.138604575276143e-06,
      "loss": 0.695,
      "step": 2398
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7057355020350179,
      "learning_rate": 8.136998149816426e-06,
      "loss": 0.5822,
      "step": 2399
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6883007978932285,
      "learning_rate": 8.135391190158604e-06,
      "loss": 0.5306,
      "step": 2400
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1434335361647263,
      "learning_rate": 8.133783696576328e-06,
      "loss": 0.7271,
      "step": 2401
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9714201781879716,
      "learning_rate": 8.132175669343334e-06,
      "loss": 0.7127,
      "step": 2402
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6714920337965887,
      "learning_rate": 8.130567108733456e-06,
      "loss": 0.5244,
      "step": 2403
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7969258703829071,
      "learning_rate": 8.128958015020615e-06,
      "loss": 0.5628,
      "step": 2404
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6185889595083937,
      "learning_rate": 8.127348388478817e-06,
      "loss": 0.4908,
      "step": 2405
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6214940054709104,
      "learning_rate": 8.125738229382172e-06,
      "loss": 0.5299,
      "step": 2406
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6481716959167375,
      "learning_rate": 8.124127538004868e-06,
      "loss": 0.5694,
      "step": 2407
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8864681221105619,
      "learning_rate": 8.122516314621194e-06,
      "loss": 0.6452,
      "step": 2408
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6381478772664615,
      "learning_rate": 8.120904559505521e-06,
      "loss": 0.5594,
      "step": 2409
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6471198915573602,
      "learning_rate": 8.119292272932313e-06,
      "loss": 0.5577,
      "step": 2410
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9268059780946492,
      "learning_rate": 8.117679455176129e-06,
      "loss": 0.5967,
      "step": 2411
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.671799763037132,
      "learning_rate": 8.116066106511614e-06,
      "loss": 0.5659,
      "step": 2412
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8725333658708626,
      "learning_rate": 8.114452227213504e-06,
      "loss": 0.6023,
      "step": 2413
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7179522017545644,
      "learning_rate": 8.112837817556622e-06,
      "loss": 0.5396,
      "step": 2414
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.956163501847583,
      "learning_rate": 8.111222877815893e-06,
      "loss": 0.6173,
      "step": 2415
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6409828894773908,
      "learning_rate": 8.10960740826632e-06,
      "loss": 0.5101,
      "step": 2416
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.73718109085478,
      "learning_rate": 8.107991409183e-06,
      "loss": 0.5397,
      "step": 2417
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7972367992417654,
      "learning_rate": 8.106374880841122e-06,
      "loss": 0.609,
      "step": 2418
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0962035078179866,
      "learning_rate": 8.104757823515963e-06,
      "loss": 0.6826,
      "step": 2419
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8999997635617961,
      "learning_rate": 8.103140237482893e-06,
      "loss": 0.607,
      "step": 2420
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7633029670947203,
      "learning_rate": 8.10152212301737e-06,
      "loss": 0.5786,
      "step": 2421
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8641991505547325,
      "learning_rate": 8.09990348039494e-06,
      "loss": 0.6185,
      "step": 2422
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6753976871296726,
      "learning_rate": 8.09828430989124e-06,
      "loss": 0.5593,
      "step": 2423
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3453134465895997,
      "learning_rate": 8.096664611782e-06,
      "loss": 0.6555,
      "step": 2424
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6283961223166129,
      "learning_rate": 8.09504438634304e-06,
      "loss": 0.5187,
      "step": 2425
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8093742672212454,
      "learning_rate": 8.093423633850264e-06,
      "loss": 0.6538,
      "step": 2426
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7708124981369415,
      "learning_rate": 8.09180235457967e-06,
      "loss": 0.5891,
      "step": 2427
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8635398702545797,
      "learning_rate": 8.090180548807345e-06,
      "loss": 0.6664,
      "step": 2428
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8315911275206549,
      "learning_rate": 8.088558216809466e-06,
      "loss": 0.6638,
      "step": 2429
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9211971888802181,
      "learning_rate": 8.086935358862298e-06,
      "loss": 0.7258,
      "step": 2430
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6950157196344704,
      "learning_rate": 8.085311975242197e-06,
      "loss": 0.5816,
      "step": 2431
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8014085377556148,
      "learning_rate": 8.083688066225612e-06,
      "loss": 0.6427,
      "step": 2432
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5068086702278343,
      "learning_rate": 8.082063632089074e-06,
      "loss": 0.6234,
      "step": 2433
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0546594702262562,
      "learning_rate": 8.080438673109208e-06,
      "loss": 0.5803,
      "step": 2434
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7247851083692188,
      "learning_rate": 8.078813189562726e-06,
      "loss": 0.5556,
      "step": 2435
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9504107091406003,
      "learning_rate": 8.077187181726435e-06,
      "loss": 0.6036,
      "step": 2436
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7090271250440942,
      "learning_rate": 8.075560649877225e-06,
      "loss": 0.5441,
      "step": 2437
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9825085742197898,
      "learning_rate": 8.073933594292076e-06,
      "loss": 0.6299,
      "step": 2438
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8955139156672639,
      "learning_rate": 8.07230601524806e-06,
      "loss": 0.6486,
      "step": 2439
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1949433903896163,
      "learning_rate": 8.070677913022337e-06,
      "loss": 0.675,
      "step": 2440
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7250620937795847,
      "learning_rate": 8.069049287892156e-06,
      "loss": 0.5646,
      "step": 2441
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6894576515771151,
      "learning_rate": 8.067420140134853e-06,
      "loss": 0.5368,
      "step": 2442
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9022798191370169,
      "learning_rate": 8.065790470027858e-06,
      "loss": 0.651,
      "step": 2443
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6533579034497073,
      "learning_rate": 8.064160277848683e-06,
      "loss": 0.5669,
      "step": 2444
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8932472826635292,
      "learning_rate": 8.062529563874936e-06,
      "loss": 0.6869,
      "step": 2445
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8594734332609194,
      "learning_rate": 8.06089832838431e-06,
      "loss": 0.6371,
      "step": 2446
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8717000748731434,
      "learning_rate": 8.059266571654587e-06,
      "loss": 0.5834,
      "step": 2447
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6808450695348428,
      "learning_rate": 8.057634293963638e-06,
      "loss": 0.5526,
      "step": 2448
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7191729396321469,
      "learning_rate": 8.056001495589423e-06,
      "loss": 0.55,
      "step": 2449
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.757146767073832,
      "learning_rate": 8.054368176809989e-06,
      "loss": 0.5379,
      "step": 2450
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.637281541931795,
      "learning_rate": 8.052734337903475e-06,
      "loss": 0.5413,
      "step": 2451
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9473006707620221,
      "learning_rate": 8.051099979148107e-06,
      "loss": 0.688,
      "step": 2452
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9958919657111568,
      "learning_rate": 8.049465100822199e-06,
      "loss": 0.6938,
      "step": 2453
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6531987531351056,
      "learning_rate": 8.047829703204153e-06,
      "loss": 0.5455,
      "step": 2454
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7334787860884254,
      "learning_rate": 8.04619378657246e-06,
      "loss": 0.6432,
      "step": 2455
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7826736634056337,
      "learning_rate": 8.044557351205701e-06,
      "loss": 0.5645,
      "step": 2456
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8053293519694975,
      "learning_rate": 8.042920397382545e-06,
      "loss": 0.5477,
      "step": 2457
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9279243409733561,
      "learning_rate": 8.041282925381744e-06,
      "loss": 0.6708,
      "step": 2458
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7008979599487158,
      "learning_rate": 8.039644935482146e-06,
      "loss": 0.5515,
      "step": 2459
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8258988150015644,
      "learning_rate": 8.03800642796268e-06,
      "loss": 0.5415,
      "step": 2460
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7278754952623744,
      "learning_rate": 8.03636740310237e-06,
      "loss": 0.5939,
      "step": 2461
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8589164665755079,
      "learning_rate": 8.034727861180325e-06,
      "loss": 0.6438,
      "step": 2462
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6572701827942246,
      "learning_rate": 8.033087802475738e-06,
      "loss": 0.5033,
      "step": 2463
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6961925136844825,
      "learning_rate": 8.031447227267899e-06,
      "loss": 0.5616,
      "step": 2464
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.729922186668422,
      "learning_rate": 8.029806135836179e-06,
      "loss": 0.5667,
      "step": 2465
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6895181293323871,
      "learning_rate": 8.028164528460038e-06,
      "loss": 0.5536,
      "step": 2466
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8285958684695105,
      "learning_rate": 8.026522405419024e-06,
      "loss": 0.6401,
      "step": 2467
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8209750437508495,
      "learning_rate": 8.024879766992775e-06,
      "loss": 0.5852,
      "step": 2468
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8493622450233307,
      "learning_rate": 8.023236613461015e-06,
      "loss": 0.6564,
      "step": 2469
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6537449367996405,
      "learning_rate": 8.021592945103556e-06,
      "loss": 0.5613,
      "step": 2470
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.696603681320776,
      "learning_rate": 8.019948762200296e-06,
      "loss": 0.6168,
      "step": 2471
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7477846669270151,
      "learning_rate": 8.018304065031224e-06,
      "loss": 0.5684,
      "step": 2472
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7825913723700246,
      "learning_rate": 8.016658853876414e-06,
      "loss": 0.6334,
      "step": 2473
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6292176457888397,
      "learning_rate": 8.015013129016028e-06,
      "loss": 0.5221,
      "step": 2474
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8695789854112458,
      "learning_rate": 8.013366890730317e-06,
      "loss": 0.6467,
      "step": 2475
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.829989412277105,
      "learning_rate": 8.011720139299616e-06,
      "loss": 0.7213,
      "step": 2476
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9624281026563432,
      "learning_rate": 8.010072875004353e-06,
      "loss": 0.6433,
      "step": 2477
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7952852016003795,
      "learning_rate": 8.00842509812504e-06,
      "loss": 0.5972,
      "step": 2478
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6316703587172406,
      "learning_rate": 8.00677680894227e-06,
      "loss": 0.505,
      "step": 2479
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.80696919028627,
      "learning_rate": 8.005128007736735e-06,
      "loss": 0.6106,
      "step": 2480
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8837760032650265,
      "learning_rate": 8.003478694789209e-06,
      "loss": 0.6366,
      "step": 2481
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8860760198477845,
      "learning_rate": 8.00182887038055e-06,
      "loss": 0.7186,
      "step": 2482
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7813272915667994,
      "learning_rate": 8.00017853479171e-06,
      "loss": 0.5827,
      "step": 2483
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9872855712179546,
      "learning_rate": 7.998527688303717e-06,
      "loss": 0.6395,
      "step": 2484
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7117307247343383,
      "learning_rate": 7.9968763311977e-06,
      "loss": 0.5503,
      "step": 2485
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7423196946569927,
      "learning_rate": 7.995224463754864e-06,
      "loss": 0.5885,
      "step": 2486
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8474004325991016,
      "learning_rate": 7.993572086256505e-06,
      "loss": 0.5756,
      "step": 2487
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7914307702218254,
      "learning_rate": 7.991919198984008e-06,
      "loss": 0.6561,
      "step": 2488
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6602603759540381,
      "learning_rate": 7.990265802218841e-06,
      "loss": 0.5606,
      "step": 2489
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7346298851190962,
      "learning_rate": 7.98861189624256e-06,
      "loss": 0.6043,
      "step": 2490
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9204659928277439,
      "learning_rate": 7.986957481336807e-06,
      "loss": 0.5506,
      "step": 2491
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8077814019873388,
      "learning_rate": 7.985302557783313e-06,
      "loss": 0.6049,
      "step": 2492
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9393025971159807,
      "learning_rate": 7.983647125863896e-06,
      "loss": 0.6196,
      "step": 2493
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7132024985184515,
      "learning_rate": 7.981991185860455e-06,
      "loss": 0.6178,
      "step": 2494
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6914206090420442,
      "learning_rate": 7.980334738054981e-06,
      "loss": 0.5919,
      "step": 2495
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.79141234078157,
      "learning_rate": 7.97867778272955e-06,
      "loss": 0.5487,
      "step": 2496
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8503196578459087,
      "learning_rate": 7.977020320166326e-06,
      "loss": 0.6473,
      "step": 2497
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9325666661039893,
      "learning_rate": 7.975362350647555e-06,
      "loss": 0.6439,
      "step": 2498
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6462498372332367,
      "learning_rate": 7.973703874455572e-06,
      "loss": 0.567,
      "step": 2499
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7122219694881514,
      "learning_rate": 7.972044891872801e-06,
      "loss": 0.5,
      "step": 2500
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7732411768919837,
      "learning_rate": 7.970385403181747e-06,
      "loss": 0.5749,
      "step": 2501
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8991509340178808,
      "learning_rate": 7.968725408665002e-06,
      "loss": 0.5696,
      "step": 2502
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7761067203987014,
      "learning_rate": 7.967064908605248e-06,
      "loss": 0.5586,
      "step": 2503
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8454747433864903,
      "learning_rate": 7.965403903285254e-06,
      "loss": 0.584,
      "step": 2504
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6238616540530499,
      "learning_rate": 7.963742392987866e-06,
      "loss": 0.5403,
      "step": 2505
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8425217019040718,
      "learning_rate": 7.962080377996025e-06,
      "loss": 0.5984,
      "step": 2506
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8158533627725938,
      "learning_rate": 7.960417858592754e-06,
      "loss": 0.6689,
      "step": 2507
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9281872264549673,
      "learning_rate": 7.958754835061163e-06,
      "loss": 0.6559,
      "step": 2508
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8827825259015066,
      "learning_rate": 7.957091307684448e-06,
      "loss": 0.6167,
      "step": 2509
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8150032148072945,
      "learning_rate": 7.95542727674589e-06,
      "loss": 0.6545,
      "step": 2510
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6698875166365278,
      "learning_rate": 7.953762742528857e-06,
      "loss": 0.528,
      "step": 2511
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7163075212287794,
      "learning_rate": 7.952097705316801e-06,
      "loss": 0.6209,
      "step": 2512
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7621034156605405,
      "learning_rate": 7.95043216539326e-06,
      "loss": 0.5259,
      "step": 2513
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6826220118007611,
      "learning_rate": 7.948766123041857e-06,
      "loss": 0.5081,
      "step": 2514
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8563664187016866,
      "learning_rate": 7.947099578546306e-06,
      "loss": 0.6523,
      "step": 2515
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8670941949433756,
      "learning_rate": 7.945432532190399e-06,
      "loss": 0.6752,
      "step": 2516
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.877482645838485,
      "learning_rate": 7.943764984258016e-06,
      "loss": 0.6668,
      "step": 2517
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.899598297456225,
      "learning_rate": 7.942096935033123e-06,
      "loss": 0.655,
      "step": 2518
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7832144233710502,
      "learning_rate": 7.940428384799774e-06,
      "loss": 0.6215,
      "step": 2519
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7310364127174431,
      "learning_rate": 7.938759333842104e-06,
      "loss": 0.6088,
      "step": 2520
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7229043869381659,
      "learning_rate": 7.937089782444336e-06,
      "loss": 0.5604,
      "step": 2521
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.960971123990552,
      "learning_rate": 7.935419730890777e-06,
      "loss": 0.6554,
      "step": 2522
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7503397463291318,
      "learning_rate": 7.933749179465818e-06,
      "loss": 0.6154,
      "step": 2523
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9152919278244538,
      "learning_rate": 7.932078128453935e-06,
      "loss": 0.6879,
      "step": 2524
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8546445546531586,
      "learning_rate": 7.930406578139696e-06,
      "loss": 0.6837,
      "step": 2525
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8380134771721715,
      "learning_rate": 7.928734528807743e-06,
      "loss": 0.6548,
      "step": 2526
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7924546971997317,
      "learning_rate": 7.927061980742813e-06,
      "loss": 0.5924,
      "step": 2527
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6821382187546627,
      "learning_rate": 7.925388934229722e-06,
      "loss": 0.524,
      "step": 2528
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6883932018359311,
      "learning_rate": 7.92371538955337e-06,
      "loss": 0.5638,
      "step": 2529
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8231348135440547,
      "learning_rate": 7.922041346998746e-06,
      "loss": 0.5555,
      "step": 2530
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.60726826564693,
      "learning_rate": 7.920366806850924e-06,
      "loss": 0.5225,
      "step": 2531
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7899820717684318,
      "learning_rate": 7.918691769395059e-06,
      "loss": 0.5634,
      "step": 2532
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8267961106501716,
      "learning_rate": 7.91701623491639e-06,
      "loss": 0.6542,
      "step": 2533
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7075588782026214,
      "learning_rate": 7.915340203700248e-06,
      "loss": 0.5534,
      "step": 2534
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7568078402070907,
      "learning_rate": 7.91366367603204e-06,
      "loss": 0.6049,
      "step": 2535
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6559061209506393,
      "learning_rate": 7.911986652197263e-06,
      "loss": 0.507,
      "step": 2536
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8298837424790314,
      "learning_rate": 7.910309132481497e-06,
      "loss": 0.6374,
      "step": 2537
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8974549372589856,
      "learning_rate": 7.908631117170402e-06,
      "loss": 0.5655,
      "step": 2538
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6551685153596557,
      "learning_rate": 7.906952606549732e-06,
      "loss": 0.5331,
      "step": 2539
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8996817778918511,
      "learning_rate": 7.905273600905317e-06,
      "loss": 0.5992,
      "step": 2540
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6614138677097244,
      "learning_rate": 7.903594100523074e-06,
      "loss": 0.4518,
      "step": 2541
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6637871928608828,
      "learning_rate": 7.901914105689007e-06,
      "loss": 0.5384,
      "step": 2542
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7325192268408454,
      "learning_rate": 7.900233616689196e-06,
      "loss": 0.5604,
      "step": 2543
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7146358875951913,
      "learning_rate": 7.898552633809819e-06,
      "loss": 0.5827,
      "step": 2544
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7205279077807724,
      "learning_rate": 7.896871157337123e-06,
      "loss": 0.5487,
      "step": 2545
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7275630420373732,
      "learning_rate": 7.895189187557447e-06,
      "loss": 0.5678,
      "step": 2546
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6610505797029942,
      "learning_rate": 7.893506724757215e-06,
      "loss": 0.5356,
      "step": 2547
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6632201705253028,
      "learning_rate": 7.891823769222935e-06,
      "loss": 0.5015,
      "step": 2548
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7311937552921745,
      "learning_rate": 7.890140321241191e-06,
      "loss": 0.5673,
      "step": 2549
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5856291982631545,
      "learning_rate": 7.88845638109866e-06,
      "loss": 0.496,
      "step": 2550
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8342177773865391,
      "learning_rate": 7.886771949082099e-06,
      "loss": 0.5613,
      "step": 2551
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9125768272986223,
      "learning_rate": 7.885087025478349e-06,
      "loss": 0.6657,
      "step": 2552
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6725524406232308,
      "learning_rate": 7.883401610574338e-06,
      "loss": 0.5629,
      "step": 2553
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8386107467603523,
      "learning_rate": 7.881715704657069e-06,
      "loss": 0.6623,
      "step": 2554
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6312653752267728,
      "learning_rate": 7.880029308013636e-06,
      "loss": 0.4883,
      "step": 2555
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9468639063815736,
      "learning_rate": 7.878342420931218e-06,
      "loss": 0.648,
      "step": 2556
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6443327226938073,
      "learning_rate": 7.87665504369707e-06,
      "loss": 0.5272,
      "step": 2557
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6679444751410026,
      "learning_rate": 7.87496717659854e-06,
      "loss": 0.4959,
      "step": 2558
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7396494512111418,
      "learning_rate": 7.873278819923047e-06,
      "loss": 0.548,
      "step": 2559
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8451742426384206,
      "learning_rate": 7.871589973958108e-06,
      "loss": 0.6472,
      "step": 2560
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8730460960016027,
      "learning_rate": 7.869900638991312e-06,
      "loss": 0.6132,
      "step": 2561
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7674033136995915,
      "learning_rate": 7.868210815310333e-06,
      "loss": 0.6321,
      "step": 2562
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7491348633229841,
      "learning_rate": 7.866520503202936e-06,
      "loss": 0.5676,
      "step": 2563
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8737321680038387,
      "learning_rate": 7.86482970295696e-06,
      "loss": 0.6832,
      "step": 2564
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8625761542615646,
      "learning_rate": 7.863138414860329e-06,
      "loss": 0.6435,
      "step": 2565
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6893637920448848,
      "learning_rate": 7.861446639201055e-06,
      "loss": 0.5706,
      "step": 2566
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8259452774939636,
      "learning_rate": 7.85975437626723e-06,
      "loss": 0.647,
      "step": 2567
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7399999907724197,
      "learning_rate": 7.858061626347027e-06,
      "loss": 0.53,
      "step": 2568
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8105529923507633,
      "learning_rate": 7.856368389728703e-06,
      "loss": 0.6058,
      "step": 2569
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7992308674734614,
      "learning_rate": 7.8546746667006e-06,
      "loss": 0.6668,
      "step": 2570
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7824888784709066,
      "learning_rate": 7.852980457551141e-06,
      "loss": 0.553,
      "step": 2571
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8133884195215355,
      "learning_rate": 7.851285762568832e-06,
      "loss": 0.5796,
      "step": 2572
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6657023880623612,
      "learning_rate": 7.849590582042264e-06,
      "loss": 0.521,
      "step": 2573
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6955755630905265,
      "learning_rate": 7.847894916260105e-06,
      "loss": 0.5466,
      "step": 2574
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7488901507444516,
      "learning_rate": 7.84619876551111e-06,
      "loss": 0.6626,
      "step": 2575
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6487639796479855,
      "learning_rate": 7.844502130084118e-06,
      "loss": 0.4951,
      "step": 2576
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6568164520554413,
      "learning_rate": 7.842805010268048e-06,
      "loss": 0.5083,
      "step": 2577
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9332953661221384,
      "learning_rate": 7.8411074063519e-06,
      "loss": 0.666,
      "step": 2578
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9396683036746047,
      "learning_rate": 7.83940931862476e-06,
      "loss": 0.6999,
      "step": 2579
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6311110308948953,
      "learning_rate": 7.837710747375791e-06,
      "loss": 0.5263,
      "step": 2580
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9121499150221932,
      "learning_rate": 7.836011692894248e-06,
      "loss": 0.607,
      "step": 2581
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.891271255627214,
      "learning_rate": 7.834312155469457e-06,
      "loss": 0.6441,
      "step": 2582
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7215152006747193,
      "learning_rate": 7.832612135390834e-06,
      "loss": 0.5639,
      "step": 2583
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6482651857803328,
      "learning_rate": 7.830911632947874e-06,
      "loss": 0.5489,
      "step": 2584
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9852562284290151,
      "learning_rate": 7.829210648430156e-06,
      "loss": 0.6084,
      "step": 2585
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6424167107507144,
      "learning_rate": 7.82750918212734e-06,
      "loss": 0.5218,
      "step": 2586
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7007343923974474,
      "learning_rate": 7.825807234329164e-06,
      "loss": 0.5362,
      "step": 2587
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7282085946018152,
      "learning_rate": 7.824104805325456e-06,
      "loss": 0.5064,
      "step": 2588
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8006967154599658,
      "learning_rate": 7.822401895406122e-06,
      "loss": 0.4994,
      "step": 2589
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8297582044554498,
      "learning_rate": 7.820698504861148e-06,
      "loss": 0.6315,
      "step": 2590
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6279406260961634,
      "learning_rate": 7.818994633980606e-06,
      "loss": 0.4887,
      "step": 2591
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7815953777160795,
      "learning_rate": 7.817290283054645e-06,
      "loss": 0.5239,
      "step": 2592
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6659167228692938,
      "learning_rate": 7.8155854523735e-06,
      "loss": 0.5618,
      "step": 2593
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9146085296280144,
      "learning_rate": 7.813880142227484e-06,
      "loss": 0.5996,
      "step": 2594
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7767742804006268,
      "learning_rate": 7.812174352906994e-06,
      "loss": 0.6261,
      "step": 2595
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9300097972902812,
      "learning_rate": 7.810468084702509e-06,
      "loss": 0.5839,
      "step": 2596
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7701663458094374,
      "learning_rate": 7.80876133790459e-06,
      "loss": 0.5762,
      "step": 2597
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9072707067662261,
      "learning_rate": 7.807054112803873e-06,
      "loss": 0.647,
      "step": 2598
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7365818612922479,
      "learning_rate": 7.805346409691087e-06,
      "loss": 0.5958,
      "step": 2599
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2363445052865716,
      "learning_rate": 7.803638228857033e-06,
      "loss": 0.6707,
      "step": 2600
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7119821784150594,
      "learning_rate": 7.801929570592596e-06,
      "loss": 0.5439,
      "step": 2601
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6598534839689683,
      "learning_rate": 7.800220435188745e-06,
      "loss": 0.5131,
      "step": 2602
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8441879458884096,
      "learning_rate": 7.798510822936524e-06,
      "loss": 0.5649,
      "step": 2603
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7967374297291847,
      "learning_rate": 7.796800734127067e-06,
      "loss": 0.6435,
      "step": 2604
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6684031028252342,
      "learning_rate": 7.79509016905158e-06,
      "loss": 0.5081,
      "step": 2605
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6965217701833712,
      "learning_rate": 7.793379128001357e-06,
      "loss": 0.5189,
      "step": 2606
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7937954420407329,
      "learning_rate": 7.79166761126777e-06,
      "loss": 0.6644,
      "step": 2607
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7879431334860658,
      "learning_rate": 7.789955619142274e-06,
      "loss": 0.5937,
      "step": 2608
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9502952661475601,
      "learning_rate": 7.788243151916398e-06,
      "loss": 0.7026,
      "step": 2609
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7153132251224847,
      "learning_rate": 7.786530209881764e-06,
      "loss": 0.5298,
      "step": 2610
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6663791801858784,
      "learning_rate": 7.784816793330061e-06,
      "loss": 0.5836,
      "step": 2611
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8051280634364077,
      "learning_rate": 7.783102902553074e-06,
      "loss": 0.5348,
      "step": 2612
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7296436912052862,
      "learning_rate": 7.781388537842655e-06,
      "loss": 0.6022,
      "step": 2613
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5982526136725665,
      "learning_rate": 7.779673699490745e-06,
      "loss": 0.5316,
      "step": 2614
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8859201612994138,
      "learning_rate": 7.777958387789363e-06,
      "loss": 0.6874,
      "step": 2615
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.741578824655153,
      "learning_rate": 7.776242603030607e-06,
      "loss": 0.589,
      "step": 2616
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7815747806402055,
      "learning_rate": 7.77452634550666e-06,
      "loss": 0.5679,
      "step": 2617
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9514552926162047,
      "learning_rate": 7.772809615509778e-06,
      "loss": 0.636,
      "step": 2618
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9527660278391786,
      "learning_rate": 7.77109241333231e-06,
      "loss": 0.5986,
      "step": 2619
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7058014015975158,
      "learning_rate": 7.76937473926667e-06,
      "loss": 0.5832,
      "step": 2620
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7616638271468806,
      "learning_rate": 7.767656593605361e-06,
      "loss": 0.5832,
      "step": 2621
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8496370494565307,
      "learning_rate": 7.765937976640971e-06,
      "loss": 0.6752,
      "step": 2622
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6780181517885135,
      "learning_rate": 7.764218888666157e-06,
      "loss": 0.5719,
      "step": 2623
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.917307437254972,
      "learning_rate": 7.762499329973665e-06,
      "loss": 0.5332,
      "step": 2624
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6286513399673823,
      "learning_rate": 7.760779300856316e-06,
      "loss": 0.5466,
      "step": 2625
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7623334595597904,
      "learning_rate": 7.759058801607014e-06,
      "loss": 0.5539,
      "step": 2626
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6519166843010614,
      "learning_rate": 7.75733783251874e-06,
      "loss": 0.5794,
      "step": 2627
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7068296690458556,
      "learning_rate": 7.755616393884562e-06,
      "loss": 0.5674,
      "step": 2628
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7175341821364184,
      "learning_rate": 7.753894485997617e-06,
      "loss": 0.5667,
      "step": 2629
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7641779419448566,
      "learning_rate": 7.752172109151132e-06,
      "loss": 0.5459,
      "step": 2630
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7954156821645183,
      "learning_rate": 7.750449263638409e-06,
      "loss": 0.6731,
      "step": 2631
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7725568454526498,
      "learning_rate": 7.74872594975283e-06,
      "loss": 0.6309,
      "step": 2632
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.872619278687467,
      "learning_rate": 7.747002167787858e-06,
      "loss": 0.5876,
      "step": 2633
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7117154358653245,
      "learning_rate": 7.745277918037035e-06,
      "loss": 0.55,
      "step": 2634
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.585826210737778,
      "learning_rate": 7.743553200793983e-06,
      "loss": 0.5113,
      "step": 2635
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0611140337280847,
      "learning_rate": 7.741828016352402e-06,
      "loss": 0.6573,
      "step": 2636
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7773519884842081,
      "learning_rate": 7.740102365006075e-06,
      "loss": 0.5905,
      "step": 2637
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8298155813853204,
      "learning_rate": 7.73837624704886e-06,
      "loss": 0.6691,
      "step": 2638
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8438407192665045,
      "learning_rate": 7.7366496627747e-06,
      "loss": 0.6326,
      "step": 2639
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9777680300044581,
      "learning_rate": 7.734922612477611e-06,
      "loss": 0.5894,
      "step": 2640
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7666751165110746,
      "learning_rate": 7.733195096451696e-06,
      "loss": 0.5562,
      "step": 2641
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9476262775001394,
      "learning_rate": 7.731467114991128e-06,
      "loss": 0.6626,
      "step": 2642
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8878137461203477,
      "learning_rate": 7.729738668390167e-06,
      "loss": 0.7162,
      "step": 2643
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8375041323196158,
      "learning_rate": 7.728009756943149e-06,
      "loss": 0.5794,
      "step": 2644
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6922060112985475,
      "learning_rate": 7.726280380944489e-06,
      "loss": 0.562,
      "step": 2645
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6551729313080378,
      "learning_rate": 7.724550540688683e-06,
      "loss": 0.5369,
      "step": 2646
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6891749134127959,
      "learning_rate": 7.722820236470302e-06,
      "loss": 0.558,
      "step": 2647
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.726680985435801,
      "learning_rate": 7.721089468584005e-06,
      "loss": 0.5829,
      "step": 2648
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8354331783022282,
      "learning_rate": 7.719358237324516e-06,
      "loss": 0.5797,
      "step": 2649
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8946148413226056,
      "learning_rate": 7.717626542986652e-06,
      "loss": 0.6946,
      "step": 2650
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7937513114640031,
      "learning_rate": 7.715894385865299e-06,
      "loss": 0.6061,
      "step": 2651
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7332761043630776,
      "learning_rate": 7.714161766255425e-06,
      "loss": 0.5246,
      "step": 2652
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7397786301881935,
      "learning_rate": 7.712428684452079e-06,
      "loss": 0.6019,
      "step": 2653
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.77137686310079,
      "learning_rate": 7.710695140750384e-06,
      "loss": 0.5535,
      "step": 2654
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7307719167737979,
      "learning_rate": 7.708961135445549e-06,
      "loss": 0.5648,
      "step": 2655
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8138713320314339,
      "learning_rate": 7.707226668832854e-06,
      "loss": 0.6749,
      "step": 2656
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7127461887665362,
      "learning_rate": 7.705491741207661e-06,
      "loss": 0.5622,
      "step": 2657
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7307288235631265,
      "learning_rate": 7.703756352865408e-06,
      "loss": 0.5328,
      "step": 2658
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7257445412833696,
      "learning_rate": 7.702020504101618e-06,
      "loss": 0.6048,
      "step": 2659
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.715546508454551,
      "learning_rate": 7.700284195211883e-06,
      "loss": 0.5824,
      "step": 2660
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7443185337086762,
      "learning_rate": 7.698547426491883e-06,
      "loss": 0.5566,
      "step": 2661
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7393450034687399,
      "learning_rate": 7.69681019823737e-06,
      "loss": 0.6106,
      "step": 2662
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7490774446747605,
      "learning_rate": 7.695072510744174e-06,
      "loss": 0.6337,
      "step": 2663
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.726061307116932,
      "learning_rate": 7.693334364308208e-06,
      "loss": 0.5574,
      "step": 2664
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8610484167206152,
      "learning_rate": 7.691595759225457e-06,
      "loss": 0.6321,
      "step": 2665
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1912650606538906,
      "learning_rate": 7.68985669579199e-06,
      "loss": 0.5986,
      "step": 2666
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6857360125297636,
      "learning_rate": 7.688117174303951e-06,
      "loss": 0.5652,
      "step": 2667
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8003695506270507,
      "learning_rate": 7.68637719505756e-06,
      "loss": 0.6612,
      "step": 2668
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7283821617771326,
      "learning_rate": 7.68463675834912e-06,
      "loss": 0.5464,
      "step": 2669
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8199876486185335,
      "learning_rate": 7.68289586447501e-06,
      "loss": 0.5653,
      "step": 2670
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7057179652499728,
      "learning_rate": 7.68115451373168e-06,
      "loss": 0.5521,
      "step": 2671
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8235515349311006,
      "learning_rate": 7.679412706415672e-06,
      "loss": 0.6092,
      "step": 2672
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8636819956537833,
      "learning_rate": 7.677670442823594e-06,
      "loss": 0.6093,
      "step": 2673
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7080627750988319,
      "learning_rate": 7.675927723252134e-06,
      "loss": 0.5801,
      "step": 2674
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8708700839746261,
      "learning_rate": 7.674184547998063e-06,
      "loss": 0.6093,
      "step": 2675
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8913291843622851,
      "learning_rate": 7.672440917358221e-06,
      "loss": 0.6604,
      "step": 2676
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9449274162922053,
      "learning_rate": 7.670696831629531e-06,
      "loss": 0.6289,
      "step": 2677
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6919000466214699,
      "learning_rate": 7.668952291108996e-06,
      "loss": 0.5046,
      "step": 2678
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7020000035327029,
      "learning_rate": 7.667207296093688e-06,
      "loss": 0.5862,
      "step": 2679
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.88971465457254,
      "learning_rate": 7.665461846880766e-06,
      "loss": 0.6598,
      "step": 2680
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7005520549385662,
      "learning_rate": 7.663715943767461e-06,
      "loss": 0.5893,
      "step": 2681
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6739883084933995,
      "learning_rate": 7.661969587051079e-06,
      "loss": 0.5181,
      "step": 2682
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9180342638820376,
      "learning_rate": 7.66022277702901e-06,
      "loss": 0.7337,
      "step": 2683
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7918545103119662,
      "learning_rate": 7.658475513998716e-06,
      "loss": 0.5704,
      "step": 2684
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8428144287265061,
      "learning_rate": 7.656727798257737e-06,
      "loss": 0.6585,
      "step": 2685
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6611212100572257,
      "learning_rate": 7.65497963010369e-06,
      "loss": 0.5521,
      "step": 2686
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.767935941411705,
      "learning_rate": 7.653231009834276e-06,
      "loss": 0.56,
      "step": 2687
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7576161741805942,
      "learning_rate": 7.651481937747258e-06,
      "loss": 0.5223,
      "step": 2688
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7571483621672602,
      "learning_rate": 7.649732414140491e-06,
      "loss": 0.5856,
      "step": 2689
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8427817003990717,
      "learning_rate": 7.647982439311898e-06,
      "loss": 0.5714,
      "step": 2690
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7665876080198591,
      "learning_rate": 7.646232013559482e-06,
      "loss": 0.6659,
      "step": 2691
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8484365654735585,
      "learning_rate": 7.644481137181323e-06,
      "loss": 0.64,
      "step": 2692
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6733679821941454,
      "learning_rate": 7.642729810475576e-06,
      "loss": 0.5212,
      "step": 2693
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8771972006029759,
      "learning_rate": 7.640978033740473e-06,
      "loss": 0.5415,
      "step": 2694
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9257431839033528,
      "learning_rate": 7.639225807274327e-06,
      "loss": 0.6464,
      "step": 2695
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9159739177431039,
      "learning_rate": 7.637473131375517e-06,
      "loss": 0.6526,
      "step": 2696
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7275415689700523,
      "learning_rate": 7.635720006342513e-06,
      "loss": 0.6221,
      "step": 2697
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8171859241625635,
      "learning_rate": 7.63396643247385e-06,
      "loss": 0.6199,
      "step": 2698
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8177450800477294,
      "learning_rate": 7.632212410068142e-06,
      "loss": 0.6168,
      "step": 2699
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7171593163823642,
      "learning_rate": 7.630457939424086e-06,
      "loss": 0.5828,
      "step": 2700
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8668162516708091,
      "learning_rate": 7.6287030208404446e-06,
      "loss": 0.6431,
      "step": 2701
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8757720485570994,
      "learning_rate": 7.626947654616066e-06,
      "loss": 0.6018,
      "step": 2702
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6997164772348926,
      "learning_rate": 7.625191841049868e-06,
      "loss": 0.62,
      "step": 2703
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.812722705207381,
      "learning_rate": 7.623435580440849e-06,
      "loss": 0.6075,
      "step": 2704
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.391656340481501,
      "learning_rate": 7.6216788730880806e-06,
      "loss": 0.6465,
      "step": 2705
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7922823100985256,
      "learning_rate": 7.619921719290715e-06,
      "loss": 0.6256,
      "step": 2706
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6967721956254086,
      "learning_rate": 7.618164119347973e-06,
      "loss": 0.5802,
      "step": 2707
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7735384700934813,
      "learning_rate": 7.616406073559157e-06,
      "loss": 0.6124,
      "step": 2708
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7328953130306884,
      "learning_rate": 7.614647582223645e-06,
      "loss": 0.5486,
      "step": 2709
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7580614630146327,
      "learning_rate": 7.612888645640888e-06,
      "loss": 0.6099,
      "step": 2710
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7422254501911365,
      "learning_rate": 7.611129264110415e-06,
      "loss": 0.5647,
      "step": 2711
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8964542805292356,
      "learning_rate": 7.6093694379318315e-06,
      "loss": 0.6675,
      "step": 2712
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9092064325050946,
      "learning_rate": 7.607609167404817e-06,
      "loss": 0.6781,
      "step": 2713
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.79206317898653,
      "learning_rate": 7.605848452829128e-06,
      "loss": 0.6226,
      "step": 2714
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7478648086548831,
      "learning_rate": 7.604087294504594e-06,
      "loss": 0.5787,
      "step": 2715
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8645050906102298,
      "learning_rate": 7.602325692731124e-06,
      "loss": 0.6491,
      "step": 2716
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7465382285852344,
      "learning_rate": 7.6005636478086984e-06,
      "loss": 0.5789,
      "step": 2717
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7159849857769799,
      "learning_rate": 7.598801160037377e-06,
      "loss": 0.5498,
      "step": 2718
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7605018557718445,
      "learning_rate": 7.597038229717292e-06,
      "loss": 0.5538,
      "step": 2719
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9528717637188856,
      "learning_rate": 7.595274857148651e-06,
      "loss": 0.6247,
      "step": 2720
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6593706026499795,
      "learning_rate": 7.59351104263174e-06,
      "loss": 0.5774,
      "step": 2721
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6359761904607262,
      "learning_rate": 7.591746786466918e-06,
      "loss": 0.5323,
      "step": 2722
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7614280946412803,
      "learning_rate": 7.589982088954619e-06,
      "loss": 0.568,
      "step": 2723
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7848791632750642,
      "learning_rate": 7.588216950395352e-06,
      "loss": 0.5552,
      "step": 2724
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8317204207936151,
      "learning_rate": 7.5864513710897e-06,
      "loss": 0.6109,
      "step": 2725
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6667625430147582,
      "learning_rate": 7.584685351338328e-06,
      "loss": 0.5427,
      "step": 2726
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1569563083416816,
      "learning_rate": 7.582918891441966e-06,
      "loss": 0.5475,
      "step": 2727
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7892831427043902,
      "learning_rate": 7.5811519917014275e-06,
      "loss": 0.5906,
      "step": 2728
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6176416801938055,
      "learning_rate": 7.579384652417593e-06,
      "loss": 0.5425,
      "step": 2729
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7099653290600364,
      "learning_rate": 7.577616873891425e-06,
      "loss": 0.5331,
      "step": 2730
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7664426581160024,
      "learning_rate": 7.575848656423957e-06,
      "loss": 0.6165,
      "step": 2731
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7404908163675877,
      "learning_rate": 7.574080000316298e-06,
      "loss": 0.5557,
      "step": 2732
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6935629556588265,
      "learning_rate": 7.57231090586963e-06,
      "loss": 0.5074,
      "step": 2733
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6362762852049509,
      "learning_rate": 7.570541373385214e-06,
      "loss": 0.5104,
      "step": 2734
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7821687552125114,
      "learning_rate": 7.568771403164381e-06,
      "loss": 0.5898,
      "step": 2735
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7937424086711169,
      "learning_rate": 7.56700099550854e-06,
      "loss": 0.6487,
      "step": 2736
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7678714627836957,
      "learning_rate": 7.565230150719173e-06,
      "loss": 0.7374,
      "step": 2737
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6299692120238257,
      "learning_rate": 7.563458869097833e-06,
      "loss": 0.5106,
      "step": 2738
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.774927246316752,
      "learning_rate": 7.5616871509461554e-06,
      "loss": 0.5603,
      "step": 2739
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8094760484704159,
      "learning_rate": 7.559914996565845e-06,
      "loss": 0.6185,
      "step": 2740
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7135069867572567,
      "learning_rate": 7.5581424062586775e-06,
      "loss": 0.5388,
      "step": 2741
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7143339361036312,
      "learning_rate": 7.556369380326509e-06,
      "loss": 0.5359,
      "step": 2742
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6756601431276299,
      "learning_rate": 7.554595919071268e-06,
      "loss": 0.5613,
      "step": 2743
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8097902147122563,
      "learning_rate": 7.552822022794954e-06,
      "loss": 0.6186,
      "step": 2744
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6419991556514685,
      "learning_rate": 7.551047691799647e-06,
      "loss": 0.5343,
      "step": 2745
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7184236465573266,
      "learning_rate": 7.549272926387493e-06,
      "loss": 0.5848,
      "step": 2746
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9904900295460088,
      "learning_rate": 7.547497726860717e-06,
      "loss": 0.65,
      "step": 2747
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8037002343582437,
      "learning_rate": 7.54572209352162e-06,
      "loss": 0.6202,
      "step": 2748
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8032833193111435,
      "learning_rate": 7.54394602667257e-06,
      "loss": 0.559,
      "step": 2749
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7930011601371217,
      "learning_rate": 7.542169526616014e-06,
      "loss": 0.599,
      "step": 2750
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7249984697972786,
      "learning_rate": 7.540392593654473e-06,
      "loss": 0.6266,
      "step": 2751
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6278352392284934,
      "learning_rate": 7.538615228090539e-06,
      "loss": 0.5312,
      "step": 2752
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7400362046125013,
      "learning_rate": 7.536837430226878e-06,
      "loss": 0.5651,
      "step": 2753
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7442681454334864,
      "learning_rate": 7.53505920036623e-06,
      "loss": 0.513,
      "step": 2754
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8024840202920478,
      "learning_rate": 7.533280538811412e-06,
      "loss": 0.6104,
      "step": 2755
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8495531730016921,
      "learning_rate": 7.53150144586531e-06,
      "loss": 0.6881,
      "step": 2756
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8709920391686855,
      "learning_rate": 7.529721921830884e-06,
      "loss": 0.6733,
      "step": 2757
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.728838250694351,
      "learning_rate": 7.52794196701117e-06,
      "loss": 0.5883,
      "step": 2758
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9153913788268614,
      "learning_rate": 7.526161581709274e-06,
      "loss": 0.5702,
      "step": 2759
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8399246142832262,
      "learning_rate": 7.524380766228379e-06,
      "loss": 0.6209,
      "step": 2760
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7878379884857731,
      "learning_rate": 7.522599520871737e-06,
      "loss": 0.6188,
      "step": 2761
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7164761220948352,
      "learning_rate": 7.5208178459426785e-06,
      "loss": 0.5234,
      "step": 2762
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.807324703744647,
      "learning_rate": 7.5190357417446e-06,
      "loss": 0.6468,
      "step": 2763
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8131748586399106,
      "learning_rate": 7.51725320858098e-06,
      "loss": 0.6327,
      "step": 2764
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7531929692127107,
      "learning_rate": 7.515470246755363e-06,
      "loss": 0.5862,
      "step": 2765
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6796607539272923,
      "learning_rate": 7.513686856571367e-06,
      "loss": 0.5318,
      "step": 2766
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6910802424104073,
      "learning_rate": 7.5119030383326885e-06,
      "loss": 0.5495,
      "step": 2767
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6968499155138721,
      "learning_rate": 7.5101187923430906e-06,
      "loss": 0.6042,
      "step": 2768
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7758256024486233,
      "learning_rate": 7.5083341189064126e-06,
      "loss": 0.6205,
      "step": 2769
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8950642908825718,
      "learning_rate": 7.5065490183265665e-06,
      "loss": 0.6357,
      "step": 2770
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7071771462489088,
      "learning_rate": 7.5047634909075365e-06,
      "loss": 0.5672,
      "step": 2771
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9758817932018523,
      "learning_rate": 7.502977536953376e-06,
      "loss": 0.6574,
      "step": 2772
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8241684574713491,
      "learning_rate": 7.501191156768218e-06,
      "loss": 0.5251,
      "step": 2773
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7882206503512899,
      "learning_rate": 7.499404350656264e-06,
      "loss": 0.6108,
      "step": 2774
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8151266626473438,
      "learning_rate": 7.4976171189217875e-06,
      "loss": 0.722,
      "step": 2775
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8404012937128001,
      "learning_rate": 7.495829461869134e-06,
      "loss": 0.6414,
      "step": 2776
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7365351816341162,
      "learning_rate": 7.4940413798027255e-06,
      "loss": 0.5924,
      "step": 2777
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6975033031649471,
      "learning_rate": 7.492252873027054e-06,
      "loss": 0.5568,
      "step": 2778
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6434841359262528,
      "learning_rate": 7.490463941846681e-06,
      "loss": 0.5301,
      "step": 2779
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9067545162989353,
      "learning_rate": 7.488674586566245e-06,
      "loss": 0.6515,
      "step": 2780
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7846673484292362,
      "learning_rate": 7.486884807490451e-06,
      "loss": 0.6605,
      "step": 2781
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6596953844332776,
      "learning_rate": 7.485094604924085e-06,
      "loss": 0.5675,
      "step": 2782
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.139512770587114,
      "learning_rate": 7.483303979171996e-06,
      "loss": 0.6852,
      "step": 2783
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6699009441477491,
      "learning_rate": 7.481512930539112e-06,
      "loss": 0.5497,
      "step": 2784
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.809741275490068,
      "learning_rate": 7.4797214593304265e-06,
      "loss": 0.6738,
      "step": 2785
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7771526737241754,
      "learning_rate": 7.477929565851011e-06,
      "loss": 0.578,
      "step": 2786
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.704105988508471,
      "learning_rate": 7.476137250406006e-06,
      "loss": 0.5494,
      "step": 2787
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9720871354255579,
      "learning_rate": 7.474344513300622e-06,
      "loss": 0.661,
      "step": 2788
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7935676464818128,
      "learning_rate": 7.4725513548401455e-06,
      "loss": 0.5307,
      "step": 2789
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.736125146135366,
      "learning_rate": 7.470757775329932e-06,
      "loss": 0.546,
      "step": 2790
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6434049693512166,
      "learning_rate": 7.468963775075409e-06,
      "loss": 0.5257,
      "step": 2791
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7486314378965022,
      "learning_rate": 7.467169354382078e-06,
      "loss": 0.5787,
      "step": 2792
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7830765573065516,
      "learning_rate": 7.465374513555508e-06,
      "loss": 0.6406,
      "step": 2793
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7416752147609592,
      "learning_rate": 7.463579252901341e-06,
      "loss": 0.6001,
      "step": 2794
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7932544895583734,
      "learning_rate": 7.461783572725294e-06,
      "loss": 0.5725,
      "step": 2795
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8456748438592332,
      "learning_rate": 7.459987473333151e-06,
      "loss": 0.7023,
      "step": 2796
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7125644413675578,
      "learning_rate": 7.45819095503077e-06,
      "loss": 0.5158,
      "step": 2797
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9033570139068714,
      "learning_rate": 7.456394018124078e-06,
      "loss": 0.7051,
      "step": 2798
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2770720614207984,
      "learning_rate": 7.454596662919076e-06,
      "loss": 0.6321,
      "step": 2799
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7295604218527582,
      "learning_rate": 7.452798889721832e-06,
      "loss": 0.5698,
      "step": 2800
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9898411589870634,
      "learning_rate": 7.451000698838491e-06,
      "loss": 0.6618,
      "step": 2801
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9119650963677739,
      "learning_rate": 7.449202090575267e-06,
      "loss": 0.5417,
      "step": 2802
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7104281812219381,
      "learning_rate": 7.447403065238441e-06,
      "loss": 0.5384,
      "step": 2803
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7205310094790782,
      "learning_rate": 7.4456036231343695e-06,
      "loss": 0.5506,
      "step": 2804
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9059139551531954,
      "learning_rate": 7.443803764569481e-06,
      "loss": 0.6102,
      "step": 2805
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9067937407859735,
      "learning_rate": 7.442003489850269e-06,
      "loss": 0.6438,
      "step": 2806
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7852243507239536,
      "learning_rate": 7.440202799283305e-06,
      "loss": 0.6652,
      "step": 2807
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7721161633766079,
      "learning_rate": 7.438401693175225e-06,
      "loss": 0.5445,
      "step": 2808
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8972149443350992,
      "learning_rate": 7.436600171832739e-06,
      "loss": 0.5881,
      "step": 2809
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7626009422103235,
      "learning_rate": 7.434798235562628e-06,
      "loss": 0.5941,
      "step": 2810
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8047749854107277,
      "learning_rate": 7.432995884671744e-06,
      "loss": 0.6144,
      "step": 2811
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4678456824114459,
      "learning_rate": 7.4311931194670085e-06,
      "loss": 0.6417,
      "step": 2812
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6604353020511844,
      "learning_rate": 7.429389940255412e-06,
      "loss": 0.5274,
      "step": 2813
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8529674817005919,
      "learning_rate": 7.427586347344017e-06,
      "loss": 0.6289,
      "step": 2814
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8013202131373539,
      "learning_rate": 7.425782341039957e-06,
      "loss": 0.596,
      "step": 2815
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6834557147795206,
      "learning_rate": 7.4239779216504386e-06,
      "loss": 0.5578,
      "step": 2816
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7633287135547377,
      "learning_rate": 7.422173089482732e-06,
      "loss": 0.5902,
      "step": 2817
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9158569476634496,
      "learning_rate": 7.420367844844181e-06,
      "loss": 0.6439,
      "step": 2818
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8827029514428754,
      "learning_rate": 7.418562188042202e-06,
      "loss": 0.5777,
      "step": 2819
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8158684938546891,
      "learning_rate": 7.416756119384278e-06,
      "loss": 0.6359,
      "step": 2820
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7436591213506316,
      "learning_rate": 7.414949639177965e-06,
      "loss": 0.6018,
      "step": 2821
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.792487060961103,
      "learning_rate": 7.413142747730888e-06,
      "loss": 0.5646,
      "step": 2822
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1969064272659133,
      "learning_rate": 7.411335445350739e-06,
      "loss": 0.6117,
      "step": 2823
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8680410099423544,
      "learning_rate": 7.409527732345286e-06,
      "loss": 0.6409,
      "step": 2824
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9210056257054146,
      "learning_rate": 7.407719609022364e-06,
      "loss": 0.6288,
      "step": 2825
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8640229418824921,
      "learning_rate": 7.405911075689873e-06,
      "loss": 0.7012,
      "step": 2826
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7635452820306983,
      "learning_rate": 7.404102132655793e-06,
      "loss": 0.5999,
      "step": 2827
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7809114951757538,
      "learning_rate": 7.402292780228164e-06,
      "loss": 0.6313,
      "step": 2828
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8133700076302167,
      "learning_rate": 7.400483018715102e-06,
      "loss": 0.5951,
      "step": 2829
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.885618376770053,
      "learning_rate": 7.39867284842479e-06,
      "loss": 0.6154,
      "step": 2830
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6922785924001391,
      "learning_rate": 7.396862269665481e-06,
      "loss": 0.5355,
      "step": 2831
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8011579540325787,
      "learning_rate": 7.395051282745498e-06,
      "loss": 0.6297,
      "step": 2832
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9363355576810274,
      "learning_rate": 7.393239887973233e-06,
      "loss": 0.6647,
      "step": 2833
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6980248750014387,
      "learning_rate": 7.391428085657149e-06,
      "loss": 0.5771,
      "step": 2834
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9281539487708826,
      "learning_rate": 7.389615876105773e-06,
      "loss": 0.6259,
      "step": 2835
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6357150834755935,
      "learning_rate": 7.387803259627708e-06,
      "loss": 0.5345,
      "step": 2836
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8004713244179662,
      "learning_rate": 7.385990236531626e-06,
      "loss": 0.661,
      "step": 2837
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9092593636169148,
      "learning_rate": 7.384176807126263e-06,
      "loss": 0.5931,
      "step": 2838
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8641341664554995,
      "learning_rate": 7.382362971720428e-06,
      "loss": 0.5792,
      "step": 2839
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9384934754070131,
      "learning_rate": 7.380548730622998e-06,
      "loss": 0.6101,
      "step": 2840
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8993188027728529,
      "learning_rate": 7.3787340841429204e-06,
      "loss": 0.6773,
      "step": 2841
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8216448945664718,
      "learning_rate": 7.376919032589209e-06,
      "loss": 0.6545,
      "step": 2842
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9490359401112175,
      "learning_rate": 7.375103576270949e-06,
      "loss": 0.6019,
      "step": 2843
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7424189196626539,
      "learning_rate": 7.373287715497294e-06,
      "loss": 0.5986,
      "step": 2844
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8599509592323239,
      "learning_rate": 7.371471450577466e-06,
      "loss": 0.6992,
      "step": 2845
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6385025053064425,
      "learning_rate": 7.369654781820754e-06,
      "loss": 0.5393,
      "step": 2846
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0448492836003456,
      "learning_rate": 7.36783770953652e-06,
      "loss": 0.6766,
      "step": 2847
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8375821318025848,
      "learning_rate": 7.366020234034194e-06,
      "loss": 0.6283,
      "step": 2848
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6567250037850638,
      "learning_rate": 7.364202355623269e-06,
      "loss": 0.5475,
      "step": 2849
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0388567737752157,
      "learning_rate": 7.362384074613313e-06,
      "loss": 0.7117,
      "step": 2850
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.201138998696804,
      "learning_rate": 7.360565391313961e-06,
      "loss": 0.6528,
      "step": 2851
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7902597459524227,
      "learning_rate": 7.358746306034914e-06,
      "loss": 0.5434,
      "step": 2852
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7008384269371887,
      "learning_rate": 7.356926819085946e-06,
      "loss": 0.55,
      "step": 2853
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7799511169318565,
      "learning_rate": 7.355106930776894e-06,
      "loss": 0.6287,
      "step": 2854
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7599687231764983,
      "learning_rate": 7.353286641417667e-06,
      "loss": 0.6309,
      "step": 2855
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.990798707368135,
      "learning_rate": 7.351465951318243e-06,
      "loss": 0.6853,
      "step": 2856
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.679323887927624,
      "learning_rate": 7.349644860788663e-06,
      "loss": 0.5334,
      "step": 2857
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8631613997105384,
      "learning_rate": 7.347823370139042e-06,
      "loss": 0.6657,
      "step": 2858
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6484232606824437,
      "learning_rate": 7.346001479679561e-06,
      "loss": 0.5463,
      "step": 2859
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6429189109587077,
      "learning_rate": 7.344179189720468e-06,
      "loss": 0.5535,
      "step": 2860
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6226618855258182,
      "learning_rate": 7.342356500572081e-06,
      "loss": 0.5343,
      "step": 2861
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9298565980352995,
      "learning_rate": 7.340533412544784e-06,
      "loss": 0.6588,
      "step": 2862
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.151576818973023,
      "learning_rate": 7.338709925949031e-06,
      "loss": 0.6652,
      "step": 2863
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7102917109206107,
      "learning_rate": 7.336886041095342e-06,
      "loss": 0.532,
      "step": 2864
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8113055560444782,
      "learning_rate": 7.335061758294304e-06,
      "loss": 0.641,
      "step": 2865
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6465447535820217,
      "learning_rate": 7.3332370778565765e-06,
      "loss": 0.5574,
      "step": 2866
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6648166402403834,
      "learning_rate": 7.331412000092882e-06,
      "loss": 0.5098,
      "step": 2867
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7895565954237493,
      "learning_rate": 7.329586525314013e-06,
      "loss": 0.5931,
      "step": 2868
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7414527944765841,
      "learning_rate": 7.327760653830827e-06,
      "loss": 0.5588,
      "step": 2869
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6896143202134559,
      "learning_rate": 7.325934385954253e-06,
      "loss": 0.6228,
      "step": 2870
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6476335485398594,
      "learning_rate": 7.3241077219952835e-06,
      "loss": 0.505,
      "step": 2871
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9408007461651444,
      "learning_rate": 7.322280662264982e-06,
      "loss": 0.6682,
      "step": 2872
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9091270413358847,
      "learning_rate": 7.3204532070744786e-06,
      "loss": 0.5484,
      "step": 2873
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7436389426827509,
      "learning_rate": 7.318625356734966e-06,
      "loss": 0.5657,
      "step": 2874
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8365383818303793,
      "learning_rate": 7.31679711155771e-06,
      "loss": 0.6025,
      "step": 2875
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6783044647999716,
      "learning_rate": 7.314968471854045e-06,
      "loss": 0.5881,
      "step": 2876
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8138430098964579,
      "learning_rate": 7.313139437935363e-06,
      "loss": 0.7131,
      "step": 2877
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6648183022913393,
      "learning_rate": 7.311310010113136e-06,
      "loss": 0.5293,
      "step": 2878
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8661903257434097,
      "learning_rate": 7.309480188698891e-06,
      "loss": 0.6796,
      "step": 2879
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7955237115574195,
      "learning_rate": 7.3076499740042304e-06,
      "loss": 0.5831,
      "step": 2880
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7005981900877916,
      "learning_rate": 7.30581936634082e-06,
      "loss": 0.5195,
      "step": 2881
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7125403564423088,
      "learning_rate": 7.303988366020395e-06,
      "loss": 0.5774,
      "step": 2882
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7966605262025139,
      "learning_rate": 7.3021569733547525e-06,
      "loss": 0.5672,
      "step": 2883
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6408581123666159,
      "learning_rate": 7.300325188655762e-06,
      "loss": 0.5491,
      "step": 2884
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6729588686163938,
      "learning_rate": 7.298493012235356e-06,
      "loss": 0.5882,
      "step": 2885
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8267576202984049,
      "learning_rate": 7.296660444405535e-06,
      "loss": 0.6398,
      "step": 2886
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7966558166896005,
      "learning_rate": 7.2948274854783675e-06,
      "loss": 0.5778,
      "step": 2887
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8561943791015211,
      "learning_rate": 7.292994135765986e-06,
      "loss": 0.6957,
      "step": 2888
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6623849906121865,
      "learning_rate": 7.291160395580592e-06,
      "loss": 0.5477,
      "step": 2889
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7343714851941893,
      "learning_rate": 7.2893262652344525e-06,
      "loss": 0.5849,
      "step": 2890
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7873218477879738,
      "learning_rate": 7.287491745039899e-06,
      "loss": 0.6311,
      "step": 2891
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.839586644181958,
      "learning_rate": 7.285656835309331e-06,
      "loss": 0.6209,
      "step": 2892
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8396634681771563,
      "learning_rate": 7.283821536355217e-06,
      "loss": 0.6205,
      "step": 2893
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8624888896223887,
      "learning_rate": 7.281985848490088e-06,
      "loss": 0.6439,
      "step": 2894
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6606634156738503,
      "learning_rate": 7.280149772026543e-06,
      "loss": 0.5578,
      "step": 2895
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6403054329211018,
      "learning_rate": 7.278313307277246e-06,
      "loss": 0.601,
      "step": 2896
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8262560993345395,
      "learning_rate": 7.276476454554929e-06,
      "loss": 0.5662,
      "step": 2897
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.708835024781072,
      "learning_rate": 7.2746392141723855e-06,
      "loss": 0.5806,
      "step": 2898
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7789276040053643,
      "learning_rate": 7.272801586442483e-06,
      "loss": 0.6009,
      "step": 2899
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6546894783018286,
      "learning_rate": 7.270963571678147e-06,
      "loss": 0.5633,
      "step": 2900
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7976681468572578,
      "learning_rate": 7.269125170192374e-06,
      "loss": 0.5243,
      "step": 2901
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9280675913614975,
      "learning_rate": 7.267286382298223e-06,
      "loss": 0.6668,
      "step": 2902
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7159934928959768,
      "learning_rate": 7.265447208308823e-06,
      "loss": 0.5767,
      "step": 2903
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8084192931223748,
      "learning_rate": 7.2636076485373645e-06,
      "loss": 0.6601,
      "step": 2904
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7838871066062002,
      "learning_rate": 7.261767703297106e-06,
      "loss": 0.5528,
      "step": 2905
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7651466819633849,
      "learning_rate": 7.25992737290137e-06,
      "loss": 0.5644,
      "step": 2906
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7475776654515275,
      "learning_rate": 7.2580866576635455e-06,
      "loss": 0.5759,
      "step": 2907
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.97313575108032,
      "learning_rate": 7.256245557897087e-06,
      "loss": 0.6327,
      "step": 2908
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7019378626324887,
      "learning_rate": 7.2544040739155175e-06,
      "loss": 0.5846,
      "step": 2909
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7030979096116602,
      "learning_rate": 7.252562206032419e-06,
      "loss": 0.5168,
      "step": 2910
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9477283408724446,
      "learning_rate": 7.250719954561445e-06,
      "loss": 0.6217,
      "step": 2911
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8233896715207245,
      "learning_rate": 7.24887731981631e-06,
      "loss": 0.6198,
      "step": 2912
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7470560269126937,
      "learning_rate": 7.247034302110796e-06,
      "loss": 0.5991,
      "step": 2913
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7791388102540987,
      "learning_rate": 7.24519090175875e-06,
      "loss": 0.6573,
      "step": 2914
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9035331631095982,
      "learning_rate": 7.243347119074083e-06,
      "loss": 0.6181,
      "step": 2915
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6795602372065908,
      "learning_rate": 7.241502954370774e-06,
      "loss": 0.5515,
      "step": 2916
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6716714562267163,
      "learning_rate": 7.239658407962862e-06,
      "loss": 0.5555,
      "step": 2917
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9917691432338173,
      "learning_rate": 7.2378134801644575e-06,
      "loss": 0.6633,
      "step": 2918
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7413779978681652,
      "learning_rate": 7.235968171289729e-06,
      "loss": 0.561,
      "step": 2919
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7652866883129045,
      "learning_rate": 7.234122481652916e-06,
      "loss": 0.5384,
      "step": 2920
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9967114250440547,
      "learning_rate": 7.232276411568319e-06,
      "loss": 0.6926,
      "step": 2921
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7318013497361314,
      "learning_rate": 7.230429961350305e-06,
      "loss": 0.5927,
      "step": 2922
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.717276143424521,
      "learning_rate": 7.228583131313304e-06,
      "loss": 0.5191,
      "step": 2923
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7155339257913605,
      "learning_rate": 7.226735921771815e-06,
      "loss": 0.558,
      "step": 2924
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7376311403813524,
      "learning_rate": 7.224888333040394e-06,
      "loss": 0.6196,
      "step": 2925
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6774890991823183,
      "learning_rate": 7.2230403654336715e-06,
      "loss": 0.5225,
      "step": 2926
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7249595976177491,
      "learning_rate": 7.221192019266332e-06,
      "loss": 0.5864,
      "step": 2927
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7036284081710583,
      "learning_rate": 7.219343294853133e-06,
      "loss": 0.5383,
      "step": 2928
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6407480774273943,
      "learning_rate": 7.2174941925088915e-06,
      "loss": 0.5635,
      "step": 2929
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9036098755703411,
      "learning_rate": 7.215644712548491e-06,
      "loss": 0.639,
      "step": 2930
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8668411908393069,
      "learning_rate": 7.213794855286877e-06,
      "loss": 0.6745,
      "step": 2931
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8156597128858952,
      "learning_rate": 7.211944621039065e-06,
      "loss": 0.6251,
      "step": 2932
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7649428027318845,
      "learning_rate": 7.210094010120124e-06,
      "loss": 0.5865,
      "step": 2933
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6756427023930736,
      "learning_rate": 7.208243022845199e-06,
      "loss": 0.5635,
      "step": 2934
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9203702234857714,
      "learning_rate": 7.206391659529494e-06,
      "loss": 0.6179,
      "step": 2935
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8042192424233775,
      "learning_rate": 7.204539920488275e-06,
      "loss": 0.5802,
      "step": 2936
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9942020614469318,
      "learning_rate": 7.202687806036874e-06,
      "loss": 0.6994,
      "step": 2937
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7070935986584366,
      "learning_rate": 7.200835316490687e-06,
      "loss": 0.5932,
      "step": 2938
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8814074724996127,
      "learning_rate": 7.198982452165174e-06,
      "loss": 0.6463,
      "step": 2939
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6661383112331489,
      "learning_rate": 7.197129213375858e-06,
      "loss": 0.5369,
      "step": 2940
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8050566964842777,
      "learning_rate": 7.195275600438328e-06,
      "loss": 0.6541,
      "step": 2941
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.856675902555081,
      "learning_rate": 7.193421613668231e-06,
      "loss": 0.7227,
      "step": 2942
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9589320120285684,
      "learning_rate": 7.1915672533812855e-06,
      "loss": 0.6704,
      "step": 2943
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.80216088362574,
      "learning_rate": 7.189712519893267e-06,
      "loss": 0.5809,
      "step": 2944
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7093480385216203,
      "learning_rate": 7.1878574135200204e-06,
      "loss": 0.5389,
      "step": 2945
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6413752649400862,
      "learning_rate": 7.186001934577448e-06,
      "loss": 0.5348,
      "step": 2946
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6690118492461999,
      "learning_rate": 7.184146083381519e-06,
      "loss": 0.6138,
      "step": 2947
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7668010369648064,
      "learning_rate": 7.182289860248269e-06,
      "loss": 0.543,
      "step": 2948
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8273713241864384,
      "learning_rate": 7.180433265493788e-06,
      "loss": 0.6426,
      "step": 2949
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8591190933924621,
      "learning_rate": 7.178576299434239e-06,
      "loss": 0.6759,
      "step": 2950
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6911908997598203,
      "learning_rate": 7.176718962385842e-06,
      "loss": 0.581,
      "step": 2951
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8646514389701477,
      "learning_rate": 7.174861254664882e-06,
      "loss": 0.6688,
      "step": 2952
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6958365536625196,
      "learning_rate": 7.173003176587708e-06,
      "loss": 0.5448,
      "step": 2953
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6588055074421009,
      "learning_rate": 7.171144728470731e-06,
      "loss": 0.5518,
      "step": 2954
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8246773105027293,
      "learning_rate": 7.169285910630427e-06,
      "loss": 0.5615,
      "step": 2955
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.757300244285086,
      "learning_rate": 7.167426723383331e-06,
      "loss": 0.5736,
      "step": 2956
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8375233005988142,
      "learning_rate": 7.165567167046044e-06,
      "loss": 0.605,
      "step": 2957
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6707128095407431,
      "learning_rate": 7.16370724193523e-06,
      "loss": 0.558,
      "step": 2958
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7536820099014473,
      "learning_rate": 7.161846948367613e-06,
      "loss": 0.5662,
      "step": 2959
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9359011354478194,
      "learning_rate": 7.159986286659983e-06,
      "loss": 0.7473,
      "step": 2960
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.90758992217512,
      "learning_rate": 7.158125257129192e-06,
      "loss": 0.6664,
      "step": 2961
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9845409959856751,
      "learning_rate": 7.156263860092152e-06,
      "loss": 0.6374,
      "step": 2962
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8893635388579256,
      "learning_rate": 7.15440209586584e-06,
      "loss": 0.6815,
      "step": 2963
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8687825036573555,
      "learning_rate": 7.152539964767295e-06,
      "loss": 0.6312,
      "step": 2964
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.803700756845121,
      "learning_rate": 7.150677467113621e-06,
      "loss": 0.5827,
      "step": 2965
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7049773100509169,
      "learning_rate": 7.148814603221978e-06,
      "loss": 0.5445,
      "step": 2966
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7573555676458228,
      "learning_rate": 7.1469513734095955e-06,
      "loss": 0.6377,
      "step": 2967
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7777461335177506,
      "learning_rate": 7.14508777799376e-06,
      "loss": 0.5789,
      "step": 2968
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7135669149533928,
      "learning_rate": 7.143223817291825e-06,
      "loss": 0.559,
      "step": 2969
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0027785554945337,
      "learning_rate": 7.1413594916212e-06,
      "loss": 0.6783,
      "step": 2970
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9135368783533783,
      "learning_rate": 7.139494801299363e-06,
      "loss": 0.6151,
      "step": 2971
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.69814885290027,
      "learning_rate": 7.13762974664385e-06,
      "loss": 0.6754,
      "step": 2972
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9057520587358571,
      "learning_rate": 7.135764327972261e-06,
      "loss": 0.6243,
      "step": 2973
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7018201853203253,
      "learning_rate": 7.133898545602257e-06,
      "loss": 0.5772,
      "step": 2974
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7735526587878031,
      "learning_rate": 7.132032399851562e-06,
      "loss": 0.6132,
      "step": 2975
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8228602001091586,
      "learning_rate": 7.130165891037959e-06,
      "loss": 0.6294,
      "step": 2976
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6759064904761289,
      "learning_rate": 7.128299019479298e-06,
      "loss": 0.4742,
      "step": 2977
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7559012628964766,
      "learning_rate": 7.126431785493486e-06,
      "loss": 0.5535,
      "step": 2978
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7869832363043596,
      "learning_rate": 7.124564189398495e-06,
      "loss": 0.5967,
      "step": 2979
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7895781880525654,
      "learning_rate": 7.122696231512355e-06,
      "loss": 0.5846,
      "step": 2980
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8597787266529932,
      "learning_rate": 7.120827912153161e-06,
      "loss": 0.5833,
      "step": 2981
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.843680876080643,
      "learning_rate": 7.1189592316390685e-06,
      "loss": 0.7085,
      "step": 2982
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8383654361633295,
      "learning_rate": 7.117090190288294e-06,
      "loss": 0.643,
      "step": 2983
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6872504404666091,
      "learning_rate": 7.115220788419116e-06,
      "loss": 0.6007,
      "step": 2984
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8326014425177376,
      "learning_rate": 7.113351026349874e-06,
      "loss": 0.6527,
      "step": 2985
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8221375517019802,
      "learning_rate": 7.111480904398968e-06,
      "loss": 0.638,
      "step": 2986
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6888734970277014,
      "learning_rate": 7.109610422884863e-06,
      "loss": 0.5807,
      "step": 2987
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8722924451178358,
      "learning_rate": 7.107739582126079e-06,
      "loss": 0.6522,
      "step": 2988
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6622645554734401,
      "learning_rate": 7.105868382441203e-06,
      "loss": 0.4967,
      "step": 2989
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7541751695646751,
      "learning_rate": 7.10399682414888e-06,
      "loss": 0.5876,
      "step": 2990
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.701777870712533,
      "learning_rate": 7.102124907567817e-06,
      "loss": 0.5317,
      "step": 2991
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6615256579651861,
      "learning_rate": 7.100252633016782e-06,
      "loss": 0.5595,
      "step": 2992
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8252558841635048,
      "learning_rate": 7.098380000814604e-06,
      "loss": 0.6203,
      "step": 2993
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.862270861427249,
      "learning_rate": 7.096507011280174e-06,
      "loss": 0.6588,
      "step": 2994
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6865980275228062,
      "learning_rate": 7.0946336647324395e-06,
      "loss": 0.5632,
      "step": 2995
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8258391478606953,
      "learning_rate": 7.092759961490415e-06,
      "loss": 0.6238,
      "step": 2996
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8701166937030154,
      "learning_rate": 7.09088590187317e-06,
      "loss": 0.6292,
      "step": 2997
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8037308928188226,
      "learning_rate": 7.089011486199839e-06,
      "loss": 0.5649,
      "step": 2998
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9051578679306854,
      "learning_rate": 7.087136714789615e-06,
      "loss": 0.568,
      "step": 2999
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7969085168896913,
      "learning_rate": 7.085261587961754e-06,
      "loss": 0.5811,
      "step": 3000
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8198312366243342,
      "learning_rate": 7.083386106035566e-06,
      "loss": 0.684,
      "step": 3001
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7210405659644403,
      "learning_rate": 7.081510269330431e-06,
      "loss": 0.5961,
      "step": 3002
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9613916846184977,
      "learning_rate": 7.079634078165779e-06,
      "loss": 0.6756,
      "step": 3003
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6785757163276998,
      "learning_rate": 7.077757532861111e-06,
      "loss": 0.5701,
      "step": 3004
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8426624974124951,
      "learning_rate": 7.075880633735981e-06,
      "loss": 0.6101,
      "step": 3005
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7403011533148561,
      "learning_rate": 7.074003381110004e-06,
      "loss": 0.5289,
      "step": 3006
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8080912879893964,
      "learning_rate": 7.07212577530286e-06,
      "loss": 0.5245,
      "step": 3007
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7603773416727762,
      "learning_rate": 7.070247816634282e-06,
      "loss": 0.6549,
      "step": 3008
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6176757502518891,
      "learning_rate": 7.068369505424071e-06,
      "loss": 0.5248,
      "step": 3009
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8684138846327495,
      "learning_rate": 7.06649084199208e-06,
      "loss": 0.6572,
      "step": 3010
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7587533120760415,
      "learning_rate": 7.064611826658226e-06,
      "loss": 0.565,
      "step": 3011
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.211897124888826,
      "learning_rate": 7.062732459742488e-06,
      "loss": 0.7096,
      "step": 3012
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7770537992997335,
      "learning_rate": 7.060852741564902e-06,
      "loss": 0.5265,
      "step": 3013
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7649425612924766,
      "learning_rate": 7.0589726724455635e-06,
      "loss": 0.5549,
      "step": 3014
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8725765030621405,
      "learning_rate": 7.057092252704629e-06,
      "loss": 0.6385,
      "step": 3015
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9611956530411195,
      "learning_rate": 7.055211482662314e-06,
      "loss": 0.6128,
      "step": 3016
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7014431993207055,
      "learning_rate": 7.053330362638895e-06,
      "loss": 0.5523,
      "step": 3017
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7096150088676013,
      "learning_rate": 7.051448892954707e-06,
      "loss": 0.558,
      "step": 3018
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.692644979852246,
      "learning_rate": 7.0495670739301435e-06,
      "loss": 0.5391,
      "step": 3019
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6333665728426101,
      "learning_rate": 7.04768490588566e-06,
      "loss": 0.5205,
      "step": 3020
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7845396951098721,
      "learning_rate": 7.045802389141769e-06,
      "loss": 0.5915,
      "step": 3021
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8729908592073734,
      "learning_rate": 7.043919524019045e-06,
      "loss": 0.666,
      "step": 3022
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7697630140273541,
      "learning_rate": 7.042036310838121e-06,
      "loss": 0.623,
      "step": 3023
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6887642417593378,
      "learning_rate": 7.040152749919684e-06,
      "loss": 0.5462,
      "step": 3024
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8556995734980393,
      "learning_rate": 7.03826884158449e-06,
      "loss": 0.6267,
      "step": 3025
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7395549806069585,
      "learning_rate": 7.036384586153347e-06,
      "loss": 0.5459,
      "step": 3026
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8450834446911724,
      "learning_rate": 7.0344999839471215e-06,
      "loss": 0.6404,
      "step": 3027
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7110333313762182,
      "learning_rate": 7.032615035286746e-06,
      "loss": 0.5624,
      "step": 3028
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6904359686086238,
      "learning_rate": 7.0307297404932054e-06,
      "loss": 0.5581,
      "step": 3029
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7725053120319834,
      "learning_rate": 7.028844099887545e-06,
      "loss": 0.6169,
      "step": 3030
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1161044161946523,
      "learning_rate": 7.026958113790872e-06,
      "loss": 0.5549,
      "step": 3031
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8500349776705105,
      "learning_rate": 7.0250717825243475e-06,
      "loss": 0.6614,
      "step": 3032
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6471553978570826,
      "learning_rate": 7.0231851064091964e-06,
      "loss": 0.5426,
      "step": 3033
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7874830851454462,
      "learning_rate": 7.021298085766699e-06,
      "loss": 0.6075,
      "step": 3034
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7465750236562145,
      "learning_rate": 7.019410720918195e-06,
      "loss": 0.5118,
      "step": 3035
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8756801487259661,
      "learning_rate": 7.017523012185084e-06,
      "loss": 0.6584,
      "step": 3036
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8806619712880692,
      "learning_rate": 7.015634959888822e-06,
      "loss": 0.6879,
      "step": 3037
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7409501660262899,
      "learning_rate": 7.013746564350925e-06,
      "loss": 0.5824,
      "step": 3038
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6864269455368546,
      "learning_rate": 7.011857825892967e-06,
      "loss": 0.5815,
      "step": 3039
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7403561125378395,
      "learning_rate": 7.009968744836581e-06,
      "loss": 0.5183,
      "step": 3040
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7044296311078336,
      "learning_rate": 7.0080793215034575e-06,
      "loss": 0.5316,
      "step": 3041
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.85886460842654,
      "learning_rate": 7.006189556215346e-06,
      "loss": 0.5988,
      "step": 3042
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7751996573303258,
      "learning_rate": 7.004299449294052e-06,
      "loss": 0.6244,
      "step": 3043
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6418762495525675,
      "learning_rate": 7.002409001061443e-06,
      "loss": 0.4667,
      "step": 3044
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8914649877271704,
      "learning_rate": 7.000518211839442e-06,
      "loss": 0.6646,
      "step": 3045
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6789236560892044,
      "learning_rate": 6.998627081950031e-06,
      "loss": 0.5189,
      "step": 3046
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7058168389931408,
      "learning_rate": 6.99673561171525e-06,
      "loss": 0.6283,
      "step": 3047
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7630030832539432,
      "learning_rate": 6.994843801457195e-06,
      "loss": 0.5512,
      "step": 3048
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7953126952077831,
      "learning_rate": 6.992951651498023e-06,
      "loss": 0.5903,
      "step": 3049
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8646574910949234,
      "learning_rate": 6.9910591621599475e-06,
      "loss": 0.5875,
      "step": 3050
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.045980948661036,
      "learning_rate": 6.9891663337652406e-06,
      "loss": 0.664,
      "step": 3051
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.722237762580419,
      "learning_rate": 6.987273166636228e-06,
      "loss": 0.5038,
      "step": 3052
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1306625855487964,
      "learning_rate": 6.9853796610952996e-06,
      "loss": 0.6601,
      "step": 3053
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6150494391570147,
      "learning_rate": 6.983485817464898e-06,
      "loss": 0.5215,
      "step": 3054
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6399278371217657,
      "learning_rate": 6.981591636067525e-06,
      "loss": 0.531,
      "step": 3055
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.977642676846304,
      "learning_rate": 6.979697117225741e-06,
      "loss": 0.6353,
      "step": 3056
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8147357744747524,
      "learning_rate": 6.9778022612621634e-06,
      "loss": 0.5678,
      "step": 3057
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7627617329262633,
      "learning_rate": 6.975907068499463e-06,
      "loss": 0.5923,
      "step": 3058
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7276258102130468,
      "learning_rate": 6.9740115392603755e-06,
      "loss": 0.6029,
      "step": 3059
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7416330702505679,
      "learning_rate": 6.972115673867688e-06,
      "loss": 0.6355,
      "step": 3060
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7778534726874673,
      "learning_rate": 6.970219472644245e-06,
      "loss": 0.5545,
      "step": 3061
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8056824387964002,
      "learning_rate": 6.968322935912952e-06,
      "loss": 0.6321,
      "step": 3062
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8631674383596292,
      "learning_rate": 6.96642606399677e-06,
      "loss": 0.6895,
      "step": 3063
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8182784743662215,
      "learning_rate": 6.964528857218716e-06,
      "loss": 0.7138,
      "step": 3064
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7573460239146297,
      "learning_rate": 6.962631315901861e-06,
      "loss": 0.5998,
      "step": 3065
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7837527974687174,
      "learning_rate": 6.960733440369341e-06,
      "loss": 0.6131,
      "step": 3066
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7597360214399947,
      "learning_rate": 6.958835230944343e-06,
      "loss": 0.553,
      "step": 3067
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7856010228720411,
      "learning_rate": 6.9569366879501135e-06,
      "loss": 0.5427,
      "step": 3068
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5964918127289016,
      "learning_rate": 6.955037811709951e-06,
      "loss": 0.5037,
      "step": 3069
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7511345645795674,
      "learning_rate": 6.9531386025472185e-06,
      "loss": 0.5628,
      "step": 3070
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6575948986607947,
      "learning_rate": 6.951239060785328e-06,
      "loss": 0.5468,
      "step": 3071
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6612274073302472,
      "learning_rate": 6.949339186747754e-06,
      "loss": 0.5976,
      "step": 3072
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8792149834096752,
      "learning_rate": 6.947438980758024e-06,
      "loss": 0.674,
      "step": 3073
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7321475312312102,
      "learning_rate": 6.945538443139724e-06,
      "loss": 0.5579,
      "step": 3074
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8522100475420138,
      "learning_rate": 6.943637574216495e-06,
      "loss": 0.5569,
      "step": 3075
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6551523813967494,
      "learning_rate": 6.9417363743120385e-06,
      "loss": 0.524,
      "step": 3076
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.989402426249599,
      "learning_rate": 6.939834843750104e-06,
      "loss": 0.6502,
      "step": 3077
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7951736653668456,
      "learning_rate": 6.937932982854505e-06,
      "loss": 0.6458,
      "step": 3078
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8513931480773506,
      "learning_rate": 6.936030791949109e-06,
      "loss": 0.5716,
      "step": 3079
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8654912063052131,
      "learning_rate": 6.9341282713578375e-06,
      "loss": 0.6142,
      "step": 3080
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6190609515458507,
      "learning_rate": 6.932225421404672e-06,
      "loss": 0.5135,
      "step": 3081
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7160370389087627,
      "learning_rate": 6.930322242413646e-06,
      "loss": 0.5766,
      "step": 3082
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8354032388748516,
      "learning_rate": 6.928418734708853e-06,
      "loss": 0.5804,
      "step": 3083
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5921561514179442,
      "learning_rate": 6.9265148986144395e-06,
      "loss": 0.5094,
      "step": 3084
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8469265363888865,
      "learning_rate": 6.9246107344546084e-06,
      "loss": 0.5703,
      "step": 3085
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8162632074634929,
      "learning_rate": 6.922706242553619e-06,
      "loss": 0.5606,
      "step": 3086
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7893259432336682,
      "learning_rate": 6.920801423235788e-06,
      "loss": 0.5926,
      "step": 3087
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7208294458872869,
      "learning_rate": 6.918896276825485e-06,
      "loss": 0.5843,
      "step": 3088
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.969547270473866,
      "learning_rate": 6.916990803647138e-06,
      "loss": 0.6715,
      "step": 3089
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7333606861879639,
      "learning_rate": 6.915085004025228e-06,
      "loss": 0.5656,
      "step": 3090
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7243864518832337,
      "learning_rate": 6.913178878284292e-06,
      "loss": 0.5723,
      "step": 3091
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7733485646023331,
      "learning_rate": 6.911272426748925e-06,
      "loss": 0.58,
      "step": 3092
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8290283947921372,
      "learning_rate": 6.909365649743776e-06,
      "loss": 0.5705,
      "step": 3093
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8644394568238655,
      "learning_rate": 6.907458547593548e-06,
      "loss": 0.6574,
      "step": 3094
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.658465575996566,
      "learning_rate": 6.9055511206230005e-06,
      "loss": 0.5469,
      "step": 3095
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8493209645300295,
      "learning_rate": 6.90364336915695e-06,
      "loss": 0.6327,
      "step": 3096
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.92504291754693,
      "learning_rate": 6.901735293520265e-06,
      "loss": 0.557,
      "step": 3097
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6396218634339993,
      "learning_rate": 6.8998268940378725e-06,
      "loss": 0.5226,
      "step": 3098
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8850390742956666,
      "learning_rate": 6.897918171034751e-06,
      "loss": 0.638,
      "step": 3099
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8775568763224451,
      "learning_rate": 6.896009124835939e-06,
      "loss": 0.6182,
      "step": 3100
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7875971341491276,
      "learning_rate": 6.894099755766522e-06,
      "loss": 0.5804,
      "step": 3101
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8304681731763327,
      "learning_rate": 6.892190064151651e-06,
      "loss": 0.6095,
      "step": 3102
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6532507961761137,
      "learning_rate": 6.890280050316527e-06,
      "loss": 0.5862,
      "step": 3103
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6184833781789275,
      "learning_rate": 6.8883697145864e-06,
      "loss": 0.5317,
      "step": 3104
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7444624422064668,
      "learning_rate": 6.886459057286585e-06,
      "loss": 0.6293,
      "step": 3105
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.738324939179867,
      "learning_rate": 6.884548078742446e-06,
      "loss": 0.5459,
      "step": 3106
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8875548189622121,
      "learning_rate": 6.882636779279401e-06,
      "loss": 0.6171,
      "step": 3107
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8910214080001175,
      "learning_rate": 6.8807251592229276e-06,
      "loss": 0.6506,
      "step": 3108
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7878766016929267,
      "learning_rate": 6.878813218898552e-06,
      "loss": 0.6775,
      "step": 3109
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.943945346482362,
      "learning_rate": 6.876900958631858e-06,
      "loss": 0.66,
      "step": 3110
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6527262444038725,
      "learning_rate": 6.874988378748484e-06,
      "loss": 0.5501,
      "step": 3111
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6653928895498751,
      "learning_rate": 6.873075479574123e-06,
      "loss": 0.578,
      "step": 3112
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8089927798537804,
      "learning_rate": 6.871162261434522e-06,
      "loss": 0.5665,
      "step": 3113
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6763072751204212,
      "learning_rate": 6.869248724655481e-06,
      "loss": 0.5076,
      "step": 3114
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.826047655370385,
      "learning_rate": 6.867334869562856e-06,
      "loss": 0.6616,
      "step": 3115
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.595283618930961,
      "learning_rate": 6.865420696482556e-06,
      "loss": 0.4852,
      "step": 3116
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7854427702934368,
      "learning_rate": 6.863506205740545e-06,
      "loss": 0.6097,
      "step": 3117
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8498000284568843,
      "learning_rate": 6.861591397662841e-06,
      "loss": 0.6396,
      "step": 3118
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6949193333856496,
      "learning_rate": 6.8596762725755174e-06,
      "loss": 0.5008,
      "step": 3119
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7310086649370663,
      "learning_rate": 6.857760830804698e-06,
      "loss": 0.5324,
      "step": 3120
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8035859394807092,
      "learning_rate": 6.855845072676561e-06,
      "loss": 0.5492,
      "step": 3121
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8315343140182654,
      "learning_rate": 6.853928998517345e-06,
      "loss": 0.5999,
      "step": 3122
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9001319960101497,
      "learning_rate": 6.852012608653333e-06,
      "loss": 0.6652,
      "step": 3123
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7925692888896836,
      "learning_rate": 6.8500959034108685e-06,
      "loss": 0.6486,
      "step": 3124
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.793768615901906,
      "learning_rate": 6.8481788831163456e-06,
      "loss": 0.6117,
      "step": 3125
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6864897108882693,
      "learning_rate": 6.8462615480962135e-06,
      "loss": 0.565,
      "step": 3126
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9300878639242727,
      "learning_rate": 6.844343898676974e-06,
      "loss": 0.6411,
      "step": 3127
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7449617326643326,
      "learning_rate": 6.842425935185182e-06,
      "loss": 0.6166,
      "step": 3128
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8810017578701566,
      "learning_rate": 6.840507657947448e-06,
      "loss": 0.6504,
      "step": 3129
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8958492042116712,
      "learning_rate": 6.8385890672904335e-06,
      "loss": 0.6901,
      "step": 3130
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7009467942245847,
      "learning_rate": 6.836670163540855e-06,
      "loss": 0.5546,
      "step": 3131
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7918060625086003,
      "learning_rate": 6.834750947025483e-06,
      "loss": 0.7054,
      "step": 3132
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8232011534323697,
      "learning_rate": 6.832831418071138e-06,
      "loss": 0.6045,
      "step": 3133
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6564152366542801,
      "learning_rate": 6.8309115770046986e-06,
      "loss": 0.5502,
      "step": 3134
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.321622089122117,
      "learning_rate": 6.828991424153091e-06,
      "loss": 0.6455,
      "step": 3135
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8484232465784994,
      "learning_rate": 6.827070959843298e-06,
      "loss": 0.6354,
      "step": 3136
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7288481206610785,
      "learning_rate": 6.825150184402355e-06,
      "loss": 0.5481,
      "step": 3137
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9364694896455233,
      "learning_rate": 6.823229098157349e-06,
      "loss": 0.6366,
      "step": 3138
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6479771240738635,
      "learning_rate": 6.8213077014354225e-06,
      "loss": 0.5284,
      "step": 3139
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8417858710911886,
      "learning_rate": 6.81938599456377e-06,
      "loss": 0.5908,
      "step": 3140
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8163345219001725,
      "learning_rate": 6.817463977869635e-06,
      "loss": 0.5524,
      "step": 3141
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6774325183800121,
      "learning_rate": 6.81554165168032e-06,
      "loss": 0.5332,
      "step": 3142
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1067585780349518,
      "learning_rate": 6.813619016323173e-06,
      "loss": 0.6145,
      "step": 3143
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6624982518864575,
      "learning_rate": 6.8116960721256045e-06,
      "loss": 0.4881,
      "step": 3144
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7622283633145275,
      "learning_rate": 6.809772819415068e-06,
      "loss": 0.5974,
      "step": 3145
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6784652595666629,
      "learning_rate": 6.8078492585190745e-06,
      "loss": 0.5697,
      "step": 3146
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7587648289455629,
      "learning_rate": 6.805925389765187e-06,
      "loss": 0.554,
      "step": 3147
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6714839990117176,
      "learning_rate": 6.8040012134810195e-06,
      "loss": 0.5742,
      "step": 3148
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9266579357296365,
      "learning_rate": 6.802076729994239e-06,
      "loss": 0.6707,
      "step": 3149
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6987203582529693,
      "learning_rate": 6.800151939632566e-06,
      "loss": 0.5117,
      "step": 3150
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7686564940678177,
      "learning_rate": 6.798226842723771e-06,
      "loss": 0.6231,
      "step": 3151
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8959511075127252,
      "learning_rate": 6.796301439595679e-06,
      "loss": 0.6638,
      "step": 3152
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6326212021463321,
      "learning_rate": 6.794375730576164e-06,
      "loss": 0.5458,
      "step": 3153
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7906085799100441,
      "learning_rate": 6.792449715993157e-06,
      "loss": 0.5985,
      "step": 3154
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7422869145777646,
      "learning_rate": 6.7905233961746355e-06,
      "loss": 0.5693,
      "step": 3155
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.765429292131019,
      "learning_rate": 6.788596771448634e-06,
      "loss": 0.6078,
      "step": 3156
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7580896572283728,
      "learning_rate": 6.786669842143236e-06,
      "loss": 0.5373,
      "step": 3157
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9719416053772618,
      "learning_rate": 6.784742608586577e-06,
      "loss": 0.5904,
      "step": 3158
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8082674605117065,
      "learning_rate": 6.782815071106845e-06,
      "loss": 0.5575,
      "step": 3159
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7208816270972793,
      "learning_rate": 6.780887230032281e-06,
      "loss": 0.5805,
      "step": 3160
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.777265275770535,
      "learning_rate": 6.778959085691173e-06,
      "loss": 0.6106,
      "step": 3161
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7120851283795944,
      "learning_rate": 6.777030638411868e-06,
      "loss": 0.5555,
      "step": 3162
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7330718151822807,
      "learning_rate": 6.775101888522756e-06,
      "loss": 0.5327,
      "step": 3163
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7822749721695045,
      "learning_rate": 6.773172836352286e-06,
      "loss": 0.5624,
      "step": 3164
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8139946793221429,
      "learning_rate": 6.771243482228955e-06,
      "loss": 0.6261,
      "step": 3165
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6741856946053867,
      "learning_rate": 6.769313826481312e-06,
      "loss": 0.5016,
      "step": 3166
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6408114836203054,
      "learning_rate": 6.767383869437957e-06,
      "loss": 0.5221,
      "step": 3167
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9087656316098638,
      "learning_rate": 6.765453611427542e-06,
      "loss": 0.6566,
      "step": 3168
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8002643323377501,
      "learning_rate": 6.763523052778768e-06,
      "loss": 0.551,
      "step": 3169
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9425315303180866,
      "learning_rate": 6.76159219382039e-06,
      "loss": 0.5786,
      "step": 3170
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8154677819308739,
      "learning_rate": 6.759661034881214e-06,
      "loss": 0.6527,
      "step": 3171
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7606896127080065,
      "learning_rate": 6.7577295762900955e-06,
      "loss": 0.5681,
      "step": 3172
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8168234133648932,
      "learning_rate": 6.755797818375943e-06,
      "loss": 0.6529,
      "step": 3173
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7869517117903847,
      "learning_rate": 6.7538657614677126e-06,
      "loss": 0.5873,
      "step": 3174
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7060304129302237,
      "learning_rate": 6.7519334058944164e-06,
      "loss": 0.574,
      "step": 3175
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8240468508270222,
      "learning_rate": 6.750000751985112e-06,
      "loss": 0.7005,
      "step": 3176
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9855506761178785,
      "learning_rate": 6.74806780006891e-06,
      "loss": 0.6806,
      "step": 3177
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7414134873010033,
      "learning_rate": 6.746134550474974e-06,
      "loss": 0.5949,
      "step": 3178
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8196916582070289,
      "learning_rate": 6.7442010035325145e-06,
      "loss": 0.598,
      "step": 3179
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7835816954025843,
      "learning_rate": 6.742267159570796e-06,
      "loss": 0.608,
      "step": 3180
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6880365226354992,
      "learning_rate": 6.74033301891913e-06,
      "loss": 0.5792,
      "step": 3181
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8101969525469045,
      "learning_rate": 6.738398581906882e-06,
      "loss": 0.6056,
      "step": 3182
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6730490399273185,
      "learning_rate": 6.736463848863466e-06,
      "loss": 0.5552,
      "step": 3183
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6877744571552816,
      "learning_rate": 6.734528820118347e-06,
      "loss": 0.4969,
      "step": 3184
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8892511414983605,
      "learning_rate": 6.732593496001038e-06,
      "loss": 0.638,
      "step": 3185
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7853927907555196,
      "learning_rate": 6.730657876841109e-06,
      "loss": 0.5874,
      "step": 3186
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6989471215876331,
      "learning_rate": 6.728721962968173e-06,
      "loss": 0.541,
      "step": 3187
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.905229393985051,
      "learning_rate": 6.726785754711897e-06,
      "loss": 0.6402,
      "step": 3188
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8800627840978599,
      "learning_rate": 6.724849252401995e-06,
      "loss": 0.6725,
      "step": 3189
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5064218391673445,
      "learning_rate": 6.722912456368236e-06,
      "loss": 0.6043,
      "step": 3190
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1388228133498755,
      "learning_rate": 6.720975366940434e-06,
      "loss": 0.6612,
      "step": 3191
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8481654457044425,
      "learning_rate": 6.719037984448457e-06,
      "loss": 0.6518,
      "step": 3192
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8249185178484697,
      "learning_rate": 6.71710030922222e-06,
      "loss": 0.6168,
      "step": 3193
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7432676805376397,
      "learning_rate": 6.715162341591687e-06,
      "loss": 0.6568,
      "step": 3194
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8706312118773447,
      "learning_rate": 6.713224081886877e-06,
      "loss": 0.5694,
      "step": 3195
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9438094996926238,
      "learning_rate": 6.711285530437855e-06,
      "loss": 0.6302,
      "step": 3196
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5750488512150372,
      "learning_rate": 6.709346687574735e-06,
      "loss": 0.4629,
      "step": 3197
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8274666511416832,
      "learning_rate": 6.7074075536276795e-06,
      "loss": 0.6379,
      "step": 3198
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6714846933104798,
      "learning_rate": 6.705468128926907e-06,
      "loss": 0.51,
      "step": 3199
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.933527083577643,
      "learning_rate": 6.7035284138026805e-06,
      "loss": 0.6875,
      "step": 3200
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8556577020225966,
      "learning_rate": 6.7015884085853114e-06,
      "loss": 0.5819,
      "step": 3201
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8434954864676036,
      "learning_rate": 6.699648113605164e-06,
      "loss": 0.6641,
      "step": 3202
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8088761939282472,
      "learning_rate": 6.697707529192648e-06,
      "loss": 0.557,
      "step": 3203
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.63629947691648,
      "learning_rate": 6.695766655678229e-06,
      "loss": 0.5523,
      "step": 3204
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6983738726228311,
      "learning_rate": 6.693825493392412e-06,
      "loss": 0.571,
      "step": 3205
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6849477062333319,
      "learning_rate": 6.691884042665761e-06,
      "loss": 0.5244,
      "step": 3206
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6167514872185222,
      "learning_rate": 6.689942303828883e-06,
      "loss": 0.5291,
      "step": 3207
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7602023220591377,
      "learning_rate": 6.688000277212436e-06,
      "loss": 0.5538,
      "step": 3208
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7131202697258787,
      "learning_rate": 6.686057963147129e-06,
      "loss": 0.5583,
      "step": 3209
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.877502324096378,
      "learning_rate": 6.684115361963715e-06,
      "loss": 0.6692,
      "step": 3210
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7791310511627406,
      "learning_rate": 6.682172473993e-06,
      "loss": 0.5799,
      "step": 3211
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7326601545534721,
      "learning_rate": 6.680229299565838e-06,
      "loss": 0.6027,
      "step": 3212
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7525460036744519,
      "learning_rate": 6.67828583901313e-06,
      "loss": 0.618,
      "step": 3213
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.812527373634474,
      "learning_rate": 6.676342092665827e-06,
      "loss": 0.6174,
      "step": 3214
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8206434490483032,
      "learning_rate": 6.674398060854931e-06,
      "loss": 0.6305,
      "step": 3215
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8654762681317059,
      "learning_rate": 6.672453743911489e-06,
      "loss": 0.6324,
      "step": 3216
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8078821289059402,
      "learning_rate": 6.670509142166597e-06,
      "loss": 0.5754,
      "step": 3217
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.928748710887052,
      "learning_rate": 6.668564255951402e-06,
      "loss": 0.641,
      "step": 3218
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2234072340891011,
      "learning_rate": 6.666619085597097e-06,
      "loss": 0.7043,
      "step": 3219
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7736439174914075,
      "learning_rate": 6.664673631434925e-06,
      "loss": 0.5805,
      "step": 3220
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.774103313814348,
      "learning_rate": 6.662727893796176e-06,
      "loss": 0.5273,
      "step": 3221
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9769044715362657,
      "learning_rate": 6.6607818730121876e-06,
      "loss": 0.6268,
      "step": 3222
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6377835877454072,
      "learning_rate": 6.65883556941435e-06,
      "loss": 0.5199,
      "step": 3223
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.683265051113322,
      "learning_rate": 6.656888983334095e-06,
      "loss": 0.5405,
      "step": 3224
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9362381677066479,
      "learning_rate": 6.654942115102908e-06,
      "loss": 0.6376,
      "step": 3225
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7305770537409942,
      "learning_rate": 6.6529949650523195e-06,
      "loss": 0.5437,
      "step": 3226
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9851928756111706,
      "learning_rate": 6.65104753351391e-06,
      "loss": 0.6024,
      "step": 3227
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7763081241754975,
      "learning_rate": 6.649099820819305e-06,
      "loss": 0.5895,
      "step": 3228
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8984791885716891,
      "learning_rate": 6.647151827300181e-06,
      "loss": 0.6462,
      "step": 3229
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6651638815892492,
      "learning_rate": 6.645203553288262e-06,
      "loss": 0.5904,
      "step": 3230
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.730746526323111,
      "learning_rate": 6.643254999115316e-06,
      "loss": 0.593,
      "step": 3231
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7611130732725617,
      "learning_rate": 6.641306165113164e-06,
      "loss": 0.5295,
      "step": 3232
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9163102768906297,
      "learning_rate": 6.639357051613671e-06,
      "loss": 0.651,
      "step": 3233
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6690449005103866,
      "learning_rate": 6.637407658948749e-06,
      "loss": 0.5268,
      "step": 3234
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6694288100444343,
      "learning_rate": 6.635457987450364e-06,
      "loss": 0.5868,
      "step": 3235
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1820348620862644,
      "learning_rate": 6.633508037450519e-06,
      "loss": 0.6284,
      "step": 3236
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0090864571388356,
      "learning_rate": 6.631557809281275e-06,
      "loss": 0.7001,
      "step": 3237
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7752070413352607,
      "learning_rate": 6.6296073032747325e-06,
      "loss": 0.6348,
      "step": 3238
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6734603115598812,
      "learning_rate": 6.627656519763044e-06,
      "loss": 0.5358,
      "step": 3239
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9471781638965864,
      "learning_rate": 6.625705459078407e-06,
      "loss": 0.6525,
      "step": 3240
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6999674640456416,
      "learning_rate": 6.623754121553065e-06,
      "loss": 0.5641,
      "step": 3241
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8919684115933687,
      "learning_rate": 6.621802507519314e-06,
      "loss": 0.5666,
      "step": 3242
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7821390216633386,
      "learning_rate": 6.619850617309491e-06,
      "loss": 0.6135,
      "step": 3243
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7809944463346813,
      "learning_rate": 6.617898451255984e-06,
      "loss": 0.6113,
      "step": 3244
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8128425655159538,
      "learning_rate": 6.6159460096912254e-06,
      "loss": 0.6486,
      "step": 3245
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7135439547230367,
      "learning_rate": 6.613993292947697e-06,
      "loss": 0.5485,
      "step": 3246
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6700150938470216,
      "learning_rate": 6.612040301357925e-06,
      "loss": 0.5958,
      "step": 3247
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8256094375685624,
      "learning_rate": 6.6100870352544835e-06,
      "loss": 0.5116,
      "step": 3248
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8378796317064201,
      "learning_rate": 6.608133494969993e-06,
      "loss": 0.6825,
      "step": 3249
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7573670340308144,
      "learning_rate": 6.606179680837122e-06,
      "loss": 0.5677,
      "step": 3250
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6109933129894055,
      "learning_rate": 6.604225593188585e-06,
      "loss": 0.5067,
      "step": 3251
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7893168109840055,
      "learning_rate": 6.602271232357142e-06,
      "loss": 0.6396,
      "step": 3252
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8049400645892669,
      "learning_rate": 6.6003165986756015e-06,
      "loss": 0.6086,
      "step": 3253
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8224229186815304,
      "learning_rate": 6.598361692476813e-06,
      "loss": 0.5751,
      "step": 3254
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6918534181976074,
      "learning_rate": 6.5964065140936825e-06,
      "loss": 0.5364,
      "step": 3255
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7556663170682492,
      "learning_rate": 6.594451063859154e-06,
      "loss": 0.595,
      "step": 3256
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8494250958296593,
      "learning_rate": 6.592495342106219e-06,
      "loss": 0.691,
      "step": 3257
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8384121881684764,
      "learning_rate": 6.5905393491679185e-06,
      "loss": 0.6377,
      "step": 3258
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7236954015988676,
      "learning_rate": 6.5885830853773365e-06,
      "loss": 0.5868,
      "step": 3259
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7435237522385246,
      "learning_rate": 6.586626551067605e-06,
      "loss": 0.5769,
      "step": 3260
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8433511768074987,
      "learning_rate": 6.584669746571902e-06,
      "loss": 0.6555,
      "step": 3261
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6734027508286957,
      "learning_rate": 6.582712672223449e-06,
      "loss": 0.5369,
      "step": 3262
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8051481259987183,
      "learning_rate": 6.580755328355517e-06,
      "loss": 0.5823,
      "step": 3263
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6553404341817208,
      "learning_rate": 6.578797715301422e-06,
      "loss": 0.5357,
      "step": 3264
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9108072738939391,
      "learning_rate": 6.576839833394522e-06,
      "loss": 0.6937,
      "step": 3265
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7943175938690936,
      "learning_rate": 6.574881682968226e-06,
      "loss": 0.6032,
      "step": 3266
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7722330966012264,
      "learning_rate": 6.572923264355987e-06,
      "loss": 0.5858,
      "step": 3267
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8251817943039168,
      "learning_rate": 6.570964577891303e-06,
      "loss": 0.6818,
      "step": 3268
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7406711665473202,
      "learning_rate": 6.569005623907717e-06,
      "loss": 0.5626,
      "step": 3269
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7480439126692651,
      "learning_rate": 6.567046402738818e-06,
      "loss": 0.5739,
      "step": 3270
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6859661587202633,
      "learning_rate": 6.565086914718244e-06,
      "loss": 0.5747,
      "step": 3271
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9746964468080379,
      "learning_rate": 6.563127160179672e-06,
      "loss": 0.6827,
      "step": 3272
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.76062411408561,
      "learning_rate": 6.56116713945683e-06,
      "loss": 0.5569,
      "step": 3273
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6387530834451175,
      "learning_rate": 6.559206852883487e-06,
      "loss": 0.5463,
      "step": 3274
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7755272404170621,
      "learning_rate": 6.557246300793462e-06,
      "loss": 0.6121,
      "step": 3275
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6536370141932599,
      "learning_rate": 6.555285483520615e-06,
      "loss": 0.5257,
      "step": 3276
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.727177598312349,
      "learning_rate": 6.553324401398853e-06,
      "loss": 0.5832,
      "step": 3277
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8894720674093568,
      "learning_rate": 6.551363054762128e-06,
      "loss": 0.595,
      "step": 3278
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.659201630162912,
      "learning_rate": 6.549401443944436e-06,
      "loss": 0.5,
      "step": 3279
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6681382593528573,
      "learning_rate": 6.54743956927982e-06,
      "loss": 0.5729,
      "step": 3280
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6739172559160317,
      "learning_rate": 6.5454774311023676e-06,
      "loss": 0.5589,
      "step": 3281
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7622521489361416,
      "learning_rate": 6.543515029746208e-06,
      "loss": 0.5779,
      "step": 3282
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.030900091439409,
      "learning_rate": 6.541552365545518e-06,
      "loss": 0.6408,
      "step": 3283
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8581221801617982,
      "learning_rate": 6.5395894388345215e-06,
      "loss": 0.6784,
      "step": 3284
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7754308508905653,
      "learning_rate": 6.5376262499474814e-06,
      "loss": 0.5708,
      "step": 3285
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6681913473673582,
      "learning_rate": 6.535662799218712e-06,
      "loss": 0.5517,
      "step": 3286
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7848583184110279,
      "learning_rate": 6.533699086982566e-06,
      "loss": 0.631,
      "step": 3287
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7394092222256954,
      "learning_rate": 6.531735113573442e-06,
      "loss": 0.5935,
      "step": 3288
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8655978981898818,
      "learning_rate": 6.529770879325787e-06,
      "loss": 0.615,
      "step": 3289
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8389382955357877,
      "learning_rate": 6.527806384574088e-06,
      "loss": 0.6685,
      "step": 3290
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7119620542518433,
      "learning_rate": 6.525841629652878e-06,
      "loss": 0.533,
      "step": 3291
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6721143699624627,
      "learning_rate": 6.523876614896735e-06,
      "loss": 0.5628,
      "step": 3292
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6497921544202933,
      "learning_rate": 6.521911340640279e-06,
      "loss": 0.5232,
      "step": 3293
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7619317433627457,
      "learning_rate": 6.519945807218178e-06,
      "loss": 0.5049,
      "step": 3294
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8092639201925059,
      "learning_rate": 6.51798001496514e-06,
      "loss": 0.6662,
      "step": 3295
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6759734252859513,
      "learning_rate": 6.516013964215919e-06,
      "loss": 0.5307,
      "step": 3296
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8361731753538412,
      "learning_rate": 6.514047655305314e-06,
      "loss": 0.6509,
      "step": 3297
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8467167305197648,
      "learning_rate": 6.5120810885681676e-06,
      "loss": 0.6301,
      "step": 3298
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8954484946169039,
      "learning_rate": 6.510114264339364e-06,
      "loss": 0.5608,
      "step": 3299
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1282540994970105,
      "learning_rate": 6.508147182953834e-06,
      "loss": 0.5877,
      "step": 3300
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7537054709330002,
      "learning_rate": 6.506179844746549e-06,
      "loss": 0.6493,
      "step": 3301
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7565455803528692,
      "learning_rate": 6.504212250052529e-06,
      "loss": 0.5631,
      "step": 3302
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6349168407238737,
      "learning_rate": 6.5022443992068344e-06,
      "loss": 0.4754,
      "step": 3303
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8856287452165285,
      "learning_rate": 6.500276292544567e-06,
      "loss": 0.7341,
      "step": 3304
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7118507733962567,
      "learning_rate": 6.498307930400878e-06,
      "loss": 0.5452,
      "step": 3305
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6796914620581587,
      "learning_rate": 6.496339313110958e-06,
      "loss": 0.5468,
      "step": 3306
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8632277497757777,
      "learning_rate": 6.49437044101004e-06,
      "loss": 0.6439,
      "step": 3307
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9885064830006294,
      "learning_rate": 6.492401314433404e-06,
      "loss": 0.6547,
      "step": 3308
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7076459076699314,
      "learning_rate": 6.490431933716373e-06,
      "loss": 0.5183,
      "step": 3309
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6985180715895505,
      "learning_rate": 6.488462299194311e-06,
      "loss": 0.4945,
      "step": 3310
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8221277645352243,
      "learning_rate": 6.4864924112026265e-06,
      "loss": 0.6279,
      "step": 3311
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.659295492727098,
      "learning_rate": 6.48452227007677e-06,
      "loss": 0.5417,
      "step": 3312
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7079286192522138,
      "learning_rate": 6.4825518761522354e-06,
      "loss": 0.5624,
      "step": 3313
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7416336802491766,
      "learning_rate": 6.480581229764563e-06,
      "loss": 0.5624,
      "step": 3314
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5957354324484512,
      "learning_rate": 6.4786103312493305e-06,
      "loss": 0.4875,
      "step": 3315
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6026476741783775,
      "learning_rate": 6.476639180942165e-06,
      "loss": 0.5253,
      "step": 3316
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8939101132024457,
      "learning_rate": 6.474667779178728e-06,
      "loss": 0.6539,
      "step": 3317
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9139456793076355,
      "learning_rate": 6.472696126294733e-06,
      "loss": 0.6894,
      "step": 3318
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.882269295476322,
      "learning_rate": 6.470724222625929e-06,
      "loss": 0.6457,
      "step": 3319
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7659107548275412,
      "learning_rate": 6.468752068508112e-06,
      "loss": 0.5761,
      "step": 3320
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7347419684856539,
      "learning_rate": 6.46677966427712e-06,
      "loss": 0.6477,
      "step": 3321
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8403944827663434,
      "learning_rate": 6.464807010268831e-06,
      "loss": 0.6459,
      "step": 3322
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7556831214070445,
      "learning_rate": 6.462834106819168e-06,
      "loss": 0.528,
      "step": 3323
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7648250583968272,
      "learning_rate": 6.460860954264096e-06,
      "loss": 0.5894,
      "step": 3324
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8058757702569989,
      "learning_rate": 6.458887552939624e-06,
      "loss": 0.6275,
      "step": 3325
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7227942204377473,
      "learning_rate": 6.4569139031818e-06,
      "loss": 0.5899,
      "step": 3326
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8873906739070164,
      "learning_rate": 6.454940005326715e-06,
      "loss": 0.6917,
      "step": 3327
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7138882900276078,
      "learning_rate": 6.452965859710506e-06,
      "loss": 0.5414,
      "step": 3328
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7871329931534004,
      "learning_rate": 6.450991466669347e-06,
      "loss": 0.6504,
      "step": 3329
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.767741618757116,
      "learning_rate": 6.4490168265394595e-06,
      "loss": 0.5878,
      "step": 3330
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.952277623646695,
      "learning_rate": 6.4470419396571e-06,
      "loss": 0.6844,
      "step": 3331
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7524692268715799,
      "learning_rate": 6.445066806358575e-06,
      "loss": 0.5322,
      "step": 3332
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8826213828307394,
      "learning_rate": 6.4430914269802266e-06,
      "loss": 0.561,
      "step": 3333
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7756530450059665,
      "learning_rate": 6.441115801858442e-06,
      "loss": 0.5603,
      "step": 3334
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.11808653159282,
      "learning_rate": 6.43913993132965e-06,
      "loss": 0.6623,
      "step": 3335
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.706092270802816,
      "learning_rate": 6.437163815730321e-06,
      "loss": 0.5891,
      "step": 3336
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9657030209033668,
      "learning_rate": 6.435187455396966e-06,
      "loss": 0.6943,
      "step": 3337
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6363654338641853,
      "learning_rate": 6.43321085066614e-06,
      "loss": 0.5245,
      "step": 3338
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6787614067552249,
      "learning_rate": 6.431234001874435e-06,
      "loss": 0.539,
      "step": 3339
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6755614114080194,
      "learning_rate": 6.429256909358491e-06,
      "loss": 0.5731,
      "step": 3340
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6462196449743945,
      "learning_rate": 6.427279573454985e-06,
      "loss": 0.5647,
      "step": 3341
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6656767428094686,
      "learning_rate": 6.4253019945006375e-06,
      "loss": 0.5521,
      "step": 3342
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8294814349002999,
      "learning_rate": 6.423324172832209e-06,
      "loss": 0.605,
      "step": 3343
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6274353004528871,
      "learning_rate": 6.421346108786503e-06,
      "loss": 0.5693,
      "step": 3344
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8493797436396938,
      "learning_rate": 6.4193678027003605e-06,
      "loss": 0.6726,
      "step": 3345
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.680754667365435,
      "learning_rate": 6.417389254910669e-06,
      "loss": 0.56,
      "step": 3346
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8288774622844942,
      "learning_rate": 6.415410465754353e-06,
      "loss": 0.6532,
      "step": 3347
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.645400257288062,
      "learning_rate": 6.41343143556838e-06,
      "loss": 0.5221,
      "step": 3348
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8546403108356332,
      "learning_rate": 6.41145216468976e-06,
      "loss": 0.6671,
      "step": 3349
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6901347228111183,
      "learning_rate": 6.409472653455539e-06,
      "loss": 0.5282,
      "step": 3350
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7821090471580274,
      "learning_rate": 6.40749290220281e-06,
      "loss": 0.6305,
      "step": 3351
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8824899382906244,
      "learning_rate": 6.405512911268702e-06,
      "loss": 0.705,
      "step": 3352
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8471521120975053,
      "learning_rate": 6.403532680990389e-06,
      "loss": 0.6539,
      "step": 3353
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8162824895821622,
      "learning_rate": 6.401552211705082e-06,
      "loss": 0.6476,
      "step": 3354
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.690591417763809,
      "learning_rate": 6.399571503750034e-06,
      "loss": 0.5559,
      "step": 3355
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6625257415758892,
      "learning_rate": 6.39759055746254e-06,
      "loss": 0.5546,
      "step": 3356
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7010552369601221,
      "learning_rate": 6.395609373179934e-06,
      "loss": 0.5756,
      "step": 3357
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.922348560184355,
      "learning_rate": 6.3936279512395896e-06,
      "loss": 0.6047,
      "step": 3358
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6125535300753597,
      "learning_rate": 6.3916462919789255e-06,
      "loss": 0.5127,
      "step": 3359
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9984193379668501,
      "learning_rate": 6.389664395735395e-06,
      "loss": 0.6656,
      "step": 3360
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6765207552568084,
      "learning_rate": 6.387682262846498e-06,
      "loss": 0.5235,
      "step": 3361
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7356430724978521,
      "learning_rate": 6.385699893649766e-06,
      "loss": 0.549,
      "step": 3362
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7855497414438462,
      "learning_rate": 6.383717288482778e-06,
      "loss": 0.5793,
      "step": 3363
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6289425045496839,
      "learning_rate": 6.381734447683152e-06,
      "loss": 0.5347,
      "step": 3364
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.720860354892167,
      "learning_rate": 6.379751371588545e-06,
      "loss": 0.5343,
      "step": 3365
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6207474409677359,
      "learning_rate": 6.377768060536652e-06,
      "loss": 0.4902,
      "step": 3366
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9166928641723509,
      "learning_rate": 6.375784514865213e-06,
      "loss": 0.6777,
      "step": 3367
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6994130833180454,
      "learning_rate": 6.373800734912002e-06,
      "loss": 0.5226,
      "step": 3368
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8994609275271955,
      "learning_rate": 6.3718167210148406e-06,
      "loss": 0.594,
      "step": 3369
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8162041282330985,
      "learning_rate": 6.369832473511582e-06,
      "loss": 0.5705,
      "step": 3370
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8803012764198941,
      "learning_rate": 6.367847992740124e-06,
      "loss": 0.6564,
      "step": 3371
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8266403969820719,
      "learning_rate": 6.3658632790384026e-06,
      "loss": 0.5699,
      "step": 3372
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7893194416831567,
      "learning_rate": 6.363878332744395e-06,
      "loss": 0.6486,
      "step": 3373
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7913241913923671,
      "learning_rate": 6.361893154196116e-06,
      "loss": 0.6453,
      "step": 3374
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7243517419311286,
      "learning_rate": 6.35990774373162e-06,
      "loss": 0.644,
      "step": 3375
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6772032902068967,
      "learning_rate": 6.357922101689004e-06,
      "loss": 0.562,
      "step": 3376
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7607506820521205,
      "learning_rate": 6.3559362284064e-06,
      "loss": 0.5552,
      "step": 3377
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6513314158573268,
      "learning_rate": 6.353950124221982e-06,
      "loss": 0.5318,
      "step": 3378
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7386738028267484,
      "learning_rate": 6.351963789473965e-06,
      "loss": 0.5461,
      "step": 3379
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8219645023411236,
      "learning_rate": 6.349977224500598e-06,
      "loss": 0.6774,
      "step": 3380
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7122036293136335,
      "learning_rate": 6.347990429640171e-06,
      "loss": 0.5348,
      "step": 3381
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8694200500595037,
      "learning_rate": 6.34600340523102e-06,
      "loss": 0.6679,
      "step": 3382
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7834177254505951,
      "learning_rate": 6.344016151611512e-06,
      "loss": 0.6534,
      "step": 3383
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9621036791079701,
      "learning_rate": 6.342028669120055e-06,
      "loss": 0.6365,
      "step": 3384
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6445637695576225,
      "learning_rate": 6.340040958095096e-06,
      "loss": 0.5114,
      "step": 3385
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7002892693212692,
      "learning_rate": 6.338053018875123e-06,
      "loss": 0.5668,
      "step": 3386
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6427012913203892,
      "learning_rate": 6.3360648517986605e-06,
      "loss": 0.5089,
      "step": 3387
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8309286469572716,
      "learning_rate": 6.334076457204273e-06,
      "loss": 0.5901,
      "step": 3388
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8904877536179938,
      "learning_rate": 6.332087835430563e-06,
      "loss": 0.6104,
      "step": 3389
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6848042494746492,
      "learning_rate": 6.330098986816171e-06,
      "loss": 0.5645,
      "step": 3390
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9170523402598549,
      "learning_rate": 6.32810991169978e-06,
      "loss": 0.5533,
      "step": 3391
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8303882046461772,
      "learning_rate": 6.326120610420107e-06,
      "loss": 0.6303,
      "step": 3392
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.739424781398264,
      "learning_rate": 6.324131083315909e-06,
      "loss": 0.5918,
      "step": 3393
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8269463540364971,
      "learning_rate": 6.322141330725981e-06,
      "loss": 0.5703,
      "step": 3394
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8718846837649499,
      "learning_rate": 6.320151352989159e-06,
      "loss": 0.6841,
      "step": 3395
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9390321651150065,
      "learning_rate": 6.318161150444315e-06,
      "loss": 0.6061,
      "step": 3396
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8404559344639178,
      "learning_rate": 6.31617072343036e-06,
      "loss": 0.6101,
      "step": 3397
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8958865869468765,
      "learning_rate": 6.314180072286243e-06,
      "loss": 0.6653,
      "step": 3398
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6204165929426746,
      "learning_rate": 6.312189197350951e-06,
      "loss": 0.5059,
      "step": 3399
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8980621324416335,
      "learning_rate": 6.310198098963508e-06,
      "loss": 0.6779,
      "step": 3400
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7141825189434798,
      "learning_rate": 6.308206777462978e-06,
      "loss": 0.5351,
      "step": 3401
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9631986788553436,
      "learning_rate": 6.306215233188464e-06,
      "loss": 0.6848,
      "step": 3402
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.732145689973894,
      "learning_rate": 6.304223466479102e-06,
      "loss": 0.5444,
      "step": 3403
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7909167756925085,
      "learning_rate": 6.302231477674072e-06,
      "loss": 0.6223,
      "step": 3404
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.957639817787375,
      "learning_rate": 6.300239267112589e-06,
      "loss": 0.5978,
      "step": 3405
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.95503846064445,
      "learning_rate": 6.298246835133903e-06,
      "loss": 0.6449,
      "step": 3406
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7192761248867364,
      "learning_rate": 6.296254182077306e-06,
      "loss": 0.5416,
      "step": 3407
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8972144061795673,
      "learning_rate": 6.294261308282126e-06,
      "loss": 0.6646,
      "step": 3408
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9425585174614167,
      "learning_rate": 6.292268214087729e-06,
      "loss": 0.6346,
      "step": 3409
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.873319820796118,
      "learning_rate": 6.290274899833517e-06,
      "loss": 0.5603,
      "step": 3410
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9847792236937688,
      "learning_rate": 6.288281365858933e-06,
      "loss": 0.6632,
      "step": 3411
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8235799173340377,
      "learning_rate": 6.286287612503454e-06,
      "loss": 0.6541,
      "step": 3412
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7281333806282612,
      "learning_rate": 6.284293640106593e-06,
      "loss": 0.5721,
      "step": 3413
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7333538500503372,
      "learning_rate": 6.282299449007907e-06,
      "loss": 0.5071,
      "step": 3414
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8151021698022166,
      "learning_rate": 6.280305039546983e-06,
      "loss": 0.5713,
      "step": 3415
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.849450270845276,
      "learning_rate": 6.2783104120634484e-06,
      "loss": 0.5955,
      "step": 3416
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6026603441416055,
      "learning_rate": 6.276315566896969e-06,
      "loss": 0.5136,
      "step": 3417
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9111912869427533,
      "learning_rate": 6.274320504387245e-06,
      "loss": 0.5972,
      "step": 3418
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9638892838407052,
      "learning_rate": 6.2723252248740144e-06,
      "loss": 0.6435,
      "step": 3419
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.79699660822151,
      "learning_rate": 6.270329728697052e-06,
      "loss": 0.5577,
      "step": 3420
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8060204312928541,
      "learning_rate": 6.268334016196172e-06,
      "loss": 0.6052,
      "step": 3421
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8386635644919266,
      "learning_rate": 6.2663380877112225e-06,
      "loss": 0.5949,
      "step": 3422
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7819148664666169,
      "learning_rate": 6.264341943582088e-06,
      "loss": 0.6193,
      "step": 3423
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6867306027453859,
      "learning_rate": 6.262345584148694e-06,
      "loss": 0.5386,
      "step": 3424
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9066875185878596,
      "learning_rate": 6.260349009750997e-06,
      "loss": 0.5989,
      "step": 3425
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6457118958726119,
      "learning_rate": 6.258352220728995e-06,
      "loss": 0.5305,
      "step": 3426
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7885289603227443,
      "learning_rate": 6.256355217422718e-06,
      "loss": 0.5986,
      "step": 3427
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6247774194476489,
      "learning_rate": 6.254358000172237e-06,
      "loss": 0.5434,
      "step": 3428
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6663443836330664,
      "learning_rate": 6.252360569317656e-06,
      "loss": 0.5354,
      "step": 3429
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7236541518246997,
      "learning_rate": 6.2503629251991175e-06,
      "loss": 0.5963,
      "step": 3430
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7827945251263246,
      "learning_rate": 6.248365068156799e-06,
      "loss": 0.6051,
      "step": 3431
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7711966207571743,
      "learning_rate": 6.246366998530914e-06,
      "loss": 0.5494,
      "step": 3432
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6320784557289187,
      "learning_rate": 6.244368716661714e-06,
      "loss": 0.4818,
      "step": 3433
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.861065143332177,
      "learning_rate": 6.242370222889485e-06,
      "loss": 0.6408,
      "step": 3434
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8819630192489708,
      "learning_rate": 6.240371517554551e-06,
      "loss": 0.6368,
      "step": 3435
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6440729660424318,
      "learning_rate": 6.238372600997268e-06,
      "loss": 0.4757,
      "step": 3436
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6787641672683281,
      "learning_rate": 6.236373473558033e-06,
      "loss": 0.6229,
      "step": 3437
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8124451914813133,
      "learning_rate": 6.234374135577276e-06,
      "loss": 0.6297,
      "step": 3438
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6803340316552556,
      "learning_rate": 6.232374587395464e-06,
      "loss": 0.5954,
      "step": 3439
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7906846307228111,
      "learning_rate": 6.230374829353097e-06,
      "loss": 0.6785,
      "step": 3440
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8365586694505374,
      "learning_rate": 6.228374861790716e-06,
      "loss": 0.6049,
      "step": 3441
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6486857884705961,
      "learning_rate": 6.226374685048893e-06,
      "loss": 0.543,
      "step": 3442
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8592707801027841,
      "learning_rate": 6.224374299468237e-06,
      "loss": 0.6745,
      "step": 3443
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6773468169413863,
      "learning_rate": 6.222373705389394e-06,
      "loss": 0.5703,
      "step": 3444
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6473954419876244,
      "learning_rate": 6.220372903153042e-06,
      "loss": 0.5055,
      "step": 3445
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7308717183475844,
      "learning_rate": 6.218371893099901e-06,
      "loss": 0.5543,
      "step": 3446
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7599032713463516,
      "learning_rate": 6.216370675570718e-06,
      "loss": 0.6492,
      "step": 3447
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7425814312916679,
      "learning_rate": 6.214369250906281e-06,
      "loss": 0.5869,
      "step": 3448
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7837146390059074,
      "learning_rate": 6.212367619447409e-06,
      "loss": 0.6351,
      "step": 3449
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7578870373314055,
      "learning_rate": 6.210365781534965e-06,
      "loss": 0.5805,
      "step": 3450
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7432446663353631,
      "learning_rate": 6.208363737509837e-06,
      "loss": 0.5496,
      "step": 3451
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8906430499601756,
      "learning_rate": 6.206361487712954e-06,
      "loss": 0.6802,
      "step": 3452
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7684925724382321,
      "learning_rate": 6.204359032485276e-06,
      "loss": 0.5581,
      "step": 3453
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7727500789615049,
      "learning_rate": 6.202356372167804e-06,
      "loss": 0.5574,
      "step": 3454
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.668092793539768,
      "learning_rate": 6.200353507101566e-06,
      "loss": 0.5629,
      "step": 3455
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6625074376548281,
      "learning_rate": 6.198350437627631e-06,
      "loss": 0.5524,
      "step": 3456
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6894049680306608,
      "learning_rate": 6.196347164087101e-06,
      "loss": 0.5352,
      "step": 3457
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7509889646163803,
      "learning_rate": 6.1943436868211125e-06,
      "loss": 0.5783,
      "step": 3458
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6650465814104387,
      "learning_rate": 6.192340006170836e-06,
      "loss": 0.4934,
      "step": 3459
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.798882358357191,
      "learning_rate": 6.1903361224774795e-06,
      "loss": 0.6602,
      "step": 3460
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7390406362853507,
      "learning_rate": 6.188332036082281e-06,
      "loss": 0.5621,
      "step": 3461
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7529336428422556,
      "learning_rate": 6.186327747326518e-06,
      "loss": 0.6065,
      "step": 3462
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7058463708579121,
      "learning_rate": 6.184323256551498e-06,
      "loss": 0.5965,
      "step": 3463
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7549298935127112,
      "learning_rate": 6.182318564098565e-06,
      "loss": 0.5849,
      "step": 3464
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.726983957325392,
      "learning_rate": 6.180313670309098e-06,
      "loss": 0.5603,
      "step": 3465
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8024173155728049,
      "learning_rate": 6.17830857552451e-06,
      "loss": 0.6692,
      "step": 3466
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1304922894097105,
      "learning_rate": 6.176303280086246e-06,
      "loss": 0.6522,
      "step": 3467
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6998991038356294,
      "learning_rate": 6.17429778433579e-06,
      "loss": 0.5295,
      "step": 3468
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7902387494963926,
      "learning_rate": 6.172292088614654e-06,
      "loss": 0.6315,
      "step": 3469
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9392059131825539,
      "learning_rate": 6.170286193264388e-06,
      "loss": 0.7119,
      "step": 3470
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8974577064555453,
      "learning_rate": 6.168280098626576e-06,
      "loss": 0.6494,
      "step": 3471
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6901552558902487,
      "learning_rate": 6.166273805042834e-06,
      "loss": 0.5736,
      "step": 3472
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6521369446437644,
      "learning_rate": 6.1642673128548125e-06,
      "loss": 0.5279,
      "step": 3473
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7331700511723845,
      "learning_rate": 6.162260622404197e-06,
      "loss": 0.5311,
      "step": 3474
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7277537223983758,
      "learning_rate": 6.160253734032706e-06,
      "loss": 0.5502,
      "step": 3475
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8679190623864572,
      "learning_rate": 6.158246648082092e-06,
      "loss": 0.5371,
      "step": 3476
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9594178210859036,
      "learning_rate": 6.1562393648941395e-06,
      "loss": 0.6422,
      "step": 3477
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9163289569881978,
      "learning_rate": 6.154231884810669e-06,
      "loss": 0.6952,
      "step": 3478
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6283955424196445,
      "learning_rate": 6.152224208173533e-06,
      "loss": 0.5313,
      "step": 3479
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6533490352790934,
      "learning_rate": 6.150216335324619e-06,
      "loss": 0.5438,
      "step": 3480
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.82758793010938,
      "learning_rate": 6.148208266605845e-06,
      "loss": 0.5395,
      "step": 3481
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6167459873546352,
      "learning_rate": 6.146200002359166e-06,
      "loss": 0.5223,
      "step": 3482
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6970961302424825,
      "learning_rate": 6.144191542926567e-06,
      "loss": 0.5578,
      "step": 3483
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8276963716690393,
      "learning_rate": 6.142182888650069e-06,
      "loss": 0.5246,
      "step": 3484
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7320010005430326,
      "learning_rate": 6.1401740398717244e-06,
      "loss": 0.547,
      "step": 3485
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6787153450105404,
      "learning_rate": 6.138164996933619e-06,
      "loss": 0.5548,
      "step": 3486
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7637242521512037,
      "learning_rate": 6.136155760177871e-06,
      "loss": 0.5794,
      "step": 3487
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7691962063488876,
      "learning_rate": 6.134146329946634e-06,
      "loss": 0.6215,
      "step": 3488
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7946886967264549,
      "learning_rate": 6.132136706582092e-06,
      "loss": 0.6434,
      "step": 3489
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8027517099848307,
      "learning_rate": 6.130126890426464e-06,
      "loss": 0.6437,
      "step": 3490
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.668247338188106,
      "learning_rate": 6.128116881822e-06,
      "loss": 0.5752,
      "step": 3491
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7353780788653977,
      "learning_rate": 6.126106681110983e-06,
      "loss": 0.639,
      "step": 3492
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6923545859244065,
      "learning_rate": 6.12409628863573e-06,
      "loss": 0.5658,
      "step": 3493
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7263948063964855,
      "learning_rate": 6.122085704738591e-06,
      "loss": 0.526,
      "step": 3494
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9644986152593055,
      "learning_rate": 6.120074929761945e-06,
      "loss": 0.6848,
      "step": 3495
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8983537765060773,
      "learning_rate": 6.118063964048209e-06,
      "loss": 0.6447,
      "step": 3496
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7597788384948722,
      "learning_rate": 6.1160528079398265e-06,
      "loss": 0.5619,
      "step": 3497
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7889025067585369,
      "learning_rate": 6.114041461779278e-06,
      "loss": 0.6753,
      "step": 3498
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7463419371673947,
      "learning_rate": 6.112029925909074e-06,
      "loss": 0.5794,
      "step": 3499
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8118772213049076,
      "learning_rate": 6.110018200671759e-06,
      "loss": 0.6497,
      "step": 3500
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9611855445993419,
      "learning_rate": 6.108006286409909e-06,
      "loss": 0.6602,
      "step": 3501
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8162934389970316,
      "learning_rate": 6.105994183466131e-06,
      "loss": 0.5714,
      "step": 3502
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8014887738897735,
      "learning_rate": 6.103981892183066e-06,
      "loss": 0.5224,
      "step": 3503
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7646375086847529,
      "learning_rate": 6.101969412903386e-06,
      "loss": 0.5377,
      "step": 3504
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7568546424930852,
      "learning_rate": 6.099956745969794e-06,
      "loss": 0.6212,
      "step": 3505
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8655740577414859,
      "learning_rate": 6.097943891725029e-06,
      "loss": 0.6782,
      "step": 3506
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6090674379090918,
      "learning_rate": 6.095930850511857e-06,
      "loss": 0.5136,
      "step": 3507
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8328219945031706,
      "learning_rate": 6.093917622673079e-06,
      "loss": 0.6468,
      "step": 3508
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8695867484489413,
      "learning_rate": 6.091904208551525e-06,
      "loss": 0.6391,
      "step": 3509
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6817016831374344,
      "learning_rate": 6.08989060849006e-06,
      "loss": 0.5745,
      "step": 3510
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8323457265388752,
      "learning_rate": 6.08787682283158e-06,
      "loss": 0.6416,
      "step": 3511
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.903831229927447,
      "learning_rate": 6.085862851919009e-06,
      "loss": 0.6574,
      "step": 3512
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8432976116986718,
      "learning_rate": 6.083848696095308e-06,
      "loss": 0.5857,
      "step": 3513
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9373099001804219,
      "learning_rate": 6.081834355703464e-06,
      "loss": 0.6941,
      "step": 3514
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7065550748271121,
      "learning_rate": 6.079819831086502e-06,
      "loss": 0.5065,
      "step": 3515
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8233747924960846,
      "learning_rate": 6.07780512258747e-06,
      "loss": 0.6472,
      "step": 3516
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8131299143327212,
      "learning_rate": 6.075790230549456e-06,
      "loss": 0.5998,
      "step": 3517
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7168539855919053,
      "learning_rate": 6.073775155315572e-06,
      "loss": 0.5982,
      "step": 3518
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8659408625404263,
      "learning_rate": 6.071759897228966e-06,
      "loss": 0.6074,
      "step": 3519
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8790918891615503,
      "learning_rate": 6.069744456632815e-06,
      "loss": 0.6684,
      "step": 3520
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7471126658622433,
      "learning_rate": 6.067728833870326e-06,
      "loss": 0.4933,
      "step": 3521
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7094652301624043,
      "learning_rate": 6.065713029284741e-06,
      "loss": 0.5737,
      "step": 3522
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6730662547245935,
      "learning_rate": 6.063697043219331e-06,
      "loss": 0.518,
      "step": 3523
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8879363140401194,
      "learning_rate": 6.061680876017395e-06,
      "loss": 0.7149,
      "step": 3524
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0623829632061756,
      "learning_rate": 6.059664528022267e-06,
      "loss": 0.6261,
      "step": 3525
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7835680848472724,
      "learning_rate": 6.057647999577308e-06,
      "loss": 0.5941,
      "step": 3526
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9069880232593553,
      "learning_rate": 6.055631291025915e-06,
      "loss": 0.6029,
      "step": 3527
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9621275008705984,
      "learning_rate": 6.05361440271151e-06,
      "loss": 0.5603,
      "step": 3528
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.063859495364726,
      "learning_rate": 6.051597334977548e-06,
      "loss": 0.689,
      "step": 3529
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6157204691487683,
      "learning_rate": 6.049580088167516e-06,
      "loss": 0.5488,
      "step": 3530
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9396825773868592,
      "learning_rate": 6.047562662624929e-06,
      "loss": 0.6223,
      "step": 3531
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8865459072128999,
      "learning_rate": 6.045545058693334e-06,
      "loss": 0.7247,
      "step": 3532
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8405742156976307,
      "learning_rate": 6.043527276716308e-06,
      "loss": 0.6303,
      "step": 3533
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7471576610798449,
      "learning_rate": 6.041509317037457e-06,
      "loss": 0.6235,
      "step": 3534
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7565404706543675,
      "learning_rate": 6.039491180000421e-06,
      "loss": 0.6127,
      "step": 3535
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2189207973463183,
      "learning_rate": 6.037472865948867e-06,
      "loss": 0.6186,
      "step": 3536
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6560022488743839,
      "learning_rate": 6.035454375226492e-06,
      "loss": 0.5372,
      "step": 3537
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6975606129924329,
      "learning_rate": 6.033435708177023e-06,
      "loss": 0.5889,
      "step": 3538
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8169788988518024,
      "learning_rate": 6.031416865144219e-06,
      "loss": 0.6314,
      "step": 3539
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7501275138246684,
      "learning_rate": 6.02939784647187e-06,
      "loss": 0.5622,
      "step": 3540
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7102187583015673,
      "learning_rate": 6.02737865250379e-06,
      "loss": 0.5458,
      "step": 3541
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8597501825146079,
      "learning_rate": 6.025359283583831e-06,
      "loss": 0.5504,
      "step": 3542
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7867146017004523,
      "learning_rate": 6.023339740055865e-06,
      "loss": 0.5491,
      "step": 3543
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7323822664437999,
      "learning_rate": 6.0213200222638044e-06,
      "loss": 0.5192,
      "step": 3544
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8149291308931363,
      "learning_rate": 6.019300130551584e-06,
      "loss": 0.6214,
      "step": 3545
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7467867566560927,
      "learning_rate": 6.0172800652631706e-06,
      "loss": 0.5965,
      "step": 3546
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7830340190980437,
      "learning_rate": 6.015259826742558e-06,
      "loss": 0.6661,
      "step": 3547
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7917803039558575,
      "learning_rate": 6.013239415333776e-06,
      "loss": 0.6149,
      "step": 3548
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6567729714115069,
      "learning_rate": 6.011218831380876e-06,
      "loss": 0.5462,
      "step": 3549
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7646356726247474,
      "learning_rate": 6.009198075227944e-06,
      "loss": 0.6005,
      "step": 3550
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7154563381946574,
      "learning_rate": 6.0071771472190945e-06,
      "loss": 0.5355,
      "step": 3551
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.796096215999717,
      "learning_rate": 6.005156047698469e-06,
      "loss": 0.6325,
      "step": 3552
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6272674578695968,
      "learning_rate": 6.00313477701024e-06,
      "loss": 0.5258,
      "step": 3553
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7569466487664999,
      "learning_rate": 6.001113335498609e-06,
      "loss": 0.5954,
      "step": 3554
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.758959025451322,
      "learning_rate": 5.999091723507807e-06,
      "loss": 0.5895,
      "step": 3555
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7970607749674867,
      "learning_rate": 5.997069941382093e-06,
      "loss": 0.6455,
      "step": 3556
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7594809189802804,
      "learning_rate": 5.995047989465755e-06,
      "loss": 0.5598,
      "step": 3557
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6312521034906191,
      "learning_rate": 5.993025868103111e-06,
      "loss": 0.5183,
      "step": 3558
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9532511883000315,
      "learning_rate": 5.9910035776385075e-06,
      "loss": 0.5961,
      "step": 3559
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8288996676251822,
      "learning_rate": 5.988981118416317e-06,
      "loss": 0.6242,
      "step": 3560
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7732535077261793,
      "learning_rate": 5.986958490780946e-06,
      "loss": 0.5715,
      "step": 3561
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8134869982776157,
      "learning_rate": 5.984935695076826e-06,
      "loss": 0.6379,
      "step": 3562
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8345727494486916,
      "learning_rate": 5.982912731648418e-06,
      "loss": 0.6459,
      "step": 3563
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8724717029035415,
      "learning_rate": 5.980889600840213e-06,
      "loss": 0.5759,
      "step": 3564
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8496740271347862,
      "learning_rate": 5.978866302996726e-06,
      "loss": 0.6059,
      "step": 3565
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8211642048893611,
      "learning_rate": 5.976842838462508e-06,
      "loss": 0.6547,
      "step": 3566
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6050944808059421,
      "learning_rate": 5.974819207582128e-06,
      "loss": 0.5166,
      "step": 3567
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6430056041447777,
      "learning_rate": 5.9727954107001954e-06,
      "loss": 0.5545,
      "step": 3568
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8209134745003611,
      "learning_rate": 5.970771448161338e-06,
      "loss": 0.6467,
      "step": 3569
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7266106273924304,
      "learning_rate": 5.968747320310216e-06,
      "loss": 0.5014,
      "step": 3570
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6973875445855897,
      "learning_rate": 5.966723027491518e-06,
      "loss": 0.6057,
      "step": 3571
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9649722368068999,
      "learning_rate": 5.964698570049959e-06,
      "loss": 0.6122,
      "step": 3572
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9742032066653912,
      "learning_rate": 5.962673948330284e-06,
      "loss": 0.623,
      "step": 3573
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7682866362668807,
      "learning_rate": 5.960649162677266e-06,
      "loss": 0.5726,
      "step": 3574
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6804961138309907,
      "learning_rate": 5.958624213435702e-06,
      "loss": 0.5019,
      "step": 3575
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8143525669200827,
      "learning_rate": 5.956599100950421e-06,
      "loss": 0.6582,
      "step": 3576
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7860851262431328,
      "learning_rate": 5.954573825566279e-06,
      "loss": 0.5912,
      "step": 3577
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9165967135411455,
      "learning_rate": 5.95254838762816e-06,
      "loss": 0.6606,
      "step": 3578
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7400573386980802,
      "learning_rate": 5.950522787480975e-06,
      "loss": 0.5749,
      "step": 3579
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6843701652023471,
      "learning_rate": 5.94849702546966e-06,
      "loss": 0.5082,
      "step": 3580
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8108917329924451,
      "learning_rate": 5.946471101939184e-06,
      "loss": 0.5671,
      "step": 3581
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7649257744135524,
      "learning_rate": 5.9444450172345395e-06,
      "loss": 0.5747,
      "step": 3582
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7988511007666568,
      "learning_rate": 5.942418771700747e-06,
      "loss": 0.6589,
      "step": 3583
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8680266578569742,
      "learning_rate": 5.940392365682857e-06,
      "loss": 0.6409,
      "step": 3584
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6526629781748411,
      "learning_rate": 5.938365799525942e-06,
      "loss": 0.5161,
      "step": 3585
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7125939111667973,
      "learning_rate": 5.936339073575109e-06,
      "loss": 0.5344,
      "step": 3586
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9008213988987009,
      "learning_rate": 5.934312188175486e-06,
      "loss": 0.5582,
      "step": 3587
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7812539747149474,
      "learning_rate": 5.932285143672231e-06,
      "loss": 0.5783,
      "step": 3588
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8865067131235267,
      "learning_rate": 5.930257940410529e-06,
      "loss": 0.6586,
      "step": 3589
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.965208857151136,
      "learning_rate": 5.928230578735589e-06,
      "loss": 0.6186,
      "step": 3590
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8038857255786516,
      "learning_rate": 5.926203058992652e-06,
      "loss": 0.537,
      "step": 3591
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6253481993289098,
      "learning_rate": 5.924175381526985e-06,
      "loss": 0.5141,
      "step": 3592
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7419573127139674,
      "learning_rate": 5.922147546683878e-06,
      "loss": 0.564,
      "step": 3593
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7691705754525756,
      "learning_rate": 5.920119554808651e-06,
      "loss": 0.6022,
      "step": 3594
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9148032511283726,
      "learning_rate": 5.918091406246649e-06,
      "loss": 0.5691,
      "step": 3595
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7702565412128316,
      "learning_rate": 5.916063101343245e-06,
      "loss": 0.5514,
      "step": 3596
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6798873863850917,
      "learning_rate": 5.914034640443838e-06,
      "loss": 0.5357,
      "step": 3597
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7089581364282972,
      "learning_rate": 5.912006023893855e-06,
      "loss": 0.519,
      "step": 3598
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7303638554687092,
      "learning_rate": 5.909977252038747e-06,
      "loss": 0.5549,
      "step": 3599
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6741791288689153,
      "learning_rate": 5.907948325223993e-06,
      "loss": 0.4961,
      "step": 3600
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8632256466442529,
      "learning_rate": 5.905919243795099e-06,
      "loss": 0.6696,
      "step": 3601
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8726088211824585,
      "learning_rate": 5.903890008097594e-06,
      "loss": 0.6737,
      "step": 3602
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6866104334058141,
      "learning_rate": 5.9018606184770365e-06,
      "loss": 0.5441,
      "step": 3603
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9807599591662677,
      "learning_rate": 5.8998310752790135e-06,
      "loss": 0.5523,
      "step": 3604
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.83283830030592,
      "learning_rate": 5.897801378849132e-06,
      "loss": 0.6965,
      "step": 3605
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7485611876684435,
      "learning_rate": 5.895771529533028e-06,
      "loss": 0.6036,
      "step": 3606
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6532469783350606,
      "learning_rate": 5.893741527676366e-06,
      "loss": 0.5709,
      "step": 3607
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6491188015727933,
      "learning_rate": 5.891711373624832e-06,
      "loss": 0.5138,
      "step": 3608
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8034476961357356,
      "learning_rate": 5.889681067724141e-06,
      "loss": 0.59,
      "step": 3609
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8370396089461789,
      "learning_rate": 5.8876506103200325e-06,
      "loss": 0.5926,
      "step": 3610
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.746726257539987,
      "learning_rate": 5.885620001758273e-06,
      "loss": 0.5696,
      "step": 3611
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9319247931733304,
      "learning_rate": 5.883589242384653e-06,
      "loss": 0.6088,
      "step": 3612
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.739532771285927,
      "learning_rate": 5.881558332544991e-06,
      "loss": 0.5481,
      "step": 3613
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8861879647178765,
      "learning_rate": 5.879527272585128e-06,
      "loss": 0.6978,
      "step": 3614
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8324688284164715,
      "learning_rate": 5.877496062850934e-06,
      "loss": 0.6547,
      "step": 3615
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8422378637422568,
      "learning_rate": 5.875464703688303e-06,
      "loss": 0.6437,
      "step": 3616
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6164181540392585,
      "learning_rate": 5.873433195443152e-06,
      "loss": 0.5136,
      "step": 3617
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7213991595217203,
      "learning_rate": 5.871401538461428e-06,
      "loss": 0.56,
      "step": 3618
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6862213664480401,
      "learning_rate": 5.869369733089099e-06,
      "loss": 0.5537,
      "step": 3619
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6691614913578832,
      "learning_rate": 5.867337779672162e-06,
      "loss": 0.582,
      "step": 3620
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.980547989571627,
      "learning_rate": 5.865305678556638e-06,
      "loss": 0.5774,
      "step": 3621
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6774321067152484,
      "learning_rate": 5.863273430088571e-06,
      "loss": 0.5179,
      "step": 3622
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7428194786787314,
      "learning_rate": 5.861241034614032e-06,
      "loss": 0.5455,
      "step": 3623
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6530955408777839,
      "learning_rate": 5.859208492479118e-06,
      "loss": 0.5359,
      "step": 3624
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7003092039834383,
      "learning_rate": 5.857175804029949e-06,
      "loss": 0.5771,
      "step": 3625
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8069255311729716,
      "learning_rate": 5.8551429696126695e-06,
      "loss": 0.6595,
      "step": 3626
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7190154324593653,
      "learning_rate": 5.853109989573451e-06,
      "loss": 0.5661,
      "step": 3627
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.929649690606898,
      "learning_rate": 5.8510768642584904e-06,
      "loss": 0.6358,
      "step": 3628
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6993286664300414,
      "learning_rate": 5.8490435940140055e-06,
      "loss": 0.5727,
      "step": 3629
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.716654918268465,
      "learning_rate": 5.847010179186241e-06,
      "loss": 0.5707,
      "step": 3630
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7015636461180884,
      "learning_rate": 5.844976620121467e-06,
      "loss": 0.5473,
      "step": 3631
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6588378491063771,
      "learning_rate": 5.842942917165978e-06,
      "loss": 0.5246,
      "step": 3632
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7050502995673144,
      "learning_rate": 5.840909070666092e-06,
      "loss": 0.5319,
      "step": 3633
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6964475889484161,
      "learning_rate": 5.838875080968153e-06,
      "loss": 0.5193,
      "step": 3634
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6001425658259114,
      "learning_rate": 5.836840948418525e-06,
      "loss": 0.513,
      "step": 3635
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8071858862935555,
      "learning_rate": 5.834806673363604e-06,
      "loss": 0.6224,
      "step": 3636
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7774387032493759,
      "learning_rate": 5.832772256149802e-06,
      "loss": 0.5276,
      "step": 3637
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7016516287517406,
      "learning_rate": 5.830737697123561e-06,
      "loss": 0.5535,
      "step": 3638
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8270310853968084,
      "learning_rate": 5.828702996631344e-06,
      "loss": 0.6636,
      "step": 3639
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7536480354992842,
      "learning_rate": 5.82666815501964e-06,
      "loss": 0.5495,
      "step": 3640
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7281284638789438,
      "learning_rate": 5.824633172634963e-06,
      "loss": 0.5387,
      "step": 3641
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9235042834152511,
      "learning_rate": 5.822598049823845e-06,
      "loss": 0.7081,
      "step": 3642
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8715182184684162,
      "learning_rate": 5.820562786932851e-06,
      "loss": 0.6445,
      "step": 3643
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8064603342498802,
      "learning_rate": 5.818527384308564e-06,
      "loss": 0.6191,
      "step": 3644
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8411031428683867,
      "learning_rate": 5.816491842297587e-06,
      "loss": 0.5927,
      "step": 3645
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8069680573343667,
      "learning_rate": 5.814456161246557e-06,
      "loss": 0.6584,
      "step": 3646
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8145084948138652,
      "learning_rate": 5.812420341502128e-06,
      "loss": 0.6138,
      "step": 3647
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8024624359240449,
      "learning_rate": 5.81038438341098e-06,
      "loss": 0.5643,
      "step": 3648
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.865736303200125,
      "learning_rate": 5.808348287319813e-06,
      "loss": 0.5542,
      "step": 3649
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9799358162718407,
      "learning_rate": 5.806312053575353e-06,
      "loss": 0.6078,
      "step": 3650
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7188163395401812,
      "learning_rate": 5.804275682524352e-06,
      "loss": 0.5836,
      "step": 3651
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.926139714498065,
      "learning_rate": 5.80223917451358e-06,
      "loss": 0.6583,
      "step": 3652
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8419297860870829,
      "learning_rate": 5.8002025298898345e-06,
      "loss": 0.6676,
      "step": 3653
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7609251570265859,
      "learning_rate": 5.798165748999935e-06,
      "loss": 0.537,
      "step": 3654
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7383794341613403,
      "learning_rate": 5.796128832190723e-06,
      "loss": 0.5641,
      "step": 3655
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0003546097222928,
      "learning_rate": 5.794091779809066e-06,
      "loss": 0.6535,
      "step": 3656
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8619725279897564,
      "learning_rate": 5.79205459220185e-06,
      "loss": 0.6203,
      "step": 3657
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8293298685765608,
      "learning_rate": 5.790017269715989e-06,
      "loss": 0.6131,
      "step": 3658
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8474196137794393,
      "learning_rate": 5.787979812698418e-06,
      "loss": 0.6504,
      "step": 3659
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.746441019142809,
      "learning_rate": 5.785942221496093e-06,
      "loss": 0.6754,
      "step": 3660
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9270722329816211,
      "learning_rate": 5.783904496455997e-06,
      "loss": 0.6662,
      "step": 3661
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.692865919234672,
      "learning_rate": 5.78186663792513e-06,
      "loss": 0.5406,
      "step": 3662
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8327666950022071,
      "learning_rate": 5.779828646250522e-06,
      "loss": 0.652,
      "step": 3663
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7862746699614179,
      "learning_rate": 5.7777905217792184e-06,
      "loss": 0.6654,
      "step": 3664
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6382482925009612,
      "learning_rate": 5.775752264858292e-06,
      "loss": 0.5361,
      "step": 3665
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7728858052443739,
      "learning_rate": 5.773713875834837e-06,
      "loss": 0.6085,
      "step": 3666
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7020843335874097,
      "learning_rate": 5.77167535505597e-06,
      "loss": 0.6171,
      "step": 3667
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6462785973893841,
      "learning_rate": 5.76963670286883e-06,
      "loss": 0.5527,
      "step": 3668
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7726695942855915,
      "learning_rate": 5.7675979196205755e-06,
      "loss": 0.6092,
      "step": 3669
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7688289089436985,
      "learning_rate": 5.765559005658393e-06,
      "loss": 0.6061,
      "step": 3670
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8096084056787889,
      "learning_rate": 5.763519961329487e-06,
      "loss": 0.6468,
      "step": 3671
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7428833079841595,
      "learning_rate": 5.761480786981086e-06,
      "loss": 0.5737,
      "step": 3672
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8306835704735783,
      "learning_rate": 5.75944148296044e-06,
      "loss": 0.6355,
      "step": 3673
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7059843220118877,
      "learning_rate": 5.757402049614821e-06,
      "loss": 0.4727,
      "step": 3674
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.962968332672543,
      "learning_rate": 5.755362487291525e-06,
      "loss": 0.6483,
      "step": 3675
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8336546228762126,
      "learning_rate": 5.753322796337864e-06,
      "loss": 0.5706,
      "step": 3676
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8543824541382932,
      "learning_rate": 5.751282977101181e-06,
      "loss": 0.5984,
      "step": 3677
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.728187292212504,
      "learning_rate": 5.7492430299288335e-06,
      "loss": 0.5305,
      "step": 3678
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6553315642963978,
      "learning_rate": 5.747202955168204e-06,
      "loss": 0.5587,
      "step": 3679
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8134879986716608,
      "learning_rate": 5.745162753166696e-06,
      "loss": 0.6057,
      "step": 3680
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6850167887067071,
      "learning_rate": 5.743122424271732e-06,
      "loss": 0.4995,
      "step": 3681
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6567386437740192,
      "learning_rate": 5.741081968830762e-06,
      "loss": 0.5408,
      "step": 3682
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7803618123945372,
      "learning_rate": 5.739041387191253e-06,
      "loss": 0.6484,
      "step": 3683
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6474000448146338,
      "learning_rate": 5.737000679700696e-06,
      "loss": 0.5322,
      "step": 3684
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8754652193278403,
      "learning_rate": 5.7349598467066e-06,
      "loss": 0.6128,
      "step": 3685
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8896300820246065,
      "learning_rate": 5.7329188885565e-06,
      "loss": 0.64,
      "step": 3686
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7158370670589107,
      "learning_rate": 5.730877805597947e-06,
      "loss": 0.5523,
      "step": 3687
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.702051930924746,
      "learning_rate": 5.72883659817852e-06,
      "loss": 0.5614,
      "step": 3688
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8428446053224009,
      "learning_rate": 5.726795266645813e-06,
      "loss": 0.5947,
      "step": 3689
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8403297534339824,
      "learning_rate": 5.724753811347442e-06,
      "loss": 0.6163,
      "step": 3690
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.828982464796016,
      "learning_rate": 5.7227122326310495e-06,
      "loss": 0.6314,
      "step": 3691
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.744373009461017,
      "learning_rate": 5.7206705308442925e-06,
      "loss": 0.6471,
      "step": 3692
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6206768764700785,
      "learning_rate": 5.718628706334851e-06,
      "loss": 0.4867,
      "step": 3693
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7250433412225785,
      "learning_rate": 5.716586759450428e-06,
      "loss": 0.5746,
      "step": 3694
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8780566759722332,
      "learning_rate": 5.7145446905387445e-06,
      "loss": 0.6113,
      "step": 3695
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7808163405403723,
      "learning_rate": 5.712502499947545e-06,
      "loss": 0.5113,
      "step": 3696
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.746959986955119,
      "learning_rate": 5.710460188024593e-06,
      "loss": 0.4809,
      "step": 3697
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7000890412654375,
      "learning_rate": 5.708417755117672e-06,
      "loss": 0.5736,
      "step": 3698
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8515011529441295,
      "learning_rate": 5.706375201574588e-06,
      "loss": 0.5765,
      "step": 3699
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9734989516968692,
      "learning_rate": 5.704332527743166e-06,
      "loss": 0.6231,
      "step": 3700
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7890209810814113,
      "learning_rate": 5.70228973397125e-06,
      "loss": 0.6893,
      "step": 3701
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8471272572479925,
      "learning_rate": 5.700246820606711e-06,
      "loss": 0.6464,
      "step": 3702
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6482022032900072,
      "learning_rate": 5.698203787997434e-06,
      "loss": 0.4692,
      "step": 3703
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.763134369242582,
      "learning_rate": 5.6961606364913246e-06,
      "loss": 0.6232,
      "step": 3704
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6657332618017631,
      "learning_rate": 5.694117366436312e-06,
      "loss": 0.5883,
      "step": 3705
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8109451359751612,
      "learning_rate": 5.692073978180342e-06,
      "loss": 0.6369,
      "step": 3706
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7031437301592737,
      "learning_rate": 5.690030472071385e-06,
      "loss": 0.5359,
      "step": 3707
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6981914501462341,
      "learning_rate": 5.687986848457427e-06,
      "loss": 0.5198,
      "step": 3708
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8266180031887741,
      "learning_rate": 5.685943107686476e-06,
      "loss": 0.6243,
      "step": 3709
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7214386634363655,
      "learning_rate": 5.68389925010656e-06,
      "loss": 0.5538,
      "step": 3710
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8108786760433367,
      "learning_rate": 5.681855276065725e-06,
      "loss": 0.6193,
      "step": 3711
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7756792116709265,
      "learning_rate": 5.679811185912041e-06,
      "loss": 0.5724,
      "step": 3712
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6840163115464162,
      "learning_rate": 5.677766979993593e-06,
      "loss": 0.5462,
      "step": 3713
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7608859089870624,
      "learning_rate": 5.675722658658491e-06,
      "loss": 0.645,
      "step": 3714
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7715700043634773,
      "learning_rate": 5.67367822225486e-06,
      "loss": 0.6193,
      "step": 3715
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9998495163177669,
      "learning_rate": 5.671633671130845e-06,
      "loss": 0.6676,
      "step": 3716
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8124420524609709,
      "learning_rate": 5.669589005634613e-06,
      "loss": 0.6133,
      "step": 3717
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.700959602686,
      "learning_rate": 5.66754422611435e-06,
      "loss": 0.5528,
      "step": 3718
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8339189274409947,
      "learning_rate": 5.665499332918258e-06,
      "loss": 0.6655,
      "step": 3719
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6197495959599929,
      "learning_rate": 5.663454326394564e-06,
      "loss": 0.5088,
      "step": 3720
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9491850234070007,
      "learning_rate": 5.661409206891511e-06,
      "loss": 0.6651,
      "step": 3721
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7665697388040785,
      "learning_rate": 5.65936397475736e-06,
      "loss": 0.5678,
      "step": 3722
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.698943193700711,
      "learning_rate": 5.657318630340391e-06,
      "loss": 0.5073,
      "step": 3723
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7364758846080469,
      "learning_rate": 5.65527317398891e-06,
      "loss": 0.5439,
      "step": 3724
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9065666002066101,
      "learning_rate": 5.653227606051233e-06,
      "loss": 0.5655,
      "step": 3725
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7155611546844571,
      "learning_rate": 5.651181926875699e-06,
      "loss": 0.5664,
      "step": 3726
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8047046981373644,
      "learning_rate": 5.649136136810669e-06,
      "loss": 0.658,
      "step": 3727
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8810795909433238,
      "learning_rate": 5.647090236204516e-06,
      "loss": 0.6619,
      "step": 3728
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8406971967581554,
      "learning_rate": 5.645044225405638e-06,
      "loss": 0.6436,
      "step": 3729
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6474976185117174,
      "learning_rate": 5.642998104762447e-06,
      "loss": 0.5148,
      "step": 3730
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.658248919949886,
      "learning_rate": 5.6409518746233785e-06,
      "loss": 0.5838,
      "step": 3731
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9169710061876848,
      "learning_rate": 5.6389055353368826e-06,
      "loss": 0.6476,
      "step": 3732
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.18774585626679,
      "learning_rate": 5.63685908725143e-06,
      "loss": 0.6249,
      "step": 3733
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8975256633068401,
      "learning_rate": 5.634812530715511e-06,
      "loss": 0.6952,
      "step": 3734
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6350085756192091,
      "learning_rate": 5.63276586607763e-06,
      "loss": 0.5354,
      "step": 3735
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.882510878299978,
      "learning_rate": 5.630719093686314e-06,
      "loss": 0.6703,
      "step": 3736
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9157707863830172,
      "learning_rate": 5.628672213890109e-06,
      "loss": 0.6799,
      "step": 3737
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7780857816730271,
      "learning_rate": 5.626625227037574e-06,
      "loss": 0.5319,
      "step": 3738
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7691684915914921,
      "learning_rate": 5.624578133477291e-06,
      "loss": 0.6242,
      "step": 3739
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6847686824295655,
      "learning_rate": 5.6225309335578595e-06,
      "loss": 0.5954,
      "step": 3740
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6943218906562972,
      "learning_rate": 5.620483627627895e-06,
      "loss": 0.5649,
      "step": 3741
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8789568332691308,
      "learning_rate": 5.618436216036033e-06,
      "loss": 0.6696,
      "step": 3742
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7092032572782793,
      "learning_rate": 5.616388699130925e-06,
      "loss": 0.5506,
      "step": 3743
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7687916989324135,
      "learning_rate": 5.614341077261243e-06,
      "loss": 0.6322,
      "step": 3744
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6988407095155904,
      "learning_rate": 5.6122933507756775e-06,
      "loss": 0.5843,
      "step": 3745
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7417827124668309,
      "learning_rate": 5.610245520022933e-06,
      "loss": 0.5817,
      "step": 3746
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8245333582737887,
      "learning_rate": 5.608197585351732e-06,
      "loss": 0.5937,
      "step": 3747
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8124088065892567,
      "learning_rate": 5.60614954711082e-06,
      "loss": 0.6652,
      "step": 3748
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7706309684712072,
      "learning_rate": 5.604101405648955e-06,
      "loss": 0.6225,
      "step": 3749
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8296362414086256,
      "learning_rate": 5.602053161314913e-06,
      "loss": 0.6552,
      "step": 3750
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8039178612599128,
      "learning_rate": 5.6000048144574914e-06,
      "loss": 0.6294,
      "step": 3751
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7064968717210245,
      "learning_rate": 5.5979563654255e-06,
      "loss": 0.5353,
      "step": 3752
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8593535427876091,
      "learning_rate": 5.595907814567768e-06,
      "loss": 0.6637,
      "step": 3753
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7354470629092521,
      "learning_rate": 5.593859162233145e-06,
      "loss": 0.6243,
      "step": 3754
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8225093057451864,
      "learning_rate": 5.5918104087704925e-06,
      "loss": 0.593,
      "step": 3755
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7065847780582107,
      "learning_rate": 5.589761554528692e-06,
      "loss": 0.6409,
      "step": 3756
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8257092559323711,
      "learning_rate": 5.587712599856644e-06,
      "loss": 0.6515,
      "step": 3757
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8230713519589216,
      "learning_rate": 5.585663545103263e-06,
      "loss": 0.5718,
      "step": 3758
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7767947931328489,
      "learning_rate": 5.583614390617481e-06,
      "loss": 0.6196,
      "step": 3759
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6540839347134032,
      "learning_rate": 5.581565136748248e-06,
      "loss": 0.5232,
      "step": 3760
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6320646035950085,
      "learning_rate": 5.579515783844532e-06,
      "loss": 0.4825,
      "step": 3761
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.75485151863857,
      "learning_rate": 5.577466332255316e-06,
      "loss": 0.5334,
      "step": 3762
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6606289169897692,
      "learning_rate": 5.575416782329598e-06,
      "loss": 0.5083,
      "step": 3763
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7666985575707803,
      "learning_rate": 5.573367134416397e-06,
      "loss": 0.5701,
      "step": 3764
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8263293495191297,
      "learning_rate": 5.5713173888647456e-06,
      "loss": 0.5969,
      "step": 3765
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7483891732093688,
      "learning_rate": 5.569267546023696e-06,
      "loss": 0.6352,
      "step": 3766
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7130565347986373,
      "learning_rate": 5.5672176062423145e-06,
      "loss": 0.5474,
      "step": 3767
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7917777041224456,
      "learning_rate": 5.565167569869683e-06,
      "loss": 0.6348,
      "step": 3768
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9286400057524099,
      "learning_rate": 5.563117437254902e-06,
      "loss": 0.716,
      "step": 3769
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7352809625167775,
      "learning_rate": 5.561067208747089e-06,
      "loss": 0.5747,
      "step": 3770
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.656574420159467,
      "learning_rate": 5.559016884695374e-06,
      "loss": 0.5745,
      "step": 3771
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.788370359305701,
      "learning_rate": 5.556966465448907e-06,
      "loss": 0.6095,
      "step": 3772
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8243774848418189,
      "learning_rate": 5.554915951356854e-06,
      "loss": 0.6384,
      "step": 3773
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8239082845827175,
      "learning_rate": 5.552865342768397e-06,
      "loss": 0.6129,
      "step": 3774
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6731821369513035,
      "learning_rate": 5.55081464003273e-06,
      "loss": 0.4963,
      "step": 3775
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7018677559325273,
      "learning_rate": 5.54876384349907e-06,
      "loss": 0.5437,
      "step": 3776
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7747102564655425,
      "learning_rate": 5.546712953516643e-06,
      "loss": 0.5551,
      "step": 3777
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7403891080268005,
      "learning_rate": 5.544661970434696e-06,
      "loss": 0.6009,
      "step": 3778
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.78713833005115,
      "learning_rate": 5.54261089460249e-06,
      "loss": 0.5962,
      "step": 3779
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.851454242443393,
      "learning_rate": 5.540559726369301e-06,
      "loss": 0.667,
      "step": 3780
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8756948364494134,
      "learning_rate": 5.5385084660844225e-06,
      "loss": 0.6565,
      "step": 3781
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7583746715174433,
      "learning_rate": 5.536457114097163e-06,
      "loss": 0.5506,
      "step": 3782
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8351209786717441,
      "learning_rate": 5.534405670756845e-06,
      "loss": 0.6051,
      "step": 3783
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8032521502231091,
      "learning_rate": 5.5323541364128085e-06,
      "loss": 0.6124,
      "step": 3784
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8057952881010237,
      "learning_rate": 5.530302511414407e-06,
      "loss": 0.6433,
      "step": 3785
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6034973781951312,
      "learning_rate": 5.5282507961110146e-06,
      "loss": 0.5121,
      "step": 3786
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7045495916877996,
      "learning_rate": 5.526198990852014e-06,
      "loss": 0.6005,
      "step": 3787
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7881856535656534,
      "learning_rate": 5.524147095986808e-06,
      "loss": 0.6133,
      "step": 3788
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7077175798428397,
      "learning_rate": 5.522095111864813e-06,
      "loss": 0.5731,
      "step": 3789
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8790488114558154,
      "learning_rate": 5.5200430388354575e-06,
      "loss": 0.6433,
      "step": 3790
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7401839731625945,
      "learning_rate": 5.517990877248192e-06,
      "loss": 0.6429,
      "step": 3791
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7820097556621365,
      "learning_rate": 5.515938627452475e-06,
      "loss": 0.5196,
      "step": 3792
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7716335975875197,
      "learning_rate": 5.513886289797785e-06,
      "loss": 0.5808,
      "step": 3793
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6753117916435982,
      "learning_rate": 5.511833864633615e-06,
      "loss": 0.5428,
      "step": 3794
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6823198521790625,
      "learning_rate": 5.509781352309468e-06,
      "loss": 0.5685,
      "step": 3795
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1811555544161623,
      "learning_rate": 5.507728753174868e-06,
      "loss": 0.6959,
      "step": 3796
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6995180925161227,
      "learning_rate": 5.505676067579352e-06,
      "loss": 0.541,
      "step": 3797
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6959581559943104,
      "learning_rate": 5.503623295872468e-06,
      "loss": 0.5404,
      "step": 3798
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6682403949262287,
      "learning_rate": 5.5015704384037845e-06,
      "loss": 0.5605,
      "step": 3799
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8349675473208893,
      "learning_rate": 5.49951749552288e-06,
      "loss": 0.5985,
      "step": 3800
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6385448061715642,
      "learning_rate": 5.497464467579351e-06,
      "loss": 0.5018,
      "step": 3801
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.895833405280133,
      "learning_rate": 5.495411354922806e-06,
      "loss": 0.6358,
      "step": 3802
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7391780279163028,
      "learning_rate": 5.493358157902867e-06,
      "loss": 0.5718,
      "step": 3803
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7170843813999257,
      "learning_rate": 5.491304876869175e-06,
      "loss": 0.531,
      "step": 3804
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6868656287426803,
      "learning_rate": 5.48925151217138e-06,
      "loss": 0.5787,
      "step": 3805
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6632978398798797,
      "learning_rate": 5.487198064159149e-06,
      "loss": 0.508,
      "step": 3806
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7827899347295381,
      "learning_rate": 5.485144533182164e-06,
      "loss": 0.5992,
      "step": 3807
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8217609233260615,
      "learning_rate": 5.483090919590119e-06,
      "loss": 0.6143,
      "step": 3808
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7670942768175887,
      "learning_rate": 5.481037223732724e-06,
      "loss": 0.54,
      "step": 3809
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8146122915050482,
      "learning_rate": 5.4789834459597006e-06,
      "loss": 0.636,
      "step": 3810
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7870990608877836,
      "learning_rate": 5.476929586620786e-06,
      "loss": 0.6198,
      "step": 3811
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8919310737708784,
      "learning_rate": 5.4748756460657304e-06,
      "loss": 0.6187,
      "step": 3812
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6856255791581137,
      "learning_rate": 5.472821624644301e-06,
      "loss": 0.5791,
      "step": 3813
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8707993824599897,
      "learning_rate": 5.470767522706274e-06,
      "loss": 0.6232,
      "step": 3814
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6780172687583828,
      "learning_rate": 5.468713340601441e-06,
      "loss": 0.5409,
      "step": 3815
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.764302449066546,
      "learning_rate": 5.4666590786796115e-06,
      "loss": 0.599,
      "step": 3816
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9392956569368222,
      "learning_rate": 5.464604737290602e-06,
      "loss": 0.6255,
      "step": 3817
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7641492922132964,
      "learning_rate": 5.462550316784246e-06,
      "loss": 0.6148,
      "step": 3818
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7938853659027516,
      "learning_rate": 5.46049581751039e-06,
      "loss": 0.5942,
      "step": 3819
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7277916557270209,
      "learning_rate": 5.458441239818893e-06,
      "loss": 0.6048,
      "step": 3820
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7843583052764689,
      "learning_rate": 5.45638658405963e-06,
      "loss": 0.6311,
      "step": 3821
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7348089455768977,
      "learning_rate": 5.454331850582485e-06,
      "loss": 0.541,
      "step": 3822
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6537547719923894,
      "learning_rate": 5.4522770397373605e-06,
      "loss": 0.4931,
      "step": 3823
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8041038312236473,
      "learning_rate": 5.450222151874166e-06,
      "loss": 0.6436,
      "step": 3824
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8592330235699351,
      "learning_rate": 5.448167187342831e-06,
      "loss": 0.648,
      "step": 3825
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0003520613880121,
      "learning_rate": 5.446112146493292e-06,
      "loss": 0.6357,
      "step": 3826
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7606521319132613,
      "learning_rate": 5.444057029675503e-06,
      "loss": 0.6044,
      "step": 3827
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9011213144168897,
      "learning_rate": 5.442001837239427e-06,
      "loss": 0.6733,
      "step": 3828
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6389917283686845,
      "learning_rate": 5.439946569535043e-06,
      "loss": 0.5717,
      "step": 3829
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7747034608140921,
      "learning_rate": 5.437891226912342e-06,
      "loss": 0.59,
      "step": 3830
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7045123141173107,
      "learning_rate": 5.435835809721326e-06,
      "loss": 0.5806,
      "step": 3831
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7782449551892204,
      "learning_rate": 5.433780318312013e-06,
      "loss": 0.5608,
      "step": 3832
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7209084161895974,
      "learning_rate": 5.4317247530344305e-06,
      "loss": 0.6196,
      "step": 3833
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8260617435213631,
      "learning_rate": 5.429669114238621e-06,
      "loss": 0.6048,
      "step": 3834
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9573921073955799,
      "learning_rate": 5.427613402274637e-06,
      "loss": 0.6328,
      "step": 3835
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8305749950914513,
      "learning_rate": 5.425557617492546e-06,
      "loss": 0.6535,
      "step": 3836
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8060792540361774,
      "learning_rate": 5.423501760242425e-06,
      "loss": 0.6649,
      "step": 3837
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8337030013744501,
      "learning_rate": 5.4214458308743674e-06,
      "loss": 0.6251,
      "step": 3838
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7064456100201285,
      "learning_rate": 5.419389829738475e-06,
      "loss": 0.541,
      "step": 3839
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8077887818439091,
      "learning_rate": 5.417333757184864e-06,
      "loss": 0.6126,
      "step": 3840
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7944744784863351,
      "learning_rate": 5.415277613563661e-06,
      "loss": 0.5822,
      "step": 3841
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7359226148266003,
      "learning_rate": 5.413221399225007e-06,
      "loss": 0.5364,
      "step": 3842
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8207786284640401,
      "learning_rate": 5.411165114519054e-06,
      "loss": 0.6316,
      "step": 3843
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.761351414277866,
      "learning_rate": 5.409108759795966e-06,
      "loss": 0.5533,
      "step": 3844
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8620611040195203,
      "learning_rate": 5.4070523354059175e-06,
      "loss": 0.6326,
      "step": 3845
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8067164645720434,
      "learning_rate": 5.404995841699097e-06,
      "loss": 0.6788,
      "step": 3846
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8938055268343883,
      "learning_rate": 5.402939279025705e-06,
      "loss": 0.5842,
      "step": 3847
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6327188598711672,
      "learning_rate": 5.40088264773595e-06,
      "loss": 0.4597,
      "step": 3848
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8038640277830794,
      "learning_rate": 5.398825948180058e-06,
      "loss": 0.6055,
      "step": 3849
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6903374320625769,
      "learning_rate": 5.396769180708261e-06,
      "loss": 0.5273,
      "step": 3850
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7052734244364458,
      "learning_rate": 5.3947123456708065e-06,
      "loss": 0.5497,
      "step": 3851
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7319679782592236,
      "learning_rate": 5.39265544341795e-06,
      "loss": 0.6111,
      "step": 3852
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6286419621787235,
      "learning_rate": 5.390598474299964e-06,
      "loss": 0.547,
      "step": 3853
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6495615993824007,
      "learning_rate": 5.388541438667126e-06,
      "loss": 0.5687,
      "step": 3854
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8190019595768467,
      "learning_rate": 5.386484336869728e-06,
      "loss": 0.6582,
      "step": 3855
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6144310144825965,
      "learning_rate": 5.384427169258075e-06,
      "loss": 0.5553,
      "step": 3856
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7190871083020112,
      "learning_rate": 5.382369936182481e-06,
      "loss": 0.5912,
      "step": 3857
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7019300242830034,
      "learning_rate": 5.380312637993269e-06,
      "loss": 0.5412,
      "step": 3858
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6354299612327892,
      "learning_rate": 5.3782552750407765e-06,
      "loss": 0.5138,
      "step": 3859
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6702103895728937,
      "learning_rate": 5.376197847675353e-06,
      "loss": 0.6079,
      "step": 3860
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.696995486897744,
      "learning_rate": 5.374140356247354e-06,
      "loss": 0.4903,
      "step": 3861
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7737769559787069,
      "learning_rate": 5.37208280110715e-06,
      "loss": 0.6091,
      "step": 3862
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8294111635474973,
      "learning_rate": 5.370025182605122e-06,
      "loss": 0.6243,
      "step": 3863
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.793935295795038,
      "learning_rate": 5.36796750109166e-06,
      "loss": 0.5845,
      "step": 3864
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.691177235719243,
      "learning_rate": 5.365909756917165e-06,
      "loss": 0.4849,
      "step": 3865
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.773087409054588,
      "learning_rate": 5.363851950432052e-06,
      "loss": 0.5812,
      "step": 3866
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8831648602839782,
      "learning_rate": 5.361794081986741e-06,
      "loss": 0.6536,
      "step": 3867
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7150638310158256,
      "learning_rate": 5.3597361519316685e-06,
      "loss": 0.5346,
      "step": 3868
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7623917603921809,
      "learning_rate": 5.357678160617276e-06,
      "loss": 0.5575,
      "step": 3869
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1889003241608633,
      "learning_rate": 5.355620108394018e-06,
      "loss": 0.6372,
      "step": 3870
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.830736440456483,
      "learning_rate": 5.353561995612362e-06,
      "loss": 0.6866,
      "step": 3871
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.80264355673953,
      "learning_rate": 5.351503822622781e-06,
      "loss": 0.6549,
      "step": 3872
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8737518549648303,
      "learning_rate": 5.349445589775761e-06,
      "loss": 0.6902,
      "step": 3873
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6790303833821116,
      "learning_rate": 5.347387297421798e-06,
      "loss": 0.5769,
      "step": 3874
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7295110762831493,
      "learning_rate": 5.345328945911397e-06,
      "loss": 0.5727,
      "step": 3875
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8017729015067593,
      "learning_rate": 5.343270535595074e-06,
      "loss": 0.587,
      "step": 3876
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7476190672272506,
      "learning_rate": 5.341212066823356e-06,
      "loss": 0.5732,
      "step": 3877
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6405887601608288,
      "learning_rate": 5.339153539946777e-06,
      "loss": 0.5507,
      "step": 3878
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7271641143245126,
      "learning_rate": 5.337094955315882e-06,
      "loss": 0.5139,
      "step": 3879
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7242961103500672,
      "learning_rate": 5.33503631328123e-06,
      "loss": 0.6501,
      "step": 3880
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8466096687254789,
      "learning_rate": 5.3329776141933835e-06,
      "loss": 0.6748,
      "step": 3881
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8880092196494722,
      "learning_rate": 5.330918858402918e-06,
      "loss": 0.6523,
      "step": 3882
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5984737574694566,
      "learning_rate": 5.328860046260418e-06,
      "loss": 0.5417,
      "step": 3883
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6077989362682091,
      "learning_rate": 5.326801178116478e-06,
      "loss": 0.5563,
      "step": 3884
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8019998428435321,
      "learning_rate": 5.324742254321703e-06,
      "loss": 0.6176,
      "step": 3885
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6628284387210959,
      "learning_rate": 5.322683275226705e-06,
      "loss": 0.5714,
      "step": 3886
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9531470850931998,
      "learning_rate": 5.3206242411821075e-06,
      "loss": 0.6002,
      "step": 3887
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8075602127205689,
      "learning_rate": 5.318565152538541e-06,
      "loss": 0.589,
      "step": 3888
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8293248742864335,
      "learning_rate": 5.316506009646648e-06,
      "loss": 0.6868,
      "step": 3889
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.728632371409959,
      "learning_rate": 5.314446812857079e-06,
      "loss": 0.5762,
      "step": 3890
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.736799829621883,
      "learning_rate": 5.312387562520494e-06,
      "loss": 0.5123,
      "step": 3891
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8132818832545302,
      "learning_rate": 5.31032825898756e-06,
      "loss": 0.6795,
      "step": 3892
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6425187666420311,
      "learning_rate": 5.308268902608959e-06,
      "loss": 0.5505,
      "step": 3893
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7895670791913938,
      "learning_rate": 5.306209493735373e-06,
      "loss": 0.6207,
      "step": 3894
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8313051619781185,
      "learning_rate": 5.304150032717502e-06,
      "loss": 0.5407,
      "step": 3895
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6736623302126086,
      "learning_rate": 5.302090519906048e-06,
      "loss": 0.5638,
      "step": 3896
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7223814770812205,
      "learning_rate": 5.300030955651726e-06,
      "loss": 0.6144,
      "step": 3897
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6612983729165133,
      "learning_rate": 5.297971340305258e-06,
      "loss": 0.517,
      "step": 3898
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.911059572482929,
      "learning_rate": 5.295911674217374e-06,
      "loss": 0.6769,
      "step": 3899
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.798125451736803,
      "learning_rate": 5.2938519577388155e-06,
      "loss": 0.5625,
      "step": 3900
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2277441434243586,
      "learning_rate": 5.291792191220331e-06,
      "loss": 0.6652,
      "step": 3901
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7965989597396412,
      "learning_rate": 5.289732375012675e-06,
      "loss": 0.6189,
      "step": 3902
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7284296983113183,
      "learning_rate": 5.287672509466614e-06,
      "loss": 0.5619,
      "step": 3903
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7153552105582052,
      "learning_rate": 5.285612594932922e-06,
      "loss": 0.5161,
      "step": 3904
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7823567827528392,
      "learning_rate": 5.283552631762381e-06,
      "loss": 0.625,
      "step": 3905
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.917451724370807,
      "learning_rate": 5.281492620305779e-06,
      "loss": 0.6016,
      "step": 3906
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9107262281377362,
      "learning_rate": 5.279432560913918e-06,
      "loss": 0.6555,
      "step": 3907
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7420049487936115,
      "learning_rate": 5.277372453937602e-06,
      "loss": 0.5721,
      "step": 3908
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7589901278008877,
      "learning_rate": 5.275312299727647e-06,
      "loss": 0.5808,
      "step": 3909
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7160711210356638,
      "learning_rate": 5.273252098634875e-06,
      "loss": 0.6231,
      "step": 3910
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.04335638137264,
      "learning_rate": 5.271191851010119e-06,
      "loss": 0.6267,
      "step": 3911
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8458939871562218,
      "learning_rate": 5.2691315572042155e-06,
      "loss": 0.6284,
      "step": 3912
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7248244845522537,
      "learning_rate": 5.26707121756801e-06,
      "loss": 0.5666,
      "step": 3913
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8704400287327139,
      "learning_rate": 5.26501083245236e-06,
      "loss": 0.667,
      "step": 3914
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6763817543633236,
      "learning_rate": 5.262950402208127e-06,
      "loss": 0.492,
      "step": 3915
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7897791455427242,
      "learning_rate": 5.2608899271861765e-06,
      "loss": 0.6871,
      "step": 3916
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8215969220604853,
      "learning_rate": 5.258829407737392e-06,
      "loss": 0.6357,
      "step": 3917
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7954659706004132,
      "learning_rate": 5.2567688442126555e-06,
      "loss": 0.5363,
      "step": 3918
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8153359060392984,
      "learning_rate": 5.25470823696286e-06,
      "loss": 0.6415,
      "step": 3919
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6230678789586339,
      "learning_rate": 5.252647586338905e-06,
      "loss": 0.5272,
      "step": 3920
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7671432237615076,
      "learning_rate": 5.250586892691698e-06,
      "loss": 0.6401,
      "step": 3921
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7208126738008377,
      "learning_rate": 5.2485261563721535e-06,
      "loss": 0.5089,
      "step": 3922
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7145080882827488,
      "learning_rate": 5.246465377731195e-06,
      "loss": 0.5437,
      "step": 3923
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6905274673987536,
      "learning_rate": 5.244404557119749e-06,
      "loss": 0.5508,
      "step": 3924
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8106874109486311,
      "learning_rate": 5.242343694888753e-06,
      "loss": 0.614,
      "step": 3925
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6957217086755179,
      "learning_rate": 5.240282791389151e-06,
      "loss": 0.5402,
      "step": 3926
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.630372581445786,
      "learning_rate": 5.238221846971892e-06,
      "loss": 0.4862,
      "step": 3927
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7731970985067274,
      "learning_rate": 5.236160861987935e-06,
      "loss": 0.583,
      "step": 3928
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0071684811360497,
      "learning_rate": 5.234099836788245e-06,
      "loss": 0.615,
      "step": 3929
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9171601406914759,
      "learning_rate": 5.23203877172379e-06,
      "loss": 0.6445,
      "step": 3930
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6087526357967677,
      "learning_rate": 5.229977667145549e-06,
      "loss": 0.5302,
      "step": 3931
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7878824035891563,
      "learning_rate": 5.2279165234045085e-06,
      "loss": 0.6143,
      "step": 3932
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5930690498674795,
      "learning_rate": 5.225855340851657e-06,
      "loss": 0.4817,
      "step": 3933
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7331913905266806,
      "learning_rate": 5.223794119837996e-06,
      "loss": 0.6417,
      "step": 3934
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7410897486652996,
      "learning_rate": 5.221732860714525e-06,
      "loss": 0.6344,
      "step": 3935
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6356543120402645,
      "learning_rate": 5.21967156383226e-06,
      "loss": 0.5281,
      "step": 3936
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9882021085591083,
      "learning_rate": 5.217610229542216e-06,
      "loss": 0.6515,
      "step": 3937
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6596919112488022,
      "learning_rate": 5.2155488581954176e-06,
      "loss": 0.5049,
      "step": 3938
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.853979833761869,
      "learning_rate": 5.213487450142892e-06,
      "loss": 0.6217,
      "step": 3939
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8151279619920629,
      "learning_rate": 5.2114260057356794e-06,
      "loss": 0.5773,
      "step": 3940
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9487051719291467,
      "learning_rate": 5.209364525324819e-06,
      "loss": 0.6526,
      "step": 3941
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7740052051441528,
      "learning_rate": 5.207303009261363e-06,
      "loss": 0.5138,
      "step": 3942
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8509034195940681,
      "learning_rate": 5.205241457896364e-06,
      "loss": 0.5932,
      "step": 3943
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8272147455548723,
      "learning_rate": 5.2031798715808824e-06,
      "loss": 0.6764,
      "step": 3944
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8485951340071919,
      "learning_rate": 5.2011182506659856e-06,
      "loss": 0.6642,
      "step": 3945
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8327932607190893,
      "learning_rate": 5.199056595502745e-06,
      "loss": 0.6205,
      "step": 3946
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6901673640092363,
      "learning_rate": 5.196994906442241e-06,
      "loss": 0.5469,
      "step": 3947
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8409752221760088,
      "learning_rate": 5.194933183835556e-06,
      "loss": 0.6335,
      "step": 3948
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6610313833884395,
      "learning_rate": 5.192871428033782e-06,
      "loss": 0.5382,
      "step": 3949
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7261820462916982,
      "learning_rate": 5.190809639388012e-06,
      "loss": 0.5379,
      "step": 3950
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7112033018841851,
      "learning_rate": 5.1887478182493474e-06,
      "loss": 0.5569,
      "step": 3951
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9168185585416618,
      "learning_rate": 5.186685964968896e-06,
      "loss": 0.6542,
      "step": 3952
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8977435848892936,
      "learning_rate": 5.184624079897771e-06,
      "loss": 0.6782,
      "step": 3953
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.775057102507285,
      "learning_rate": 5.182562163387088e-06,
      "loss": 0.6172,
      "step": 3954
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7438685462146467,
      "learning_rate": 5.1805002157879704e-06,
      "loss": 0.5664,
      "step": 3955
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7787555138334914,
      "learning_rate": 5.1784382374515464e-06,
      "loss": 0.5688,
      "step": 3956
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7980526028213298,
      "learning_rate": 5.17637622872895e-06,
      "loss": 0.6386,
      "step": 3957
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9173565476943023,
      "learning_rate": 5.174314189971319e-06,
      "loss": 0.6318,
      "step": 3958
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7140606973586655,
      "learning_rate": 5.172252121529797e-06,
      "loss": 0.5956,
      "step": 3959
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.819159583163676,
      "learning_rate": 5.170190023755534e-06,
      "loss": 0.5257,
      "step": 3960
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6222897543955713,
      "learning_rate": 5.168127896999683e-06,
      "loss": 0.5028,
      "step": 3961
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8116308612826724,
      "learning_rate": 5.166065741613402e-06,
      "loss": 0.6755,
      "step": 3962
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8079616096038881,
      "learning_rate": 5.164003557947855e-06,
      "loss": 0.6514,
      "step": 3963
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6325325189809083,
      "learning_rate": 5.161941346354211e-06,
      "loss": 0.5586,
      "step": 3964
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7598035162863783,
      "learning_rate": 5.159879107183643e-06,
      "loss": 0.5387,
      "step": 3965
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8244041636725825,
      "learning_rate": 5.157816840787328e-06,
      "loss": 0.5731,
      "step": 3966
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7636594459504319,
      "learning_rate": 5.155754547516451e-06,
      "loss": 0.5652,
      "step": 3967
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7687513322591288,
      "learning_rate": 5.153692227722195e-06,
      "loss": 0.6231,
      "step": 3968
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6365027273253177,
      "learning_rate": 5.151629881755754e-06,
      "loss": 0.5229,
      "step": 3969
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6660627820368301,
      "learning_rate": 5.149567509968325e-06,
      "loss": 0.5635,
      "step": 3970
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8977479084855744,
      "learning_rate": 5.147505112711106e-06,
      "loss": 0.6474,
      "step": 3971
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6242025286809467,
      "learning_rate": 5.1454426903353035e-06,
      "loss": 0.4784,
      "step": 3972
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.668758006157135,
      "learning_rate": 5.143380243192128e-06,
      "loss": 0.5372,
      "step": 3973
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6521677659179348,
      "learning_rate": 5.141317771632789e-06,
      "loss": 0.5372,
      "step": 3974
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7291272260705304,
      "learning_rate": 5.139255276008506e-06,
      "loss": 0.6172,
      "step": 3975
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7316764635573091,
      "learning_rate": 5.1371927566704995e-06,
      "loss": 0.5432,
      "step": 3976
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7201679313009776,
      "learning_rate": 5.135130213969996e-06,
      "loss": 0.5819,
      "step": 3977
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.808797444175203,
      "learning_rate": 5.133067648258227e-06,
      "loss": 0.5254,
      "step": 3978
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.838837036615046,
      "learning_rate": 5.131005059886421e-06,
      "loss": 0.6337,
      "step": 3979
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7475109946373851,
      "learning_rate": 5.128942449205819e-06,
      "loss": 0.6117,
      "step": 3980
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8451944146525131,
      "learning_rate": 5.12687981656766e-06,
      "loss": 0.6911,
      "step": 3981
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6693917000374486,
      "learning_rate": 5.124817162323192e-06,
      "loss": 0.5234,
      "step": 3982
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7726321828292547,
      "learning_rate": 5.122754486823661e-06,
      "loss": 0.4918,
      "step": 3983
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6928701631120032,
      "learning_rate": 5.12069179042032e-06,
      "loss": 0.5528,
      "step": 3984
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7047977773124475,
      "learning_rate": 5.118629073464424e-06,
      "loss": 0.5386,
      "step": 3985
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6385016494572069,
      "learning_rate": 5.116566336307233e-06,
      "loss": 0.5165,
      "step": 3986
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.775412533633928,
      "learning_rate": 5.11450357930001e-06,
      "loss": 0.5892,
      "step": 3987
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8511416255850703,
      "learning_rate": 5.112440802794019e-06,
      "loss": 0.5522,
      "step": 3988
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7634150618667501,
      "learning_rate": 5.110378007140532e-06,
      "loss": 0.5609,
      "step": 3989
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8048193148634439,
      "learning_rate": 5.108315192690819e-06,
      "loss": 0.5602,
      "step": 3990
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8709184298970912,
      "learning_rate": 5.106252359796158e-06,
      "loss": 0.655,
      "step": 3991
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6233364534175336,
      "learning_rate": 5.104189508807826e-06,
      "loss": 0.5237,
      "step": 3992
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7559971459861031,
      "learning_rate": 5.102126640077108e-06,
      "loss": 0.6135,
      "step": 3993
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9911347383831909,
      "learning_rate": 5.100063753955284e-06,
      "loss": 0.6973,
      "step": 3994
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8274758276903637,
      "learning_rate": 5.0980008507936475e-06,
      "loss": 0.5894,
      "step": 3995
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6534841013227789,
      "learning_rate": 5.095937930943486e-06,
      "loss": 0.5459,
      "step": 3996
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6049952907772924,
      "learning_rate": 5.093874994756095e-06,
      "loss": 0.518,
      "step": 3997
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.653584927243782,
      "learning_rate": 5.091812042582771e-06,
      "loss": 0.5568,
      "step": 3998
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7597314378676668,
      "learning_rate": 5.089749074774811e-06,
      "loss": 0.6313,
      "step": 3999
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8448182292971246,
      "learning_rate": 5.08768609168352e-06,
      "loss": 0.5565,
      "step": 4000
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7368078283415932,
      "learning_rate": 5.085623093660201e-06,
      "loss": 0.5917,
      "step": 4001
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8385262255143276,
      "learning_rate": 5.08356008105616e-06,
      "loss": 0.6545,
      "step": 4002
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8479285076706935,
      "learning_rate": 5.08149705422271e-06,
      "loss": 0.6013,
      "step": 4003
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8424998573119318,
      "learning_rate": 5.079434013511159e-06,
      "loss": 0.6561,
      "step": 4004
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6573910500537073,
      "learning_rate": 5.077370959272823e-06,
      "loss": 0.5385,
      "step": 4005
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6955431099832903,
      "learning_rate": 5.07530789185902e-06,
      "loss": 0.5317,
      "step": 4006
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7157400746031785,
      "learning_rate": 5.073244811621064e-06,
      "loss": 0.5331,
      "step": 4007
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8321937434217676,
      "learning_rate": 5.071181718910283e-06,
      "loss": 0.5805,
      "step": 4008
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7235452608282076,
      "learning_rate": 5.069118614077996e-06,
      "loss": 0.4927,
      "step": 4009
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7483213653726687,
      "learning_rate": 5.067055497475529e-06,
      "loss": 0.5403,
      "step": 4010
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7078573562250018,
      "learning_rate": 5.064992369454209e-06,
      "loss": 0.4875,
      "step": 4011
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7502973305168704,
      "learning_rate": 5.062929230365367e-06,
      "loss": 0.5196,
      "step": 4012
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7040531099196715,
      "learning_rate": 5.060866080560332e-06,
      "loss": 0.545,
      "step": 4013
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7487422882476922,
      "learning_rate": 5.058802920390437e-06,
      "loss": 0.5427,
      "step": 4014
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6754838839546053,
      "learning_rate": 5.056739750207017e-06,
      "loss": 0.5902,
      "step": 4015
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7864563497390438,
      "learning_rate": 5.054676570361409e-06,
      "loss": 0.5393,
      "step": 4016
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7313864239976106,
      "learning_rate": 5.0526133812049505e-06,
      "loss": 0.5961,
      "step": 4017
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.024762915717553,
      "learning_rate": 5.0505501830889814e-06,
      "loss": 0.6347,
      "step": 4018
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.2535746641185765,
      "learning_rate": 5.048486976364843e-06,
      "loss": 0.6139,
      "step": 4019
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0739846441391503,
      "learning_rate": 5.046423761383878e-06,
      "loss": 0.6202,
      "step": 4020
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8000080541817809,
      "learning_rate": 5.0443605384974296e-06,
      "loss": 0.6333,
      "step": 4021
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6463793909051258,
      "learning_rate": 5.0422973080568435e-06,
      "loss": 0.5135,
      "step": 4022
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7829491687888811,
      "learning_rate": 5.040234070413465e-06,
      "loss": 0.6345,
      "step": 4023
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6191956525147597,
      "learning_rate": 5.038170825918646e-06,
      "loss": 0.5401,
      "step": 4024
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9346662196034498,
      "learning_rate": 5.036107574923732e-06,
      "loss": 0.6128,
      "step": 4025
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6952086486395497,
      "learning_rate": 5.034044317780074e-06,
      "loss": 0.5532,
      "step": 4026
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9625641551589321,
      "learning_rate": 5.031981054839025e-06,
      "loss": 0.6646,
      "step": 4027
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6335104692879268,
      "learning_rate": 5.029917786451933e-06,
      "loss": 0.5411,
      "step": 4028
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7783873207886681,
      "learning_rate": 5.027854512970156e-06,
      "loss": 0.5774,
      "step": 4029
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6910372647508887,
      "learning_rate": 5.025791234745045e-06,
      "loss": 0.5559,
      "step": 4030
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8510773266034817,
      "learning_rate": 5.023727952127955e-06,
      "loss": 0.6604,
      "step": 4031
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8000334262615656,
      "learning_rate": 5.021664665470242e-06,
      "loss": 0.662,
      "step": 4032
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7307104036133295,
      "learning_rate": 5.019601375123261e-06,
      "loss": 0.5373,
      "step": 4033
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.83977828553352,
      "learning_rate": 5.017538081438371e-06,
      "loss": 0.6239,
      "step": 4034
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6962055082394998,
      "learning_rate": 5.0154747847669296e-06,
      "loss": 0.5451,
      "step": 4035
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6816986305335738,
      "learning_rate": 5.0134114854602915e-06,
      "loss": 0.57,
      "step": 4036
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7454360993146433,
      "learning_rate": 5.011348183869818e-06,
      "loss": 0.535,
      "step": 4037
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7795950484634164,
      "learning_rate": 5.009284880346867e-06,
      "loss": 0.5475,
      "step": 4038
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7770615466112519,
      "learning_rate": 5.007221575242798e-06,
      "loss": 0.624,
      "step": 4039
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7679775422861651,
      "learning_rate": 5.0051582689089695e-06,
      "loss": 0.5998,
      "step": 4040
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8314630627406452,
      "learning_rate": 5.003094961696742e-06,
      "loss": 0.5632,
      "step": 4041
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9268488565529598,
      "learning_rate": 5.001031653957475e-06,
      "loss": 0.6281,
      "step": 4042
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.748857160195666,
      "learning_rate": 4.998968346042526e-06,
      "loss": 0.6174,
      "step": 4043
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8810042213745155,
      "learning_rate": 4.996905038303258e-06,
      "loss": 0.5827,
      "step": 4044
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7684608887343929,
      "learning_rate": 4.9948417310910305e-06,
      "loss": 0.5226,
      "step": 4045
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9062057746167201,
      "learning_rate": 4.992778424757203e-06,
      "loss": 0.6814,
      "step": 4046
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7155257495629138,
      "learning_rate": 4.990715119653133e-06,
      "loss": 0.5959,
      "step": 4047
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7682000075359732,
      "learning_rate": 4.988651816130183e-06,
      "loss": 0.543,
      "step": 4048
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7386597792514806,
      "learning_rate": 4.986588514539709e-06,
      "loss": 0.549,
      "step": 4049
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7158236979402866,
      "learning_rate": 4.984525215233073e-06,
      "loss": 0.527,
      "step": 4050
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7074382393112482,
      "learning_rate": 4.9824619185616306e-06,
      "loss": 0.5111,
      "step": 4051
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7774182121636132,
      "learning_rate": 4.980398624876739e-06,
      "loss": 0.5791,
      "step": 4052
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6800399341005464,
      "learning_rate": 4.978335334529759e-06,
      "loss": 0.5469,
      "step": 4053
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8545608272312367,
      "learning_rate": 4.976272047872046e-06,
      "loss": 0.5345,
      "step": 4054
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9342929609182672,
      "learning_rate": 4.974208765254956e-06,
      "loss": 0.6256,
      "step": 4055
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.101260152172225,
      "learning_rate": 4.972145487029845e-06,
      "loss": 0.7008,
      "step": 4056
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.739271931462951,
      "learning_rate": 4.9700822135480666e-06,
      "loss": 0.5344,
      "step": 4057
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6450279287802339,
      "learning_rate": 4.968018945160976e-06,
      "loss": 0.5501,
      "step": 4058
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6967132575754956,
      "learning_rate": 4.965955682219926e-06,
      "loss": 0.5568,
      "step": 4059
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6796344480069014,
      "learning_rate": 4.9638924250762685e-06,
      "loss": 0.5643,
      "step": 4060
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.900453374746162,
      "learning_rate": 4.961829174081355e-06,
      "loss": 0.5704,
      "step": 4061
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7243762624036385,
      "learning_rate": 4.959765929586535e-06,
      "loss": 0.5977,
      "step": 4062
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8218576654450279,
      "learning_rate": 4.957702691943158e-06,
      "loss": 0.6586,
      "step": 4063
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6362813540723148,
      "learning_rate": 4.955639461502572e-06,
      "loss": 0.5356,
      "step": 4064
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8889743155565952,
      "learning_rate": 4.953576238616124e-06,
      "loss": 0.6222,
      "step": 4065
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9365538256952871,
      "learning_rate": 4.951513023635158e-06,
      "loss": 0.6536,
      "step": 4066
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8165656205255383,
      "learning_rate": 4.949449816911019e-06,
      "loss": 0.6798,
      "step": 4067
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6560026407121607,
      "learning_rate": 4.94738661879505e-06,
      "loss": 0.5158,
      "step": 4068
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6847872701073555,
      "learning_rate": 4.945323429638592e-06,
      "loss": 0.5346,
      "step": 4069
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7298798212031871,
      "learning_rate": 4.9432602497929845e-06,
      "loss": 0.5085,
      "step": 4070
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.787514214704033,
      "learning_rate": 4.941197079609565e-06,
      "loss": 0.6202,
      "step": 4071
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7622857318901765,
      "learning_rate": 4.93913391943967e-06,
      "loss": 0.6352,
      "step": 4072
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8887235871144169,
      "learning_rate": 4.937070769634635e-06,
      "loss": 0.6245,
      "step": 4073
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.752672774301807,
      "learning_rate": 4.935007630545793e-06,
      "loss": 0.5261,
      "step": 4074
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7244996374002073,
      "learning_rate": 4.932944502524472e-06,
      "loss": 0.5714,
      "step": 4075
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6936494046491752,
      "learning_rate": 4.930881385922006e-06,
      "loss": 0.5683,
      "step": 4076
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8723750629791313,
      "learning_rate": 4.928818281089719e-06,
      "loss": 0.6662,
      "step": 4077
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.855739373273639,
      "learning_rate": 4.926755188378937e-06,
      "loss": 0.6208,
      "step": 4078
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7514287348334113,
      "learning_rate": 4.924692108140983e-06,
      "loss": 0.6109,
      "step": 4079
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7919236261826393,
      "learning_rate": 4.922629040727178e-06,
      "loss": 0.6223,
      "step": 4080
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6330712334112418,
      "learning_rate": 4.920565986488843e-06,
      "loss": 0.5371,
      "step": 4081
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6656315723042943,
      "learning_rate": 4.918502945777292e-06,
      "loss": 0.5302,
      "step": 4082
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8731822284215348,
      "learning_rate": 4.9164399189438406e-06,
      "loss": 0.6296,
      "step": 4083
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.692620820443639,
      "learning_rate": 4.914376906339801e-06,
      "loss": 0.525,
      "step": 4084
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.739448994586261,
      "learning_rate": 4.912313908316481e-06,
      "loss": 0.5613,
      "step": 4085
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6844504504137746,
      "learning_rate": 4.91025092522519e-06,
      "loss": 0.5387,
      "step": 4086
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7086653042114045,
      "learning_rate": 4.908187957417231e-06,
      "loss": 0.5402,
      "step": 4087
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2915150228772212,
      "learning_rate": 4.906125005243906e-06,
      "loss": 0.6434,
      "step": 4088
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0018152581955475,
      "learning_rate": 4.904062069056515e-06,
      "loss": 0.6857,
      "step": 4089
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8192208571986294,
      "learning_rate": 4.901999149206354e-06,
      "loss": 0.5908,
      "step": 4090
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7035363698070811,
      "learning_rate": 4.899936246044717e-06,
      "loss": 0.5796,
      "step": 4091
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6165506434615436,
      "learning_rate": 4.897873359922895e-06,
      "loss": 0.4763,
      "step": 4092
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8200177504226985,
      "learning_rate": 4.895810491192175e-06,
      "loss": 0.6052,
      "step": 4093
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8229489488822884,
      "learning_rate": 4.893747640203844e-06,
      "loss": 0.6062,
      "step": 4094
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.652692222138843,
      "learning_rate": 4.891684807309183e-06,
      "loss": 0.5482,
      "step": 4095
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8795349466324024,
      "learning_rate": 4.88962199285947e-06,
      "loss": 0.6674,
      "step": 4096
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7052708115141737,
      "learning_rate": 4.887559197205982e-06,
      "loss": 0.5504,
      "step": 4097
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6918782353209547,
      "learning_rate": 4.8854964206999925e-06,
      "loss": 0.5698,
      "step": 4098
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7323308689131311,
      "learning_rate": 4.883433663692768e-06,
      "loss": 0.5483,
      "step": 4099
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7550902263089855,
      "learning_rate": 4.8813709265355766e-06,
      "loss": 0.5846,
      "step": 4100
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8750850486718925,
      "learning_rate": 4.879308209579681e-06,
      "loss": 0.6008,
      "step": 4101
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.839526391064823,
      "learning_rate": 4.87724551317634e-06,
      "loss": 0.5571,
      "step": 4102
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7771312600937295,
      "learning_rate": 4.87518283767681e-06,
      "loss": 0.5953,
      "step": 4103
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8337602557345495,
      "learning_rate": 4.873120183432341e-06,
      "loss": 0.5363,
      "step": 4104
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9783892627739132,
      "learning_rate": 4.871057550794182e-06,
      "loss": 0.6017,
      "step": 4105
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6867520589318331,
      "learning_rate": 4.8689949401135805e-06,
      "loss": 0.5719,
      "step": 4106
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8579573019105908,
      "learning_rate": 4.866932351741776e-06,
      "loss": 0.5899,
      "step": 4107
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7553796620952964,
      "learning_rate": 4.8648697860300045e-06,
      "loss": 0.5752,
      "step": 4108
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6501571168851404,
      "learning_rate": 4.862807243329502e-06,
      "loss": 0.4775,
      "step": 4109
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7502625536494554,
      "learning_rate": 4.8607447239914966e-06,
      "loss": 0.4845,
      "step": 4110
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1260136222716974,
      "learning_rate": 4.858682228367213e-06,
      "loss": 0.642,
      "step": 4111
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9334970014842207,
      "learning_rate": 4.856619756807874e-06,
      "loss": 0.5651,
      "step": 4112
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8255293683964169,
      "learning_rate": 4.854557309664697e-06,
      "loss": 0.6214,
      "step": 4113
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6042962114920335,
      "learning_rate": 4.852494887288895e-06,
      "loss": 0.5141,
      "step": 4114
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9514881845680646,
      "learning_rate": 4.850432490031677e-06,
      "loss": 0.6715,
      "step": 4115
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1647363006443063,
      "learning_rate": 4.848370118244247e-06,
      "loss": 0.6361,
      "step": 4116
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7391372996014577,
      "learning_rate": 4.846307772277808e-06,
      "loss": 0.5305,
      "step": 4117
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8531997857899264,
      "learning_rate": 4.844245452483551e-06,
      "loss": 0.658,
      "step": 4118
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6608306684098285,
      "learning_rate": 4.842183159212673e-06,
      "loss": 0.5569,
      "step": 4119
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7316685742795973,
      "learning_rate": 4.840120892816359e-06,
      "loss": 0.5802,
      "step": 4120
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8848493996257852,
      "learning_rate": 4.838058653645791e-06,
      "loss": 0.6008,
      "step": 4121
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.681842390475123,
      "learning_rate": 4.835996442052147e-06,
      "loss": 0.5788,
      "step": 4122
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7990535059080215,
      "learning_rate": 4.8339342583866e-06,
      "loss": 0.6447,
      "step": 4123
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9179589949333559,
      "learning_rate": 4.831872103000321e-06,
      "loss": 0.6416,
      "step": 4124
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6833154187757796,
      "learning_rate": 4.829809976244469e-06,
      "loss": 0.5572,
      "step": 4125
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6647438949110751,
      "learning_rate": 4.8277478784702056e-06,
      "loss": 0.56,
      "step": 4126
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9807225034051716,
      "learning_rate": 4.825685810028683e-06,
      "loss": 0.6181,
      "step": 4127
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.808616393443888,
      "learning_rate": 4.823623771271053e-06,
      "loss": 0.6407,
      "step": 4128
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9570981678184455,
      "learning_rate": 4.821561762548456e-06,
      "loss": 0.6153,
      "step": 4129
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6445356911638195,
      "learning_rate": 4.819499784212032e-06,
      "loss": 0.5625,
      "step": 4130
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6331674292315457,
      "learning_rate": 4.817437836612914e-06,
      "loss": 0.5242,
      "step": 4131
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7534944480206083,
      "learning_rate": 4.8153759201022305e-06,
      "loss": 0.6274,
      "step": 4132
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8580250627691435,
      "learning_rate": 4.813314035031105e-06,
      "loss": 0.6101,
      "step": 4133
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7862684191226618,
      "learning_rate": 4.811252181750654e-06,
      "loss": 0.5187,
      "step": 4134
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6926710365606015,
      "learning_rate": 4.809190360611991e-06,
      "loss": 0.5261,
      "step": 4135
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8000754993874579,
      "learning_rate": 4.807128571966221e-06,
      "loss": 0.6078,
      "step": 4136
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8498234927990753,
      "learning_rate": 4.805066816164446e-06,
      "loss": 0.6682,
      "step": 4137
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6575167296419226,
      "learning_rate": 4.803005093557761e-06,
      "loss": 0.5142,
      "step": 4138
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7581219805452567,
      "learning_rate": 4.800943404497257e-06,
      "loss": 0.5524,
      "step": 4139
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8137812092162776,
      "learning_rate": 4.798881749334017e-06,
      "loss": 0.5301,
      "step": 4140
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7244622361576895,
      "learning_rate": 4.79682012841912e-06,
      "loss": 0.5535,
      "step": 4141
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6706564469193905,
      "learning_rate": 4.794758542103639e-06,
      "loss": 0.5473,
      "step": 4142
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9196743076637054,
      "learning_rate": 4.792696990738639e-06,
      "loss": 0.5995,
      "step": 4143
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0376612947006751,
      "learning_rate": 4.790635474675183e-06,
      "loss": 0.6848,
      "step": 4144
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7442264398578571,
      "learning_rate": 4.788573994264321e-06,
      "loss": 0.5235,
      "step": 4145
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7879620646661468,
      "learning_rate": 4.786512549857109e-06,
      "loss": 0.5406,
      "step": 4146
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7363211574087645,
      "learning_rate": 4.784451141804584e-06,
      "loss": 0.5993,
      "step": 4147
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6084256955526857,
      "learning_rate": 4.782389770457785e-06,
      "loss": 0.5174,
      "step": 4148
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8886147830023661,
      "learning_rate": 4.7803284361677405e-06,
      "loss": 0.6087,
      "step": 4149
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8320970327971176,
      "learning_rate": 4.778267139285474e-06,
      "loss": 0.6966,
      "step": 4150
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8453912392042926,
      "learning_rate": 4.776205880162005e-06,
      "loss": 0.6263,
      "step": 4151
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7209206546214638,
      "learning_rate": 4.774144659148343e-06,
      "loss": 0.5564,
      "step": 4152
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8315453701652941,
      "learning_rate": 4.772083476595492e-06,
      "loss": 0.5745,
      "step": 4153
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7994989416708732,
      "learning_rate": 4.770022332854451e-06,
      "loss": 0.636,
      "step": 4154
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6321210320547559,
      "learning_rate": 4.767961228276211e-06,
      "loss": 0.4907,
      "step": 4155
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7873716729677722,
      "learning_rate": 4.765900163211756e-06,
      "loss": 0.617,
      "step": 4156
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7058287296756292,
      "learning_rate": 4.763839138012065e-06,
      "loss": 0.5601,
      "step": 4157
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.80190079540078,
      "learning_rate": 4.7617781530281085e-06,
      "loss": 0.6593,
      "step": 4158
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7599906031130926,
      "learning_rate": 4.75971720861085e-06,
      "loss": 0.5067,
      "step": 4159
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8914924312843188,
      "learning_rate": 4.757656305111248e-06,
      "loss": 0.6147,
      "step": 4160
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8646992925410405,
      "learning_rate": 4.755595442880253e-06,
      "loss": 0.6275,
      "step": 4161
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6821930676796342,
      "learning_rate": 4.753534622268807e-06,
      "loss": 0.5324,
      "step": 4162
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6402092434188056,
      "learning_rate": 4.7514738436278465e-06,
      "loss": 0.4989,
      "step": 4163
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8822096973442393,
      "learning_rate": 4.749413107308302e-06,
      "loss": 0.6481,
      "step": 4164
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8880714360486498,
      "learning_rate": 4.747352413661095e-06,
      "loss": 0.6724,
      "step": 4165
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5886914657939684,
      "learning_rate": 4.745291763037141e-06,
      "loss": 0.5034,
      "step": 4166
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7872696460328779,
      "learning_rate": 4.7432311557873444e-06,
      "loss": 0.6334,
      "step": 4167
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7871243209929952,
      "learning_rate": 4.741170592262608e-06,
      "loss": 0.6144,
      "step": 4168
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8299558773385124,
      "learning_rate": 4.739110072813823e-06,
      "loss": 0.6586,
      "step": 4169
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7560711007689411,
      "learning_rate": 4.737049597791875e-06,
      "loss": 0.6227,
      "step": 4170
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7621097540827885,
      "learning_rate": 4.7349891675476404e-06,
      "loss": 0.6528,
      "step": 4171
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7226277006442361,
      "learning_rate": 4.732928782431991e-06,
      "loss": 0.5208,
      "step": 4172
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7703314032836279,
      "learning_rate": 4.730868442795786e-06,
      "loss": 0.671,
      "step": 4173
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7643178458922305,
      "learning_rate": 4.728808148989882e-06,
      "loss": 0.6007,
      "step": 4174
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7069434530417544,
      "learning_rate": 4.726747901365125e-06,
      "loss": 0.5177,
      "step": 4175
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8840323008373085,
      "learning_rate": 4.724687700272355e-06,
      "loss": 0.6241,
      "step": 4176
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8524444547142161,
      "learning_rate": 4.7226275460623985e-06,
      "loss": 0.6308,
      "step": 4177
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7282595187109793,
      "learning_rate": 4.7205674390860834e-06,
      "loss": 0.5175,
      "step": 4178
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7263627008817285,
      "learning_rate": 4.718507379694222e-06,
      "loss": 0.534,
      "step": 4179
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6248345499944433,
      "learning_rate": 4.716447368237621e-06,
      "loss": 0.5183,
      "step": 4180
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6897538293194313,
      "learning_rate": 4.71438740506708e-06,
      "loss": 0.5253,
      "step": 4181
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8227693698540856,
      "learning_rate": 4.712327490533387e-06,
      "loss": 0.5827,
      "step": 4182
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6922792568525376,
      "learning_rate": 4.710267624987326e-06,
      "loss": 0.5837,
      "step": 4183
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9211946554064792,
      "learning_rate": 4.7082078087796706e-06,
      "loss": 0.6284,
      "step": 4184
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7742322407013416,
      "learning_rate": 4.706148042261185e-06,
      "loss": 0.5746,
      "step": 4185
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6622490015545,
      "learning_rate": 4.7040883257826266e-06,
      "loss": 0.5731,
      "step": 4186
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7216346112681995,
      "learning_rate": 4.702028659694744e-06,
      "loss": 0.6083,
      "step": 4187
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6494127389970505,
      "learning_rate": 4.699969044348276e-06,
      "loss": 0.5231,
      "step": 4188
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7268336785111906,
      "learning_rate": 4.697909480093954e-06,
      "loss": 0.5199,
      "step": 4189
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6600914464233625,
      "learning_rate": 4.6958499672824996e-06,
      "loss": 0.5412,
      "step": 4190
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9927621903345056,
      "learning_rate": 4.693790506264627e-06,
      "loss": 0.6956,
      "step": 4191
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9086227675597718,
      "learning_rate": 4.691731097391043e-06,
      "loss": 0.6723,
      "step": 4192
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8441032132839217,
      "learning_rate": 4.6896717410124404e-06,
      "loss": 0.6084,
      "step": 4193
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7710085844145217,
      "learning_rate": 4.687612437479508e-06,
      "loss": 0.5634,
      "step": 4194
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9517107933846263,
      "learning_rate": 4.685553187142923e-06,
      "loss": 0.6268,
      "step": 4195
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9225947092171035,
      "learning_rate": 4.683493990353354e-06,
      "loss": 0.6226,
      "step": 4196
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7977823772216441,
      "learning_rate": 4.681434847461461e-06,
      "loss": 0.6272,
      "step": 4197
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6397954052173853,
      "learning_rate": 4.679375758817894e-06,
      "loss": 0.53,
      "step": 4198
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.758027755916074,
      "learning_rate": 4.677316724773296e-06,
      "loss": 0.5419,
      "step": 4199
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.805995659523113,
      "learning_rate": 4.675257745678298e-06,
      "loss": 0.5214,
      "step": 4200
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7623516212334143,
      "learning_rate": 4.673198821883523e-06,
      "loss": 0.5943,
      "step": 4201
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6844731682817229,
      "learning_rate": 4.671139953739584e-06,
      "loss": 0.5775,
      "step": 4202
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6400697253229971,
      "learning_rate": 4.669081141597083e-06,
      "loss": 0.538,
      "step": 4203
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6232927825322789,
      "learning_rate": 4.667022385806618e-06,
      "loss": 0.5355,
      "step": 4204
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8829371692400296,
      "learning_rate": 4.664963686718772e-06,
      "loss": 0.5787,
      "step": 4205
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7526636968478414,
      "learning_rate": 4.662905044684119e-06,
      "loss": 0.5598,
      "step": 4206
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8491422713028841,
      "learning_rate": 4.660846460053225e-06,
      "loss": 0.5434,
      "step": 4207
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7709862400718848,
      "learning_rate": 4.6587879331766465e-06,
      "loss": 0.5639,
      "step": 4208
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9292947724551804,
      "learning_rate": 4.6567294644049275e-06,
      "loss": 0.5923,
      "step": 4209
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6984239894360695,
      "learning_rate": 4.654671054088604e-06,
      "loss": 0.568,
      "step": 4210
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5855105801908731,
      "learning_rate": 4.652612702578203e-06,
      "loss": 0.5205,
      "step": 4211
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.975030417128812,
      "learning_rate": 4.65055441022424e-06,
      "loss": 0.6098,
      "step": 4212
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.712596075126336,
      "learning_rate": 4.648496177377221e-06,
      "loss": 0.6229,
      "step": 4213
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6613009493405987,
      "learning_rate": 4.64643800438764e-06,
      "loss": 0.4978,
      "step": 4214
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6615734185842508,
      "learning_rate": 4.644379891605983e-06,
      "loss": 0.5352,
      "step": 4215
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0248420183593918,
      "learning_rate": 4.642321839382726e-06,
      "loss": 0.586,
      "step": 4216
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6693688003932176,
      "learning_rate": 4.640263848068333e-06,
      "loss": 0.5078,
      "step": 4217
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0003929349676586,
      "learning_rate": 4.63820591801326e-06,
      "loss": 0.5979,
      "step": 4218
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6612912943875484,
      "learning_rate": 4.63614804956795e-06,
      "loss": 0.5316,
      "step": 4219
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6997957341277706,
      "learning_rate": 4.634090243082837e-06,
      "loss": 0.5513,
      "step": 4220
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7981191881569423,
      "learning_rate": 4.632032498908343e-06,
      "loss": 0.5346,
      "step": 4221
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7398219267470332,
      "learning_rate": 4.62997481739488e-06,
      "loss": 0.6506,
      "step": 4222
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6775551987599228,
      "learning_rate": 4.627917198892852e-06,
      "loss": 0.5305,
      "step": 4223
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7484709186870059,
      "learning_rate": 4.625859643752648e-06,
      "loss": 0.5805,
      "step": 4224
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.663959302220853,
      "learning_rate": 4.62380215232465e-06,
      "loss": 0.5636,
      "step": 4225
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8098130906079022,
      "learning_rate": 4.621744724959225e-06,
      "loss": 0.5691,
      "step": 4226
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8157104117008908,
      "learning_rate": 4.619687362006734e-06,
      "loss": 0.6222,
      "step": 4227
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7426790421916588,
      "learning_rate": 4.617630063817522e-06,
      "loss": 0.4971,
      "step": 4228
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7699024957101612,
      "learning_rate": 4.615572830741926e-06,
      "loss": 0.5253,
      "step": 4229
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8602413979088466,
      "learning_rate": 4.613515663130273e-06,
      "loss": 0.5974,
      "step": 4230
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6950353640061578,
      "learning_rate": 4.611458561332876e-06,
      "loss": 0.5366,
      "step": 4231
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.877582513364586,
      "learning_rate": 4.609401525700039e-06,
      "loss": 0.5723,
      "step": 4232
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8956010969356248,
      "learning_rate": 4.607344556582052e-06,
      "loss": 0.6101,
      "step": 4233
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7394549044676947,
      "learning_rate": 4.605287654329197e-06,
      "loss": 0.5248,
      "step": 4234
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8387306343856894,
      "learning_rate": 4.603230819291742e-06,
      "loss": 0.5788,
      "step": 4235
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7233643904928372,
      "learning_rate": 4.601174051819945e-06,
      "loss": 0.5437,
      "step": 4236
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7533710005842715,
      "learning_rate": 4.599117352264052e-06,
      "loss": 0.5077,
      "step": 4237
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7965672752490813,
      "learning_rate": 4.597060720974298e-06,
      "loss": 0.5525,
      "step": 4238
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.70387357182827,
      "learning_rate": 4.595004158300904e-06,
      "loss": 0.5491,
      "step": 4239
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8188663480296382,
      "learning_rate": 4.592947664594085e-06,
      "loss": 0.68,
      "step": 4240
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8396903003478393,
      "learning_rate": 4.590891240204037e-06,
      "loss": 0.6159,
      "step": 4241
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.968591130232146,
      "learning_rate": 4.588834885480946e-06,
      "loss": 0.6472,
      "step": 4242
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8045201236616198,
      "learning_rate": 4.586778600774994e-06,
      "loss": 0.5603,
      "step": 4243
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6976429329288849,
      "learning_rate": 4.5847223864363406e-06,
      "loss": 0.5553,
      "step": 4244
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8433699324387849,
      "learning_rate": 4.582666242815138e-06,
      "loss": 0.6371,
      "step": 4245
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7927997737236845,
      "learning_rate": 4.580610170261526e-06,
      "loss": 0.6021,
      "step": 4246
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7848123587632546,
      "learning_rate": 4.578554169125633e-06,
      "loss": 0.6351,
      "step": 4247
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8202415591065428,
      "learning_rate": 4.576498239757574e-06,
      "loss": 0.6087,
      "step": 4248
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6976202333638307,
      "learning_rate": 4.5744423825074545e-06,
      "loss": 0.5412,
      "step": 4249
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8654372934442717,
      "learning_rate": 4.572386597725363e-06,
      "loss": 0.6182,
      "step": 4250
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7821929778206282,
      "learning_rate": 4.5703308857613795e-06,
      "loss": 0.549,
      "step": 4251
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8559460081716068,
      "learning_rate": 4.568275246965569e-06,
      "loss": 0.6377,
      "step": 4252
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6575443589344072,
      "learning_rate": 4.5662196816879866e-06,
      "loss": 0.535,
      "step": 4253
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6187164243379435,
      "learning_rate": 4.564164190278674e-06,
      "loss": 0.5234,
      "step": 4254
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7225818388747259,
      "learning_rate": 4.562108773087659e-06,
      "loss": 0.5264,
      "step": 4255
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6340733376867966,
      "learning_rate": 4.560053430464958e-06,
      "loss": 0.5075,
      "step": 4256
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7316111365959566,
      "learning_rate": 4.557998162760574e-06,
      "loss": 0.5818,
      "step": 4257
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7981144418366527,
      "learning_rate": 4.555942970324498e-06,
      "loss": 0.5854,
      "step": 4258
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7182520923699315,
      "learning_rate": 4.553887853506709e-06,
      "loss": 0.5827,
      "step": 4259
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6450647639668683,
      "learning_rate": 4.551832812657171e-06,
      "loss": 0.5216,
      "step": 4260
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.641867659700773,
      "learning_rate": 4.549777848125834e-06,
      "loss": 0.5416,
      "step": 4261
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9670865181996074,
      "learning_rate": 4.54772296026264e-06,
      "loss": 0.6679,
      "step": 4262
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.684732296961073,
      "learning_rate": 4.545668149417515e-06,
      "loss": 0.56,
      "step": 4263
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6449500738330984,
      "learning_rate": 4.543613415940371e-06,
      "loss": 0.5345,
      "step": 4264
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1918378576783348,
      "learning_rate": 4.541558760181108e-06,
      "loss": 0.6161,
      "step": 4265
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7378955636160488,
      "learning_rate": 4.53950418248961e-06,
      "loss": 0.5991,
      "step": 4266
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7680941347583717,
      "learning_rate": 4.537449683215754e-06,
      "loss": 0.5975,
      "step": 4267
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.871898713417721,
      "learning_rate": 4.535395262709398e-06,
      "loss": 0.6081,
      "step": 4268
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8247345272844328,
      "learning_rate": 4.5333409213203885e-06,
      "loss": 0.6472,
      "step": 4269
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9495176533479497,
      "learning_rate": 4.531286659398558e-06,
      "loss": 0.6238,
      "step": 4270
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6644655768280376,
      "learning_rate": 4.529232477293727e-06,
      "loss": 0.5453,
      "step": 4271
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7292651735979029,
      "learning_rate": 4.527178375355701e-06,
      "loss": 0.5272,
      "step": 4272
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8402002434211814,
      "learning_rate": 4.525124353934271e-06,
      "loss": 0.5777,
      "step": 4273
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8825379302849763,
      "learning_rate": 4.523070413379216e-06,
      "loss": 0.6504,
      "step": 4274
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8167405971518923,
      "learning_rate": 4.521016554040301e-06,
      "loss": 0.6795,
      "step": 4275
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8976887544850193,
      "learning_rate": 4.518962776267278e-06,
      "loss": 0.6136,
      "step": 4276
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8124866265258778,
      "learning_rate": 4.516909080409882e-06,
      "loss": 0.6481,
      "step": 4277
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6936794794611817,
      "learning_rate": 4.514855466817837e-06,
      "loss": 0.5548,
      "step": 4278
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7250953358677796,
      "learning_rate": 4.512801935840852e-06,
      "loss": 0.527,
      "step": 4279
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7268198807137963,
      "learning_rate": 4.510748487828621e-06,
      "loss": 0.5515,
      "step": 4280
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6980211816613745,
      "learning_rate": 4.508695123130827e-06,
      "loss": 0.5493,
      "step": 4281
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7026408716876937,
      "learning_rate": 4.506641842097134e-06,
      "loss": 0.5328,
      "step": 4282
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8673192725498058,
      "learning_rate": 4.5045886450771954e-06,
      "loss": 0.6134,
      "step": 4283
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8659812536920948,
      "learning_rate": 4.5025355324206495e-06,
      "loss": 0.6315,
      "step": 4284
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8689227197290147,
      "learning_rate": 4.5004825044771205e-06,
      "loss": 0.6875,
      "step": 4285
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6850586781361362,
      "learning_rate": 4.498429561596217e-06,
      "loss": 0.5257,
      "step": 4286
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.837268635347408,
      "learning_rate": 4.4963767041275336e-06,
      "loss": 0.6089,
      "step": 4287
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.835248129339369,
      "learning_rate": 4.49432393242065e-06,
      "loss": 0.6253,
      "step": 4288
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8530761989285995,
      "learning_rate": 4.492271246825133e-06,
      "loss": 0.6113,
      "step": 4289
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8664589337880827,
      "learning_rate": 4.490218647690534e-06,
      "loss": 0.6804,
      "step": 4290
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8726275259486077,
      "learning_rate": 4.488166135366388e-06,
      "loss": 0.6084,
      "step": 4291
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7980753409698471,
      "learning_rate": 4.4861137102022165e-06,
      "loss": 0.4971,
      "step": 4292
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8014274827688181,
      "learning_rate": 4.484061372547527e-06,
      "loss": 0.6543,
      "step": 4293
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.892402177590191,
      "learning_rate": 4.482009122751811e-06,
      "loss": 0.6047,
      "step": 4294
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9968851672742536,
      "learning_rate": 4.479956961164544e-06,
      "loss": 0.6578,
      "step": 4295
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.631712436125498,
      "learning_rate": 4.47790488813519e-06,
      "loss": 0.5045,
      "step": 4296
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7964008122827447,
      "learning_rate": 4.4758529040131934e-06,
      "loss": 0.6333,
      "step": 4297
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6713894780671378,
      "learning_rate": 4.473801009147987e-06,
      "loss": 0.5069,
      "step": 4298
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7975667380015566,
      "learning_rate": 4.471749203888987e-06,
      "loss": 0.5779,
      "step": 4299
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8390345092664563,
      "learning_rate": 4.4696974885855945e-06,
      "loss": 0.6623,
      "step": 4300
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6421372468223091,
      "learning_rate": 4.467645863587193e-06,
      "loss": 0.4968,
      "step": 4301
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8953854383897983,
      "learning_rate": 4.465594329243157e-06,
      "loss": 0.6599,
      "step": 4302
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8752159435031448,
      "learning_rate": 4.463542885902839e-06,
      "loss": 0.6304,
      "step": 4303
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7839989703639463,
      "learning_rate": 4.461491533915579e-06,
      "loss": 0.5016,
      "step": 4304
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.665687292643882,
      "learning_rate": 4.4594402736307e-06,
      "loss": 0.4915,
      "step": 4305
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7011905377440351,
      "learning_rate": 4.457389105397511e-06,
      "loss": 0.5429,
      "step": 4306
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8399208018122137,
      "learning_rate": 4.455338029565306e-06,
      "loss": 0.6619,
      "step": 4307
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8548219064805802,
      "learning_rate": 4.4532870464833585e-06,
      "loss": 0.6475,
      "step": 4308
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6642099735601585,
      "learning_rate": 4.451236156500932e-06,
      "loss": 0.5323,
      "step": 4309
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8238343367257817,
      "learning_rate": 4.449185359967271e-06,
      "loss": 0.7038,
      "step": 4310
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7450769461965905,
      "learning_rate": 4.447134657231605e-06,
      "loss": 0.5496,
      "step": 4311
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7836224243294251,
      "learning_rate": 4.445084048643147e-06,
      "loss": 0.5544,
      "step": 4312
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8155307267419714,
      "learning_rate": 4.443033534551094e-06,
      "loss": 0.6542,
      "step": 4313
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7207284168490987,
      "learning_rate": 4.440983115304627e-06,
      "loss": 0.5948,
      "step": 4314
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9337629737871321,
      "learning_rate": 4.438932791252913e-06,
      "loss": 0.7644,
      "step": 4315
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9049565928982354,
      "learning_rate": 4.436882562745099e-06,
      "loss": 0.6371,
      "step": 4316
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8313283436625327,
      "learning_rate": 4.434832430130319e-06,
      "loss": 0.6796,
      "step": 4317
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.872529747207513,
      "learning_rate": 4.432782393757687e-06,
      "loss": 0.6456,
      "step": 4318
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.898426694309625,
      "learning_rate": 4.430732453976305e-06,
      "loss": 0.594,
      "step": 4319
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7241863911057442,
      "learning_rate": 4.428682611135255e-06,
      "loss": 0.6491,
      "step": 4320
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6795563495998798,
      "learning_rate": 4.426632865583605e-06,
      "loss": 0.5932,
      "step": 4321
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8702003187395987,
      "learning_rate": 4.424583217670405e-06,
      "loss": 0.6853,
      "step": 4322
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.819095505720782,
      "learning_rate": 4.4225336677446875e-06,
      "loss": 0.5969,
      "step": 4323
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7352696441835315,
      "learning_rate": 4.42048421615547e-06,
      "loss": 0.6246,
      "step": 4324
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7780170497139071,
      "learning_rate": 4.418434863251754e-06,
      "loss": 0.5706,
      "step": 4325
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9237510905134435,
      "learning_rate": 4.416385609382522e-06,
      "loss": 0.6308,
      "step": 4326
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7634558482389361,
      "learning_rate": 4.414336454896739e-06,
      "loss": 0.6874,
      "step": 4327
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8966712398318945,
      "learning_rate": 4.412287400143358e-06,
      "loss": 0.5595,
      "step": 4328
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8721660430327461,
      "learning_rate": 4.41023844547131e-06,
      "loss": 0.7074,
      "step": 4329
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.798209988546998,
      "learning_rate": 4.40818959122951e-06,
      "loss": 0.6548,
      "step": 4330
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6215329889662746,
      "learning_rate": 4.406140837766858e-06,
      "loss": 0.5668,
      "step": 4331
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7887135662123005,
      "learning_rate": 4.404092185432234e-06,
      "loss": 0.639,
      "step": 4332
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1032231106437858,
      "learning_rate": 4.4020436345745035e-06,
      "loss": 0.6213,
      "step": 4333
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6681236292645945,
      "learning_rate": 4.399995185542512e-06,
      "loss": 0.5561,
      "step": 4334
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5892457216553397,
      "learning_rate": 4.3979468386850884e-06,
      "loss": 0.5103,
      "step": 4335
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7951539859174098,
      "learning_rate": 4.395898594351048e-06,
      "loss": 0.6113,
      "step": 4336
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9406727358474832,
      "learning_rate": 4.393850452889182e-06,
      "loss": 0.6753,
      "step": 4337
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9938822358645597,
      "learning_rate": 4.39180241464827e-06,
      "loss": 0.7064,
      "step": 4338
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6620203618799757,
      "learning_rate": 4.389754479977071e-06,
      "loss": 0.5229,
      "step": 4339
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8543077631730749,
      "learning_rate": 4.387706649224324e-06,
      "loss": 0.6289,
      "step": 4340
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6652160346965879,
      "learning_rate": 4.385658922738758e-06,
      "loss": 0.5096,
      "step": 4341
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7520137932327667,
      "learning_rate": 4.3836113008690776e-06,
      "loss": 0.6489,
      "step": 4342
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8398007389218389,
      "learning_rate": 4.38156378396397e-06,
      "loss": 0.6537,
      "step": 4343
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7955293278487024,
      "learning_rate": 4.379516372372108e-06,
      "loss": 0.6084,
      "step": 4344
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8455922624241003,
      "learning_rate": 4.377469066442143e-06,
      "loss": 0.6375,
      "step": 4345
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8594604085067442,
      "learning_rate": 4.375421866522711e-06,
      "loss": 0.5891,
      "step": 4346
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6950798836528627,
      "learning_rate": 4.373374772962428e-06,
      "loss": 0.556,
      "step": 4347
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7154165337187443,
      "learning_rate": 4.371327786109891e-06,
      "loss": 0.5628,
      "step": 4348
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.663563750111381,
      "learning_rate": 4.369280906313686e-06,
      "loss": 0.4827,
      "step": 4349
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7803592721361096,
      "learning_rate": 4.36723413392237e-06,
      "loss": 0.5781,
      "step": 4350
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.899832149414895,
      "learning_rate": 4.36518746928449e-06,
      "loss": 0.6147,
      "step": 4351
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9273090795043808,
      "learning_rate": 4.3631409127485696e-06,
      "loss": 0.6419,
      "step": 4352
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8563376047287822,
      "learning_rate": 4.361094464663118e-06,
      "loss": 0.6328,
      "step": 4353
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.6377745341816885,
      "learning_rate": 4.359048125376622e-06,
      "loss": 0.6369,
      "step": 4354
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7840415007860636,
      "learning_rate": 4.3570018952375535e-06,
      "loss": 0.5988,
      "step": 4355
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.631625727697405,
      "learning_rate": 4.3549557745943635e-06,
      "loss": 0.4872,
      "step": 4356
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.683577231303051,
      "learning_rate": 4.352909763795485e-06,
      "loss": 0.5207,
      "step": 4357
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8484486050390335,
      "learning_rate": 4.350863863189333e-06,
      "loss": 0.6315,
      "step": 4358
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7162493208386852,
      "learning_rate": 4.3488180731243e-06,
      "loss": 0.5671,
      "step": 4359
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8012950055965488,
      "learning_rate": 4.346772393948767e-06,
      "loss": 0.5329,
      "step": 4360
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7140012722654178,
      "learning_rate": 4.34472682601109e-06,
      "loss": 0.5461,
      "step": 4361
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.773054766575202,
      "learning_rate": 4.342681369659608e-06,
      "loss": 0.5389,
      "step": 4362
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8367063710005508,
      "learning_rate": 4.340636025242641e-06,
      "loss": 0.6522,
      "step": 4363
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9129576504524181,
      "learning_rate": 4.33859079310849e-06,
      "loss": 0.5837,
      "step": 4364
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8391651759626918,
      "learning_rate": 4.336545673605435e-06,
      "loss": 0.5885,
      "step": 4365
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9764985524117383,
      "learning_rate": 4.334500667081742e-06,
      "loss": 0.6235,
      "step": 4366
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.999775309604442,
      "learning_rate": 4.332455773885651e-06,
      "loss": 0.6399,
      "step": 4367
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7513709360270177,
      "learning_rate": 4.330410994365388e-06,
      "loss": 0.6117,
      "step": 4368
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6964864276009453,
      "learning_rate": 4.3283663288691555e-06,
      "loss": 0.5478,
      "step": 4369
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8829931466511156,
      "learning_rate": 4.326321777745142e-06,
      "loss": 0.6725,
      "step": 4370
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8564211092020534,
      "learning_rate": 4.32427734134151e-06,
      "loss": 0.6497,
      "step": 4371
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8081600591293844,
      "learning_rate": 4.322233020006407e-06,
      "loss": 0.498,
      "step": 4372
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7116148782644764,
      "learning_rate": 4.32018881408796e-06,
      "loss": 0.5756,
      "step": 4373
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7435320597155147,
      "learning_rate": 4.318144723934276e-06,
      "loss": 0.5371,
      "step": 4374
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.831035640799675,
      "learning_rate": 4.316100749893443e-06,
      "loss": 0.6098,
      "step": 4375
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.822788003880369,
      "learning_rate": 4.3140568923135265e-06,
      "loss": 0.6174,
      "step": 4376
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7033814927788082,
      "learning_rate": 4.312013151542575e-06,
      "loss": 0.557,
      "step": 4377
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.820223218945259,
      "learning_rate": 4.309969527928616e-06,
      "loss": 0.678,
      "step": 4378
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8075742775220537,
      "learning_rate": 4.307926021819659e-06,
      "loss": 0.6013,
      "step": 4379
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.828809443171805,
      "learning_rate": 4.30588263356369e-06,
      "loss": 0.5721,
      "step": 4380
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8607363203360329,
      "learning_rate": 4.303839363508677e-06,
      "loss": 0.706,
      "step": 4381
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.982740544125849,
      "learning_rate": 4.301796212002568e-06,
      "loss": 0.6997,
      "step": 4382
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7879554640148345,
      "learning_rate": 4.29975317939329e-06,
      "loss": 0.5897,
      "step": 4383
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8628538771348504,
      "learning_rate": 4.297710266028751e-06,
      "loss": 0.6441,
      "step": 4384
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8058890579576748,
      "learning_rate": 4.295667472256836e-06,
      "loss": 0.6415,
      "step": 4385
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6715754210015439,
      "learning_rate": 4.293624798425414e-06,
      "loss": 0.5201,
      "step": 4386
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8451684900489934,
      "learning_rate": 4.291582244882329e-06,
      "loss": 0.5653,
      "step": 4387
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8188198937442785,
      "learning_rate": 4.289539811975408e-06,
      "loss": 0.6201,
      "step": 4388
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.775092320627096,
      "learning_rate": 4.287497500052456e-06,
      "loss": 0.6265,
      "step": 4389
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7084126950507942,
      "learning_rate": 4.285455309461256e-06,
      "loss": 0.534,
      "step": 4390
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.900616535504914,
      "learning_rate": 4.283413240549574e-06,
      "loss": 0.6659,
      "step": 4391
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8208774478706352,
      "learning_rate": 4.2813712936651506e-06,
      "loss": 0.6328,
      "step": 4392
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.782433050104911,
      "learning_rate": 4.27932946915571e-06,
      "loss": 0.6253,
      "step": 4393
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6842266867942965,
      "learning_rate": 4.277287767368952e-06,
      "loss": 0.5875,
      "step": 4394
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7663245092156283,
      "learning_rate": 4.275246188652559e-06,
      "loss": 0.6032,
      "step": 4395
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6752798482582338,
      "learning_rate": 4.2732047333541895e-06,
      "loss": 0.524,
      "step": 4396
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9313095502623406,
      "learning_rate": 4.271163401821482e-06,
      "loss": 0.6069,
      "step": 4397
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8019830734169505,
      "learning_rate": 4.269122194402054e-06,
      "loss": 0.6264,
      "step": 4398
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8457642123745607,
      "learning_rate": 4.267081111443501e-06,
      "loss": 0.5841,
      "step": 4399
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6067332774236164,
      "learning_rate": 4.2650401532934005e-06,
      "loss": 0.4988,
      "step": 4400
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6746976738983865,
      "learning_rate": 4.262999320299306e-06,
      "loss": 0.6042,
      "step": 4401
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9593765997588732,
      "learning_rate": 4.260958612808748e-06,
      "loss": 0.6084,
      "step": 4402
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9496935712497462,
      "learning_rate": 4.258918031169239e-06,
      "loss": 0.6593,
      "step": 4403
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7501932363545909,
      "learning_rate": 4.256877575728269e-06,
      "loss": 0.5859,
      "step": 4404
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6681831403497059,
      "learning_rate": 4.254837246833307e-06,
      "loss": 0.5585,
      "step": 4405
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8352713747574878,
      "learning_rate": 4.252797044831797e-06,
      "loss": 0.5999,
      "step": 4406
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0033806169981134,
      "learning_rate": 4.250756970071167e-06,
      "loss": 0.6508,
      "step": 4407
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7415248405937985,
      "learning_rate": 4.24871702289882e-06,
      "loss": 0.5617,
      "step": 4408
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6800380634485673,
      "learning_rate": 4.246677203662137e-06,
      "loss": 0.5526,
      "step": 4409
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6752305504198819,
      "learning_rate": 4.2446375127084775e-06,
      "loss": 0.5597,
      "step": 4410
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.860725233434424,
      "learning_rate": 4.24259795038518e-06,
      "loss": 0.6521,
      "step": 4411
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6410028005612902,
      "learning_rate": 4.2405585170395605e-06,
      "loss": 0.5173,
      "step": 4412
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.597169745449015,
      "learning_rate": 4.2385192130189145e-06,
      "loss": 0.5476,
      "step": 4413
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6978752809713807,
      "learning_rate": 4.236480038670514e-06,
      "loss": 0.5287,
      "step": 4414
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8754603037400586,
      "learning_rate": 4.234440994341608e-06,
      "loss": 0.5991,
      "step": 4415
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6800498802828441,
      "learning_rate": 4.232402080379426e-06,
      "loss": 0.4805,
      "step": 4416
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8957000909929884,
      "learning_rate": 4.230363297131173e-06,
      "loss": 0.6186,
      "step": 4417
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8366311584683085,
      "learning_rate": 4.228324644944032e-06,
      "loss": 0.6803,
      "step": 4418
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7781682995555774,
      "learning_rate": 4.2262861241651635e-06,
      "loss": 0.623,
      "step": 4419
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8073638100477304,
      "learning_rate": 4.224247735141709e-06,
      "loss": 0.5973,
      "step": 4420
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9235060005182137,
      "learning_rate": 4.222209478220783e-06,
      "loss": 0.5849,
      "step": 4421
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6945462825103523,
      "learning_rate": 4.220171353749481e-06,
      "loss": 0.5358,
      "step": 4422
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7069245518762349,
      "learning_rate": 4.218133362074872e-06,
      "loss": 0.5561,
      "step": 4423
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7018026728986374,
      "learning_rate": 4.216095503544006e-06,
      "loss": 0.5183,
      "step": 4424
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7844546732012173,
      "learning_rate": 4.214057778503908e-06,
      "loss": 0.5966,
      "step": 4425
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8848703650925667,
      "learning_rate": 4.212020187301584e-06,
      "loss": 0.6518,
      "step": 4426
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6448911228036215,
      "learning_rate": 4.209982730284012e-06,
      "loss": 0.5281,
      "step": 4427
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6361294523369796,
      "learning_rate": 4.2079454077981516e-06,
      "loss": 0.5347,
      "step": 4428
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7748772916124865,
      "learning_rate": 4.205908220190938e-06,
      "loss": 0.581,
      "step": 4429
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8100475158397745,
      "learning_rate": 4.203871167809279e-06,
      "loss": 0.5431,
      "step": 4430
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8065475573830592,
      "learning_rate": 4.201834251000068e-06,
      "loss": 0.6559,
      "step": 4431
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7141958428718586,
      "learning_rate": 4.199797470110168e-06,
      "loss": 0.52,
      "step": 4432
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9433393396723571,
      "learning_rate": 4.197760825486423e-06,
      "loss": 0.5945,
      "step": 4433
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6421198110735689,
      "learning_rate": 4.195724317475651e-06,
      "loss": 0.4773,
      "step": 4434
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7918566182329413,
      "learning_rate": 4.193687946424649e-06,
      "loss": 0.6136,
      "step": 4435
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.795441187698482,
      "learning_rate": 4.1916517126801905e-06,
      "loss": 0.6243,
      "step": 4436
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6914072720914571,
      "learning_rate": 4.189615616589024e-06,
      "loss": 0.5451,
      "step": 4437
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6738469166382922,
      "learning_rate": 4.1875796584978725e-06,
      "loss": 0.4631,
      "step": 4438
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6544194302610971,
      "learning_rate": 4.185543838753444e-06,
      "loss": 0.5412,
      "step": 4439
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7756213319689607,
      "learning_rate": 4.183508157702415e-06,
      "loss": 0.5198,
      "step": 4440
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8754822293947353,
      "learning_rate": 4.18147261569144e-06,
      "loss": 0.6343,
      "step": 4441
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7879180220291384,
      "learning_rate": 4.1794372130671515e-06,
      "loss": 0.5442,
      "step": 4442
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9695887602577561,
      "learning_rate": 4.177401950176156e-06,
      "loss": 0.6694,
      "step": 4443
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0284798486476336,
      "learning_rate": 4.175366827365041e-06,
      "loss": 0.5992,
      "step": 4444
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7250527780498202,
      "learning_rate": 4.173331844980362e-06,
      "loss": 0.5693,
      "step": 4445
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7051701526998766,
      "learning_rate": 4.171297003368658e-06,
      "loss": 0.5209,
      "step": 4446
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.800320758976769,
      "learning_rate": 4.169262302876443e-06,
      "loss": 0.6316,
      "step": 4447
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7064458499898304,
      "learning_rate": 4.167227743850201e-06,
      "loss": 0.5428,
      "step": 4448
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7263998875079912,
      "learning_rate": 4.165193326636397e-06,
      "loss": 0.5444,
      "step": 4449
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7264158283065071,
      "learning_rate": 4.163159051581475e-06,
      "loss": 0.5255,
      "step": 4450
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7908346881017806,
      "learning_rate": 4.161124919031848e-06,
      "loss": 0.6021,
      "step": 4451
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8809202758921088,
      "learning_rate": 4.159090929333907e-06,
      "loss": 0.6834,
      "step": 4452
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.640701368942295,
      "learning_rate": 4.157057082834022e-06,
      "loss": 0.5602,
      "step": 4453
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8437108871495803,
      "learning_rate": 4.1550233798785335e-06,
      "loss": 0.602,
      "step": 4454
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8765935116171523,
      "learning_rate": 4.1529898208137604e-06,
      "loss": 0.6847,
      "step": 4455
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8698150831041311,
      "learning_rate": 4.150956405985996e-06,
      "loss": 0.6159,
      "step": 4456
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9008177394925169,
      "learning_rate": 4.14892313574151e-06,
      "loss": 0.613,
      "step": 4457
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8694715830956062,
      "learning_rate": 4.146890010426549e-06,
      "loss": 0.644,
      "step": 4458
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8256535438576119,
      "learning_rate": 4.1448570303873305e-06,
      "loss": 0.6722,
      "step": 4459
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.772849376790894,
      "learning_rate": 4.142824195970052e-06,
      "loss": 0.6084,
      "step": 4460
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.829374809555207,
      "learning_rate": 4.140791507520883e-06,
      "loss": 0.6482,
      "step": 4461
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7964565266071935,
      "learning_rate": 4.1387589653859686e-06,
      "loss": 0.6163,
      "step": 4462
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.678221986857363,
      "learning_rate": 4.13672656991143e-06,
      "loss": 0.5125,
      "step": 4463
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7854990822333536,
      "learning_rate": 4.134694321443363e-06,
      "loss": 0.624,
      "step": 4464
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6059465308241474,
      "learning_rate": 4.132662220327838e-06,
      "loss": 0.5079,
      "step": 4465
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9156875796548971,
      "learning_rate": 4.130630266910902e-06,
      "loss": 0.6074,
      "step": 4466
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7225565365590488,
      "learning_rate": 4.128598461538574e-06,
      "loss": 0.5731,
      "step": 4467
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7980905953476304,
      "learning_rate": 4.1265668045568495e-06,
      "loss": 0.61,
      "step": 4468
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7936195448809921,
      "learning_rate": 4.124535296311699e-06,
      "loss": 0.6294,
      "step": 4469
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6857743393778188,
      "learning_rate": 4.122503937149066e-06,
      "loss": 0.5671,
      "step": 4470
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8509053202680728,
      "learning_rate": 4.120472727414872e-06,
      "loss": 0.617,
      "step": 4471
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6936505336685249,
      "learning_rate": 4.11844166745501e-06,
      "loss": 0.5831,
      "step": 4472
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6665547368207002,
      "learning_rate": 4.1164107576153475e-06,
      "loss": 0.5106,
      "step": 4473
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7040820145719161,
      "learning_rate": 4.1143799982417276e-06,
      "loss": 0.4656,
      "step": 4474
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.03498755693866,
      "learning_rate": 4.112349389679968e-06,
      "loss": 0.6186,
      "step": 4475
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6291052651966531,
      "learning_rate": 4.11031893227586e-06,
      "loss": 0.4835,
      "step": 4476
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0405536087578553,
      "learning_rate": 4.108288626375169e-06,
      "loss": 0.6228,
      "step": 4477
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8495624840035347,
      "learning_rate": 4.106258472323635e-06,
      "loss": 0.5264,
      "step": 4478
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9311194202820979,
      "learning_rate": 4.104228470466973e-06,
      "loss": 0.6418,
      "step": 4479
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7403730629951322,
      "learning_rate": 4.1021986211508695e-06,
      "loss": 0.5616,
      "step": 4480
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8997188331422484,
      "learning_rate": 4.100168924720989e-06,
      "loss": 0.6169,
      "step": 4481
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8682297629657586,
      "learning_rate": 4.098139381522964e-06,
      "loss": 0.6557,
      "step": 4482
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.667710364478011,
      "learning_rate": 4.096109991902407e-06,
      "loss": 0.4878,
      "step": 4483
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6826195080322754,
      "learning_rate": 4.094080756204904e-06,
      "loss": 0.5451,
      "step": 4484
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9368518362681154,
      "learning_rate": 4.092051674776008e-06,
      "loss": 0.5787,
      "step": 4485
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8312111914194017,
      "learning_rate": 4.090022747961254e-06,
      "loss": 0.6435,
      "step": 4486
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7570072202531515,
      "learning_rate": 4.087993976106146e-06,
      "loss": 0.5958,
      "step": 4487
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6610870430350286,
      "learning_rate": 4.085965359556163e-06,
      "loss": 0.4931,
      "step": 4488
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6291958819416069,
      "learning_rate": 4.083936898656756e-06,
      "loss": 0.5615,
      "step": 4489
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9192216162257776,
      "learning_rate": 4.081908593753353e-06,
      "loss": 0.6523,
      "step": 4490
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7016562543049941,
      "learning_rate": 4.079880445191351e-06,
      "loss": 0.5255,
      "step": 4491
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7144108847668628,
      "learning_rate": 4.0778524533161235e-06,
      "loss": 0.5683,
      "step": 4492
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8957817686677506,
      "learning_rate": 4.075824618473016e-06,
      "loss": 0.6105,
      "step": 4493
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8456968990107605,
      "learning_rate": 4.0737969410073485e-06,
      "loss": 0.634,
      "step": 4494
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8938367414031712,
      "learning_rate": 4.071769421264413e-06,
      "loss": 0.5933,
      "step": 4495
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6861271589415892,
      "learning_rate": 4.0697420595894735e-06,
      "loss": 0.4993,
      "step": 4496
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7872958407726934,
      "learning_rate": 4.067714856327771e-06,
      "loss": 0.5295,
      "step": 4497
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8401578552600698,
      "learning_rate": 4.065687811824516e-06,
      "loss": 0.6081,
      "step": 4498
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6303556128274284,
      "learning_rate": 4.0636609264248926e-06,
      "loss": 0.5833,
      "step": 4499
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7449710690106373,
      "learning_rate": 4.061634200474059e-06,
      "loss": 0.5435,
      "step": 4500
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6394832923598611,
      "learning_rate": 4.059607634317146e-06,
      "loss": 0.5495,
      "step": 4501
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6202441411313765,
      "learning_rate": 4.057581228299254e-06,
      "loss": 0.539,
      "step": 4502
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.839989360612665,
      "learning_rate": 4.055554982765463e-06,
      "loss": 0.6995,
      "step": 4503
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8550292360545774,
      "learning_rate": 4.053528898060818e-06,
      "loss": 0.5956,
      "step": 4504
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.777474829516731,
      "learning_rate": 4.051502974530341e-06,
      "loss": 0.528,
      "step": 4505
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6639671499491367,
      "learning_rate": 4.049477212519028e-06,
      "loss": 0.5206,
      "step": 4506
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6337231399568349,
      "learning_rate": 4.047451612371841e-06,
      "loss": 0.5314,
      "step": 4507
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8099585713264955,
      "learning_rate": 4.045426174433722e-06,
      "loss": 0.5947,
      "step": 4508
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0325086468550524,
      "learning_rate": 4.043400899049581e-06,
      "loss": 0.6728,
      "step": 4509
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8121420007393971,
      "learning_rate": 4.041375786564299e-06,
      "loss": 0.5988,
      "step": 4510
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5994445604614119,
      "learning_rate": 4.039350837322736e-06,
      "loss": 0.4898,
      "step": 4511
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7664533881416199,
      "learning_rate": 4.037326051669717e-06,
      "loss": 0.5529,
      "step": 4512
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7339234094451136,
      "learning_rate": 4.0353014299500425e-06,
      "loss": 0.541,
      "step": 4513
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.834292436273063,
      "learning_rate": 4.033276972508484e-06,
      "loss": 0.685,
      "step": 4514
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6373930343283828,
      "learning_rate": 4.031252679689785e-06,
      "loss": 0.5358,
      "step": 4515
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6723670265566082,
      "learning_rate": 4.029228551838664e-06,
      "loss": 0.5836,
      "step": 4516
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8286246667984236,
      "learning_rate": 4.027204589299805e-06,
      "loss": 0.5702,
      "step": 4517
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7497012728183629,
      "learning_rate": 4.0251807924178725e-06,
      "loss": 0.5202,
      "step": 4518
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8300544722357185,
      "learning_rate": 4.023157161537494e-06,
      "loss": 0.649,
      "step": 4519
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8741021675353209,
      "learning_rate": 4.021133697003275e-06,
      "loss": 0.6232,
      "step": 4520
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8095469614215342,
      "learning_rate": 4.019110399159789e-06,
      "loss": 0.6644,
      "step": 4521
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7679272331147581,
      "learning_rate": 4.017087268351583e-06,
      "loss": 0.5326,
      "step": 4522
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.808127212469725,
      "learning_rate": 4.0150643049231744e-06,
      "loss": 0.552,
      "step": 4523
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8087816256788897,
      "learning_rate": 4.013041509219055e-06,
      "loss": 0.6077,
      "step": 4524
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7055910271872686,
      "learning_rate": 4.011018881583684e-06,
      "loss": 0.4965,
      "step": 4525
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7706770867690466,
      "learning_rate": 4.008996422361495e-06,
      "loss": 0.6467,
      "step": 4526
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7721696724865775,
      "learning_rate": 4.006974131896891e-06,
      "loss": 0.581,
      "step": 4527
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7927763138443585,
      "learning_rate": 4.004952010534247e-06,
      "loss": 0.6204,
      "step": 4528
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8948284011904157,
      "learning_rate": 4.002930058617909e-06,
      "loss": 0.6197,
      "step": 4529
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.821729284960434,
      "learning_rate": 4.000908276492195e-06,
      "loss": 0.653,
      "step": 4530
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6143805701987523,
      "learning_rate": 3.9988866645013925e-06,
      "loss": 0.4926,
      "step": 4531
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8311299352373979,
      "learning_rate": 3.996865222989762e-06,
      "loss": 0.615,
      "step": 4532
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8094146121329434,
      "learning_rate": 3.994843952301533e-06,
      "loss": 0.6042,
      "step": 4533
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8648595247341165,
      "learning_rate": 3.992822852780908e-06,
      "loss": 0.65,
      "step": 4534
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.896037340784818,
      "learning_rate": 3.9908019247720584e-06,
      "loss": 0.635,
      "step": 4535
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6806823867919641,
      "learning_rate": 3.988781168619125e-06,
      "loss": 0.5329,
      "step": 4536
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6895522545326527,
      "learning_rate": 3.986760584666226e-06,
      "loss": 0.5124,
      "step": 4537
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.632482621101068,
      "learning_rate": 3.984740173257443e-06,
      "loss": 0.5045,
      "step": 4538
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7743533515950725,
      "learning_rate": 3.982719934736832e-06,
      "loss": 0.5986,
      "step": 4539
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8637524743488536,
      "learning_rate": 3.9806998694484185e-06,
      "loss": 0.6008,
      "step": 4540
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7504590681424177,
      "learning_rate": 3.978679977736197e-06,
      "loss": 0.5439,
      "step": 4541
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6844718007071564,
      "learning_rate": 3.976660259944136e-06,
      "loss": 0.5508,
      "step": 4542
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8007898019090697,
      "learning_rate": 3.974640716416172e-06,
      "loss": 0.5395,
      "step": 4543
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.853710962479426,
      "learning_rate": 3.972621347496212e-06,
      "loss": 0.6688,
      "step": 4544
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7813937002883821,
      "learning_rate": 3.970602153528133e-06,
      "loss": 0.5704,
      "step": 4545
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7037140207088858,
      "learning_rate": 3.968583134855782e-06,
      "loss": 0.5257,
      "step": 4546
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6949108202317474,
      "learning_rate": 3.966564291822979e-06,
      "loss": 0.5454,
      "step": 4547
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6074549916156692,
      "learning_rate": 3.9645456247735116e-06,
      "loss": 0.4829,
      "step": 4548
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6722261071294441,
      "learning_rate": 3.962527134051135e-06,
      "loss": 0.5357,
      "step": 4549
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7552968197412698,
      "learning_rate": 3.96050881999958e-06,
      "loss": 0.567,
      "step": 4550
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7285588351863173,
      "learning_rate": 3.958490682962543e-06,
      "loss": 0.4673,
      "step": 4551
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6788614589431907,
      "learning_rate": 3.956472723283693e-06,
      "loss": 0.5365,
      "step": 4552
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9948833611993003,
      "learning_rate": 3.954454941306667e-06,
      "loss": 0.6932,
      "step": 4553
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6456013077423708,
      "learning_rate": 3.952437337375072e-06,
      "loss": 0.4828,
      "step": 4554
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.712819145937391,
      "learning_rate": 3.9504199118324845e-06,
      "loss": 0.5281,
      "step": 4555
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9378994491712823,
      "learning_rate": 3.948402665022452e-06,
      "loss": 0.6541,
      "step": 4556
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.85946904511879,
      "learning_rate": 3.94638559728849e-06,
      "loss": 0.6165,
      "step": 4557
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8486353526538747,
      "learning_rate": 3.944368708974086e-06,
      "loss": 0.6487,
      "step": 4558
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6359170716143816,
      "learning_rate": 3.9423520004226916e-06,
      "loss": 0.547,
      "step": 4559
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.647376383410015,
      "learning_rate": 3.940335471977733e-06,
      "loss": 0.5506,
      "step": 4560
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0956950623384702,
      "learning_rate": 3.938319123982605e-06,
      "loss": 0.5261,
      "step": 4561
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9461874589705112,
      "learning_rate": 3.93630295678067e-06,
      "loss": 0.6038,
      "step": 4562
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8790905280523886,
      "learning_rate": 3.934286970715259e-06,
      "loss": 0.6461,
      "step": 4563
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7887680735411222,
      "learning_rate": 3.932271166129674e-06,
      "loss": 0.5799,
      "step": 4564
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.86126683838304,
      "learning_rate": 3.930255543367187e-06,
      "loss": 0.5982,
      "step": 4565
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8141051902610317,
      "learning_rate": 3.928240102771036e-06,
      "loss": 0.5239,
      "step": 4566
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.046501032184945,
      "learning_rate": 3.926224844684429e-06,
      "loss": 0.6756,
      "step": 4567
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6638123677878982,
      "learning_rate": 3.924209769450545e-06,
      "loss": 0.5314,
      "step": 4568
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9880581748134359,
      "learning_rate": 3.92219487741253e-06,
      "loss": 0.7062,
      "step": 4569
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8314958282450264,
      "learning_rate": 3.920180168913499e-06,
      "loss": 0.5755,
      "step": 4570
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7342076988686367,
      "learning_rate": 3.9181656442965356e-06,
      "loss": 0.5916,
      "step": 4571
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6965393431030459,
      "learning_rate": 3.916151303904693e-06,
      "loss": 0.5236,
      "step": 4572
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.706958773422459,
      "learning_rate": 3.914137148080991e-06,
      "loss": 0.5486,
      "step": 4573
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8366107787131191,
      "learning_rate": 3.9121231771684205e-06,
      "loss": 0.5979,
      "step": 4574
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9636515129332259,
      "learning_rate": 3.91010939150994e-06,
      "loss": 0.6533,
      "step": 4575
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7284865865019958,
      "learning_rate": 3.9080957914484756e-06,
      "loss": 0.568,
      "step": 4576
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.962932449502878,
      "learning_rate": 3.906082377326922e-06,
      "loss": 0.6931,
      "step": 4577
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6857622066104377,
      "learning_rate": 3.9040691494881445e-06,
      "loss": 0.5394,
      "step": 4578
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8038457589699521,
      "learning_rate": 3.902056108274972e-06,
      "loss": 0.543,
      "step": 4579
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.82037980807607,
      "learning_rate": 3.900043254030207e-06,
      "loss": 0.6635,
      "step": 4580
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6766786092908463,
      "learning_rate": 3.898030587096615e-06,
      "loss": 0.5371,
      "step": 4581
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8474691008789372,
      "learning_rate": 3.8960181078169355e-06,
      "loss": 0.629,
      "step": 4582
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6418338157282201,
      "learning_rate": 3.89400581653387e-06,
      "loss": 0.545,
      "step": 4583
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3789183579528237,
      "learning_rate": 3.891993713590093e-06,
      "loss": 0.6892,
      "step": 4584
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7522069662142868,
      "learning_rate": 3.889981799328242e-06,
      "loss": 0.5264,
      "step": 4585
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6971872882571089,
      "learning_rate": 3.887970074090928e-06,
      "loss": 0.5322,
      "step": 4586
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6600219331143145,
      "learning_rate": 3.885958538220724e-06,
      "loss": 0.5353,
      "step": 4587
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7505581797861914,
      "learning_rate": 3.883947192060176e-06,
      "loss": 0.576,
      "step": 4588
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9315520977338261,
      "learning_rate": 3.8819360359517936e-06,
      "loss": 0.6647,
      "step": 4589
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9030072230898907,
      "learning_rate": 3.8799250702380565e-06,
      "loss": 0.6471,
      "step": 4590
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6306693346595053,
      "learning_rate": 3.877914295261411e-06,
      "loss": 0.5247,
      "step": 4591
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6777270719991172,
      "learning_rate": 3.875903711364271e-06,
      "loss": 0.5738,
      "step": 4592
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7870825650535681,
      "learning_rate": 3.873893318889019e-06,
      "loss": 0.5917,
      "step": 4593
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6608604149554316,
      "learning_rate": 3.8718831181780015e-06,
      "loss": 0.5092,
      "step": 4594
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7626094806052015,
      "learning_rate": 3.869873109573537e-06,
      "loss": 0.5792,
      "step": 4595
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8066414470826253,
      "learning_rate": 3.867863293417909e-06,
      "loss": 0.6802,
      "step": 4596
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7414355852365169,
      "learning_rate": 3.865853670053368e-06,
      "loss": 0.5681,
      "step": 4597
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7070149058058068,
      "learning_rate": 3.86384423982213e-06,
      "loss": 0.542,
      "step": 4598
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7060132998178545,
      "learning_rate": 3.861835003066383e-06,
      "loss": 0.5763,
      "step": 4599
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6679425118377987,
      "learning_rate": 3.859825960128277e-06,
      "loss": 0.5571,
      "step": 4600
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6982893625550693,
      "learning_rate": 3.857817111349932e-06,
      "loss": 0.5489,
      "step": 4601
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6738943386830591,
      "learning_rate": 3.855808457073434e-06,
      "loss": 0.5376,
      "step": 4602
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7562338356194049,
      "learning_rate": 3.853799997640836e-06,
      "loss": 0.5156,
      "step": 4603
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8299941146368418,
      "learning_rate": 3.8517917333941554e-06,
      "loss": 0.6037,
      "step": 4604
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7066288611132916,
      "learning_rate": 3.8497836646753826e-06,
      "loss": 0.513,
      "step": 4605
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7717550489671091,
      "learning_rate": 3.847775791826468e-06,
      "loss": 0.608,
      "step": 4606
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6906201239731287,
      "learning_rate": 3.8457681151893315e-06,
      "loss": 0.5773,
      "step": 4607
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9282421546294841,
      "learning_rate": 3.843760635105861e-06,
      "loss": 0.6021,
      "step": 4608
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7750930057114463,
      "learning_rate": 3.841753351917909e-06,
      "loss": 0.5486,
      "step": 4609
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8456932131654048,
      "learning_rate": 3.839746265967295e-06,
      "loss": 0.647,
      "step": 4610
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8985318556376093,
      "learning_rate": 3.837739377595804e-06,
      "loss": 0.6597,
      "step": 4611
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.759034253056153,
      "learning_rate": 3.835732687145189e-06,
      "loss": 0.5717,
      "step": 4612
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6056031392324246,
      "learning_rate": 3.833726194957168e-06,
      "loss": 0.5248,
      "step": 4613
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7926161510391273,
      "learning_rate": 3.831719901373426e-06,
      "loss": 0.5793,
      "step": 4614
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7834046507953263,
      "learning_rate": 3.829713806735613e-06,
      "loss": 0.5877,
      "step": 4615
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8392907713347888,
      "learning_rate": 3.827707911385348e-06,
      "loss": 0.5155,
      "step": 4616
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.669347155304091,
      "learning_rate": 3.825702215664212e-06,
      "loss": 0.5324,
      "step": 4617
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7501519524702971,
      "learning_rate": 3.8236967199137546e-06,
      "loss": 0.5442,
      "step": 4618
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6389394535667118,
      "learning_rate": 3.821691424475492e-06,
      "loss": 0.5189,
      "step": 4619
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7239492452737994,
      "learning_rate": 3.8196863296909035e-06,
      "loss": 0.6101,
      "step": 4620
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6756528275476881,
      "learning_rate": 3.817681435901436e-06,
      "loss": 0.5359,
      "step": 4621
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9951737071097115,
      "learning_rate": 3.815676743448504e-06,
      "loss": 0.6635,
      "step": 4622
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7063950037658637,
      "learning_rate": 3.8136722526734837e-06,
      "loss": 0.5435,
      "step": 4623
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7982005672170439,
      "learning_rate": 3.81166796391772e-06,
      "loss": 0.6796,
      "step": 4624
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7428193680315647,
      "learning_rate": 3.8096638775225226e-06,
      "loss": 0.5335,
      "step": 4625
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7715154433110496,
      "learning_rate": 3.8076599938291652e-06,
      "loss": 0.5542,
      "step": 4626
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7798028238396731,
      "learning_rate": 3.805656313178889e-06,
      "loss": 0.6276,
      "step": 4627
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6196570671219331,
      "learning_rate": 3.803652835912901e-06,
      "loss": 0.5211,
      "step": 4628
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6441885082942713,
      "learning_rate": 3.8016495623723715e-06,
      "loss": 0.5431,
      "step": 4629
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8861784866832928,
      "learning_rate": 3.7996464928984365e-06,
      "loss": 0.6413,
      "step": 4630
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.893278252797954,
      "learning_rate": 3.797643627832199e-06,
      "loss": 0.6715,
      "step": 4631
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8204789881194132,
      "learning_rate": 3.795640967514726e-06,
      "loss": 0.522,
      "step": 4632
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9456295565789163,
      "learning_rate": 3.793638512287049e-06,
      "loss": 0.6476,
      "step": 4633
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.907713088082132,
      "learning_rate": 3.7916362624901643e-06,
      "loss": 0.6215,
      "step": 4634
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6440151261356034,
      "learning_rate": 3.7896342184650363e-06,
      "loss": 0.5188,
      "step": 4635
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7843371391700882,
      "learning_rate": 3.7876323805525917e-06,
      "loss": 0.5082,
      "step": 4636
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7105943494472945,
      "learning_rate": 3.7856307490937228e-06,
      "loss": 0.5924,
      "step": 4637
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7352941509804366,
      "learning_rate": 3.7836293244292853e-06,
      "loss": 0.5724,
      "step": 4638
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6709134273438605,
      "learning_rate": 3.7816281069001026e-06,
      "loss": 0.5087,
      "step": 4639
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7024961024364729,
      "learning_rate": 3.7796270968469595e-06,
      "loss": 0.5335,
      "step": 4640
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7754288675854669,
      "learning_rate": 3.7776262946106086e-06,
      "loss": 0.5844,
      "step": 4641
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8400529841645835,
      "learning_rate": 3.7756257005317655e-06,
      "loss": 0.5998,
      "step": 4642
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6636386635724815,
      "learning_rate": 3.7736253149511095e-06,
      "loss": 0.4426,
      "step": 4643
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6966823459396856,
      "learning_rate": 3.771625138209286e-06,
      "loss": 0.5491,
      "step": 4644
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.879565056877042,
      "learning_rate": 3.7696251706469044e-06,
      "loss": 0.6567,
      "step": 4645
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6371345073659821,
      "learning_rate": 3.7676254126045384e-06,
      "loss": 0.5631,
      "step": 4646
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.782881297588299,
      "learning_rate": 3.7656258644227256e-06,
      "loss": 0.5461,
      "step": 4647
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8501299172349109,
      "learning_rate": 3.7636265264419684e-06,
      "loss": 0.6774,
      "step": 4648
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.729508395051032,
      "learning_rate": 3.7616273990027343e-06,
      "loss": 0.5737,
      "step": 4649
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9136969319057946,
      "learning_rate": 3.7596284824454522e-06,
      "loss": 0.6399,
      "step": 4650
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8765847610149756,
      "learning_rate": 3.7576297771105176e-06,
      "loss": 0.6602,
      "step": 4651
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8372393351581214,
      "learning_rate": 3.755631283338287e-06,
      "loss": 0.6276,
      "step": 4652
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.759296036998442,
      "learning_rate": 3.753633001469087e-06,
      "loss": 0.5784,
      "step": 4653
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9399949063765041,
      "learning_rate": 3.7516349318432022e-06,
      "loss": 0.6788,
      "step": 4654
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6104913000524868,
      "learning_rate": 3.7496370748008837e-06,
      "loss": 0.5438,
      "step": 4655
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8440338008111886,
      "learning_rate": 3.7476394306823443e-06,
      "loss": 0.6089,
      "step": 4656
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8037526191815576,
      "learning_rate": 3.745641999827764e-06,
      "loss": 0.5764,
      "step": 4657
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8958301210932439,
      "learning_rate": 3.7436447825772827e-06,
      "loss": 0.5668,
      "step": 4658
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8646863037424776,
      "learning_rate": 3.741647779271006e-06,
      "loss": 0.6235,
      "step": 4659
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9298616874351711,
      "learning_rate": 3.739650990249003e-06,
      "loss": 0.6228,
      "step": 4660
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.844443054430272,
      "learning_rate": 3.737654415851307e-06,
      "loss": 0.6376,
      "step": 4661
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8534129319078696,
      "learning_rate": 3.7356580564179125e-06,
      "loss": 0.6048,
      "step": 4662
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7368352764416887,
      "learning_rate": 3.7336619122887784e-06,
      "loss": 0.5357,
      "step": 4663
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6922930551973379,
      "learning_rate": 3.7316659838038293e-06,
      "loss": 0.5609,
      "step": 4664
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9932109609537716,
      "learning_rate": 3.7296702713029493e-06,
      "loss": 0.542,
      "step": 4665
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9321548113401201,
      "learning_rate": 3.727674775125987e-06,
      "loss": 0.6282,
      "step": 4666
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.888255319414392,
      "learning_rate": 3.7256794956127564e-06,
      "loss": 0.6004,
      "step": 4667
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7902035241214458,
      "learning_rate": 3.7236844331030316e-06,
      "loss": 0.5519,
      "step": 4668
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6890897742740402,
      "learning_rate": 3.721689587936551e-06,
      "loss": 0.5673,
      "step": 4669
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8579854987054736,
      "learning_rate": 3.7196949604530174e-06,
      "loss": 0.6468,
      "step": 4670
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9347135475111792,
      "learning_rate": 3.7177005509920925e-06,
      "loss": 0.66,
      "step": 4671
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8330090276775827,
      "learning_rate": 3.715706359893406e-06,
      "loss": 0.6495,
      "step": 4672
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8124484948509604,
      "learning_rate": 3.7137123874965462e-06,
      "loss": 0.537,
      "step": 4673
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7229605628048922,
      "learning_rate": 3.711718634141067e-06,
      "loss": 0.5305,
      "step": 4674
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7190834794394976,
      "learning_rate": 3.7097251001664824e-06,
      "loss": 0.5458,
      "step": 4675
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7924097644349875,
      "learning_rate": 3.707731785912272e-06,
      "loss": 0.5758,
      "step": 4676
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8025604062291016,
      "learning_rate": 3.705738691717875e-06,
      "loss": 0.6324,
      "step": 4677
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0967434466102395,
      "learning_rate": 3.703745817922696e-06,
      "loss": 0.6176,
      "step": 4678
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6557323406313835,
      "learning_rate": 3.701753164866098e-06,
      "loss": 0.5483,
      "step": 4679
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6099912962178521,
      "learning_rate": 3.699760732887413e-06,
      "loss": 0.5105,
      "step": 4680
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6938329244872425,
      "learning_rate": 3.6977685223259285e-06,
      "loss": 0.5101,
      "step": 4681
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6415441312676198,
      "learning_rate": 3.6957765335208983e-06,
      "loss": 0.5614,
      "step": 4682
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5911340111928757,
      "learning_rate": 3.6937847668115377e-06,
      "loss": 0.4684,
      "step": 4683
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6856324722840981,
      "learning_rate": 3.691793222537023e-06,
      "loss": 0.5009,
      "step": 4684
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7456400266005688,
      "learning_rate": 3.6898019010364934e-06,
      "loss": 0.5343,
      "step": 4685
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6818814852539161,
      "learning_rate": 3.6878108026490505e-06,
      "loss": 0.5622,
      "step": 4686
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.82951989671035,
      "learning_rate": 3.685819927713758e-06,
      "loss": 0.5651,
      "step": 4687
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6892505032378883,
      "learning_rate": 3.6838292765696403e-06,
      "loss": 0.5902,
      "step": 4688
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7555375188398477,
      "learning_rate": 3.6818388495556856e-06,
      "loss": 0.5895,
      "step": 4689
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6415078040758984,
      "learning_rate": 3.679848647010842e-06,
      "loss": 0.5481,
      "step": 4690
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7633098209620045,
      "learning_rate": 3.6778586692740205e-06,
      "loss": 0.5728,
      "step": 4691
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7461049811139693,
      "learning_rate": 3.675868916684093e-06,
      "loss": 0.5494,
      "step": 4692
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8512081032029335,
      "learning_rate": 3.673879389579895e-06,
      "loss": 0.5674,
      "step": 4693
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6281200941907052,
      "learning_rate": 3.671890088300222e-06,
      "loss": 0.5172,
      "step": 4694
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7935366897133953,
      "learning_rate": 3.66990101318383e-06,
      "loss": 0.5418,
      "step": 4695
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6703064248489337,
      "learning_rate": 3.6679121645694392e-06,
      "loss": 0.5116,
      "step": 4696
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9829618803020502,
      "learning_rate": 3.6659235427957286e-06,
      "loss": 0.6556,
      "step": 4697
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8332846806246345,
      "learning_rate": 3.663935148201341e-06,
      "loss": 0.5956,
      "step": 4698
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.675293595226352,
      "learning_rate": 3.6619469811248786e-06,
      "loss": 0.5396,
      "step": 4699
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7926402427488014,
      "learning_rate": 3.659959041904905e-06,
      "loss": 0.6698,
      "step": 4700
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6435200795986242,
      "learning_rate": 3.657971330879947e-06,
      "loss": 0.5134,
      "step": 4701
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8759783237199781,
      "learning_rate": 3.655983848388489e-06,
      "loss": 0.5587,
      "step": 4702
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5970316159027351,
      "learning_rate": 3.6539965947689804e-06,
      "loss": 0.492,
      "step": 4703
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8362960272663376,
      "learning_rate": 3.652009570359829e-06,
      "loss": 0.6446,
      "step": 4704
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6712304329028754,
      "learning_rate": 3.6500227754994046e-06,
      "loss": 0.5997,
      "step": 4705
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7885645641086384,
      "learning_rate": 3.648036210526037e-06,
      "loss": 0.579,
      "step": 4706
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8104668961366499,
      "learning_rate": 3.646049875778019e-06,
      "loss": 0.6482,
      "step": 4707
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8037702546272149,
      "learning_rate": 3.644063771593601e-06,
      "loss": 0.5602,
      "step": 4708
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8399692509109183,
      "learning_rate": 3.6420778983109974e-06,
      "loss": 0.6097,
      "step": 4709
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9974522283518326,
      "learning_rate": 3.640092256268381e-06,
      "loss": 0.6135,
      "step": 4710
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8759427678970045,
      "learning_rate": 3.638106845803886e-06,
      "loss": 0.6302,
      "step": 4711
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8720126711295777,
      "learning_rate": 3.6361216672556067e-06,
      "loss": 0.6297,
      "step": 4712
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6748568877463499,
      "learning_rate": 3.6341367209615987e-06,
      "loss": 0.5398,
      "step": 4713
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6485845582787974,
      "learning_rate": 3.6321520072598775e-06,
      "loss": 0.514,
      "step": 4714
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8883938647630483,
      "learning_rate": 3.6301675264884197e-06,
      "loss": 0.5815,
      "step": 4715
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7587650053415198,
      "learning_rate": 3.628183278985161e-06,
      "loss": 0.5781,
      "step": 4716
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.818713374105233,
      "learning_rate": 3.6261992650879986e-06,
      "loss": 0.6759,
      "step": 4717
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6274169080295308,
      "learning_rate": 3.6242154851347885e-06,
      "loss": 0.5247,
      "step": 4718
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8868389399481289,
      "learning_rate": 3.6222319394633488e-06,
      "loss": 0.6506,
      "step": 4719
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7376488209664734,
      "learning_rate": 3.6202486284114567e-06,
      "loss": 0.5648,
      "step": 4720
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.063400957229389,
      "learning_rate": 3.6182655523168493e-06,
      "loss": 0.6769,
      "step": 4721
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8040217481290075,
      "learning_rate": 3.6162827115172234e-06,
      "loss": 0.6563,
      "step": 4722
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6530126408569679,
      "learning_rate": 3.614300106350236e-06,
      "loss": 0.525,
      "step": 4723
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9042997592322346,
      "learning_rate": 3.6123177371535044e-06,
      "loss": 0.6736,
      "step": 4724
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.933224436955365,
      "learning_rate": 3.6103356042646054e-06,
      "loss": 0.5819,
      "step": 4725
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9131729759955876,
      "learning_rate": 3.608353708021076e-06,
      "loss": 0.6167,
      "step": 4726
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6846960226708735,
      "learning_rate": 3.6063720487604113e-06,
      "loss": 0.5191,
      "step": 4727
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7923886627651665,
      "learning_rate": 3.604390626820069e-06,
      "loss": 0.6139,
      "step": 4728
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6206508519151634,
      "learning_rate": 3.6024094425374625e-06,
      "loss": 0.485,
      "step": 4729
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6615512670131243,
      "learning_rate": 3.6004284962499686e-06,
      "loss": 0.5389,
      "step": 4730
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8409763260067968,
      "learning_rate": 3.5984477882949198e-06,
      "loss": 0.6355,
      "step": 4731
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7543320709288874,
      "learning_rate": 3.596467319009613e-06,
      "loss": 0.529,
      "step": 4732
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7972509596251136,
      "learning_rate": 3.5944870887312993e-06,
      "loss": 0.631,
      "step": 4733
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6931952200348175,
      "learning_rate": 3.592507097797192e-06,
      "loss": 0.5464,
      "step": 4734
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.84730794076543,
      "learning_rate": 3.5905273465444627e-06,
      "loss": 0.6511,
      "step": 4735
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0341919405731168,
      "learning_rate": 3.5885478353102428e-06,
      "loss": 0.6207,
      "step": 4736
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6754062150535651,
      "learning_rate": 3.5865685644316217e-06,
      "loss": 0.5452,
      "step": 4737
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.846293908241235,
      "learning_rate": 3.584589534245649e-06,
      "loss": 0.6017,
      "step": 4738
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.874981888161896,
      "learning_rate": 3.5826107450893337e-06,
      "loss": 0.613,
      "step": 4739
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8135745020609386,
      "learning_rate": 3.5806321972996416e-06,
      "loss": 0.5901,
      "step": 4740
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7981840046548233,
      "learning_rate": 3.5786538912135004e-06,
      "loss": 0.6247,
      "step": 4741
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8895879720504103,
      "learning_rate": 3.576675827167793e-06,
      "loss": 0.6475,
      "step": 4742
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.898437566901641,
      "learning_rate": 3.5746980054993646e-06,
      "loss": 0.6062,
      "step": 4743
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8722425470485958,
      "learning_rate": 3.5727204265450165e-06,
      "loss": 0.6639,
      "step": 4744
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6293242240164223,
      "learning_rate": 3.5707430906415108e-06,
      "loss": 0.5502,
      "step": 4745
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7281008507815523,
      "learning_rate": 3.568765998125567e-06,
      "loss": 0.5515,
      "step": 4746
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7322492096029514,
      "learning_rate": 3.5667891493338633e-06,
      "loss": 0.599,
      "step": 4747
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8091610492308802,
      "learning_rate": 3.564812544603036e-06,
      "loss": 0.6756,
      "step": 4748
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9074481453523645,
      "learning_rate": 3.5628361842696813e-06,
      "loss": 0.6557,
      "step": 4749
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0181248975968313,
      "learning_rate": 3.560860068670352e-06,
      "loss": 0.6977,
      "step": 4750
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7629335920540325,
      "learning_rate": 3.55888419814156e-06,
      "loss": 0.5502,
      "step": 4751
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6844872403287836,
      "learning_rate": 3.556908573019776e-06,
      "loss": 0.4857,
      "step": 4752
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.746712634132732,
      "learning_rate": 3.5549331936414276e-06,
      "loss": 0.5988,
      "step": 4753
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.681013144387057,
      "learning_rate": 3.5529580603428997e-06,
      "loss": 0.596,
      "step": 4754
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6851246791868632,
      "learning_rate": 3.5509831734605414e-06,
      "loss": 0.561,
      "step": 4755
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8228783616223908,
      "learning_rate": 3.549008533330652e-06,
      "loss": 0.6276,
      "step": 4756
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6257032998750763,
      "learning_rate": 3.547034140289494e-06,
      "loss": 0.468,
      "step": 4757
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9414331319022593,
      "learning_rate": 3.5450599946732846e-06,
      "loss": 0.641,
      "step": 4758
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6463721066780047,
      "learning_rate": 3.543086096818201e-06,
      "loss": 0.5247,
      "step": 4759
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7686275151203134,
      "learning_rate": 3.5411124470603763e-06,
      "loss": 0.5471,
      "step": 4760
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7552823240815967,
      "learning_rate": 3.539139045735904e-06,
      "loss": 0.5285,
      "step": 4761
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8300872376137409,
      "learning_rate": 3.5371658931808327e-06,
      "loss": 0.6134,
      "step": 4762
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.789759722416413,
      "learning_rate": 3.5351929897311698e-06,
      "loss": 0.5262,
      "step": 4763
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6988512924983046,
      "learning_rate": 3.5332203357228813e-06,
      "loss": 0.556,
      "step": 4764
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.717569948873118,
      "learning_rate": 3.5312479314918878e-06,
      "loss": 0.575,
      "step": 4765
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7087328481554697,
      "learning_rate": 3.5292757773740717e-06,
      "loss": 0.5661,
      "step": 4766
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6237070500470364,
      "learning_rate": 3.5273038737052676e-06,
      "loss": 0.4752,
      "step": 4767
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0817877986738431,
      "learning_rate": 3.525332220821272e-06,
      "loss": 0.7145,
      "step": 4768
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7802200775419803,
      "learning_rate": 3.523360819057836e-06,
      "loss": 0.5728,
      "step": 4769
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7168916473892792,
      "learning_rate": 3.5213896687506686e-06,
      "loss": 0.5682,
      "step": 4770
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7133472910923415,
      "learning_rate": 3.5194187702354377e-06,
      "loss": 0.5635,
      "step": 4771
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7420869093246777,
      "learning_rate": 3.517448123847764e-06,
      "loss": 0.5375,
      "step": 4772
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7052125350552413,
      "learning_rate": 3.5154777299232312e-06,
      "loss": 0.5782,
      "step": 4773
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7947565903628313,
      "learning_rate": 3.5135075887973747e-06,
      "loss": 0.603,
      "step": 4774
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6601721262691077,
      "learning_rate": 3.51153770080569e-06,
      "loss": 0.5774,
      "step": 4775
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7798491189633278,
      "learning_rate": 3.509568066283628e-06,
      "loss": 0.5489,
      "step": 4776
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8049342102593532,
      "learning_rate": 3.5075986855665955e-06,
      "loss": 0.6427,
      "step": 4777
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6480478624632887,
      "learning_rate": 3.5056295589899615e-06,
      "loss": 0.513,
      "step": 4778
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0040250923365985,
      "learning_rate": 3.5036606868890444e-06,
      "loss": 0.6429,
      "step": 4779
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7627620983937304,
      "learning_rate": 3.501692069599123e-06,
      "loss": 0.5413,
      "step": 4780
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9708481647245439,
      "learning_rate": 3.4997237074554334e-06,
      "loss": 0.6539,
      "step": 4781
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8584156691406621,
      "learning_rate": 3.4977556007931672e-06,
      "loss": 0.557,
      "step": 4782
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7553294821132691,
      "learning_rate": 3.495787749947471e-06,
      "loss": 0.5992,
      "step": 4783
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7935100057445086,
      "learning_rate": 3.493820155253451e-06,
      "loss": 0.5411,
      "step": 4784
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7093719651309593,
      "learning_rate": 3.4918528170461673e-06,
      "loss": 0.553,
      "step": 4785
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6253463183542285,
      "learning_rate": 3.489885735660637e-06,
      "loss": 0.5021,
      "step": 4786
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6509738502962583,
      "learning_rate": 3.4879189114318337e-06,
      "loss": 0.4994,
      "step": 4787
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7038564590150723,
      "learning_rate": 3.4859523446946865e-06,
      "loss": 0.494,
      "step": 4788
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6423683584764396,
      "learning_rate": 3.4839860357840826e-06,
      "loss": 0.5628,
      "step": 4789
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7389609376070936,
      "learning_rate": 3.482019985034861e-06,
      "loss": 0.5342,
      "step": 4790
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7882749867642856,
      "learning_rate": 3.480054192781823e-06,
      "loss": 0.5598,
      "step": 4791
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7938943532587981,
      "learning_rate": 3.478088659359722e-06,
      "loss": 0.5907,
      "step": 4792
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8265812136373645,
      "learning_rate": 3.4761233851032674e-06,
      "loss": 0.6368,
      "step": 4793
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7894155094085734,
      "learning_rate": 3.474158370347124e-06,
      "loss": 0.6046,
      "step": 4794
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7746740583737033,
      "learning_rate": 3.472193615425914e-06,
      "loss": 0.5565,
      "step": 4795
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.945757961316094,
      "learning_rate": 3.4702291206742145e-06,
      "loss": 0.6428,
      "step": 4796
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7672917925716772,
      "learning_rate": 3.468264886426559e-06,
      "loss": 0.5558,
      "step": 4797
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9071892592053592,
      "learning_rate": 3.466300913017436e-06,
      "loss": 0.6626,
      "step": 4798
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8473330708568858,
      "learning_rate": 3.464337200781289e-06,
      "loss": 0.5902,
      "step": 4799
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8493850563212643,
      "learning_rate": 3.462373750052519e-06,
      "loss": 0.616,
      "step": 4800
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6763096686896363,
      "learning_rate": 3.46041056116548e-06,
      "loss": 0.5392,
      "step": 4801
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6518321999835311,
      "learning_rate": 3.4584476344544836e-06,
      "loss": 0.5771,
      "step": 4802
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9043667572066351,
      "learning_rate": 3.456484970253794e-06,
      "loss": 0.6055,
      "step": 4803
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7155071234151839,
      "learning_rate": 3.4545225688976345e-06,
      "loss": 0.549,
      "step": 4804
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8478511896925388,
      "learning_rate": 3.452560430720181e-06,
      "loss": 0.5928,
      "step": 4805
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9009720535558737,
      "learning_rate": 3.4505985560555657e-06,
      "loss": 0.6368,
      "step": 4806
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9320451120311418,
      "learning_rate": 3.448636945237874e-06,
      "loss": 0.5805,
      "step": 4807
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7005472627380441,
      "learning_rate": 3.4466755986011484e-06,
      "loss": 0.5557,
      "step": 4808
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8033978323764367,
      "learning_rate": 3.444714516479386e-06,
      "loss": 0.6297,
      "step": 4809
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.206176008969584,
      "learning_rate": 3.442753699206539e-06,
      "loss": 0.6441,
      "step": 4810
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7966491427373297,
      "learning_rate": 3.440793147116514e-06,
      "loss": 0.5994,
      "step": 4811
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7596099901560068,
      "learning_rate": 3.438832860543172e-06,
      "loss": 0.5521,
      "step": 4812
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0325413646549357,
      "learning_rate": 3.4368728398203297e-06,
      "loss": 0.6488,
      "step": 4813
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7157274817519422,
      "learning_rate": 3.434913085281758e-06,
      "loss": 0.5207,
      "step": 4814
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7016189231796639,
      "learning_rate": 3.432953597261183e-06,
      "loss": 0.5589,
      "step": 4815
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1038534068692278,
      "learning_rate": 3.4309943760922844e-06,
      "loss": 0.659,
      "step": 4816
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7399367983175158,
      "learning_rate": 3.4290354221086985e-06,
      "loss": 0.5412,
      "step": 4817
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7766395142912038,
      "learning_rate": 3.4270767356440137e-06,
      "loss": 0.5371,
      "step": 4818
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8171759037871216,
      "learning_rate": 3.4251183170317746e-06,
      "loss": 0.648,
      "step": 4819
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.765875795243998,
      "learning_rate": 3.4231601666054793e-06,
      "loss": 0.521,
      "step": 4820
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7135420020722786,
      "learning_rate": 3.4212022846985805e-06,
      "loss": 0.5164,
      "step": 4821
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7769823514206529,
      "learning_rate": 3.419244671644484e-06,
      "loss": 0.5462,
      "step": 4822
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6670391683974994,
      "learning_rate": 3.4172873277765515e-06,
      "loss": 0.5995,
      "step": 4823
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9756655126366315,
      "learning_rate": 3.4153302534280994e-06,
      "loss": 0.6389,
      "step": 4824
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7274549141871021,
      "learning_rate": 3.4133734489323957e-06,
      "loss": 0.5256,
      "step": 4825
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6844493289585716,
      "learning_rate": 3.411416914622665e-06,
      "loss": 0.5266,
      "step": 4826
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7214241116810347,
      "learning_rate": 3.4094606508320827e-06,
      "loss": 0.5598,
      "step": 4827
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6434347799421288,
      "learning_rate": 3.4075046578937826e-06,
      "loss": 0.4939,
      "step": 4828
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7057183476839789,
      "learning_rate": 3.405548936140847e-06,
      "loss": 0.5691,
      "step": 4829
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7440769219138041,
      "learning_rate": 3.403593485906318e-06,
      "loss": 0.5462,
      "step": 4830
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8069058421927559,
      "learning_rate": 3.401638307523187e-06,
      "loss": 0.6443,
      "step": 4831
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8133869617537681,
      "learning_rate": 3.3996834013244006e-06,
      "loss": 0.5978,
      "step": 4832
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7080877040370535,
      "learning_rate": 3.3977287676428594e-06,
      "loss": 0.5852,
      "step": 4833
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7470867081227013,
      "learning_rate": 3.3957744068114168e-06,
      "loss": 0.523,
      "step": 4834
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8536832094165678,
      "learning_rate": 3.3938203191628795e-06,
      "loss": 0.6343,
      "step": 4835
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7998581367089739,
      "learning_rate": 3.391866505030009e-06,
      "loss": 0.5595,
      "step": 4836
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0619694756094007,
      "learning_rate": 3.3899129647455194e-06,
      "loss": 0.6412,
      "step": 4837
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0017191695412018,
      "learning_rate": 3.3879596986420782e-06,
      "loss": 0.6416,
      "step": 4838
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8679015960493628,
      "learning_rate": 3.3860067070523057e-06,
      "loss": 0.6365,
      "step": 4839
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7598776484629107,
      "learning_rate": 3.3840539903087766e-06,
      "loss": 0.5379,
      "step": 4840
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8279308677174302,
      "learning_rate": 3.3821015487440184e-06,
      "loss": 0.6146,
      "step": 4841
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6849055828479812,
      "learning_rate": 3.380149382690511e-06,
      "loss": 0.5576,
      "step": 4842
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7739495777487878,
      "learning_rate": 3.378197492480688e-06,
      "loss": 0.6603,
      "step": 4843
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8283635070337946,
      "learning_rate": 3.3762458784469367e-06,
      "loss": 0.5969,
      "step": 4844
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7790218080956149,
      "learning_rate": 3.374294540921596e-06,
      "loss": 0.62,
      "step": 4845
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6751833785401478,
      "learning_rate": 3.372343480236959e-06,
      "loss": 0.5433,
      "step": 4846
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7016284353913876,
      "learning_rate": 3.37039269672527e-06,
      "loss": 0.5176,
      "step": 4847
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8392908861156689,
      "learning_rate": 3.368442190718728e-06,
      "loss": 0.6295,
      "step": 4848
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6836045134325753,
      "learning_rate": 3.366491962549483e-06,
      "loss": 0.5127,
      "step": 4849
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6597338883743055,
      "learning_rate": 3.3645420125496398e-06,
      "loss": 0.5217,
      "step": 4850
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.813766879527436,
      "learning_rate": 3.362592341051253e-06,
      "loss": 0.5392,
      "step": 4851
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8460313155741838,
      "learning_rate": 3.360642948386332e-06,
      "loss": 0.6343,
      "step": 4852
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7923539545087678,
      "learning_rate": 3.3586938348868393e-06,
      "loss": 0.5164,
      "step": 4853
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.702408947430897,
      "learning_rate": 3.356745000884687e-06,
      "loss": 0.5502,
      "step": 4854
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6612461276502262,
      "learning_rate": 3.354796446711739e-06,
      "loss": 0.5071,
      "step": 4855
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8038853273832788,
      "learning_rate": 3.352848172699819e-06,
      "loss": 0.5776,
      "step": 4856
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.042661789026824,
      "learning_rate": 3.3509001791806957e-06,
      "loss": 0.5893,
      "step": 4857
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9172990715480969,
      "learning_rate": 3.3489524664860912e-06,
      "loss": 0.6139,
      "step": 4858
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7188960921970509,
      "learning_rate": 3.3470050349476814e-06,
      "loss": 0.6237,
      "step": 4859
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7113900775815523,
      "learning_rate": 3.345057884897093e-06,
      "loss": 0.5448,
      "step": 4860
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8496930468829204,
      "learning_rate": 3.3431110166659055e-06,
      "loss": 0.6478,
      "step": 4861
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8532105449038324,
      "learning_rate": 3.3411644305856507e-06,
      "loss": 0.6977,
      "step": 4862
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7631110646799267,
      "learning_rate": 3.339218126987812e-06,
      "loss": 0.567,
      "step": 4863
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6766010557626443,
      "learning_rate": 3.337272106203825e-06,
      "loss": 0.5492,
      "step": 4864
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8998840420935064,
      "learning_rate": 3.335326368565075e-06,
      "loss": 0.627,
      "step": 4865
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.858024881401816,
      "learning_rate": 3.333380914402903e-06,
      "loss": 0.5879,
      "step": 4866
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7614518312974131,
      "learning_rate": 3.3314357440485977e-06,
      "loss": 0.55,
      "step": 4867
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8000254000280547,
      "learning_rate": 3.3294908578334027e-06,
      "loss": 0.6017,
      "step": 4868
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6982594416499359,
      "learning_rate": 3.327546256088512e-06,
      "loss": 0.5669,
      "step": 4869
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6725220476244151,
      "learning_rate": 3.3256019391450696e-06,
      "loss": 0.5649,
      "step": 4870
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.989083363135568,
      "learning_rate": 3.3236579073341735e-06,
      "loss": 0.6282,
      "step": 4871
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6353686708705769,
      "learning_rate": 3.3217141609868713e-06,
      "loss": 0.5435,
      "step": 4872
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8655108665798321,
      "learning_rate": 3.319770700434164e-06,
      "loss": 0.6262,
      "step": 4873
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8672443351181393,
      "learning_rate": 3.3178275260070002e-06,
      "loss": 0.5959,
      "step": 4874
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7880321689007912,
      "learning_rate": 3.315884638036285e-06,
      "loss": 0.591,
      "step": 4875
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.865815556073217,
      "learning_rate": 3.3139420368528713e-06,
      "loss": 0.612,
      "step": 4876
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8418796330647859,
      "learning_rate": 3.3119997227875635e-06,
      "loss": 0.6577,
      "step": 4877
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9258534237585381,
      "learning_rate": 3.3100576961711173e-06,
      "loss": 0.6396,
      "step": 4878
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7562459828733519,
      "learning_rate": 3.3081159573342404e-06,
      "loss": 0.5109,
      "step": 4879
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.17661149627926,
      "learning_rate": 3.306174506607589e-06,
      "loss": 0.6467,
      "step": 4880
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6524699833054226,
      "learning_rate": 3.3042333443217733e-06,
      "loss": 0.5507,
      "step": 4881
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8411939304649696,
      "learning_rate": 3.302292470807353e-06,
      "loss": 0.6535,
      "step": 4882
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9588206798839688,
      "learning_rate": 3.300351886394838e-06,
      "loss": 0.5941,
      "step": 4883
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6656219379954181,
      "learning_rate": 3.2984115914146902e-06,
      "loss": 0.5588,
      "step": 4884
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7143392009954862,
      "learning_rate": 3.296471586197321e-06,
      "loss": 0.5988,
      "step": 4885
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.892175453269238,
      "learning_rate": 3.294531871073094e-06,
      "loss": 0.6374,
      "step": 4886
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9017819338410041,
      "learning_rate": 3.2925924463723213e-06,
      "loss": 0.6911,
      "step": 4887
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8813805627169389,
      "learning_rate": 3.290653312425267e-06,
      "loss": 0.6316,
      "step": 4888
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7540193800724198,
      "learning_rate": 3.2887144695621464e-06,
      "loss": 0.5318,
      "step": 4889
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8585230468011072,
      "learning_rate": 3.286775918113123e-06,
      "loss": 0.6621,
      "step": 4890
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8573584847627042,
      "learning_rate": 3.2848376584083134e-06,
      "loss": 0.606,
      "step": 4891
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7666590412801827,
      "learning_rate": 3.282899690777782e-06,
      "loss": 0.5697,
      "step": 4892
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6413491102721429,
      "learning_rate": 3.280962015551544e-06,
      "loss": 0.5733,
      "step": 4893
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6642561091882658,
      "learning_rate": 3.279024633059567e-06,
      "loss": 0.5173,
      "step": 4894
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6479034186218848,
      "learning_rate": 3.277087543631765e-06,
      "loss": 0.4998,
      "step": 4895
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6952032316172522,
      "learning_rate": 3.275150747598006e-06,
      "loss": 0.5558,
      "step": 4896
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7282713974997099,
      "learning_rate": 3.273214245288105e-06,
      "loss": 0.5682,
      "step": 4897
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6436940871354166,
      "learning_rate": 3.2712780370318287e-06,
      "loss": 0.5269,
      "step": 4898
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0035014834444884,
      "learning_rate": 3.269342123158892e-06,
      "loss": 0.6074,
      "step": 4899
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7674967010196748,
      "learning_rate": 3.2674065039989634e-06,
      "loss": 0.5995,
      "step": 4900
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9119643740731648,
      "learning_rate": 3.265471179881655e-06,
      "loss": 0.6521,
      "step": 4901
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.673955547997138,
      "learning_rate": 3.263536151136536e-06,
      "loss": 0.547,
      "step": 4902
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8012164055101305,
      "learning_rate": 3.26160141809312e-06,
      "loss": 0.575,
      "step": 4903
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6713412322603758,
      "learning_rate": 3.259666981080871e-06,
      "loss": 0.53,
      "step": 4904
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7579530035280362,
      "learning_rate": 3.257732840429206e-06,
      "loss": 0.579,
      "step": 4905
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9239961923841928,
      "learning_rate": 3.2557989964674864e-06,
      "loss": 0.6146,
      "step": 4906
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.80694798824872,
      "learning_rate": 3.2538654495250277e-06,
      "loss": 0.6353,
      "step": 4907
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7222510736191212,
      "learning_rate": 3.251932199931091e-06,
      "loss": 0.5571,
      "step": 4908
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7502950042134441,
      "learning_rate": 3.24999924801489e-06,
      "loss": 0.5256,
      "step": 4909
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8608629489315914,
      "learning_rate": 3.2480665941055856e-06,
      "loss": 0.6357,
      "step": 4910
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7418677042212336,
      "learning_rate": 3.2461342385322882e-06,
      "loss": 0.5605,
      "step": 4911
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7832289109113795,
      "learning_rate": 3.2442021816240587e-06,
      "loss": 0.5425,
      "step": 4912
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8632173099569125,
      "learning_rate": 3.2422704237099058e-06,
      "loss": 0.6517,
      "step": 4913
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.835718632633242,
      "learning_rate": 3.2403389651187865e-06,
      "loss": 0.6461,
      "step": 4914
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6475646693020294,
      "learning_rate": 3.2384078061796107e-06,
      "loss": 0.5699,
      "step": 4915
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6756160597692056,
      "learning_rate": 3.236476947221234e-06,
      "loss": 0.5497,
      "step": 4916
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7360523277398411,
      "learning_rate": 3.2345463885724603e-06,
      "loss": 0.5519,
      "step": 4917
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7857838372051005,
      "learning_rate": 3.2326161305620445e-06,
      "loss": 0.5953,
      "step": 4918
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0867414981901475,
      "learning_rate": 3.230686173518689e-06,
      "loss": 0.5561,
      "step": 4919
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8334901216951245,
      "learning_rate": 3.2287565177710457e-06,
      "loss": 0.6191,
      "step": 4920
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8252590404172756,
      "learning_rate": 3.226827163647715e-06,
      "loss": 0.54,
      "step": 4921
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6342083779239805,
      "learning_rate": 3.224898111477245e-06,
      "loss": 0.4946,
      "step": 4922
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8927706406972433,
      "learning_rate": 3.222969361588134e-06,
      "loss": 0.6247,
      "step": 4923
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8496782475841955,
      "learning_rate": 3.2210409143088283e-06,
      "loss": 0.5323,
      "step": 4924
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7886095191936603,
      "learning_rate": 3.219112769967721e-06,
      "loss": 0.5769,
      "step": 4925
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7990749223210991,
      "learning_rate": 3.2171849288931563e-06,
      "loss": 0.6367,
      "step": 4926
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9538821994478781,
      "learning_rate": 3.2152573914134233e-06,
      "loss": 0.6341,
      "step": 4927
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6930772956881712,
      "learning_rate": 3.2133301578567647e-06,
      "loss": 0.5206,
      "step": 4928
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7413004598599584,
      "learning_rate": 3.2114032285513673e-06,
      "loss": 0.5647,
      "step": 4929
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6464996730049951,
      "learning_rate": 3.209476603825366e-06,
      "loss": 0.5477,
      "step": 4930
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7615829799789813,
      "learning_rate": 3.207550284006846e-06,
      "loss": 0.5265,
      "step": 4931
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8651244088782334,
      "learning_rate": 3.205624269423838e-06,
      "loss": 0.6537,
      "step": 4932
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8193347925799067,
      "learning_rate": 3.2036985604043246e-06,
      "loss": 0.5612,
      "step": 4933
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8387643501603282,
      "learning_rate": 3.201773157276232e-06,
      "loss": 0.5935,
      "step": 4934
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8157075502071942,
      "learning_rate": 3.1998480603674367e-06,
      "loss": 0.562,
      "step": 4935
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8148379058600491,
      "learning_rate": 3.1979232700057635e-06,
      "loss": 0.7118,
      "step": 4936
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6905417656518047,
      "learning_rate": 3.195998786518983e-06,
      "loss": 0.5101,
      "step": 4937
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6652861432019276,
      "learning_rate": 3.1940746102348154e-06,
      "loss": 0.575,
      "step": 4938
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.723374555182776,
      "learning_rate": 3.192150741480927e-06,
      "loss": 0.5252,
      "step": 4939
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8947954162078751,
      "learning_rate": 3.1902271805849333e-06,
      "loss": 0.6618,
      "step": 4940
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6768322076991062,
      "learning_rate": 3.188303927874397e-06,
      "loss": 0.5688,
      "step": 4941
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9061095380882681,
      "learning_rate": 3.186380983676828e-06,
      "loss": 0.6529,
      "step": 4942
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8159919985696781,
      "learning_rate": 3.1844583483196832e-06,
      "loss": 0.6602,
      "step": 4943
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8603149980940482,
      "learning_rate": 3.182536022130368e-06,
      "loss": 0.6139,
      "step": 4944
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7509759478324181,
      "learning_rate": 3.1806140054362334e-06,
      "loss": 0.5302,
      "step": 4945
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.665792640991831,
      "learning_rate": 3.1786922985645796e-06,
      "loss": 0.547,
      "step": 4946
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7434816450208102,
      "learning_rate": 3.176770901842653e-06,
      "loss": 0.5891,
      "step": 4947
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7910232360326743,
      "learning_rate": 3.1748498155976483e-06,
      "loss": 0.5843,
      "step": 4948
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7675773313527777,
      "learning_rate": 3.172929040156705e-06,
      "loss": 0.6172,
      "step": 4949
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.678461169401386,
      "learning_rate": 3.1710085758469115e-06,
      "loss": 0.5332,
      "step": 4950
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7168108556638194,
      "learning_rate": 3.169088422995304e-06,
      "loss": 0.6055,
      "step": 4951
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8828118134190684,
      "learning_rate": 3.1671685819288633e-06,
      "loss": 0.7033,
      "step": 4952
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9027804332632563,
      "learning_rate": 3.1652490529745184e-06,
      "loss": 0.6137,
      "step": 4953
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7898009879510415,
      "learning_rate": 3.1633298364591457e-06,
      "loss": 0.6138,
      "step": 4954
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8787990749261377,
      "learning_rate": 3.1614109327095677e-06,
      "loss": 0.6449,
      "step": 4955
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8364346039271051,
      "learning_rate": 3.159492342052554e-06,
      "loss": 0.6083,
      "step": 4956
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7947598014173956,
      "learning_rate": 3.157574064814819e-06,
      "loss": 0.5677,
      "step": 4957
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7103119835086947,
      "learning_rate": 3.155656101323028e-06,
      "loss": 0.5601,
      "step": 4958
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9016593715419259,
      "learning_rate": 3.1537384519037873e-06,
      "loss": 0.6377,
      "step": 4959
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7496533370530231,
      "learning_rate": 3.1518211168836544e-06,
      "loss": 0.5295,
      "step": 4960
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8323363027335362,
      "learning_rate": 3.1499040965891315e-06,
      "loss": 0.6071,
      "step": 4961
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8202449582037884,
      "learning_rate": 3.1479873913466676e-06,
      "loss": 0.6298,
      "step": 4962
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.249544773907663,
      "learning_rate": 3.1460710014826555e-06,
      "loss": 0.6427,
      "step": 4963
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8824146843396334,
      "learning_rate": 3.1441549273234382e-06,
      "loss": 0.6531,
      "step": 4964
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7998270664811491,
      "learning_rate": 3.1422391691953036e-06,
      "loss": 0.5805,
      "step": 4965
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8811540580169402,
      "learning_rate": 3.1403237274244834e-06,
      "loss": 0.6774,
      "step": 4966
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0553376679768194,
      "learning_rate": 3.1384086023371597e-06,
      "loss": 0.6827,
      "step": 4967
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8419507838284247,
      "learning_rate": 3.136493794259456e-06,
      "loss": 0.6646,
      "step": 4968
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8423246887675496,
      "learning_rate": 3.1345793035174452e-06,
      "loss": 0.6069,
      "step": 4969
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8934151390438726,
      "learning_rate": 3.1326651304371457e-06,
      "loss": 0.5994,
      "step": 4970
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7114782230046063,
      "learning_rate": 3.1307512753445206e-06,
      "loss": 0.4949,
      "step": 4971
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7161345759540172,
      "learning_rate": 3.128837738565479e-06,
      "loss": 0.5715,
      "step": 4972
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8830345859693479,
      "learning_rate": 3.126924520425877e-06,
      "loss": 0.6008,
      "step": 4973
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8658218617705198,
      "learning_rate": 3.1250116212515157e-06,
      "loss": 0.5512,
      "step": 4974
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6528126243296772,
      "learning_rate": 3.123099041368143e-06,
      "loss": 0.5359,
      "step": 4975
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7308498210634851,
      "learning_rate": 3.1211867811014492e-06,
      "loss": 0.5417,
      "step": 4976
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6643036815370214,
      "learning_rate": 3.1192748407770733e-06,
      "loss": 0.5519,
      "step": 4977
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6806872322796154,
      "learning_rate": 3.117363220720599e-06,
      "loss": 0.5,
      "step": 4978
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4402475571102153,
      "learning_rate": 3.1154519212575552e-06,
      "loss": 0.6614,
      "step": 4979
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7795024490442468,
      "learning_rate": 3.1135409427134156e-06,
      "loss": 0.5519,
      "step": 4980
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7548319278432761,
      "learning_rate": 3.1116302854136003e-06,
      "loss": 0.5551,
      "step": 4981
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8129771989783109,
      "learning_rate": 3.109719949683475e-06,
      "loss": 0.6457,
      "step": 4982
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7920356526417051,
      "learning_rate": 3.1078099358483495e-06,
      "loss": 0.5976,
      "step": 4983
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7644943534496127,
      "learning_rate": 3.105900244233479e-06,
      "loss": 0.5793,
      "step": 4984
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8842847966267777,
      "learning_rate": 3.1039908751640634e-06,
      "loss": 0.6217,
      "step": 4985
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6164995935996114,
      "learning_rate": 3.1020818289652504e-06,
      "loss": 0.4674,
      "step": 4986
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6714869291699325,
      "learning_rate": 3.10017310596213e-06,
      "loss": 0.5495,
      "step": 4987
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6814550791435244,
      "learning_rate": 3.0982647064797367e-06,
      "loss": 0.5597,
      "step": 4988
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7395183397217501,
      "learning_rate": 3.0963566308430516e-06,
      "loss": 0.6085,
      "step": 4989
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8114143907821454,
      "learning_rate": 3.0944488793770007e-06,
      "loss": 0.5521,
      "step": 4990
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8431355899480643,
      "learning_rate": 3.0925414524064535e-06,
      "loss": 0.5513,
      "step": 4991
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8476631471221191,
      "learning_rate": 3.0906343502562253e-06,
      "loss": 0.6426,
      "step": 4992
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7735019108505419,
      "learning_rate": 3.088727573251076e-06,
      "loss": 0.6104,
      "step": 4993
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7070841213477257,
      "learning_rate": 3.0868211217157086e-06,
      "loss": 0.5235,
      "step": 4994
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7606065290230307,
      "learning_rate": 3.0849149959747737e-06,
      "loss": 0.5724,
      "step": 4995
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6398751491300726,
      "learning_rate": 3.083009196352863e-06,
      "loss": 0.5391,
      "step": 4996
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8130704705420126,
      "learning_rate": 3.0811037231745156e-06,
      "loss": 0.5633,
      "step": 4997
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0557818325543082,
      "learning_rate": 3.0791985767642135e-06,
      "loss": 0.6917,
      "step": 4998
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.745421372588127,
      "learning_rate": 3.077293757446381e-06,
      "loss": 0.5791,
      "step": 4999
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7845346987460242,
      "learning_rate": 3.0753892655453932e-06,
      "loss": 0.5146,
      "step": 5000
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8042884766722785,
      "learning_rate": 3.0734851013855626e-06,
      "loss": 0.5469,
      "step": 5001
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.066151029808255,
      "learning_rate": 3.0715812652911486e-06,
      "loss": 0.6093,
      "step": 5002
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7760575017502817,
      "learning_rate": 3.0696777575863557e-06,
      "loss": 0.5855,
      "step": 5003
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6491191009117421,
      "learning_rate": 3.06777457859533e-06,
      "loss": 0.5024,
      "step": 5004
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8394753303812418,
      "learning_rate": 3.065871728642164e-06,
      "loss": 0.6361,
      "step": 5005
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9293200388729286,
      "learning_rate": 3.063969208050893e-06,
      "loss": 0.6418,
      "step": 5006
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8738802466288204,
      "learning_rate": 3.062067017145497e-06,
      "loss": 0.5916,
      "step": 5007
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7516143034017935,
      "learning_rate": 3.0601651562498978e-06,
      "loss": 0.5158,
      "step": 5008
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7991567658735562,
      "learning_rate": 3.058263625687964e-06,
      "loss": 0.6048,
      "step": 5009
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8175059542807093,
      "learning_rate": 3.0563624257835055e-06,
      "loss": 0.5799,
      "step": 5010
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7887745475538136,
      "learning_rate": 3.0544615568602776e-06,
      "loss": 0.6553,
      "step": 5011
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6725125103204541,
      "learning_rate": 3.0525610192419774e-06,
      "loss": 0.5309,
      "step": 5012
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8771590516858645,
      "learning_rate": 3.0506608132522476e-06,
      "loss": 0.625,
      "step": 5013
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9509894144134095,
      "learning_rate": 3.048760939214673e-06,
      "loss": 0.5932,
      "step": 5014
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1715137299169884,
      "learning_rate": 3.046861397452784e-06,
      "loss": 0.6358,
      "step": 5015
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8515079900548325,
      "learning_rate": 3.0449621882900505e-06,
      "loss": 0.6258,
      "step": 5016
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7106745518456697,
      "learning_rate": 3.0430633120498886e-06,
      "loss": 0.5917,
      "step": 5017
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9922685448593384,
      "learning_rate": 3.041164769055658e-06,
      "loss": 0.6985,
      "step": 5018
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8122359582056352,
      "learning_rate": 3.03926655963066e-06,
      "loss": 0.5868,
      "step": 5019
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7345459321467698,
      "learning_rate": 3.0373686840981396e-06,
      "loss": 0.5774,
      "step": 5020
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8952754988396652,
      "learning_rate": 3.035471142781287e-06,
      "loss": 0.6047,
      "step": 5021
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8278630861754526,
      "learning_rate": 3.0335739360032314e-06,
      "loss": 0.601,
      "step": 5022
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6433925822755447,
      "learning_rate": 3.0316770640870487e-06,
      "loss": 0.5083,
      "step": 5023
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6709492167969169,
      "learning_rate": 3.0297805273557563e-06,
      "loss": 0.5209,
      "step": 5024
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6971545777564855,
      "learning_rate": 3.0278843261323133e-06,
      "loss": 0.5183,
      "step": 5025
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0938457062993858,
      "learning_rate": 3.025988460739625e-06,
      "loss": 0.6174,
      "step": 5026
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8599379391062648,
      "learning_rate": 3.0240929315005375e-06,
      "loss": 0.6501,
      "step": 5027
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.690843161365665,
      "learning_rate": 3.0221977387378386e-06,
      "loss": 0.5162,
      "step": 5028
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0328366809191933,
      "learning_rate": 3.02030288277426e-06,
      "loss": 0.6362,
      "step": 5029
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7278862276601206,
      "learning_rate": 3.018408363932476e-06,
      "loss": 0.5708,
      "step": 5030
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7966026810228043,
      "learning_rate": 3.016514182535104e-06,
      "loss": 0.5437,
      "step": 5031
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8116526771916039,
      "learning_rate": 3.0146203389047025e-06,
      "loss": 0.6232,
      "step": 5032
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6372993699838145,
      "learning_rate": 3.0127268333637738e-06,
      "loss": 0.5224,
      "step": 5033
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.749541015071615,
      "learning_rate": 3.010833666234762e-06,
      "loss": 0.6246,
      "step": 5034
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7418348963897718,
      "learning_rate": 3.008940837840054e-06,
      "loss": 0.6061,
      "step": 5035
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.657103502277192,
      "learning_rate": 3.0070483485019786e-06,
      "loss": 0.5691,
      "step": 5036
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6423930074719663,
      "learning_rate": 3.0051561985428073e-06,
      "loss": 0.5671,
      "step": 5037
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7647084416249336,
      "learning_rate": 3.0032643882847523e-06,
      "loss": 0.5772,
      "step": 5038
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8130263159828129,
      "learning_rate": 3.0013729180499707e-06,
      "loss": 0.6369,
      "step": 5039
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7939018654042358,
      "learning_rate": 2.9994817881605596e-06,
      "loss": 0.5443,
      "step": 5040
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7839391237657964,
      "learning_rate": 2.9975909989385587e-06,
      "loss": 0.5705,
      "step": 5041
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6741102622215539,
      "learning_rate": 2.99570055070595e-06,
      "loss": 0.528,
      "step": 5042
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8343447808665908,
      "learning_rate": 2.993810443784657e-06,
      "loss": 0.5974,
      "step": 5043
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7152191321668764,
      "learning_rate": 2.991920678496545e-06,
      "loss": 0.5714,
      "step": 5044
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6686741233385336,
      "learning_rate": 2.9900312551634216e-06,
      "loss": 0.5167,
      "step": 5045
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7503890990194079,
      "learning_rate": 2.9881421741070354e-06,
      "loss": 0.5904,
      "step": 5046
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8002828364337375,
      "learning_rate": 2.986253435649078e-06,
      "loss": 0.5152,
      "step": 5047
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8562790460800754,
      "learning_rate": 2.9843650401111807e-06,
      "loss": 0.6189,
      "step": 5048
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7761964296400664,
      "learning_rate": 2.9824769878149184e-06,
      "loss": 0.5531,
      "step": 5049
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6825114718682199,
      "learning_rate": 2.9805892790818074e-06,
      "loss": 0.5801,
      "step": 5050
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8428221778968118,
      "learning_rate": 2.978701914233303e-06,
      "loss": 0.6601,
      "step": 5051
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9128095881527932,
      "learning_rate": 2.9768148935908052e-06,
      "loss": 0.575,
      "step": 5052
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6162211458638216,
      "learning_rate": 2.9749282174756546e-06,
      "loss": 0.5089,
      "step": 5053
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0453328813565408,
      "learning_rate": 2.97304188620913e-06,
      "loss": 0.665,
      "step": 5054
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9949130244666881,
      "learning_rate": 2.971155900112457e-06,
      "loss": 0.613,
      "step": 5055
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8479103619003527,
      "learning_rate": 2.9692702595067975e-06,
      "loss": 0.5922,
      "step": 5056
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8064943202314837,
      "learning_rate": 2.967384964713256e-06,
      "loss": 0.5505,
      "step": 5057
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7629521567300589,
      "learning_rate": 2.9655000160528784e-06,
      "loss": 0.6124,
      "step": 5058
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.787166679951818,
      "learning_rate": 2.9636154138466545e-06,
      "loss": 0.5565,
      "step": 5059
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.627286840427744,
      "learning_rate": 2.9617311584155104e-06,
      "loss": 0.5607,
      "step": 5060
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7728065791669109,
      "learning_rate": 2.9598472500803153e-06,
      "loss": 0.5735,
      "step": 5061
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7847844195127908,
      "learning_rate": 2.95796368916188e-06,
      "loss": 0.5311,
      "step": 5062
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7905854136442236,
      "learning_rate": 2.9560804759809547e-06,
      "loss": 0.6023,
      "step": 5063
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.767859920261755,
      "learning_rate": 2.9541976108582303e-06,
      "loss": 0.5773,
      "step": 5064
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8606071803912296,
      "learning_rate": 2.952315094114341e-06,
      "loss": 0.5858,
      "step": 5065
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7999038678294265,
      "learning_rate": 2.9504329260698573e-06,
      "loss": 0.5702,
      "step": 5066
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8830145480605472,
      "learning_rate": 2.9485511070452945e-06,
      "loss": 0.611,
      "step": 5067
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8365157062235715,
      "learning_rate": 2.9466696373611058e-06,
      "loss": 0.64,
      "step": 5068
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6936025799379858,
      "learning_rate": 2.9447885173376865e-06,
      "loss": 0.5583,
      "step": 5069
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6403350582941926,
      "learning_rate": 2.9429077472953715e-06,
      "loss": 0.4369,
      "step": 5070
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7912088043822698,
      "learning_rate": 2.9410273275544373e-06,
      "loss": 0.6167,
      "step": 5071
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7151672215093456,
      "learning_rate": 2.939147258435099e-06,
      "loss": 0.5516,
      "step": 5072
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8734463608373647,
      "learning_rate": 2.9372675402575122e-06,
      "loss": 0.5185,
      "step": 5073
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7575540456026442,
      "learning_rate": 2.9353881733417745e-06,
      "loss": 0.5712,
      "step": 5074
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8736440726166469,
      "learning_rate": 2.933509158007921e-06,
      "loss": 0.6262,
      "step": 5075
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7239375898253136,
      "learning_rate": 2.9316304945759298e-06,
      "loss": 0.5117,
      "step": 5076
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7370688651680586,
      "learning_rate": 2.9297521833657177e-06,
      "loss": 0.5242,
      "step": 5077
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.878141668602577,
      "learning_rate": 2.9278742246971408e-06,
      "loss": 0.625,
      "step": 5078
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7011608883277275,
      "learning_rate": 2.9259966188899964e-06,
      "loss": 0.5304,
      "step": 5079
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6966904076748,
      "learning_rate": 2.9241193662640204e-06,
      "loss": 0.5495,
      "step": 5080
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6417231886938327,
      "learning_rate": 2.9222424671388907e-06,
      "loss": 0.511,
      "step": 5081
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.693462911417472,
      "learning_rate": 2.9203659218342224e-06,
      "loss": 0.4913,
      "step": 5082
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7900963828228083,
      "learning_rate": 2.9184897306695714e-06,
      "loss": 0.5753,
      "step": 5083
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8148514879713444,
      "learning_rate": 2.916613893964435e-06,
      "loss": 0.5988,
      "step": 5084
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7124544774085725,
      "learning_rate": 2.9147384120382484e-06,
      "loss": 0.515,
      "step": 5085
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6997178029566059,
      "learning_rate": 2.9128632852103854e-06,
      "loss": 0.573,
      "step": 5086
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1315708795820967,
      "learning_rate": 2.9109885138001614e-06,
      "loss": 0.658,
      "step": 5087
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9176413053769573,
      "learning_rate": 2.909114098126831e-06,
      "loss": 0.6357,
      "step": 5088
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8206812379899895,
      "learning_rate": 2.9072400385095865e-06,
      "loss": 0.5782,
      "step": 5089
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8723181580950474,
      "learning_rate": 2.9053663352675613e-06,
      "loss": 0.6193,
      "step": 5090
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7064501080282872,
      "learning_rate": 2.9034929887198274e-06,
      "loss": 0.5065,
      "step": 5091
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8481158192491489,
      "learning_rate": 2.9016199991853966e-06,
      "loss": 0.6742,
      "step": 5092
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.552527435384719,
      "learning_rate": 2.899747366983219e-06,
      "loss": 0.6098,
      "step": 5093
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6385767356469423,
      "learning_rate": 2.8978750924321843e-06,
      "loss": 0.6009,
      "step": 5094
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7869850340271355,
      "learning_rate": 2.896003175851122e-06,
      "loss": 0.6265,
      "step": 5095
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6742842764932138,
      "learning_rate": 2.8941316175587993e-06,
      "loss": 0.4911,
      "step": 5096
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7678884596916175,
      "learning_rate": 2.8922604178739234e-06,
      "loss": 0.6269,
      "step": 5097
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8435298422656635,
      "learning_rate": 2.8903895771151402e-06,
      "loss": 0.6004,
      "step": 5098
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9192396522506059,
      "learning_rate": 2.8885190956010333e-06,
      "loss": 0.6282,
      "step": 5099
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9091860817951859,
      "learning_rate": 2.8866489736501284e-06,
      "loss": 0.6585,
      "step": 5100
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7386786305836628,
      "learning_rate": 2.884779211580886e-06,
      "loss": 0.5952,
      "step": 5101
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9187346348948497,
      "learning_rate": 2.8829098097117083e-06,
      "loss": 0.6652,
      "step": 5102
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4714466866827551,
      "learning_rate": 2.8810407683609314e-06,
      "loss": 0.6128,
      "step": 5103
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.659600892770533,
      "learning_rate": 2.879172087846839e-06,
      "loss": 0.5762,
      "step": 5104
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8581151843891774,
      "learning_rate": 2.8773037684876454e-06,
      "loss": 0.5678,
      "step": 5105
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8545377593903204,
      "learning_rate": 2.8754358106015057e-06,
      "loss": 0.6054,
      "step": 5106
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6498541467492196,
      "learning_rate": 2.8735682145065143e-06,
      "loss": 0.5229,
      "step": 5107
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6756064156040719,
      "learning_rate": 2.8717009805207026e-06,
      "loss": 0.5367,
      "step": 5108
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6937353110813358,
      "learning_rate": 2.869834108962042e-06,
      "loss": 0.5493,
      "step": 5109
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7869415419065402,
      "learning_rate": 2.8679676001484397e-06,
      "loss": 0.6606,
      "step": 5110
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6309322839766587,
      "learning_rate": 2.8661014543977446e-06,
      "loss": 0.5132,
      "step": 5111
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6402842922667611,
      "learning_rate": 2.864235672027741e-06,
      "loss": 0.5028,
      "step": 5112
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8171540989138597,
      "learning_rate": 2.8623702533561514e-06,
      "loss": 0.6219,
      "step": 5113
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6338841548721944,
      "learning_rate": 2.860505198700639e-06,
      "loss": 0.5244,
      "step": 5114
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8542887720948676,
      "learning_rate": 2.8586405083788016e-06,
      "loss": 0.6742,
      "step": 5115
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7872656446119819,
      "learning_rate": 2.8567761827081777e-06,
      "loss": 0.5628,
      "step": 5116
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6402900019101818,
      "learning_rate": 2.854912222006241e-06,
      "loss": 0.5048,
      "step": 5117
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8761378112092568,
      "learning_rate": 2.8530486265904057e-06,
      "loss": 0.6786,
      "step": 5118
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1071052112602673,
      "learning_rate": 2.8511853967780234e-06,
      "loss": 0.6572,
      "step": 5119
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7029030689748211,
      "learning_rate": 2.849322532886381e-06,
      "loss": 0.5525,
      "step": 5120
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9909517105724016,
      "learning_rate": 2.847460035232706e-06,
      "loss": 0.5844,
      "step": 5121
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9442490804991749,
      "learning_rate": 2.845597904134162e-06,
      "loss": 0.6525,
      "step": 5122
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6494560441160583,
      "learning_rate": 2.8437361399078502e-06,
      "loss": 0.5463,
      "step": 5123
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6197108843004744,
      "learning_rate": 2.8418747428708103e-06,
      "loss": 0.491,
      "step": 5124
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6602929038792663,
      "learning_rate": 2.8400137133400185e-06,
      "loss": 0.5374,
      "step": 5125
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7969911576003924,
      "learning_rate": 2.8381530516323887e-06,
      "loss": 0.6665,
      "step": 5126
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7634927541488229,
      "learning_rate": 2.836292758064772e-06,
      "loss": 0.5918,
      "step": 5127
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8797409090262036,
      "learning_rate": 2.8344328329539573e-06,
      "loss": 0.5925,
      "step": 5128
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.819838885097571,
      "learning_rate": 2.832573276616669e-06,
      "loss": 0.6143,
      "step": 5129
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8976985230923279,
      "learning_rate": 2.8307140893695728e-06,
      "loss": 0.5931,
      "step": 5130
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.887058680958711,
      "learning_rate": 2.8288552715292683e-06,
      "loss": 0.5915,
      "step": 5131
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6930251114140492,
      "learning_rate": 2.8269968234122922e-06,
      "loss": 0.5194,
      "step": 5132
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8679975696069272,
      "learning_rate": 2.8251387453351185e-06,
      "loss": 0.6218,
      "step": 5133
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6601588682863456,
      "learning_rate": 2.8232810376141593e-06,
      "loss": 0.5032,
      "step": 5134
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8106950176466404,
      "learning_rate": 2.821423700565763e-06,
      "loss": 0.6254,
      "step": 5135
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.721201954560906,
      "learning_rate": 2.8195667345062135e-06,
      "loss": 0.525,
      "step": 5136
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6640771490332805,
      "learning_rate": 2.817710139751734e-06,
      "loss": 0.5378,
      "step": 5137
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9350091056659592,
      "learning_rate": 2.8158539166184816e-06,
      "loss": 0.6324,
      "step": 5138
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2512755449795776,
      "learning_rate": 2.813998065422554e-06,
      "loss": 0.6373,
      "step": 5139
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.73420194955367,
      "learning_rate": 2.8121425864799825e-06,
      "loss": 0.5597,
      "step": 5140
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7815903484537784,
      "learning_rate": 2.810287480106735e-06,
      "loss": 0.6386,
      "step": 5141
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7277170027847618,
      "learning_rate": 2.8084327466187174e-06,
      "loss": 0.5702,
      "step": 5142
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9241973326162606,
      "learning_rate": 2.8065783863317715e-06,
      "loss": 0.6356,
      "step": 5143
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7116264821011563,
      "learning_rate": 2.8047243995616757e-06,
      "loss": 0.5791,
      "step": 5144
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7644541838097464,
      "learning_rate": 2.8028707866241443e-06,
      "loss": 0.5706,
      "step": 5145
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8130506833695966,
      "learning_rate": 2.8010175478348282e-06,
      "loss": 0.5468,
      "step": 5146
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6912794421788537,
      "learning_rate": 2.7991646835093155e-06,
      "loss": 0.5142,
      "step": 5147
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7977654033615409,
      "learning_rate": 2.7973121939631285e-06,
      "loss": 0.5409,
      "step": 5148
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8529239898813732,
      "learning_rate": 2.7954600795117276e-06,
      "loss": 0.6337,
      "step": 5149
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.822981421300624,
      "learning_rate": 2.793608340470508e-06,
      "loss": 0.6254,
      "step": 5150
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7289860090207538,
      "learning_rate": 2.7917569771548027e-06,
      "loss": 0.5408,
      "step": 5151
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.856365915073162,
      "learning_rate": 2.7899059898798785e-06,
      "loss": 0.6415,
      "step": 5152
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6612251014781776,
      "learning_rate": 2.78805537896094e-06,
      "loss": 0.5362,
      "step": 5153
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0021774016020024,
      "learning_rate": 2.786205144713126e-06,
      "loss": 0.6744,
      "step": 5154
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.68576567882279,
      "learning_rate": 2.7843552874515135e-06,
      "loss": 0.5536,
      "step": 5155
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9577772696794191,
      "learning_rate": 2.7825058074911106e-06,
      "loss": 0.5977,
      "step": 5156
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6508401860666763,
      "learning_rate": 2.780656705146869e-06,
      "loss": 0.4906,
      "step": 5157
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8000308020498672,
      "learning_rate": 2.778807980733669e-06,
      "loss": 0.6598,
      "step": 5158
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8554737868033759,
      "learning_rate": 2.7769596345663306e-06,
      "loss": 0.6263,
      "step": 5159
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8441062403901539,
      "learning_rate": 2.7751116669596067e-06,
      "loss": 0.6486,
      "step": 5160
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8629860566627963,
      "learning_rate": 2.773264078228185e-06,
      "loss": 0.6245,
      "step": 5161
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7806912403688684,
      "learning_rate": 2.7714168686866947e-06,
      "loss": 0.6154,
      "step": 5162
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8886078858809823,
      "learning_rate": 2.7695700386496947e-06,
      "loss": 0.61,
      "step": 5163
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.007021275010515,
      "learning_rate": 2.7677235884316812e-06,
      "loss": 0.5861,
      "step": 5164
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6877992839372186,
      "learning_rate": 2.7658775183470845e-06,
      "loss": 0.528,
      "step": 5165
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6803157516403499,
      "learning_rate": 2.7640318287102714e-06,
      "loss": 0.53,
      "step": 5166
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6897655848117805,
      "learning_rate": 2.7621865198355437e-06,
      "loss": 0.5049,
      "step": 5167
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9584201558024223,
      "learning_rate": 2.760341592037138e-06,
      "loss": 0.611,
      "step": 5168
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7005281413971542,
      "learning_rate": 2.758497045629227e-06,
      "loss": 0.5854,
      "step": 5169
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8207485264357269,
      "learning_rate": 2.7566528809259175e-06,
      "loss": 0.6173,
      "step": 5170
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7866032594797326,
      "learning_rate": 2.7548090982412507e-06,
      "loss": 0.6176,
      "step": 5171
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8424207618641728,
      "learning_rate": 2.752965697889205e-06,
      "loss": 0.5814,
      "step": 5172
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6853146914536142,
      "learning_rate": 2.751122680183691e-06,
      "loss": 0.5239,
      "step": 5173
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7500586682933541,
      "learning_rate": 2.749280045438556e-06,
      "loss": 0.5538,
      "step": 5174
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6663553236807322,
      "learning_rate": 2.7474377939675813e-06,
      "loss": 0.5339,
      "step": 5175
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7242520750398093,
      "learning_rate": 2.7455959260844833e-06,
      "loss": 0.5136,
      "step": 5176
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8178106857860368,
      "learning_rate": 2.7437544421029132e-06,
      "loss": 0.5971,
      "step": 5177
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.902020897353715,
      "learning_rate": 2.741913342336456e-06,
      "loss": 0.6598,
      "step": 5178
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9053732133687181,
      "learning_rate": 2.740072627098632e-06,
      "loss": 0.6298,
      "step": 5179
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.680032971338818,
      "learning_rate": 2.738232296702896e-06,
      "loss": 0.5297,
      "step": 5180
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8043774415528638,
      "learning_rate": 2.7363923514626368e-06,
      "loss": 0.5773,
      "step": 5181
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6672367262272082,
      "learning_rate": 2.7345527916911785e-06,
      "loss": 0.5455,
      "step": 5182
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7299318295239927,
      "learning_rate": 2.7327136177017777e-06,
      "loss": 0.5322,
      "step": 5183
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7621754528774594,
      "learning_rate": 2.730874829807628e-06,
      "loss": 0.5954,
      "step": 5184
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8544827069777828,
      "learning_rate": 2.729036428321855e-06,
      "loss": 0.6621,
      "step": 5185
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6929800893447793,
      "learning_rate": 2.727198413557519e-06,
      "loss": 0.509,
      "step": 5186
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8062805436810909,
      "learning_rate": 2.7253607858276166e-06,
      "loss": 0.5968,
      "step": 5187
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7393975822423453,
      "learning_rate": 2.7235235454450726e-06,
      "loss": 0.5816,
      "step": 5188
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6804203910252329,
      "learning_rate": 2.7216866927227547e-06,
      "loss": 0.4719,
      "step": 5189
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7466036692471355,
      "learning_rate": 2.7198502279734572e-06,
      "loss": 0.5636,
      "step": 5190
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.755967473355392,
      "learning_rate": 2.7180141515099124e-06,
      "loss": 0.6153,
      "step": 5191
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7249784899300633,
      "learning_rate": 2.716178463644783e-06,
      "loss": 0.5196,
      "step": 5192
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6896874198975248,
      "learning_rate": 2.7143431646906693e-06,
      "loss": 0.5267,
      "step": 5193
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6242336184929248,
      "learning_rate": 2.712508254960103e-06,
      "loss": 0.5061,
      "step": 5194
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8642473176780114,
      "learning_rate": 2.71067373476555e-06,
      "loss": 0.6466,
      "step": 5195
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7998894403126326,
      "learning_rate": 2.7088396044194087e-06,
      "loss": 0.5867,
      "step": 5196
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8542562645589561,
      "learning_rate": 2.7070058642340146e-06,
      "loss": 0.645,
      "step": 5197
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8944803801071172,
      "learning_rate": 2.705172514521634e-06,
      "loss": 0.578,
      "step": 5198
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.65550469517289,
      "learning_rate": 2.7033395555944664e-06,
      "loss": 0.5166,
      "step": 5199
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7627583236404839,
      "learning_rate": 2.7015069877646462e-06,
      "loss": 0.5435,
      "step": 5200
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9197477231974598,
      "learning_rate": 2.6996748113442397e-06,
      "loss": 0.6493,
      "step": 5201
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7925732192573283,
      "learning_rate": 2.697843026645249e-06,
      "loss": 0.4922,
      "step": 5202
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7236804362081727,
      "learning_rate": 2.696011633979607e-06,
      "loss": 0.5287,
      "step": 5203
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8518712947721799,
      "learning_rate": 2.694180633659181e-06,
      "loss": 0.5865,
      "step": 5204
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8176553594601641,
      "learning_rate": 2.692350025995771e-06,
      "loss": 0.5766,
      "step": 5205
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7266282703374228,
      "learning_rate": 2.6905198113011103e-06,
      "loss": 0.5066,
      "step": 5206
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7695846170219701,
      "learning_rate": 2.688689989886867e-06,
      "loss": 0.6189,
      "step": 5207
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7964515495984981,
      "learning_rate": 2.6868605620646383e-06,
      "loss": 0.5802,
      "step": 5208
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8680629696218501,
      "learning_rate": 2.6850315281459583e-06,
      "loss": 0.6504,
      "step": 5209
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7167740949045319,
      "learning_rate": 2.6832028884422913e-06,
      "loss": 0.5239,
      "step": 5210
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7818288681704455,
      "learning_rate": 2.681374643265037e-06,
      "loss": 0.5317,
      "step": 5211
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9651876787100336,
      "learning_rate": 2.6795467929255248e-06,
      "loss": 0.6323,
      "step": 5212
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8719896724909121,
      "learning_rate": 2.6777193377350196e-06,
      "loss": 0.6158,
      "step": 5213
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8655860041581656,
      "learning_rate": 2.6758922780047157e-06,
      "loss": 0.5818,
      "step": 5214
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6499668073058882,
      "learning_rate": 2.6740656140457478e-06,
      "loss": 0.512,
      "step": 5215
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7013168277868185,
      "learning_rate": 2.672239346169173e-06,
      "loss": 0.5509,
      "step": 5216
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.059345314738029,
      "learning_rate": 2.6704134746859877e-06,
      "loss": 0.6556,
      "step": 5217
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9008223241379106,
      "learning_rate": 2.6685879999071184e-06,
      "loss": 0.6521,
      "step": 5218
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6815671362886792,
      "learning_rate": 2.666762922143424e-06,
      "loss": 0.5591,
      "step": 5219
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9576166895333427,
      "learning_rate": 2.6649382417056956e-06,
      "loss": 0.6651,
      "step": 5220
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8380433757412757,
      "learning_rate": 2.6631139589046595e-06,
      "loss": 0.5987,
      "step": 5221
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8832024017823825,
      "learning_rate": 2.6612900740509702e-06,
      "loss": 0.6165,
      "step": 5222
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7457261250272638,
      "learning_rate": 2.659466587455217e-06,
      "loss": 0.6025,
      "step": 5223
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8229459147850618,
      "learning_rate": 2.65764349942792e-06,
      "loss": 0.6468,
      "step": 5224
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7007632226870129,
      "learning_rate": 2.6558208102795322e-06,
      "loss": 0.5745,
      "step": 5225
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7521041410076824,
      "learning_rate": 2.65399852032044e-06,
      "loss": 0.5232,
      "step": 5226
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7332808431620648,
      "learning_rate": 2.6521766298609586e-06,
      "loss": 0.5684,
      "step": 5227
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6820348364164988,
      "learning_rate": 2.650355139211338e-06,
      "loss": 0.5252,
      "step": 5228
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8973651355313759,
      "learning_rate": 2.648534048681759e-06,
      "loss": 0.6312,
      "step": 5229
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9593551190644433,
      "learning_rate": 2.6467133585823334e-06,
      "loss": 0.6532,
      "step": 5230
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8515754673470484,
      "learning_rate": 2.644893069223107e-06,
      "loss": 0.6279,
      "step": 5231
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8483891426338936,
      "learning_rate": 2.643073180914055e-06,
      "loss": 0.5926,
      "step": 5232
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6915501624086446,
      "learning_rate": 2.641253693965087e-06,
      "loss": 0.5239,
      "step": 5233
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.747231630935088,
      "learning_rate": 2.639434608686041e-06,
      "loss": 0.5496,
      "step": 5234
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8164066920757889,
      "learning_rate": 2.6376159253866888e-06,
      "loss": 0.6494,
      "step": 5235
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8764125422257726,
      "learning_rate": 2.635797644376734e-06,
      "loss": 0.6638,
      "step": 5236
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7379309562064749,
      "learning_rate": 2.6339797659658097e-06,
      "loss": 0.5821,
      "step": 5237
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9269195931587623,
      "learning_rate": 2.6321622904634826e-06,
      "loss": 0.6451,
      "step": 5238
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6678045720233075,
      "learning_rate": 2.6303452181792486e-06,
      "loss": 0.5508,
      "step": 5239
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9214877741918989,
      "learning_rate": 2.6285285494225365e-06,
      "loss": 0.6372,
      "step": 5240
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7706855728317539,
      "learning_rate": 2.6267122845027084e-06,
      "loss": 0.5649,
      "step": 5241
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6460768354717873,
      "learning_rate": 2.6248964237290526e-06,
      "loss": 0.4874,
      "step": 5242
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.662705613300476,
      "learning_rate": 2.623080967410793e-06,
      "loss": 0.52,
      "step": 5243
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7990598437806741,
      "learning_rate": 2.621265915857081e-06,
      "loss": 0.6706,
      "step": 5244
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9225338268173324,
      "learning_rate": 2.619451269377003e-06,
      "loss": 0.6473,
      "step": 5245
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8533217697505727,
      "learning_rate": 2.617637028279573e-06,
      "loss": 0.7123,
      "step": 5246
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.873994190015143,
      "learning_rate": 2.615823192873738e-06,
      "loss": 0.6195,
      "step": 5247
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0139458032267308,
      "learning_rate": 2.614009763468375e-06,
      "loss": 0.6711,
      "step": 5248
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.809710035291973,
      "learning_rate": 2.6121967403722924e-06,
      "loss": 0.5229,
      "step": 5249
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6422406968118504,
      "learning_rate": 2.610384123894229e-06,
      "loss": 0.5363,
      "step": 5250
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6570753282383198,
      "learning_rate": 2.6085719143428547e-06,
      "loss": 0.5341,
      "step": 5251
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6103822686860066,
      "learning_rate": 2.6067601120267695e-06,
      "loss": 0.5225,
      "step": 5252
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7695996376033003,
      "learning_rate": 2.604948717254504e-06,
      "loss": 0.6001,
      "step": 5253
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7139670666644147,
      "learning_rate": 2.603137730334521e-06,
      "loss": 0.5494,
      "step": 5254
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9943704861211224,
      "learning_rate": 2.601327151575212e-06,
      "loss": 0.66,
      "step": 5255
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6826620922904162,
      "learning_rate": 2.5995169812849e-06,
      "loss": 0.5032,
      "step": 5256
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8977199972802585,
      "learning_rate": 2.5977072197718378e-06,
      "loss": 0.6674,
      "step": 5257
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6665536624720048,
      "learning_rate": 2.595897867344209e-06,
      "loss": 0.5717,
      "step": 5258
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8001280344675189,
      "learning_rate": 2.5940889243101288e-06,
      "loss": 0.5854,
      "step": 5259
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6987764600203765,
      "learning_rate": 2.5922803909776395e-06,
      "loss": 0.5331,
      "step": 5260
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8417843250563803,
      "learning_rate": 2.590472267654714e-06,
      "loss": 0.5828,
      "step": 5261
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7679209891435838,
      "learning_rate": 2.588664554649261e-06,
      "loss": 0.5427,
      "step": 5262
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.701675621922384,
      "learning_rate": 2.586857252269114e-06,
      "loss": 0.5174,
      "step": 5263
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1556914796524849,
      "learning_rate": 2.5850503608220357e-06,
      "loss": 0.5997,
      "step": 5264
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7293964470323818,
      "learning_rate": 2.5832438806157235e-06,
      "loss": 0.5521,
      "step": 5265
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8560915706260345,
      "learning_rate": 2.5814378119578e-06,
      "loss": 0.6487,
      "step": 5266
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8777483202910402,
      "learning_rate": 2.5796321551558207e-06,
      "loss": 0.5826,
      "step": 5267
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.711247152771286,
      "learning_rate": 2.5778269105172704e-06,
      "loss": 0.5564,
      "step": 5268
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.866180313828373,
      "learning_rate": 2.576022078349563e-06,
      "loss": 0.6673,
      "step": 5269
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0275463129549856,
      "learning_rate": 2.574217658960043e-06,
      "loss": 0.6882,
      "step": 5270
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8071827298527652,
      "learning_rate": 2.5724136526559844e-06,
      "loss": 0.5799,
      "step": 5271
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7586171614296182,
      "learning_rate": 2.5706100597445883e-06,
      "loss": 0.5682,
      "step": 5272
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8427413360315162,
      "learning_rate": 2.5688068805329915e-06,
      "loss": 0.6191,
      "step": 5273
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7337929045199099,
      "learning_rate": 2.567004115328255e-06,
      "loss": 0.6035,
      "step": 5274
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9438880339452042,
      "learning_rate": 2.5652017644373706e-06,
      "loss": 0.5848,
      "step": 5275
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9769047683541847,
      "learning_rate": 2.563399828167261e-06,
      "loss": 0.6538,
      "step": 5276
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8829275409617017,
      "learning_rate": 2.5615983068247756e-06,
      "loss": 0.6033,
      "step": 5277
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3160186572457457,
      "learning_rate": 2.5597972007166962e-06,
      "loss": 0.6373,
      "step": 5278
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8258390241947977,
      "learning_rate": 2.5579965101497306e-06,
      "loss": 0.6359,
      "step": 5279
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0385353606902328,
      "learning_rate": 2.556196235430519e-06,
      "loss": 0.6075,
      "step": 5280
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7646205983824458,
      "learning_rate": 2.55439637686563e-06,
      "loss": 0.541,
      "step": 5281
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6097114060467259,
      "learning_rate": 2.5525969347615597e-06,
      "loss": 0.5082,
      "step": 5282
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8189481212915716,
      "learning_rate": 2.550797909424734e-06,
      "loss": 0.6647,
      "step": 5283
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7689881245083069,
      "learning_rate": 2.548999301161509e-06,
      "loss": 0.5887,
      "step": 5284
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9122215422869405,
      "learning_rate": 2.5472011102781685e-06,
      "loss": 0.7127,
      "step": 5285
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6515502180777499,
      "learning_rate": 2.5454033370809263e-06,
      "loss": 0.5224,
      "step": 5286
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8691432677847745,
      "learning_rate": 2.5436059818759228e-06,
      "loss": 0.6267,
      "step": 5287
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8713768790040591,
      "learning_rate": 2.541809044969231e-06,
      "loss": 0.6434,
      "step": 5288
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7830431389617342,
      "learning_rate": 2.5400125266668497e-06,
      "loss": 0.6034,
      "step": 5289
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0200552558680356,
      "learning_rate": 2.538216427274707e-06,
      "loss": 0.6176,
      "step": 5290
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7371042727602204,
      "learning_rate": 2.5364207470986602e-06,
      "loss": 0.5725,
      "step": 5291
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7577744632607916,
      "learning_rate": 2.5346254864444943e-06,
      "loss": 0.629,
      "step": 5292
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7466468977963334,
      "learning_rate": 2.5328306456179243e-06,
      "loss": 0.5683,
      "step": 5293
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.5179946780505507,
      "learning_rate": 2.5310362249245925e-06,
      "loss": 0.6534,
      "step": 5294
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7053951999306056,
      "learning_rate": 2.5292422246700697e-06,
      "loss": 0.553,
      "step": 5295
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7006661402395418,
      "learning_rate": 2.5274486451598566e-06,
      "loss": 0.5853,
      "step": 5296
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9067466621332471,
      "learning_rate": 2.5256554866993806e-06,
      "loss": 0.5923,
      "step": 5297
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9191279986740406,
      "learning_rate": 2.523862749593997e-06,
      "loss": 0.6119,
      "step": 5298
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8216816541969479,
      "learning_rate": 2.522070434148989e-06,
      "loss": 0.6189,
      "step": 5299
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9552477996825173,
      "learning_rate": 2.5202785406695735e-06,
      "loss": 0.4926,
      "step": 5300
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8363393654229209,
      "learning_rate": 2.5184870694608885e-06,
      "loss": 0.6114,
      "step": 5301
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6484358948002191,
      "learning_rate": 2.5166960208280034e-06,
      "loss": 0.5039,
      "step": 5302
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6915461009566738,
      "learning_rate": 2.5149053950759158e-06,
      "loss": 0.5289,
      "step": 5303
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7191270531355878,
      "learning_rate": 2.5131151925095494e-06,
      "loss": 0.5573,
      "step": 5304
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0582117147089085,
      "learning_rate": 2.5113254134337573e-06,
      "loss": 0.6053,
      "step": 5305
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.710298610894339,
      "learning_rate": 2.5095360581533206e-06,
      "loss": 0.5284,
      "step": 5306
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7029492979667713,
      "learning_rate": 2.507747126972948e-06,
      "loss": 0.5334,
      "step": 5307
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1488064095390504,
      "learning_rate": 2.5059586201972753e-06,
      "loss": 0.6311,
      "step": 5308
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7779439577540818,
      "learning_rate": 2.5041705381308668e-06,
      "loss": 0.4788,
      "step": 5309
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9933332791254801,
      "learning_rate": 2.5023828810782146e-06,
      "loss": 0.5873,
      "step": 5310
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8513073559435539,
      "learning_rate": 2.500595649343738e-06,
      "loss": 0.6416,
      "step": 5311
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7639063991240539,
      "learning_rate": 2.498808843231783e-06,
      "loss": 0.5688,
      "step": 5312
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.788341030157502,
      "learning_rate": 2.497022463046625e-06,
      "loss": 0.5406,
      "step": 5313
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7494752350933808,
      "learning_rate": 2.4952365090924655e-06,
      "loss": 0.597,
      "step": 5314
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9327386042471393,
      "learning_rate": 2.4934509816734347e-06,
      "loss": 0.6244,
      "step": 5315
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.71932882645165,
      "learning_rate": 2.4916658810935883e-06,
      "loss": 0.5138,
      "step": 5316
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6806515348120529,
      "learning_rate": 2.4898812076569107e-06,
      "loss": 0.56,
      "step": 5317
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7666257793189124,
      "learning_rate": 2.4880969616673136e-06,
      "loss": 0.5445,
      "step": 5318
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7277506922752934,
      "learning_rate": 2.486313143428634e-06,
      "loss": 0.5564,
      "step": 5319
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7166121764710645,
      "learning_rate": 2.48452975324464e-06,
      "loss": 0.5502,
      "step": 5320
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6892566640871824,
      "learning_rate": 2.4827467914190225e-06,
      "loss": 0.5342,
      "step": 5321
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8367229406763523,
      "learning_rate": 2.480964258255402e-06,
      "loss": 0.6368,
      "step": 5322
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9321918716234291,
      "learning_rate": 2.479182154057325e-06,
      "loss": 0.6727,
      "step": 5323
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8156121481571531,
      "learning_rate": 2.4774004791282653e-06,
      "loss": 0.5807,
      "step": 5324
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6726538973396666,
      "learning_rate": 2.4756192337716223e-06,
      "loss": 0.5095,
      "step": 5325
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8660779631221062,
      "learning_rate": 2.4738384182907264e-06,
      "loss": 0.5562,
      "step": 5326
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9454378469220245,
      "learning_rate": 2.4720580329888303e-06,
      "loss": 0.6312,
      "step": 5327
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7752034688525052,
      "learning_rate": 2.470278078169116e-06,
      "loss": 0.5346,
      "step": 5328
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.693389887716643,
      "learning_rate": 2.4684985541346907e-06,
      "loss": 0.5288,
      "step": 5329
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7441921617972328,
      "learning_rate": 2.466719461188588e-06,
      "loss": 0.5628,
      "step": 5330
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9819225087527707,
      "learning_rate": 2.4649407996337697e-06,
      "loss": 0.6428,
      "step": 5331
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6675731053647536,
      "learning_rate": 2.463162569773123e-06,
      "loss": 0.5688,
      "step": 5332
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6435252461158815,
      "learning_rate": 2.4613847719094624e-06,
      "loss": 0.5398,
      "step": 5333
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7685754160739873,
      "learning_rate": 2.459607406345528e-06,
      "loss": 0.6097,
      "step": 5334
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8653335017815035,
      "learning_rate": 2.457830473383986e-06,
      "loss": 0.585,
      "step": 5335
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8109551090664087,
      "learning_rate": 2.4560539733274315e-06,
      "loss": 0.5861,
      "step": 5336
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.791868389071689,
      "learning_rate": 2.4542779064783822e-06,
      "loss": 0.5948,
      "step": 5337
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6937719785335831,
      "learning_rate": 2.452502273139284e-06,
      "loss": 0.5633,
      "step": 5338
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8404216725002247,
      "learning_rate": 2.4507270736125094e-06,
      "loss": 0.6407,
      "step": 5339
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7938091361582823,
      "learning_rate": 2.448952308200356e-06,
      "loss": 0.6101,
      "step": 5340
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8191442134545885,
      "learning_rate": 2.4471779772050473e-06,
      "loss": 0.5575,
      "step": 5341
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7403353315547542,
      "learning_rate": 2.4454040809287342e-06,
      "loss": 0.5348,
      "step": 5342
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8599438179354085,
      "learning_rate": 2.4436306196734933e-06,
      "loss": 0.595,
      "step": 5343
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8825854790144705,
      "learning_rate": 2.4418575937413254e-06,
      "loss": 0.6291,
      "step": 5344
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.68301862050764,
      "learning_rate": 2.4400850034341582e-06,
      "loss": 0.5152,
      "step": 5345
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6788624255747951,
      "learning_rate": 2.438312849053846e-06,
      "loss": 0.5825,
      "step": 5346
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7380081419150237,
      "learning_rate": 2.4365411309021687e-06,
      "loss": 0.5666,
      "step": 5347
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9596195705507765,
      "learning_rate": 2.434769849280831e-06,
      "loss": 0.6246,
      "step": 5348
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7342138494967124,
      "learning_rate": 2.432999004491463e-06,
      "loss": 0.5612,
      "step": 5349
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6667667016439291,
      "learning_rate": 2.431228596835622e-06,
      "loss": 0.5267,
      "step": 5350
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7303715038111931,
      "learning_rate": 2.4294586266147874e-06,
      "loss": 0.4997,
      "step": 5351
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6386560850597269,
      "learning_rate": 2.4276890941303716e-06,
      "loss": 0.5104,
      "step": 5352
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7536038991335989,
      "learning_rate": 2.425919999683704e-06,
      "loss": 0.5472,
      "step": 5353
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6570526316974628,
      "learning_rate": 2.424151343576045e-06,
      "loss": 0.4995,
      "step": 5354
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8340767752903652,
      "learning_rate": 2.4223831261085764e-06,
      "loss": 0.6018,
      "step": 5355
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7104125961206265,
      "learning_rate": 2.420615347582408e-06,
      "loss": 0.5863,
      "step": 5356
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8732035373587792,
      "learning_rate": 2.418848008298574e-06,
      "loss": 0.6282,
      "step": 5357
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6314505691052757,
      "learning_rate": 2.417081108558035e-06,
      "loss": 0.5066,
      "step": 5358
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.897846263399217,
      "learning_rate": 2.4153146486616735e-06,
      "loss": 0.6146,
      "step": 5359
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7812882208015591,
      "learning_rate": 2.4135486289103004e-06,
      "loss": 0.5895,
      "step": 5360
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7737994567895307,
      "learning_rate": 2.411783049604651e-06,
      "loss": 0.5777,
      "step": 5361
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6040148231421012,
      "learning_rate": 2.4100179110453837e-06,
      "loss": 0.4791,
      "step": 5362
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7016395308207667,
      "learning_rate": 2.408253213533082e-06,
      "loss": 0.5269,
      "step": 5363
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6626890016453932,
      "learning_rate": 2.4064889573682597e-06,
      "loss": 0.5264,
      "step": 5364
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6480264204053937,
      "learning_rate": 2.4047251428513485e-06,
      "loss": 0.5171,
      "step": 5365
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7389679512333903,
      "learning_rate": 2.4029617702827084e-06,
      "loss": 0.5549,
      "step": 5366
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7278065098474038,
      "learning_rate": 2.4011988399626234e-06,
      "loss": 0.5757,
      "step": 5367
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7669481115886791,
      "learning_rate": 2.399436352191301e-06,
      "loss": 0.6056,
      "step": 5368
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8422950530889523,
      "learning_rate": 2.3976743072688762e-06,
      "loss": 0.6551,
      "step": 5369
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8229577730294655,
      "learning_rate": 2.3959127054954055e-06,
      "loss": 0.6399,
      "step": 5370
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6441364149994218,
      "learning_rate": 2.394151547170872e-06,
      "loss": 0.5392,
      "step": 5371
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7067806606288211,
      "learning_rate": 2.392390832595183e-06,
      "loss": 0.5188,
      "step": 5372
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6278798983084514,
      "learning_rate": 2.3906305620681685e-06,
      "loss": 0.5416,
      "step": 5373
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7145489780461023,
      "learning_rate": 2.3888707358895857e-06,
      "loss": 0.5592,
      "step": 5374
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7929315338798346,
      "learning_rate": 2.3871113543591135e-06,
      "loss": 0.6039,
      "step": 5375
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6540776364410453,
      "learning_rate": 2.385352417776357e-06,
      "loss": 0.5167,
      "step": 5376
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8526947533821853,
      "learning_rate": 2.3835939264408448e-06,
      "loss": 0.6577,
      "step": 5377
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6223658010552587,
      "learning_rate": 2.3818358806520294e-06,
      "loss": 0.4936,
      "step": 5378
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.675765142899162,
      "learning_rate": 2.380078280709287e-06,
      "loss": 0.5254,
      "step": 5379
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8321936800602324,
      "learning_rate": 2.37832112691192e-06,
      "loss": 0.5635,
      "step": 5380
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6760457101828767,
      "learning_rate": 2.3765644195591523e-06,
      "loss": 0.51,
      "step": 5381
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.925354546983763,
      "learning_rate": 2.3748081589501332e-06,
      "loss": 0.5816,
      "step": 5382
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6534995243192381,
      "learning_rate": 2.373052345383934e-06,
      "loss": 0.4827,
      "step": 5383
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7723883868607132,
      "learning_rate": 2.3712969791595545e-06,
      "loss": 0.6011,
      "step": 5384
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8063777701054817,
      "learning_rate": 2.369542060575914e-06,
      "loss": 0.5332,
      "step": 5385
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8498449762263667,
      "learning_rate": 2.3677875899318563e-06,
      "loss": 0.6023,
      "step": 5386
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7524225015616848,
      "learning_rate": 2.366033567526151e-06,
      "loss": 0.5127,
      "step": 5387
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7729641200368167,
      "learning_rate": 2.3642799936574874e-06,
      "loss": 0.5518,
      "step": 5388
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.886761729221374,
      "learning_rate": 2.362526868624483e-06,
      "loss": 0.6214,
      "step": 5389
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6299389765409734,
      "learning_rate": 2.3607741927256755e-06,
      "loss": 0.4619,
      "step": 5390
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8738654729510181,
      "learning_rate": 2.3590219662595276e-06,
      "loss": 0.6325,
      "step": 5391
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8228660083102083,
      "learning_rate": 2.3572701895244252e-06,
      "loss": 0.6307,
      "step": 5392
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7887710694370695,
      "learning_rate": 2.3555188628186783e-06,
      "loss": 0.5075,
      "step": 5393
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7574997946229495,
      "learning_rate": 2.353767986440519e-06,
      "loss": 0.5575,
      "step": 5394
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.896120589584835,
      "learning_rate": 2.3520175606881027e-06,
      "loss": 0.6828,
      "step": 5395
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6178984751225683,
      "learning_rate": 2.3502675858595104e-06,
      "loss": 0.5313,
      "step": 5396
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0551036824662303,
      "learning_rate": 2.3485180622527425e-06,
      "loss": 0.6067,
      "step": 5397
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6721172700713032,
      "learning_rate": 2.346768990165726e-06,
      "loss": 0.5041,
      "step": 5398
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6817065179056067,
      "learning_rate": 2.3450203698963096e-06,
      "loss": 0.537,
      "step": 5399
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6322783417417094,
      "learning_rate": 2.3432722017422642e-06,
      "loss": 0.4949,
      "step": 5400
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8564157233654893,
      "learning_rate": 2.341524486001286e-06,
      "loss": 0.6461,
      "step": 5401
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9013760904396483,
      "learning_rate": 2.3397772229709918e-06,
      "loss": 0.6791,
      "step": 5402
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8223743532631766,
      "learning_rate": 2.3380304129489222e-06,
      "loss": 0.544,
      "step": 5403
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7741006386300008,
      "learning_rate": 2.3362840562325416e-06,
      "loss": 0.5124,
      "step": 5404
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6707223499857118,
      "learning_rate": 2.3345381531192356e-06,
      "loss": 0.5051,
      "step": 5405
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7844729973188754,
      "learning_rate": 2.3327927039063132e-06,
      "loss": 0.5407,
      "step": 5406
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8094001757424992,
      "learning_rate": 2.331047708891007e-06,
      "loss": 0.587,
      "step": 5407
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8582538697112428,
      "learning_rate": 2.329303168370471e-06,
      "loss": 0.6154,
      "step": 5408
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8423328587794595,
      "learning_rate": 2.327559082641782e-06,
      "loss": 0.6138,
      "step": 5409
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7746988149780637,
      "learning_rate": 2.3258154520019384e-06,
      "loss": 0.6068,
      "step": 5410
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6718025721432145,
      "learning_rate": 2.324072276747866e-06,
      "loss": 0.5067,
      "step": 5411
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7085910569633641,
      "learning_rate": 2.3223295571764066e-06,
      "loss": 0.5204,
      "step": 5412
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.147222270471566,
      "learning_rate": 2.3205872935843276e-06,
      "loss": 0.6157,
      "step": 5413
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7520969433114628,
      "learning_rate": 2.3188454862683195e-06,
      "loss": 0.5227,
      "step": 5414
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7507322232511526,
      "learning_rate": 2.3171041355249924e-06,
      "loss": 0.5929,
      "step": 5415
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6759347146284236,
      "learning_rate": 2.3153632416508807e-06,
      "loss": 0.5305,
      "step": 5416
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8243327845475505,
      "learning_rate": 2.3136228049424414e-06,
      "loss": 0.6164,
      "step": 5417
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7824745703095208,
      "learning_rate": 2.311882825696051e-06,
      "loss": 0.5448,
      "step": 5418
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7636758783251725,
      "learning_rate": 2.310143304208012e-06,
      "loss": 0.6491,
      "step": 5419
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9273589794626783,
      "learning_rate": 2.308404240774544e-06,
      "loss": 0.6899,
      "step": 5420
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8420805490318296,
      "learning_rate": 2.3066656356917945e-06,
      "loss": 0.653,
      "step": 5421
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7673069020917275,
      "learning_rate": 2.304927489255827e-06,
      "loss": 0.6179,
      "step": 5422
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6622773162898108,
      "learning_rate": 2.303189801762632e-06,
      "loss": 0.5102,
      "step": 5423
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7859261737358603,
      "learning_rate": 2.301452573508118e-06,
      "loss": 0.6175,
      "step": 5424
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6777604156252802,
      "learning_rate": 2.2997158047881175e-06,
      "loss": 0.5784,
      "step": 5425
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.992417508911669,
      "learning_rate": 2.297979495898384e-06,
      "loss": 0.6385,
      "step": 5426
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0902541315610164,
      "learning_rate": 2.2962436471345933e-06,
      "loss": 0.6005,
      "step": 5427
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8109023685581865,
      "learning_rate": 2.2945082587923416e-06,
      "loss": 0.5942,
      "step": 5428
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8583208956967828,
      "learning_rate": 2.2927733311671486e-06,
      "loss": 0.5033,
      "step": 5429
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6944448285715461,
      "learning_rate": 2.291038864554453e-06,
      "loss": 0.5801,
      "step": 5430
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9278828512087278,
      "learning_rate": 2.289304859249617e-06,
      "loss": 0.6054,
      "step": 5431
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6582588346169828,
      "learning_rate": 2.2875713155479233e-06,
      "loss": 0.5181,
      "step": 5432
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7181118928320781,
      "learning_rate": 2.2858382337445774e-06,
      "loss": 0.57,
      "step": 5433
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7694876357311476,
      "learning_rate": 2.284105614134704e-06,
      "loss": 0.6105,
      "step": 5434
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9694382948828406,
      "learning_rate": 2.2823734570133506e-06,
      "loss": 0.6232,
      "step": 5435
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8574219637739839,
      "learning_rate": 2.2806417626754833e-06,
      "loss": 0.6288,
      "step": 5436
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6720394743273487,
      "learning_rate": 2.2789105314159958e-06,
      "loss": 0.5513,
      "step": 5437
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8032699678936319,
      "learning_rate": 2.277179763529697e-06,
      "loss": 0.6172,
      "step": 5438
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.814665701234805,
      "learning_rate": 2.275449459311318e-06,
      "loss": 0.6312,
      "step": 5439
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7958682646619236,
      "learning_rate": 2.273719619055512e-06,
      "loss": 0.5246,
      "step": 5440
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6825506657840111,
      "learning_rate": 2.271990243056853e-06,
      "loss": 0.5402,
      "step": 5441
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8453369251157237,
      "learning_rate": 2.270261331609835e-06,
      "loss": 0.648,
      "step": 5442
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.791894998758287,
      "learning_rate": 2.268532885008874e-06,
      "loss": 0.609,
      "step": 5443
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8574678061964685,
      "learning_rate": 2.2668049035483066e-06,
      "loss": 0.6243,
      "step": 5444
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.751011965024508,
      "learning_rate": 2.2650773875223897e-06,
      "loss": 0.5689,
      "step": 5445
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7384988250916608,
      "learning_rate": 2.263350337225302e-06,
      "loss": 0.5029,
      "step": 5446
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9307559361448564,
      "learning_rate": 2.2616237529511416e-06,
      "loss": 0.6537,
      "step": 5447
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.840607566116256,
      "learning_rate": 2.2598976349939274e-06,
      "loss": 0.6159,
      "step": 5448
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8193371530842894,
      "learning_rate": 2.2581719836475994e-06,
      "loss": 0.5749,
      "step": 5449
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7346374058338675,
      "learning_rate": 2.2564467992060197e-06,
      "loss": 0.5946,
      "step": 5450
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6884897738854778,
      "learning_rate": 2.2547220819629673e-06,
      "loss": 0.5408,
      "step": 5451
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7352881447111629,
      "learning_rate": 2.2529978322121437e-06,
      "loss": 0.5771,
      "step": 5452
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7543286695585357,
      "learning_rate": 2.251274050247172e-06,
      "loss": 0.5166,
      "step": 5453
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.786852291708415,
      "learning_rate": 2.2495507363615933e-06,
      "loss": 0.5326,
      "step": 5454
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8003772149051532,
      "learning_rate": 2.2478278908488706e-06,
      "loss": 0.5575,
      "step": 5455
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6445187805096474,
      "learning_rate": 2.246105514002385e-06,
      "loss": 0.5247,
      "step": 5456
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7267553436671903,
      "learning_rate": 2.244383606115442e-06,
      "loss": 0.4903,
      "step": 5457
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7476051197273879,
      "learning_rate": 2.242662167481262e-06,
      "loss": 0.571,
      "step": 5458
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8233348815554917,
      "learning_rate": 2.2409411983929896e-06,
      "loss": 0.6096,
      "step": 5459
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6425241998175133,
      "learning_rate": 2.239220699143687e-06,
      "loss": 0.5715,
      "step": 5460
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8118638863170009,
      "learning_rate": 2.237500670026338e-06,
      "loss": 0.6421,
      "step": 5461
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7328324508731832,
      "learning_rate": 2.235781111333844e-06,
      "loss": 0.5437,
      "step": 5462
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.848002311094021,
      "learning_rate": 2.2340620233590305e-06,
      "loss": 0.6428,
      "step": 5463
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9349523828157077,
      "learning_rate": 2.2323434063946393e-06,
      "loss": 0.6424,
      "step": 5464
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7746754400526813,
      "learning_rate": 2.2306252607333328e-06,
      "loss": 0.5952,
      "step": 5465
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7244380053129705,
      "learning_rate": 2.2289075866676934e-06,
      "loss": 0.5357,
      "step": 5466
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8212711232419224,
      "learning_rate": 2.2271903844902227e-06,
      "loss": 0.6646,
      "step": 5467
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6371127234918188,
      "learning_rate": 2.225473654493341e-06,
      "loss": 0.5084,
      "step": 5468
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7929704056355397,
      "learning_rate": 2.2237573969693926e-06,
      "loss": 0.545,
      "step": 5469
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7039795280299358,
      "learning_rate": 2.2220416122106374e-06,
      "loss": 0.5714,
      "step": 5470
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9149845528810331,
      "learning_rate": 2.2203263005092547e-06,
      "loss": 0.6597,
      "step": 5471
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8047719505540758,
      "learning_rate": 2.2186114621573445e-06,
      "loss": 0.609,
      "step": 5472
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7727057797183778,
      "learning_rate": 2.216897097446926e-06,
      "loss": 0.5249,
      "step": 5473
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8220198655219986,
      "learning_rate": 2.215183206669938e-06,
      "loss": 0.6012,
      "step": 5474
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7265727539917279,
      "learning_rate": 2.2134697901182377e-06,
      "loss": 0.5682,
      "step": 5475
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.903222193124084,
      "learning_rate": 2.211756848083602e-06,
      "loss": 0.5433,
      "step": 5476
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.627560511157613,
      "learning_rate": 2.2100443808577277e-06,
      "loss": 0.508,
      "step": 5477
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8675273771548871,
      "learning_rate": 2.20833238873223e-06,
      "loss": 0.6126,
      "step": 5478
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7465687450008791,
      "learning_rate": 2.2066208719986427e-06,
      "loss": 0.6016,
      "step": 5479
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7589702958589158,
      "learning_rate": 2.2049098309484195e-06,
      "loss": 0.5619,
      "step": 5480
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7436676254364487,
      "learning_rate": 2.2031992658729332e-06,
      "loss": 0.5846,
      "step": 5481
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8179437459808657,
      "learning_rate": 2.2014891770634754e-06,
      "loss": 0.6236,
      "step": 5482
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.819481852039836,
      "learning_rate": 2.1997795648112565e-06,
      "loss": 0.5212,
      "step": 5483
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8062307125898023,
      "learning_rate": 2.1980704294074046e-06,
      "loss": 0.6598,
      "step": 5484
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8736678231296808,
      "learning_rate": 2.196361771142968e-06,
      "loss": 0.6641,
      "step": 5485
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.797231043228781,
      "learning_rate": 2.1946535903089137e-06,
      "loss": 0.5354,
      "step": 5486
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.730620179718931,
      "learning_rate": 2.192945887196128e-06,
      "loss": 0.562,
      "step": 5487
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.687500333445565,
      "learning_rate": 2.191238662095413e-06,
      "loss": 0.5318,
      "step": 5488
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8008827186949273,
      "learning_rate": 2.189531915297493e-06,
      "loss": 0.5731,
      "step": 5489
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7450514056367135,
      "learning_rate": 2.187825647093008e-06,
      "loss": 0.606,
      "step": 5490
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7562351279620019,
      "learning_rate": 2.186119857772519e-06,
      "loss": 0.5262,
      "step": 5491
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.777023742649139,
      "learning_rate": 2.1844145476265034e-06,
      "loss": 0.6402,
      "step": 5492
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7630075972679621,
      "learning_rate": 2.182709716945357e-06,
      "loss": 0.5389,
      "step": 5493
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6683080550086076,
      "learning_rate": 2.1810053660193942e-06,
      "loss": 0.5323,
      "step": 5494
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7387737613325633,
      "learning_rate": 2.179301495138851e-06,
      "loss": 0.5573,
      "step": 5495
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9339325229210625,
      "learning_rate": 2.1775981045938777e-06,
      "loss": 0.6318,
      "step": 5496
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6991522936648061,
      "learning_rate": 2.175895194674544e-06,
      "loss": 0.5574,
      "step": 5497
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7268789330806732,
      "learning_rate": 2.1741927656708367e-06,
      "loss": 0.5712,
      "step": 5498
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0366084791100336,
      "learning_rate": 2.172490817872662e-06,
      "loss": 0.6499,
      "step": 5499
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6085721067846119,
      "learning_rate": 2.1707893515698453e-06,
      "loss": 0.5262,
      "step": 5500
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8753175546708015,
      "learning_rate": 2.1690883670521268e-06,
      "loss": 0.6083,
      "step": 5501
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8832062765984533,
      "learning_rate": 2.1673878646091673e-06,
      "loss": 0.6709,
      "step": 5502
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8024769155615417,
      "learning_rate": 2.1656878445305444e-06,
      "loss": 0.6096,
      "step": 5503
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6736036244314383,
      "learning_rate": 2.163988307105754e-06,
      "loss": 0.5355,
      "step": 5504
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7993273577780864,
      "learning_rate": 2.16228925262421e-06,
      "loss": 0.6,
      "step": 5505
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.768486974951675,
      "learning_rate": 2.1605906813752425e-06,
      "loss": 0.5565,
      "step": 5506
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8202716943392994,
      "learning_rate": 2.158892593648102e-06,
      "loss": 0.6415,
      "step": 5507
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.119292165601142,
      "learning_rate": 2.157194989731954e-06,
      "loss": 0.6379,
      "step": 5508
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7937688272521658,
      "learning_rate": 2.155497869915883e-06,
      "loss": 0.5848,
      "step": 5509
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8435610489060712,
      "learning_rate": 2.153801234488891e-06,
      "loss": 0.6862,
      "step": 5510
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7357321420284718,
      "learning_rate": 2.1521050837398973e-06,
      "loss": 0.5584,
      "step": 5511
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.74735447879491,
      "learning_rate": 2.1504094179577384e-06,
      "loss": 0.547,
      "step": 5512
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7012296461408165,
      "learning_rate": 2.1487142374311687e-06,
      "loss": 0.5101,
      "step": 5513
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.725785321295535,
      "learning_rate": 2.1470195424488604e-06,
      "loss": 0.5499,
      "step": 5514
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6491729376208991,
      "learning_rate": 2.145325333299402e-06,
      "loss": 0.5594,
      "step": 5515
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7031354977511042,
      "learning_rate": 2.143631610271299e-06,
      "loss": 0.5744,
      "step": 5516
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7673997683983097,
      "learning_rate": 2.141938373652976e-06,
      "loss": 0.5722,
      "step": 5517
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8458977155063527,
      "learning_rate": 2.1402456237327723e-06,
      "loss": 0.5795,
      "step": 5518
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6702997935290794,
      "learning_rate": 2.138553360798946e-06,
      "loss": 0.55,
      "step": 5519
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7565775237821801,
      "learning_rate": 2.136861585139673e-06,
      "loss": 0.5039,
      "step": 5520
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6399293201317746,
      "learning_rate": 2.1351702970430416e-06,
      "loss": 0.506,
      "step": 5521
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6324718045819099,
      "learning_rate": 2.1334794967970645e-06,
      "loss": 0.5257,
      "step": 5522
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5075733951081995,
      "learning_rate": 2.1317891846896667e-06,
      "loss": 0.5919,
      "step": 5523
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9518755364849981,
      "learning_rate": 2.1300993610086896e-06,
      "loss": 0.5867,
      "step": 5524
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7968632079044731,
      "learning_rate": 2.1284100260418925e-06,
      "loss": 0.5386,
      "step": 5525
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8826189390342716,
      "learning_rate": 2.1267211800769526e-06,
      "loss": 0.6519,
      "step": 5526
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1235859108089383,
      "learning_rate": 2.1250328234014616e-06,
      "loss": 0.6338,
      "step": 5527
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8493132102038925,
      "learning_rate": 2.12334495630293e-06,
      "loss": 0.6474,
      "step": 5528
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8510494293402592,
      "learning_rate": 2.1216575790687832e-06,
      "loss": 0.555,
      "step": 5529
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7007531141533185,
      "learning_rate": 2.119970691986365e-06,
      "loss": 0.5941,
      "step": 5530
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8317231529744354,
      "learning_rate": 2.118284295342933e-06,
      "loss": 0.5655,
      "step": 5531
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7215323265558325,
      "learning_rate": 2.1165983894256647e-06,
      "loss": 0.539,
      "step": 5532
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8459321081897421,
      "learning_rate": 2.1149129745216517e-06,
      "loss": 0.5281,
      "step": 5533
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6855031396466958,
      "learning_rate": 2.1132280509179028e-06,
      "loss": 0.4985,
      "step": 5534
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.018494057790836,
      "learning_rate": 2.111543618901342e-06,
      "loss": 0.5436,
      "step": 5535
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8204985499815014,
      "learning_rate": 2.109859678758811e-06,
      "loss": 0.6598,
      "step": 5536
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6676819925208591,
      "learning_rate": 2.108176230777068e-06,
      "loss": 0.538,
      "step": 5537
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.742157236254718,
      "learning_rate": 2.106493275242786e-06,
      "loss": 0.6618,
      "step": 5538
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.230737151565162,
      "learning_rate": 2.104810812442555e-06,
      "loss": 0.6408,
      "step": 5539
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7185964595205173,
      "learning_rate": 2.1031288426628804e-06,
      "loss": 0.5732,
      "step": 5540
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8169661986251882,
      "learning_rate": 2.1014473661901847e-06,
      "loss": 0.64,
      "step": 5541
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.861008560614642,
      "learning_rate": 2.0997663833108055e-06,
      "loss": 0.5901,
      "step": 5542
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7666558973572319,
      "learning_rate": 2.0980858943109968e-06,
      "loss": 0.5228,
      "step": 5543
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8360299141662741,
      "learning_rate": 2.0964058994769285e-06,
      "loss": 0.6622,
      "step": 5544
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7996590397674925,
      "learning_rate": 2.0947263990946858e-06,
      "loss": 0.5166,
      "step": 5545
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7308603994951394,
      "learning_rate": 2.093047393450271e-06,
      "loss": 0.5705,
      "step": 5546
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8530056607536854,
      "learning_rate": 2.0913688828295985e-06,
      "loss": 0.6107,
      "step": 5547
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8375968253809218,
      "learning_rate": 2.089690867518505e-06,
      "loss": 0.5616,
      "step": 5548
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0095294866767712,
      "learning_rate": 2.088013347802738e-06,
      "loss": 0.656,
      "step": 5549
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7156805700817284,
      "learning_rate": 2.086336323967961e-06,
      "loss": 0.5055,
      "step": 5550
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6319153617280628,
      "learning_rate": 2.0846597962997527e-06,
      "loss": 0.5303,
      "step": 5551
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6406284759411545,
      "learning_rate": 2.08298376508361e-06,
      "loss": 0.5136,
      "step": 5552
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.65868663834691,
      "learning_rate": 2.081308230604943e-06,
      "loss": 0.5432,
      "step": 5553
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8308231429315793,
      "learning_rate": 2.079633193149077e-06,
      "loss": 0.5445,
      "step": 5554
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7480304537907051,
      "learning_rate": 2.0779586530012546e-06,
      "loss": 0.5122,
      "step": 5555
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7185200318295508,
      "learning_rate": 2.0762846104466322e-06,
      "loss": 0.4888,
      "step": 5556
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9503475403761206,
      "learning_rate": 2.074611065770281e-06,
      "loss": 0.6851,
      "step": 5557
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8489818545682053,
      "learning_rate": 2.072938019257189e-06,
      "loss": 0.6,
      "step": 5558
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8315218311277328,
      "learning_rate": 2.0712654711922585e-06,
      "loss": 0.6163,
      "step": 5559
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8100270151300474,
      "learning_rate": 2.069593421860307e-06,
      "loss": 0.5831,
      "step": 5560
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6704463448884791,
      "learning_rate": 2.067921871546067e-06,
      "loss": 0.5537,
      "step": 5561
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.787622333937641,
      "learning_rate": 2.066250820534186e-06,
      "loss": 0.6215,
      "step": 5562
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8523658204505838,
      "learning_rate": 2.0645802691092266e-06,
      "loss": 0.6071,
      "step": 5563
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8735155891505851,
      "learning_rate": 2.0629102175556666e-06,
      "loss": 0.6556,
      "step": 5564
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7920113011253676,
      "learning_rate": 2.0612406661578977e-06,
      "loss": 0.5863,
      "step": 5565
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0166283796513007,
      "learning_rate": 2.059571615200226e-06,
      "loss": 0.6218,
      "step": 5566
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6532332449733946,
      "learning_rate": 2.057903064966877e-06,
      "loss": 0.5728,
      "step": 5567
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8758672351536236,
      "learning_rate": 2.056235015741986e-06,
      "loss": 0.6422,
      "step": 5568
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7692196061873529,
      "learning_rate": 2.054567467809603e-06,
      "loss": 0.6013,
      "step": 5569
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6311065798340466,
      "learning_rate": 2.052900421453695e-06,
      "loss": 0.4586,
      "step": 5570
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6751948614991886,
      "learning_rate": 2.051233876958143e-06,
      "loss": 0.4977,
      "step": 5571
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.857990397550982,
      "learning_rate": 2.0495678346067414e-06,
      "loss": 0.6146,
      "step": 5572
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7374865128137168,
      "learning_rate": 2.0479022946832007e-06,
      "loss": 0.5535,
      "step": 5573
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7762294594216456,
      "learning_rate": 2.0462372574711442e-06,
      "loss": 0.5657,
      "step": 5574
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9246031622653919,
      "learning_rate": 2.044572723254111e-06,
      "loss": 0.6678,
      "step": 5575
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6492410073858308,
      "learning_rate": 2.042908692315553e-06,
      "loss": 0.517,
      "step": 5576
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8818051855540839,
      "learning_rate": 2.041245164938838e-06,
      "loss": 0.5703,
      "step": 5577
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7563978191187196,
      "learning_rate": 2.0395821414072473e-06,
      "loss": 0.5898,
      "step": 5578
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7980143254065868,
      "learning_rate": 2.0379196220039754e-06,
      "loss": 0.6189,
      "step": 5579
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6657208467267434,
      "learning_rate": 2.036257607012134e-06,
      "loss": 0.5971,
      "step": 5580
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7319788423600943,
      "learning_rate": 2.034596096714747e-06,
      "loss": 0.5718,
      "step": 5581
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7167999258294124,
      "learning_rate": 2.0329350913947508e-06,
      "loss": 0.5946,
      "step": 5582
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7362578288789929,
      "learning_rate": 2.031274591334998e-06,
      "loss": 0.5051,
      "step": 5583
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7259309659847831,
      "learning_rate": 2.0296145968182546e-06,
      "loss": 0.5102,
      "step": 5584
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7489182315593272,
      "learning_rate": 2.0279551081272e-06,
      "loss": 0.584,
      "step": 5585
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0315748556547173,
      "learning_rate": 2.0262961255444273e-06,
      "loss": 0.6764,
      "step": 5586
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7428832516608075,
      "learning_rate": 2.024637649352445e-06,
      "loss": 0.5302,
      "step": 5587
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6554302218803151,
      "learning_rate": 2.0229796798336744e-06,
      "loss": 0.4961,
      "step": 5588
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9014089633908532,
      "learning_rate": 2.0213222172704496e-06,
      "loss": 0.5902,
      "step": 5589
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7657701621051626,
      "learning_rate": 2.019665261945019e-06,
      "loss": 0.5819,
      "step": 5590
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8488592972026056,
      "learning_rate": 2.0180088141395465e-06,
      "loss": 0.6349,
      "step": 5591
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8566977598272854,
      "learning_rate": 2.016352874136106e-06,
      "loss": 0.5623,
      "step": 5592
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7285906131668988,
      "learning_rate": 2.0146974422166878e-06,
      "loss": 0.5278,
      "step": 5593
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6994999894113406,
      "learning_rate": 2.0130425186631945e-06,
      "loss": 0.5438,
      "step": 5594
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.849390029680417,
      "learning_rate": 2.0113881037574423e-06,
      "loss": 0.557,
      "step": 5595
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7981814800297028,
      "learning_rate": 2.009734197781161e-06,
      "loss": 0.5414,
      "step": 5596
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7446727470435076,
      "learning_rate": 2.0080808010159937e-06,
      "loss": 0.4848,
      "step": 5597
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7596826154645443,
      "learning_rate": 2.006427913743496e-06,
      "loss": 0.5934,
      "step": 5598
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7004442288235581,
      "learning_rate": 2.004775536245138e-06,
      "loss": 0.5386,
      "step": 5599
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6614905921355728,
      "learning_rate": 2.0031236688023026e-06,
      "loss": 0.5077,
      "step": 5600
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6425782973179972,
      "learning_rate": 2.0014723116962842e-06,
      "loss": 0.5273,
      "step": 5601
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7871629829663782,
      "learning_rate": 1.999821465208293e-06,
      "loss": 0.6436,
      "step": 5602
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8854727654619043,
      "learning_rate": 1.9981711296194513e-06,
      "loss": 0.6489,
      "step": 5603
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7322818852661492,
      "learning_rate": 1.996521305210793e-06,
      "loss": 0.5484,
      "step": 5604
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6440229071395928,
      "learning_rate": 1.9948719922632644e-06,
      "loss": 0.5019,
      "step": 5605
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7984422841313952,
      "learning_rate": 1.9932231910577306e-06,
      "loss": 0.6088,
      "step": 5606
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8190166531322095,
      "learning_rate": 1.9915749018749626e-06,
      "loss": 0.6077,
      "step": 5607
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8885474322932339,
      "learning_rate": 1.989927124995647e-06,
      "loss": 0.6272,
      "step": 5608
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.086076821709225,
      "learning_rate": 1.9882798607003835e-06,
      "loss": 0.6467,
      "step": 5609
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7926130013670906,
      "learning_rate": 1.9866331092696837e-06,
      "loss": 0.6205,
      "step": 5610
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7642917304703144,
      "learning_rate": 1.9849868709839733e-06,
      "loss": 0.5554,
      "step": 5611
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8194738401249654,
      "learning_rate": 1.9833411461235873e-06,
      "loss": 0.6008,
      "step": 5612
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7958188348710266,
      "learning_rate": 1.9816959349687777e-06,
      "loss": 0.6011,
      "step": 5613
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9097094828494999,
      "learning_rate": 1.9800512377997053e-06,
      "loss": 0.5062,
      "step": 5614
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8362254123356185,
      "learning_rate": 1.9784070548964463e-06,
      "loss": 0.6502,
      "step": 5615
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6584584796883908,
      "learning_rate": 1.9767633865389864e-06,
      "loss": 0.5307,
      "step": 5616
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6677739618748387,
      "learning_rate": 1.975120233007226e-06,
      "loss": 0.4968,
      "step": 5617
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8563519816298458,
      "learning_rate": 1.973477594580977e-06,
      "loss": 0.5608,
      "step": 5618
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6777861919470805,
      "learning_rate": 1.971835471539964e-06,
      "loss": 0.5156,
      "step": 5619
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9524677825433081,
      "learning_rate": 1.9701938641638225e-06,
      "loss": 0.6285,
      "step": 5620
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7441502782741423,
      "learning_rate": 1.968552772732102e-06,
      "loss": 0.5198,
      "step": 5621
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6748969466907258,
      "learning_rate": 1.966912197524263e-06,
      "loss": 0.5212,
      "step": 5622
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.701968446008239,
      "learning_rate": 1.9652721388196776e-06,
      "loss": 0.558,
      "step": 5623
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9340363911102243,
      "learning_rate": 1.9636325968976315e-06,
      "loss": 0.6975,
      "step": 5624
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8550216187316212,
      "learning_rate": 1.9619935720373224e-06,
      "loss": 0.6497,
      "step": 5625
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6786021756697905,
      "learning_rate": 1.9603550645178577e-06,
      "loss": 0.5339,
      "step": 5626
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7556227767922549,
      "learning_rate": 1.958717074618258e-06,
      "loss": 0.563,
      "step": 5627
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.721231781529318,
      "learning_rate": 1.9570796026174576e-06,
      "loss": 0.542,
      "step": 5628
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6924333771489697,
      "learning_rate": 1.9554426487942997e-06,
      "loss": 0.5345,
      "step": 5629
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6630747797073369,
      "learning_rate": 1.953806213427541e-06,
      "loss": 0.5132,
      "step": 5630
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7456758873212999,
      "learning_rate": 1.952170296795849e-06,
      "loss": 0.5844,
      "step": 5631
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.969336563163838,
      "learning_rate": 1.950534899177801e-06,
      "loss": 0.6127,
      "step": 5632
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8299354123958859,
      "learning_rate": 1.9489000208518925e-06,
      "loss": 0.5936,
      "step": 5633
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6670978958187256,
      "learning_rate": 1.947265662096525e-06,
      "loss": 0.4823,
      "step": 5634
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7462990298055201,
      "learning_rate": 1.945631823190012e-06,
      "loss": 0.5005,
      "step": 5635
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8779756079317077,
      "learning_rate": 1.9439985044105793e-06,
      "loss": 0.6414,
      "step": 5636
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7222123319899453,
      "learning_rate": 1.9423657060363633e-06,
      "loss": 0.5052,
      "step": 5637
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6755407658938087,
      "learning_rate": 1.940733428345414e-06,
      "loss": 0.5368,
      "step": 5638
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7882931307561672,
      "learning_rate": 1.939101671615691e-06,
      "loss": 0.5438,
      "step": 5639
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7707635336120291,
      "learning_rate": 1.9374704361250644e-06,
      "loss": 0.5806,
      "step": 5640
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6867621924471001,
      "learning_rate": 1.9358397221513174e-06,
      "loss": 0.5147,
      "step": 5641
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7290548431310014,
      "learning_rate": 1.934209529972144e-06,
      "loss": 0.5321,
      "step": 5642
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6836180611078975,
      "learning_rate": 1.932579859865148e-06,
      "loss": 0.5111,
      "step": 5643
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7699511169386786,
      "learning_rate": 1.9309507121078457e-06,
      "loss": 0.6183,
      "step": 5644
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7310472148076651,
      "learning_rate": 1.9293220869776647e-06,
      "loss": 0.5365,
      "step": 5645
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.646314109956493,
      "learning_rate": 1.9276939847519416e-06,
      "loss": 0.5234,
      "step": 5646
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7498776915702167,
      "learning_rate": 1.926066405707926e-06,
      "loss": 0.5675,
      "step": 5647
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7276113774197578,
      "learning_rate": 1.924439350122778e-06,
      "loss": 0.5472,
      "step": 5648
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7224335701416914,
      "learning_rate": 1.922812818273567e-06,
      "loss": 0.5524,
      "step": 5649
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7038676456571389,
      "learning_rate": 1.921186810437275e-06,
      "loss": 0.5461,
      "step": 5650
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7120893238045858,
      "learning_rate": 1.919561326890795e-06,
      "loss": 0.5564,
      "step": 5651
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9231313078999004,
      "learning_rate": 1.9179363679109286e-06,
      "loss": 0.5981,
      "step": 5652
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6492092364277938,
      "learning_rate": 1.9163119337743903e-06,
      "loss": 0.4958,
      "step": 5653
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8497709413425779,
      "learning_rate": 1.914688024757804e-06,
      "loss": 0.6097,
      "step": 5654
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6825637718435527,
      "learning_rate": 1.9130646411377046e-06,
      "loss": 0.572,
      "step": 5655
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6366616117666464,
      "learning_rate": 1.9114417831905376e-06,
      "loss": 0.4942,
      "step": 5656
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6868144409208246,
      "learning_rate": 1.9098194511926583e-06,
      "loss": 0.5264,
      "step": 5657
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7174046254360706,
      "learning_rate": 1.9081976454203317e-06,
      "loss": 0.5613,
      "step": 5658
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8514493739215985,
      "learning_rate": 1.9065763661497371e-06,
      "loss": 0.5912,
      "step": 5659
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6692516374925389,
      "learning_rate": 1.9049556136569609e-06,
      "loss": 0.5523,
      "step": 5660
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8433657183931635,
      "learning_rate": 1.9033353882179995e-06,
      "loss": 0.667,
      "step": 5661
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.736451547530937,
      "learning_rate": 1.9017156901087608e-06,
      "loss": 0.5108,
      "step": 5662
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7895222103376313,
      "learning_rate": 1.9000965196050619e-06,
      "loss": 0.5521,
      "step": 5663
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7386615692539416,
      "learning_rate": 1.8984778769826318e-06,
      "loss": 0.5559,
      "step": 5664
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6428037286901456,
      "learning_rate": 1.8968597625171076e-06,
      "loss": 0.4365,
      "step": 5665
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8420620419361232,
      "learning_rate": 1.8952421764840374e-06,
      "loss": 0.5757,
      "step": 5666
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7627420573034267,
      "learning_rate": 1.8936251191588779e-06,
      "loss": 0.6409,
      "step": 5667
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0303500950206863,
      "learning_rate": 1.892008590817e-06,
      "loss": 0.6819,
      "step": 5668
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8173379566460758,
      "learning_rate": 1.8903925917336807e-06,
      "loss": 0.5641,
      "step": 5669
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8963285199313336,
      "learning_rate": 1.8887771221841073e-06,
      "loss": 0.6493,
      "step": 5670
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6614650994531897,
      "learning_rate": 1.8871621824433772e-06,
      "loss": 0.5373,
      "step": 5671
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7086523071473418,
      "learning_rate": 1.8855477727864985e-06,
      "loss": 0.544,
      "step": 5672
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6693291638232884,
      "learning_rate": 1.8839338934883877e-06,
      "loss": 0.5207,
      "step": 5673
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6381550060212869,
      "learning_rate": 1.8823205448238718e-06,
      "loss": 0.5446,
      "step": 5674
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7632165870249248,
      "learning_rate": 1.8807077270676877e-06,
      "loss": 0.5499,
      "step": 5675
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6845272769349781,
      "learning_rate": 1.879095440494481e-06,
      "loss": 0.5752,
      "step": 5676
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.78280223419142,
      "learning_rate": 1.877483685378807e-06,
      "loss": 0.5965,
      "step": 5677
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8671549301470479,
      "learning_rate": 1.8758724619951318e-06,
      "loss": 0.6622,
      "step": 5678
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7945552051326982,
      "learning_rate": 1.8742617706178294e-06,
      "loss": 0.5273,
      "step": 5679
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6470168702476706,
      "learning_rate": 1.8726516115211834e-06,
      "loss": 0.5889,
      "step": 5680
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7128489351494735,
      "learning_rate": 1.8710419849793876e-06,
      "loss": 0.5777,
      "step": 5681
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6499759199787295,
      "learning_rate": 1.869432891266545e-06,
      "loss": 0.5316,
      "step": 5682
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7553758782877432,
      "learning_rate": 1.8678243306566662e-06,
      "loss": 0.5493,
      "step": 5683
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7197721144686456,
      "learning_rate": 1.8662163034236736e-06,
      "loss": 0.6589,
      "step": 5684
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6913415922646856,
      "learning_rate": 1.8646088098413972e-06,
      "loss": 0.5104,
      "step": 5685
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8584374625102356,
      "learning_rate": 1.8630018501835757e-06,
      "loss": 0.5788,
      "step": 5686
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8061167069357875,
      "learning_rate": 1.8613954247238587e-06,
      "loss": 0.5948,
      "step": 5687
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9396029534574472,
      "learning_rate": 1.8597895337358025e-06,
      "loss": 0.6431,
      "step": 5688
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9824609418155691,
      "learning_rate": 1.8581841774928744e-06,
      "loss": 0.6141,
      "step": 5689
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8989922645950449,
      "learning_rate": 1.856579356268448e-06,
      "loss": 0.6437,
      "step": 5690
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8252181128944357,
      "learning_rate": 1.854975070335811e-06,
      "loss": 0.689,
      "step": 5691
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6453941500709436,
      "learning_rate": 1.8533713199681552e-06,
      "loss": 0.5137,
      "step": 5692
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7042843962620681,
      "learning_rate": 1.8517681054385816e-06,
      "loss": 0.5498,
      "step": 5693
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7308239953310444,
      "learning_rate": 1.8501654270201026e-06,
      "loss": 0.6134,
      "step": 5694
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0543878889686091,
      "learning_rate": 1.8485632849856355e-06,
      "loss": 0.66,
      "step": 5695
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8490842457356459,
      "learning_rate": 1.8469616796080103e-06,
      "loss": 0.6255,
      "step": 5696
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6715084651778989,
      "learning_rate": 1.8453606111599625e-06,
      "loss": 0.5309,
      "step": 5697
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8075334364885477,
      "learning_rate": 1.8437600799141387e-06,
      "loss": 0.5936,
      "step": 5698
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7190862689231354,
      "learning_rate": 1.8421600861430911e-06,
      "loss": 0.5231,
      "step": 5699
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.774945186795371,
      "learning_rate": 1.8405606301192825e-06,
      "loss": 0.6023,
      "step": 5700
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7826488562850418,
      "learning_rate": 1.8389617121150844e-06,
      "loss": 0.4893,
      "step": 5701
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6639905414765804,
      "learning_rate": 1.8373633324027745e-06,
      "loss": 0.556,
      "step": 5702
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7772825976539545,
      "learning_rate": 1.8357654912545414e-06,
      "loss": 0.504,
      "step": 5703
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8717133441858241,
      "learning_rate": 1.8341681889424806e-06,
      "loss": 0.6046,
      "step": 5704
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6533406953403205,
      "learning_rate": 1.8325714257385956e-06,
      "loss": 0.4838,
      "step": 5705
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8867190079623544,
      "learning_rate": 1.8309752019147986e-06,
      "loss": 0.5165,
      "step": 5706
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6930943075971878,
      "learning_rate": 1.8293795177429103e-06,
      "loss": 0.5758,
      "step": 5707
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8771940293318368,
      "learning_rate": 1.8277843734946588e-06,
      "loss": 0.5621,
      "step": 5708
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8204142256420287,
      "learning_rate": 1.8261897694416807e-06,
      "loss": 0.5392,
      "step": 5709
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.853316182461607,
      "learning_rate": 1.8245957058555203e-06,
      "loss": 0.6382,
      "step": 5710
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9145768400884016,
      "learning_rate": 1.82300218300763e-06,
      "loss": 0.6386,
      "step": 5711
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7537921452643934,
      "learning_rate": 1.8214092011693702e-06,
      "loss": 0.5485,
      "step": 5712
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8968244608196801,
      "learning_rate": 1.8198167606120092e-06,
      "loss": 0.6039,
      "step": 5713
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6181273218749963,
      "learning_rate": 1.8182248616067228e-06,
      "loss": 0.5118,
      "step": 5714
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7146249506856605,
      "learning_rate": 1.8166335044245953e-06,
      "loss": 0.4944,
      "step": 5715
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.64756477037695,
      "learning_rate": 1.8150426893366157e-06,
      "loss": 0.5466,
      "step": 5716
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6875259049621283,
      "learning_rate": 1.8134524166136875e-06,
      "loss": 0.5196,
      "step": 5717
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8439288454434887,
      "learning_rate": 1.8118626865266153e-06,
      "loss": 0.5971,
      "step": 5718
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.012448265122255,
      "learning_rate": 1.8102734993461141e-06,
      "loss": 0.6801,
      "step": 5719
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0681786169429894,
      "learning_rate": 1.8086848553428054e-06,
      "loss": 0.5871,
      "step": 5720
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6564929858831655,
      "learning_rate": 1.8070967547872192e-06,
      "loss": 0.4878,
      "step": 5721
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8320255830579747,
      "learning_rate": 1.8055091979497924e-06,
      "loss": 0.5218,
      "step": 5722
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9695415907881699,
      "learning_rate": 1.8039221851008686e-06,
      "loss": 0.635,
      "step": 5723
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1058337372771763,
      "learning_rate": 1.8023357165107008e-06,
      "loss": 0.6417,
      "step": 5724
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7997477782431707,
      "learning_rate": 1.800749792449447e-06,
      "loss": 0.6322,
      "step": 5725
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8079857867508752,
      "learning_rate": 1.7991644131871744e-06,
      "loss": 0.5932,
      "step": 5726
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0541914978638376,
      "learning_rate": 1.7975795789938561e-06,
      "loss": 0.6307,
      "step": 5727
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8555694261747371,
      "learning_rate": 1.7959952901393724e-06,
      "loss": 0.6441,
      "step": 5728
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6775536377373326,
      "learning_rate": 1.7944115468935119e-06,
      "loss": 0.4459,
      "step": 5729
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8360650694277018,
      "learning_rate": 1.7928283495259696e-06,
      "loss": 0.6437,
      "step": 5730
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7423201172380518,
      "learning_rate": 1.7912456983063475e-06,
      "loss": 0.5349,
      "step": 5731
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7736263417287377,
      "learning_rate": 1.789663593504154e-06,
      "loss": 0.5603,
      "step": 5732
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7030553003782025,
      "learning_rate": 1.7880820353888057e-06,
      "loss": 0.5337,
      "step": 5733
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8423427223986165,
      "learning_rate": 1.7865010242296254e-06,
      "loss": 0.5849,
      "step": 5734
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8761016272769074,
      "learning_rate": 1.7849205602958424e-06,
      "loss": 0.6388,
      "step": 5735
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7893869665333877,
      "learning_rate": 1.7833406438565937e-06,
      "loss": 0.6712,
      "step": 5736
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9779568753960978,
      "learning_rate": 1.781761275180922e-06,
      "loss": 0.5843,
      "step": 5737
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9062874964850393,
      "learning_rate": 1.7801824545377783e-06,
      "loss": 0.6251,
      "step": 5738
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8577072889913769,
      "learning_rate": 1.7786041821960188e-06,
      "loss": 0.6355,
      "step": 5739
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5446367191344814,
      "learning_rate": 1.7770264584244069e-06,
      "loss": 0.6138,
      "step": 5740
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8252681625676862,
      "learning_rate": 1.7754492834916121e-06,
      "loss": 0.5407,
      "step": 5741
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7339013238026564,
      "learning_rate": 1.7738726576662112e-06,
      "loss": 0.4918,
      "step": 5742
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8831005128725294,
      "learning_rate": 1.7722965812166854e-06,
      "loss": 0.6366,
      "step": 5743
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7612358345724639,
      "learning_rate": 1.7707210544114278e-06,
      "loss": 0.5482,
      "step": 5744
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6624729109266706,
      "learning_rate": 1.769146077518732e-06,
      "loss": 0.5508,
      "step": 5745
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8404092433164742,
      "learning_rate": 1.7675716508068003e-06,
      "loss": 0.6385,
      "step": 5746
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8979638683332845,
      "learning_rate": 1.7659977745437417e-06,
      "loss": 0.6972,
      "step": 5747
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6900946937007272,
      "learning_rate": 1.7644244489975704e-06,
      "loss": 0.5708,
      "step": 5748
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6618594413151697,
      "learning_rate": 1.7628516744362073e-06,
      "loss": 0.5279,
      "step": 5749
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6863226774325621,
      "learning_rate": 1.7612794511274794e-06,
      "loss": 0.5272,
      "step": 5750
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6169892224350209,
      "learning_rate": 1.7597077793391205e-06,
      "loss": 0.4689,
      "step": 5751
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6776845520065562,
      "learning_rate": 1.75813665933877e-06,
      "loss": 0.5045,
      "step": 5752
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0031560179275587,
      "learning_rate": 1.7565660913939726e-06,
      "loss": 0.7,
      "step": 5753
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7101435983485632,
      "learning_rate": 1.7549960757721802e-06,
      "loss": 0.5284,
      "step": 5754
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.727339995607302,
      "learning_rate": 1.7534266127407496e-06,
      "loss": 0.559,
      "step": 5755
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8269970566953763,
      "learning_rate": 1.7518577025669441e-06,
      "loss": 0.5777,
      "step": 5756
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6925025714193013,
      "learning_rate": 1.7502893455179331e-06,
      "loss": 0.6489,
      "step": 5757
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1668924560953013,
      "learning_rate": 1.7487215418607917e-06,
      "loss": 0.6427,
      "step": 5758
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6083411271345088,
      "learning_rate": 1.7471542918625e-06,
      "loss": 0.4806,
      "step": 5759
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7524038320361798,
      "learning_rate": 1.7455875957899443e-06,
      "loss": 0.5043,
      "step": 5760
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6976115314462221,
      "learning_rate": 1.744021453909917e-06,
      "loss": 0.5355,
      "step": 5761
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6862618541635623,
      "learning_rate": 1.7424558664891156e-06,
      "loss": 0.6099,
      "step": 5762
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6536895970040975,
      "learning_rate": 1.7408908337941432e-06,
      "loss": 0.5352,
      "step": 5763
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6439201376713182,
      "learning_rate": 1.7393263560915085e-06,
      "loss": 0.5349,
      "step": 5764
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7055432760920843,
      "learning_rate": 1.7377624336476256e-06,
      "loss": 0.5227,
      "step": 5765
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8276041983160516,
      "learning_rate": 1.7361990667288153e-06,
      "loss": 0.5727,
      "step": 5766
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.653954611591255,
      "learning_rate": 1.7346362556013013e-06,
      "loss": 0.5261,
      "step": 5767
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8435987621458022,
      "learning_rate": 1.7330740005312151e-06,
      "loss": 0.615,
      "step": 5768
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8791167489564625,
      "learning_rate": 1.7315123017845902e-06,
      "loss": 0.6315,
      "step": 5769
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7381109354667061,
      "learning_rate": 1.729951159627371e-06,
      "loss": 0.526,
      "step": 5770
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6695472719059816,
      "learning_rate": 1.728390574325402e-06,
      "loss": 0.5532,
      "step": 5771
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6350351883777356,
      "learning_rate": 1.7268305461444352e-06,
      "loss": 0.5114,
      "step": 5772
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7082825651756639,
      "learning_rate": 1.7252710753501268e-06,
      "loss": 0.5876,
      "step": 5773
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.838483890857876,
      "learning_rate": 1.7237121622080383e-06,
      "loss": 0.6192,
      "step": 5774
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8239243086931671,
      "learning_rate": 1.7221538069836347e-06,
      "loss": 0.545,
      "step": 5775
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7780272279761498,
      "learning_rate": 1.7205960099422908e-06,
      "loss": 0.5177,
      "step": 5776
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6732864942160461,
      "learning_rate": 1.7190387713492824e-06,
      "loss": 0.5144,
      "step": 5777
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8146989571067235,
      "learning_rate": 1.71748209146979e-06,
      "loss": 0.5332,
      "step": 5778
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8156694451793075,
      "learning_rate": 1.7159259705689003e-06,
      "loss": 0.6011,
      "step": 5779
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6732985813150366,
      "learning_rate": 1.7143704089116049e-06,
      "loss": 0.5575,
      "step": 5780
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6638620859450522,
      "learning_rate": 1.7128154067627995e-06,
      "loss": 0.5308,
      "step": 5781
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.854219047096941,
      "learning_rate": 1.7112609643872845e-06,
      "loss": 0.6044,
      "step": 5782
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8103742896785017,
      "learning_rate": 1.7097070820497653e-06,
      "loss": 0.576,
      "step": 5783
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6812733400886559,
      "learning_rate": 1.7081537600148523e-06,
      "loss": 0.5203,
      "step": 5784
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6586724672293538,
      "learning_rate": 1.7066009985470593e-06,
      "loss": 0.5031,
      "step": 5785
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8067857053702998,
      "learning_rate": 1.7050487979108065e-06,
      "loss": 0.6057,
      "step": 5786
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.758185917171002,
      "learning_rate": 1.7034971583704163e-06,
      "loss": 0.6087,
      "step": 5787
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6078008010678202,
      "learning_rate": 1.7019460801901179e-06,
      "loss": 0.5376,
      "step": 5788
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6386876011916381,
      "learning_rate": 1.7003955636340424e-06,
      "loss": 0.5219,
      "step": 5789
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7789709881968113,
      "learning_rate": 1.698845608966228e-06,
      "loss": 0.5509,
      "step": 5790
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9005609015028593,
      "learning_rate": 1.6972962164506153e-06,
      "loss": 0.55,
      "step": 5791
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8565180971952367,
      "learning_rate": 1.6957473863510493e-06,
      "loss": 0.6239,
      "step": 5792
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7091351922628039,
      "learning_rate": 1.6941991189312795e-06,
      "loss": 0.5219,
      "step": 5793
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.020578913620658,
      "learning_rate": 1.6926514144549611e-06,
      "loss": 0.6762,
      "step": 5794
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6613993607452004,
      "learning_rate": 1.6911042731856503e-06,
      "loss": 0.5246,
      "step": 5795
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7469482789831973,
      "learning_rate": 1.6895576953868104e-06,
      "loss": 0.5373,
      "step": 5796
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7846629958867347,
      "learning_rate": 1.6880116813218072e-06,
      "loss": 0.6102,
      "step": 5797
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6577856588113831,
      "learning_rate": 1.6864662312539105e-06,
      "loss": 0.5333,
      "step": 5798
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6675088541919377,
      "learning_rate": 1.6849213454462942e-06,
      "loss": 0.5055,
      "step": 5799
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6759524192481948,
      "learning_rate": 1.6833770241620367e-06,
      "loss": 0.5166,
      "step": 5800
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6859695600609146,
      "learning_rate": 1.681833267664118e-06,
      "loss": 0.5366,
      "step": 5801
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8344824306968963,
      "learning_rate": 1.6802900762154268e-06,
      "loss": 0.5418,
      "step": 5802
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6873472538429511,
      "learning_rate": 1.6787474500787515e-06,
      "loss": 0.5015,
      "step": 5803
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6886186064694745,
      "learning_rate": 1.6772053895167845e-06,
      "loss": 0.5161,
      "step": 5804
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8980129813270382,
      "learning_rate": 1.6756638947921234e-06,
      "loss": 0.5764,
      "step": 5805
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9400516390856912,
      "learning_rate": 1.6741229661672681e-06,
      "loss": 0.6248,
      "step": 5806
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8740214578619746,
      "learning_rate": 1.6725826039046233e-06,
      "loss": 0.6338,
      "step": 5807
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7322193819624436,
      "learning_rate": 1.6710428082664959e-06,
      "loss": 0.4966,
      "step": 5808
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7879048999557077,
      "learning_rate": 1.6695035795150977e-06,
      "loss": 0.6082,
      "step": 5809
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9262489171718701,
      "learning_rate": 1.667964917912543e-06,
      "loss": 0.6156,
      "step": 5810
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.868157802716121,
      "learning_rate": 1.6664268237208496e-06,
      "loss": 0.6124,
      "step": 5811
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9629571231606969,
      "learning_rate": 1.6648892972019398e-06,
      "loss": 0.6101,
      "step": 5812
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.630257986452402,
      "learning_rate": 1.663352338617637e-06,
      "loss": 0.5182,
      "step": 5813
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7725192520495178,
      "learning_rate": 1.6618159482296704e-06,
      "loss": 0.5162,
      "step": 5814
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7957401317794008,
      "learning_rate": 1.6602801262996709e-06,
      "loss": 0.5767,
      "step": 5815
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6924976568408082,
      "learning_rate": 1.658744873089173e-06,
      "loss": 0.5333,
      "step": 5816
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8876856529141748,
      "learning_rate": 1.6572101888596142e-06,
      "loss": 0.5505,
      "step": 5817
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8824769208240716,
      "learning_rate": 1.6556760738723349e-06,
      "loss": 0.6322,
      "step": 5818
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8568894790784245,
      "learning_rate": 1.6541425283885798e-06,
      "loss": 0.6057,
      "step": 5819
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.629991114696536,
      "learning_rate": 1.6526095526694953e-06,
      "loss": 0.5234,
      "step": 5820
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8710601492173604,
      "learning_rate": 1.6510771469761306e-06,
      "loss": 0.5908,
      "step": 5821
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.78861392849983,
      "learning_rate": 1.6495453115694393e-06,
      "loss": 0.6233,
      "step": 5822
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6002065357793965,
      "learning_rate": 1.6480140467102768e-06,
      "loss": 0.5008,
      "step": 5823
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.740915385089809,
      "learning_rate": 1.646483352659401e-06,
      "loss": 0.5474,
      "step": 5824
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8145236738881111,
      "learning_rate": 1.6449532296774739e-06,
      "loss": 0.5498,
      "step": 5825
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7441417133206782,
      "learning_rate": 1.643423678025059e-06,
      "loss": 0.5837,
      "step": 5826
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7976850832151852,
      "learning_rate": 1.6418946979626216e-06,
      "loss": 0.5967,
      "step": 5827
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8760592741905344,
      "learning_rate": 1.640366289750534e-06,
      "loss": 0.5785,
      "step": 5828
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6810828055330408,
      "learning_rate": 1.638838453649067e-06,
      "loss": 0.5136,
      "step": 5829
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.737418911972885,
      "learning_rate": 1.6373111899183953e-06,
      "loss": 0.5738,
      "step": 5830
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1547721275838845,
      "learning_rate": 1.6357844988185957e-06,
      "loss": 0.5899,
      "step": 5831
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8466645058848514,
      "learning_rate": 1.6342583806096473e-06,
      "loss": 0.6259,
      "step": 5832
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9192298182706128,
      "learning_rate": 1.6327328355514327e-06,
      "loss": 0.593,
      "step": 5833
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7362650260598587,
      "learning_rate": 1.6312078639037365e-06,
      "loss": 0.6052,
      "step": 5834
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7216085822459143,
      "learning_rate": 1.6296834659262457e-06,
      "loss": 0.4819,
      "step": 5835
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9298581391264136,
      "learning_rate": 1.6281596418785485e-06,
      "loss": 0.6282,
      "step": 5836
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7903572998676937,
      "learning_rate": 1.6266363920201366e-06,
      "loss": 0.5822,
      "step": 5837
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7132577999422386,
      "learning_rate": 1.6251137166104036e-06,
      "loss": 0.5308,
      "step": 5838
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.813344706262403,
      "learning_rate": 1.6235916159086457e-06,
      "loss": 0.5209,
      "step": 5839
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7344273276920951,
      "learning_rate": 1.6220700901740604e-06,
      "loss": 0.5567,
      "step": 5840
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7681406061228656,
      "learning_rate": 1.6205491396657474e-06,
      "loss": 0.515,
      "step": 5841
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7275369113520247,
      "learning_rate": 1.6190287646427094e-06,
      "loss": 0.5396,
      "step": 5842
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8913274326521137,
      "learning_rate": 1.6175089653638504e-06,
      "loss": 0.6263,
      "step": 5843
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8751026922070858,
      "learning_rate": 1.6159897420879755e-06,
      "loss": 0.6564,
      "step": 5844
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7515926410062148,
      "learning_rate": 1.6144710950737935e-06,
      "loss": 0.5512,
      "step": 5845
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7405113135735025,
      "learning_rate": 1.6129530245799136e-06,
      "loss": 0.612,
      "step": 5846
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.666004061993584,
      "learning_rate": 1.6114355308648472e-06,
      "loss": 0.5377,
      "step": 5847
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8863925692265253,
      "learning_rate": 1.6099186141870088e-06,
      "loss": 0.633,
      "step": 5848
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9424480768504067,
      "learning_rate": 1.6084022748047123e-06,
      "loss": 0.6714,
      "step": 5849
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9512953886730859,
      "learning_rate": 1.6068865129761752e-06,
      "loss": 0.6288,
      "step": 5850
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6975746137973988,
      "learning_rate": 1.6053713289595158e-06,
      "loss": 0.5315,
      "step": 5851
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1515504508982928,
      "learning_rate": 1.6038567230127545e-06,
      "loss": 0.6141,
      "step": 5852
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.826371570181474,
      "learning_rate": 1.6023426953938115e-06,
      "loss": 0.6346,
      "step": 5853
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6818203518131212,
      "learning_rate": 1.6008292463605101e-06,
      "loss": 0.5213,
      "step": 5854
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6759748798150119,
      "learning_rate": 1.5993163761705772e-06,
      "loss": 0.522,
      "step": 5855
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8699457605382849,
      "learning_rate": 1.5978040850816373e-06,
      "loss": 0.6489,
      "step": 5856
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7831870937471817,
      "learning_rate": 1.5962923733512176e-06,
      "loss": 0.5737,
      "step": 5857
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8874041282179453,
      "learning_rate": 1.5947812412367474e-06,
      "loss": 0.6549,
      "step": 5858
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7027488829052079,
      "learning_rate": 1.5932706889955568e-06,
      "loss": 0.5686,
      "step": 5859
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8169543931240841,
      "learning_rate": 1.5917607168848764e-06,
      "loss": 0.6414,
      "step": 5860
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.805539168386371,
      "learning_rate": 1.5902513251618396e-06,
      "loss": 0.5789,
      "step": 5861
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7464710005873254,
      "learning_rate": 1.5887425140834789e-06,
      "loss": 0.5739,
      "step": 5862
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6791323714145837,
      "learning_rate": 1.5872342839067305e-06,
      "loss": 0.5299,
      "step": 5863
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7407387624007086,
      "learning_rate": 1.5857266348884293e-06,
      "loss": 0.5183,
      "step": 5864
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8076478341283453,
      "learning_rate": 1.5842195672853128e-06,
      "loss": 0.5978,
      "step": 5865
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.845242395335814,
      "learning_rate": 1.582713081354018e-06,
      "loss": 0.6081,
      "step": 5866
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7530378201272134,
      "learning_rate": 1.581207177351085e-06,
      "loss": 0.5156,
      "step": 5867
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8217957689627217,
      "learning_rate": 1.5797018555329529e-06,
      "loss": 0.6061,
      "step": 5868
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7530159261899385,
      "learning_rate": 1.5781971161559622e-06,
      "loss": 0.5697,
      "step": 5869
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9183465822912796,
      "learning_rate": 1.576692959476353e-06,
      "loss": 0.6437,
      "step": 5870
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6952600641197307,
      "learning_rate": 1.575189385750271e-06,
      "loss": 0.5488,
      "step": 5871
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7156962069325696,
      "learning_rate": 1.5736863952337567e-06,
      "loss": 0.5772,
      "step": 5872
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8216101625083886,
      "learning_rate": 1.5721839881827545e-06,
      "loss": 0.5956,
      "step": 5873
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6596400202999854,
      "learning_rate": 1.5706821648531078e-06,
      "loss": 0.5409,
      "step": 5874
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8283781922036126,
      "learning_rate": 1.569180925500563e-06,
      "loss": 0.6615,
      "step": 5875
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7008414768364053,
      "learning_rate": 1.5676802703807636e-06,
      "loss": 0.5372,
      "step": 5876
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7706524995832714,
      "learning_rate": 1.566180199749257e-06,
      "loss": 0.4792,
      "step": 5877
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7168011977678853,
      "learning_rate": 1.5646807138614895e-06,
      "loss": 0.5409,
      "step": 5878
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7735466749437391,
      "learning_rate": 1.5631818129728072e-06,
      "loss": 0.6144,
      "step": 5879
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6971762911929578,
      "learning_rate": 1.5616834973384576e-06,
      "loss": 0.5534,
      "step": 5880
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8229319732104531,
      "learning_rate": 1.5601857672135883e-06,
      "loss": 0.5871,
      "step": 5881
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6943147442626812,
      "learning_rate": 1.5586886228532472e-06,
      "loss": 0.5531,
      "step": 5882
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6520514089849233,
      "learning_rate": 1.5571920645123818e-06,
      "loss": 0.5313,
      "step": 5883
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.822397362966487,
      "learning_rate": 1.5556960924458415e-06,
      "loss": 0.5891,
      "step": 5884
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8114589771046263,
      "learning_rate": 1.5542007069083737e-06,
      "loss": 0.6081,
      "step": 5885
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8840467383353776,
      "learning_rate": 1.5527059081546263e-06,
      "loss": 0.6484,
      "step": 5886
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6515757287267319,
      "learning_rate": 1.5512116964391505e-06,
      "loss": 0.5125,
      "step": 5887
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.045024667893443,
      "learning_rate": 1.5497180720163936e-06,
      "loss": 0.6362,
      "step": 5888
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7897407381559436,
      "learning_rate": 1.5482250351407042e-06,
      "loss": 0.5537,
      "step": 5889
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8069402945209845,
      "learning_rate": 1.5467325860663307e-06,
      "loss": 0.6056,
      "step": 5890
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7421858959320319,
      "learning_rate": 1.5452407250474223e-06,
      "loss": 0.5444,
      "step": 5891
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7855589233418996,
      "learning_rate": 1.5437494523380269e-06,
      "loss": 0.6613,
      "step": 5892
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8890677963066336,
      "learning_rate": 1.542258768192093e-06,
      "loss": 0.5946,
      "step": 5893
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8994250368885927,
      "learning_rate": 1.540768672863468e-06,
      "loss": 0.5169,
      "step": 5894
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6814967753869515,
      "learning_rate": 1.5392791666059003e-06,
      "loss": 0.5616,
      "step": 5895
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8906932723136545,
      "learning_rate": 1.5377902496730367e-06,
      "loss": 0.6279,
      "step": 5896
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9241740406517359,
      "learning_rate": 1.536301922318425e-06,
      "loss": 0.6415,
      "step": 5897
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8572185864943535,
      "learning_rate": 1.5348141847955112e-06,
      "loss": 0.6624,
      "step": 5898
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8152249955762612,
      "learning_rate": 1.5333270373576414e-06,
      "loss": 0.6076,
      "step": 5899
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7544383590950826,
      "learning_rate": 1.5318404802580617e-06,
      "loss": 0.5318,
      "step": 5900
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6936164892634805,
      "learning_rate": 1.5303545137499177e-06,
      "loss": 0.5508,
      "step": 5901
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7047536383014844,
      "learning_rate": 1.528869138086253e-06,
      "loss": 0.546,
      "step": 5902
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9822219715093419,
      "learning_rate": 1.5273843535200122e-06,
      "loss": 0.663,
      "step": 5903
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7570505699788797,
      "learning_rate": 1.5259001603040385e-06,
      "loss": 0.5377,
      "step": 5904
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0753220874579519,
      "learning_rate": 1.5244165586910748e-06,
      "loss": 0.6086,
      "step": 5905
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.759454448041383,
      "learning_rate": 1.5229335489337628e-06,
      "loss": 0.6145,
      "step": 5906
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6250259531135143,
      "learning_rate": 1.5214511312846431e-06,
      "loss": 0.5148,
      "step": 5907
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7600710998725176,
      "learning_rate": 1.5199693059961567e-06,
      "loss": 0.5607,
      "step": 5908
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7350211532263026,
      "learning_rate": 1.518488073320643e-06,
      "loss": 0.516,
      "step": 5909
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8691960863679342,
      "learning_rate": 1.5170074335103401e-06,
      "loss": 0.6209,
      "step": 5910
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6507197621832734,
      "learning_rate": 1.515527386817386e-06,
      "loss": 0.5324,
      "step": 5911
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7250007694966768,
      "learning_rate": 1.5140479334938153e-06,
      "loss": 0.4998,
      "step": 5912
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9191419850429818,
      "learning_rate": 1.5125690737915666e-06,
      "loss": 0.6246,
      "step": 5913
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6594747996901251,
      "learning_rate": 1.5110908079624726e-06,
      "loss": 0.4985,
      "step": 5914
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7284651782080697,
      "learning_rate": 1.5096131362582673e-06,
      "loss": 0.5401,
      "step": 5915
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6094309389910929,
      "learning_rate": 1.508136058930582e-06,
      "loss": 0.5135,
      "step": 5916
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7395913441951243,
      "learning_rate": 1.5066595762309478e-06,
      "loss": 0.5096,
      "step": 5917
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5903688396352945,
      "learning_rate": 1.5051836884107946e-06,
      "loss": 0.5078,
      "step": 5918
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8216335941071214,
      "learning_rate": 1.5037083957214505e-06,
      "loss": 0.64,
      "step": 5919
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9325850357672525,
      "learning_rate": 1.5022336984141428e-06,
      "loss": 0.6432,
      "step": 5920
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9158450409814138,
      "learning_rate": 1.5007595967399962e-06,
      "loss": 0.606,
      "step": 5921
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8026594844365942,
      "learning_rate": 1.4992860909500357e-06,
      "loss": 0.5377,
      "step": 5922
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8382445201843135,
      "learning_rate": 1.4978131812951841e-06,
      "loss": 0.6215,
      "step": 5923
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7805690601597003,
      "learning_rate": 1.4963408680262614e-06,
      "loss": 0.5586,
      "step": 5924
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7354570274878476,
      "learning_rate": 1.494869151393989e-06,
      "loss": 0.493,
      "step": 5925
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9198073858996956,
      "learning_rate": 1.4933980316489832e-06,
      "loss": 0.6305,
      "step": 5926
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6319337327826817,
      "learning_rate": 1.4919275090417613e-06,
      "loss": 0.5297,
      "step": 5927
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7483555945780317,
      "learning_rate": 1.490457583822738e-06,
      "loss": 0.5809,
      "step": 5928
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8055112074954867,
      "learning_rate": 1.4889882562422258e-06,
      "loss": 0.6433,
      "step": 5929
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6800242159719299,
      "learning_rate": 1.4875195265504362e-06,
      "loss": 0.5297,
      "step": 5930
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6365067854231993,
      "learning_rate": 1.486051394997478e-06,
      "loss": 0.5221,
      "step": 5931
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7467987320092495,
      "learning_rate": 1.4845838618333597e-06,
      "loss": 0.5397,
      "step": 5932
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.855320509955299,
      "learning_rate": 1.4831169273079866e-06,
      "loss": 0.6086,
      "step": 5933
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6708267092816687,
      "learning_rate": 1.4816505916711615e-06,
      "loss": 0.5175,
      "step": 5934
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8761132120671505,
      "learning_rate": 1.4801848551725872e-06,
      "loss": 0.6094,
      "step": 5935
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7997342589857858,
      "learning_rate": 1.478719718061863e-06,
      "loss": 0.5497,
      "step": 5936
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7381074437220135,
      "learning_rate": 1.477255180588486e-06,
      "loss": 0.6094,
      "step": 5937
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8855118778747625,
      "learning_rate": 1.4757912430018506e-06,
      "loss": 0.6141,
      "step": 5938
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6947700741175422,
      "learning_rate": 1.4743279055512532e-06,
      "loss": 0.5302,
      "step": 5939
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.869442066721817,
      "learning_rate": 1.4728651684858835e-06,
      "loss": 0.6398,
      "step": 5940
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7379399231132954,
      "learning_rate": 1.4714030320548301e-06,
      "loss": 0.5057,
      "step": 5941
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7324847319394786,
      "learning_rate": 1.469941496507079e-06,
      "loss": 0.553,
      "step": 5942
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8215193094263105,
      "learning_rate": 1.468480562091516e-06,
      "loss": 0.6425,
      "step": 5943
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9034475166490202,
      "learning_rate": 1.4670202290569218e-06,
      "loss": 0.6544,
      "step": 5944
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8489302259656883,
      "learning_rate": 1.4655604976519767e-06,
      "loss": 0.6398,
      "step": 5945
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7457195425457996,
      "learning_rate": 1.4641013681252569e-06,
      "loss": 0.556,
      "step": 5946
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8095512196505711,
      "learning_rate": 1.4626428407252374e-06,
      "loss": 0.6222,
      "step": 5947
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6767086932911917,
      "learning_rate": 1.4611849157002899e-06,
      "loss": 0.5918,
      "step": 5948
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7454314079690785,
      "learning_rate": 1.4597275932986848e-06,
      "loss": 0.5301,
      "step": 5949
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7258126404629276,
      "learning_rate": 1.458270873768588e-06,
      "loss": 0.5468,
      "step": 5950
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.525755271359152,
      "learning_rate": 1.4568147573580637e-06,
      "loss": 0.6212,
      "step": 5951
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8272288146500956,
      "learning_rate": 1.4553592443150739e-06,
      "loss": 0.5482,
      "step": 5952
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8266659542980566,
      "learning_rate": 1.453904334887477e-06,
      "loss": 0.5841,
      "step": 5953
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.801343697910673,
      "learning_rate": 1.4524500293230287e-06,
      "loss": 0.5963,
      "step": 5954
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8870717941376552,
      "learning_rate": 1.4509963278693828e-06,
      "loss": 0.6297,
      "step": 5955
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8306591252887179,
      "learning_rate": 1.4495432307740885e-06,
      "loss": 0.6093,
      "step": 5956
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8138944828651896,
      "learning_rate": 1.4480907382845938e-06,
      "loss": 0.587,
      "step": 5957
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.640960443830883,
      "learning_rate": 1.4466388506482425e-06,
      "loss": 0.5209,
      "step": 5958
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7898516414805914,
      "learning_rate": 1.4451875681122767e-06,
      "loss": 0.5377,
      "step": 5959
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7042045011578976,
      "learning_rate": 1.443736890923834e-06,
      "loss": 0.51,
      "step": 5960
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.162246867809616,
      "learning_rate": 1.44228681932995e-06,
      "loss": 0.6127,
      "step": 5961
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6716421491730815,
      "learning_rate": 1.4408373535775572e-06,
      "loss": 0.5425,
      "step": 5962
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6633333032048389,
      "learning_rate": 1.4393884939134833e-06,
      "loss": 0.5727,
      "step": 5963
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6608236926032384,
      "learning_rate": 1.437940240584455e-06,
      "loss": 0.5267,
      "step": 5964
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6330701673801944,
      "learning_rate": 1.4364925938370926e-06,
      "loss": 0.5338,
      "step": 5965
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7413039913488412,
      "learning_rate": 1.4350455539179192e-06,
      "loss": 0.5799,
      "step": 5966
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7739790891707063,
      "learning_rate": 1.433599121073348e-06,
      "loss": 0.5901,
      "step": 5967
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8051481132758258,
      "learning_rate": 1.4321532955496925e-06,
      "loss": 0.6134,
      "step": 5968
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6783678526297705,
      "learning_rate": 1.430708077593161e-06,
      "loss": 0.5712,
      "step": 5969
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7877745873297851,
      "learning_rate": 1.429263467449859e-06,
      "loss": 0.6028,
      "step": 5970
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6191267615792932,
      "learning_rate": 1.427819465365789e-06,
      "loss": 0.5352,
      "step": 5971
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6849466786355246,
      "learning_rate": 1.426376071586848e-06,
      "loss": 0.5357,
      "step": 5972
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7603943447595155,
      "learning_rate": 1.4249332863588334e-06,
      "loss": 0.5669,
      "step": 5973
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7021987874269323,
      "learning_rate": 1.4234911099274357e-06,
      "loss": 0.506,
      "step": 5974
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8086945062698163,
      "learning_rate": 1.4220495425382418e-06,
      "loss": 0.624,
      "step": 5975
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7139204168377902,
      "learning_rate": 1.4206085844367361e-06,
      "loss": 0.5175,
      "step": 5976
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6419213776173835,
      "learning_rate": 1.419168235868299e-06,
      "loss": 0.514,
      "step": 5977
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7990206802172736,
      "learning_rate": 1.4177284970782063e-06,
      "loss": 0.5991,
      "step": 5978
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7537870528281434,
      "learning_rate": 1.4162893683116307e-06,
      "loss": 0.5055,
      "step": 5979
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6320668823822558,
      "learning_rate": 1.414850849813641e-06,
      "loss": 0.5349,
      "step": 5980
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8177142881926991,
      "learning_rate": 1.4134129418292013e-06,
      "loss": 0.5151,
      "step": 5981
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7432383161836048,
      "learning_rate": 1.4119756446031736e-06,
      "loss": 0.5462,
      "step": 5982
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7794214011334519,
      "learning_rate": 1.410538958380313e-06,
      "loss": 0.4998,
      "step": 5983
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9262414925714734,
      "learning_rate": 1.4091028834052739e-06,
      "loss": 0.5719,
      "step": 5984
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8293905770682239,
      "learning_rate": 1.4076674199226037e-06,
      "loss": 0.6042,
      "step": 5985
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6385102480554473,
      "learning_rate": 1.406232568176747e-06,
      "loss": 0.5557,
      "step": 5986
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6534442623344828,
      "learning_rate": 1.4047983284120442e-06,
      "loss": 0.5199,
      "step": 5987
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7950501398759092,
      "learning_rate": 1.403364700872732e-06,
      "loss": 0.6034,
      "step": 5988
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6783707693013732,
      "learning_rate": 1.4019316858029418e-06,
      "loss": 0.5231,
      "step": 5989
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6888939247448539,
      "learning_rate": 1.4004992834467006e-06,
      "loss": 0.5571,
      "step": 5990
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7163480834841157,
      "learning_rate": 1.3990674940479327e-06,
      "loss": 0.5919,
      "step": 5991
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6999541991361229,
      "learning_rate": 1.397636317850456e-06,
      "loss": 0.5646,
      "step": 5992
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7198008743943831,
      "learning_rate": 1.396205755097985e-06,
      "loss": 0.5538,
      "step": 5993
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8863364329317801,
      "learning_rate": 1.3947758060341304e-06,
      "loss": 0.5875,
      "step": 5994
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7468557020009613,
      "learning_rate": 1.3933464709023964e-06,
      "loss": 0.5804,
      "step": 5995
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8034963130296258,
      "learning_rate": 1.391917749946185e-06,
      "loss": 0.6229,
      "step": 5996
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8163806876327102,
      "learning_rate": 1.3904896434087905e-06,
      "loss": 0.5968,
      "step": 5997
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6789758301208867,
      "learning_rate": 1.3890621515334073e-06,
      "loss": 0.5049,
      "step": 5998
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7316420049627483,
      "learning_rate": 1.3876352745631216e-06,
      "loss": 0.5539,
      "step": 5999
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8103186910531326,
      "learning_rate": 1.3862090127409156e-06,
      "loss": 0.5707,
      "step": 6000
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7505970066378608,
      "learning_rate": 1.384783366309666e-06,
      "loss": 0.6098,
      "step": 6001
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9009395804911083,
      "learning_rate": 1.383358335512146e-06,
      "loss": 0.6242,
      "step": 6002
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6995677412940682,
      "learning_rate": 1.3819339205910237e-06,
      "loss": 0.5266,
      "step": 6003
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8429682614789905,
      "learning_rate": 1.3805101217888623e-06,
      "loss": 0.5928,
      "step": 6004
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9386287022319724,
      "learning_rate": 1.3790869393481193e-06,
      "loss": 0.6884,
      "step": 6005
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8655067538626704,
      "learning_rate": 1.3776643735111484e-06,
      "loss": 0.6475,
      "step": 6006
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6201921386862851,
      "learning_rate": 1.3762424245201972e-06,
      "loss": 0.529,
      "step": 6007
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9047269047415928,
      "learning_rate": 1.374821092617409e-06,
      "loss": 0.634,
      "step": 6008
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9033827818889651,
      "learning_rate": 1.3734003780448218e-06,
      "loss": 0.6544,
      "step": 6009
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6054456249163662,
      "learning_rate": 1.371980281044369e-06,
      "loss": 0.4544,
      "step": 6010
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6426130083170906,
      "learning_rate": 1.3705608018578776e-06,
      "loss": 0.5057,
      "step": 6011
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8335041603346685,
      "learning_rate": 1.3691419407270707e-06,
      "loss": 0.6079,
      "step": 6012
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8116543473206832,
      "learning_rate": 1.3677236978935648e-06,
      "loss": 0.5973,
      "step": 6013
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.14300805950097,
      "learning_rate": 1.3663060735988725e-06,
      "loss": 0.6529,
      "step": 6014
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6419717599229467,
      "learning_rate": 1.3648890680844007e-06,
      "loss": 0.5269,
      "step": 6015
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6714725660150257,
      "learning_rate": 1.3634726815914501e-06,
      "loss": 0.5533,
      "step": 6016
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6730175978283861,
      "learning_rate": 1.3620569143612166e-06,
      "loss": 0.5096,
      "step": 6017
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9308954989709286,
      "learning_rate": 1.3606417666347909e-06,
      "loss": 0.6835,
      "step": 6018
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.054057802258776,
      "learning_rate": 1.359227238653158e-06,
      "loss": 0.6338,
      "step": 6019
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8259535783866477,
      "learning_rate": 1.3578133306571966e-06,
      "loss": 0.6177,
      "step": 6020
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9192245356074534,
      "learning_rate": 1.3564000428876812e-06,
      "loss": 0.673,
      "step": 6021
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8488838025352896,
      "learning_rate": 1.3549873755852799e-06,
      "loss": 0.6068,
      "step": 6022
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6994533586336426,
      "learning_rate": 1.3535753289905535e-06,
      "loss": 0.5368,
      "step": 6023
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7897552446075208,
      "learning_rate": 1.3521639033439614e-06,
      "loss": 0.5736,
      "step": 6024
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7008637813195776,
      "learning_rate": 1.3507530988858541e-06,
      "loss": 0.5559,
      "step": 6025
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9772917942084419,
      "learning_rate": 1.3493429158564764e-06,
      "loss": 0.5947,
      "step": 6026
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6172940347657915,
      "learning_rate": 1.3479333544959682e-06,
      "loss": 0.5022,
      "step": 6027
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9065971768357095,
      "learning_rate": 1.3465244150443624e-06,
      "loss": 0.6059,
      "step": 6028
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6067229747553409,
      "learning_rate": 1.3451160977415872e-06,
      "loss": 0.5293,
      "step": 6029
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7250477177365416,
      "learning_rate": 1.3437084028274643e-06,
      "loss": 0.5155,
      "step": 6030
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8577505396725205,
      "learning_rate": 1.3423013305417093e-06,
      "loss": 0.6494,
      "step": 6031
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.897424399888634,
      "learning_rate": 1.340894881123932e-06,
      "loss": 0.6197,
      "step": 6032
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6553612072707354,
      "learning_rate": 1.3394890548136363e-06,
      "loss": 0.5366,
      "step": 6033
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.499327744828209,
      "learning_rate": 1.3380838518502198e-06,
      "loss": 0.6286,
      "step": 6034
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0300526700137789,
      "learning_rate": 1.3366792724729733e-06,
      "loss": 0.6168,
      "step": 6035
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8075927598443222,
      "learning_rate": 1.335275316921083e-06,
      "loss": 0.5699,
      "step": 6036
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6597717410366899,
      "learning_rate": 1.3338719854336275e-06,
      "loss": 0.5163,
      "step": 6037
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6768126377782689,
      "learning_rate": 1.332469278249579e-06,
      "loss": 0.5212,
      "step": 6038
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.733601860471724,
      "learning_rate": 1.3310671956078041e-06,
      "loss": 0.5743,
      "step": 6039
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8306856898122218,
      "learning_rate": 1.329665737747064e-06,
      "loss": 0.6057,
      "step": 6040
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7573699753924524,
      "learning_rate": 1.3282649049060108e-06,
      "loss": 0.4816,
      "step": 6041
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9017700823777901,
      "learning_rate": 1.3268646973231925e-06,
      "loss": 0.6299,
      "step": 6042
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7422807308761529,
      "learning_rate": 1.32546511523705e-06,
      "loss": 0.4954,
      "step": 6043
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6916212253163068,
      "learning_rate": 1.324066158885917e-06,
      "loss": 0.5329,
      "step": 6044
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7149522420996964,
      "learning_rate": 1.3226678285080224e-06,
      "loss": 0.59,
      "step": 6045
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8108642728190536,
      "learning_rate": 1.3212701243414856e-06,
      "loss": 0.565,
      "step": 6046
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6915305812833428,
      "learning_rate": 1.319873046624322e-06,
      "loss": 0.5244,
      "step": 6047
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7590747643299829,
      "learning_rate": 1.3184765955944395e-06,
      "loss": 0.6153,
      "step": 6048
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8422047896311641,
      "learning_rate": 1.3170807714896379e-06,
      "loss": 0.6248,
      "step": 6049
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6379720530900855,
      "learning_rate": 1.3156855745476133e-06,
      "loss": 0.5017,
      "step": 6050
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6845356019667418,
      "learning_rate": 1.3142910050059527e-06,
      "loss": 0.5532,
      "step": 6051
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6888578405648416,
      "learning_rate": 1.3128970631021365e-06,
      "loss": 0.5561,
      "step": 6052
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6884702522586105,
      "learning_rate": 1.3115037490735383e-06,
      "loss": 0.5815,
      "step": 6053
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8503523753587275,
      "learning_rate": 1.3101110631574255e-06,
      "loss": 0.6372,
      "step": 6054
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0259526281667457,
      "learning_rate": 1.3087190055909572e-06,
      "loss": 0.6539,
      "step": 6055
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6809943919954881,
      "learning_rate": 1.307327576611187e-06,
      "loss": 0.5552,
      "step": 6056
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.917301632475324,
      "learning_rate": 1.3059367764550607e-06,
      "loss": 0.6307,
      "step": 6057
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7059629439276569,
      "learning_rate": 1.304546605359417e-06,
      "loss": 0.5603,
      "step": 6058
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6432175903378218,
      "learning_rate": 1.303157063560987e-06,
      "loss": 0.503,
      "step": 6059
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8778449999656432,
      "learning_rate": 1.3017681512963958e-06,
      "loss": 0.6396,
      "step": 6060
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1758783907972818,
      "learning_rate": 1.3003798688021606e-06,
      "loss": 0.6056,
      "step": 6061
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8032840443700658,
      "learning_rate": 1.2989922163146913e-06,
      "loss": 0.5792,
      "step": 6062
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8573378852709852,
      "learning_rate": 1.297605194070291e-06,
      "loss": 0.5287,
      "step": 6063
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8955307700350046,
      "learning_rate": 1.2962188023051548e-06,
      "loss": 0.5728,
      "step": 6064
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7312917274314831,
      "learning_rate": 1.2948330412553707e-06,
      "loss": 0.5635,
      "step": 6065
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1079459711111121,
      "learning_rate": 1.2934479111569192e-06,
      "loss": 0.6707,
      "step": 6066
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7913647319789372,
      "learning_rate": 1.2920634122456744e-06,
      "loss": 0.587,
      "step": 6067
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7641170184102855,
      "learning_rate": 1.290679544757401e-06,
      "loss": 0.5632,
      "step": 6068
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6107530012059154,
      "learning_rate": 1.2892963089277582e-06,
      "loss": 0.5461,
      "step": 6069
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8022260519961586,
      "learning_rate": 1.2879137049922963e-06,
      "loss": 0.5554,
      "step": 6070
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8562519136056977,
      "learning_rate": 1.286531733186458e-06,
      "loss": 0.5647,
      "step": 6071
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7715188925563033,
      "learning_rate": 1.2851503937455785e-06,
      "loss": 0.5768,
      "step": 6072
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9621914822730264,
      "learning_rate": 1.283769686904885e-06,
      "loss": 0.5481,
      "step": 6073
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7536268707998498,
      "learning_rate": 1.2823896128994994e-06,
      "loss": 0.5855,
      "step": 6074
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7458044684248323,
      "learning_rate": 1.281010171964433e-06,
      "loss": 0.5403,
      "step": 6075
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7783001638164412,
      "learning_rate": 1.2796313643345904e-06,
      "loss": 0.5517,
      "step": 6076
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6916703893905545,
      "learning_rate": 1.278253190244768e-06,
      "loss": 0.5168,
      "step": 6077
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6027453925260882,
      "learning_rate": 1.276875649929654e-06,
      "loss": 0.5027,
      "step": 6078
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7189046753105177,
      "learning_rate": 1.2754987436238297e-06,
      "loss": 0.604,
      "step": 6079
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7314290936631835,
      "learning_rate": 1.2741224715617684e-06,
      "loss": 0.5448,
      "step": 6080
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9882015105342422,
      "learning_rate": 1.272746833977832e-06,
      "loss": 0.6289,
      "step": 6081
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1733530989755303,
      "learning_rate": 1.2713718311062817e-06,
      "loss": 0.5985,
      "step": 6082
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8205227298903993,
      "learning_rate": 1.2699974631812633e-06,
      "loss": 0.6079,
      "step": 6083
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8755323457447933,
      "learning_rate": 1.2686237304368188e-06,
      "loss": 0.5269,
      "step": 6084
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8397324892310829,
      "learning_rate": 1.267250633106879e-06,
      "loss": 0.619,
      "step": 6085
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8917042817078477,
      "learning_rate": 1.2658781714252693e-06,
      "loss": 0.5459,
      "step": 6086
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6471386347382809,
      "learning_rate": 1.2645063456257055e-06,
      "loss": 0.5207,
      "step": 6087
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8245716489820528,
      "learning_rate": 1.2631351559417947e-06,
      "loss": 0.6071,
      "step": 6088
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6960216053843672,
      "learning_rate": 1.2617646026070362e-06,
      "loss": 0.5307,
      "step": 6089
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7124612574024674,
      "learning_rate": 1.2603946858548215e-06,
      "loss": 0.5178,
      "step": 6090
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9009129538684394,
      "learning_rate": 1.2590254059184326e-06,
      "loss": 0.6917,
      "step": 6091
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.831602385656947,
      "learning_rate": 1.2576567630310444e-06,
      "loss": 0.5843,
      "step": 6092
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7843621436884425,
      "learning_rate": 1.2562887574257216e-06,
      "loss": 0.5481,
      "step": 6093
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6475240337273345,
      "learning_rate": 1.2549213893354217e-06,
      "loss": 0.5362,
      "step": 6094
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7665872381775558,
      "learning_rate": 1.253554658992993e-06,
      "loss": 0.5648,
      "step": 6095
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6887918095354533,
      "learning_rate": 1.2521885666311762e-06,
      "loss": 0.5227,
      "step": 6096
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.772910762175124,
      "learning_rate": 1.250823112482602e-06,
      "loss": 0.5388,
      "step": 6097
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8809272504283993,
      "learning_rate": 1.2494582967797931e-06,
      "loss": 0.6369,
      "step": 6098
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8148488024848227,
      "learning_rate": 1.2480941197551632e-06,
      "loss": 0.6084,
      "step": 6099
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8672361455040535,
      "learning_rate": 1.246730581641018e-06,
      "loss": 0.6437,
      "step": 6100
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6500809168369878,
      "learning_rate": 1.2453676826695531e-06,
      "loss": 0.4997,
      "step": 6101
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8634292658895205,
      "learning_rate": 1.2440054230728566e-06,
      "loss": 0.4912,
      "step": 6102
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9513206620762039,
      "learning_rate": 1.2426438030829069e-06,
      "loss": 0.6235,
      "step": 6103
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7008496110883633,
      "learning_rate": 1.2412828229315732e-06,
      "loss": 0.5486,
      "step": 6104
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7969223573520062,
      "learning_rate": 1.2399224828506174e-06,
      "loss": 0.5301,
      "step": 6105
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6922187346291012,
      "learning_rate": 1.23856278307169e-06,
      "loss": 0.6377,
      "step": 6106
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6771003714311855,
      "learning_rate": 1.2372037238263347e-06,
      "loss": 0.5347,
      "step": 6107
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7944169316741196,
      "learning_rate": 1.235845305345983e-06,
      "loss": 0.6368,
      "step": 6108
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9537648243619598,
      "learning_rate": 1.2344875278619627e-06,
      "loss": 0.6433,
      "step": 6109
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.683887256780697,
      "learning_rate": 1.2331303916054877e-06,
      "loss": 0.5161,
      "step": 6110
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6484298106276131,
      "learning_rate": 1.2317738968076637e-06,
      "loss": 0.5986,
      "step": 6111
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.668913991741974,
      "learning_rate": 1.2304180436994879e-06,
      "loss": 0.4905,
      "step": 6112
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9075030069616917,
      "learning_rate": 1.2290628325118481e-06,
      "loss": 0.6123,
      "step": 6113
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.820641780130014,
      "learning_rate": 1.2277082634755234e-06,
      "loss": 0.6551,
      "step": 6114
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7252927123222411,
      "learning_rate": 1.226354336821181e-06,
      "loss": 0.5099,
      "step": 6115
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7932228393447641,
      "learning_rate": 1.2250010527793816e-06,
      "loss": 0.5421,
      "step": 6116
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6619852549126756,
      "learning_rate": 1.2236484115805758e-06,
      "loss": 0.559,
      "step": 6117
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7134715300268349,
      "learning_rate": 1.2222964134551035e-06,
      "loss": 0.553,
      "step": 6118
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8310633908193823,
      "learning_rate": 1.220945058633196e-06,
      "loss": 0.5238,
      "step": 6119
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8964118519201913,
      "learning_rate": 1.219594347344975e-06,
      "loss": 0.6094,
      "step": 6120
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.59983042072799,
      "learning_rate": 1.218244279820453e-06,
      "loss": 0.5219,
      "step": 6121
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8496722176316378,
      "learning_rate": 1.216894856289532e-06,
      "loss": 0.663,
      "step": 6122
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7057927658769098,
      "learning_rate": 1.2155460769820048e-06,
      "loss": 0.5365,
      "step": 6123
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8619764085141581,
      "learning_rate": 1.2141979421275546e-06,
      "loss": 0.6533,
      "step": 6124
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6375408180022438,
      "learning_rate": 1.2128504519557543e-06,
      "loss": 0.4946,
      "step": 6125
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8717052967470488,
      "learning_rate": 1.2115036066960683e-06,
      "loss": 0.6652,
      "step": 6126
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7345593500747517,
      "learning_rate": 1.2101574065778498e-06,
      "loss": 0.5657,
      "step": 6127
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8940010362256054,
      "learning_rate": 1.2088118518303427e-06,
      "loss": 0.5514,
      "step": 6128
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8366021031673786,
      "learning_rate": 1.2074669426826806e-06,
      "loss": 0.6265,
      "step": 6129
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8790434193697267,
      "learning_rate": 1.206122679363888e-06,
      "loss": 0.6509,
      "step": 6130
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7848583893105114,
      "learning_rate": 1.2047790621028787e-06,
      "loss": 0.5662,
      "step": 6131
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9662591986593232,
      "learning_rate": 1.2034360911284565e-06,
      "loss": 0.6134,
      "step": 6132
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9108502169622835,
      "learning_rate": 1.202093766669316e-06,
      "loss": 0.6101,
      "step": 6133
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7168246131007312,
      "learning_rate": 1.2007520889540397e-06,
      "loss": 0.4855,
      "step": 6134
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8138976553363829,
      "learning_rate": 1.1994110582111028e-06,
      "loss": 0.6007,
      "step": 6135
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.81512892558438,
      "learning_rate": 1.1980706746688692e-06,
      "loss": 0.6187,
      "step": 6136
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8751543934466348,
      "learning_rate": 1.1967309385555915e-06,
      "loss": 0.6029,
      "step": 6137
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8723820464892906,
      "learning_rate": 1.1953918500994128e-06,
      "loss": 0.6412,
      "step": 6138
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6816850317062958,
      "learning_rate": 1.1940534095283656e-06,
      "loss": 0.4984,
      "step": 6139
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.831417523076056,
      "learning_rate": 1.192715617070373e-06,
      "loss": 0.6724,
      "step": 6140
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9538029269609437,
      "learning_rate": 1.1913784729532462e-06,
      "loss": 0.6131,
      "step": 6141
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6938557371563928,
      "learning_rate": 1.1900419774046884e-06,
      "loss": 0.5261,
      "step": 6142
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8159928255872908,
      "learning_rate": 1.1887061306522896e-06,
      "loss": 0.6192,
      "step": 6143
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7190714993213324,
      "learning_rate": 1.1873709329235305e-06,
      "loss": 0.5821,
      "step": 6144
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.808213704877347,
      "learning_rate": 1.1860363844457822e-06,
      "loss": 0.6252,
      "step": 6145
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7697640215178182,
      "learning_rate": 1.1847024854463047e-06,
      "loss": 0.6021,
      "step": 6146
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0060601802753137,
      "learning_rate": 1.1833692361522459e-06,
      "loss": 0.5979,
      "step": 6147
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7098569053601365,
      "learning_rate": 1.1820366367906445e-06,
      "loss": 0.5517,
      "step": 6148
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.679607499865355,
      "learning_rate": 1.1807046875884286e-06,
      "loss": 0.5961,
      "step": 6149
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8495271095191237,
      "learning_rate": 1.1793733887724156e-06,
      "loss": 0.6253,
      "step": 6150
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.696843863261326,
      "learning_rate": 1.1780427405693118e-06,
      "loss": 0.557,
      "step": 6151
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8612366803768845,
      "learning_rate": 1.1767127432057117e-06,
      "loss": 0.6076,
      "step": 6152
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.614657376907369,
      "learning_rate": 1.1753833969081014e-06,
      "loss": 0.5288,
      "step": 6153
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7378628876834291,
      "learning_rate": 1.1740547019028537e-06,
      "loss": 0.6064,
      "step": 6154
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6604945264808391,
      "learning_rate": 1.1727266584162317e-06,
      "loss": 0.5242,
      "step": 6155
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7281271477432399,
      "learning_rate": 1.171399266674388e-06,
      "loss": 0.6234,
      "step": 6156
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.852274518881763,
      "learning_rate": 1.1700725269033624e-06,
      "loss": 0.5503,
      "step": 6157
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7497403266504802,
      "learning_rate": 1.168746439329086e-06,
      "loss": 0.6508,
      "step": 6158
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8439379266791349,
      "learning_rate": 1.1674210041773776e-06,
      "loss": 0.6128,
      "step": 6159
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8195612851552496,
      "learning_rate": 1.1660962216739425e-06,
      "loss": 0.5712,
      "step": 6160
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.988188847067404,
      "learning_rate": 1.1647720920443816e-06,
      "loss": 0.655,
      "step": 6161
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7851582447545722,
      "learning_rate": 1.1634486155141778e-06,
      "loss": 0.5413,
      "step": 6162
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9486162560621779,
      "learning_rate": 1.1621257923087065e-06,
      "loss": 0.6147,
      "step": 6163
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9150414034760967,
      "learning_rate": 1.1608036226532292e-06,
      "loss": 0.6156,
      "step": 6164
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6821148013297105,
      "learning_rate": 1.1594821067728995e-06,
      "loss": 0.4879,
      "step": 6165
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6288667761698284,
      "learning_rate": 1.158161244892756e-06,
      "loss": 0.4646,
      "step": 6166
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6656312099453875,
      "learning_rate": 1.1568410372377293e-06,
      "loss": 0.5492,
      "step": 6167
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6785429332951738,
      "learning_rate": 1.1555214840326355e-06,
      "loss": 0.496,
      "step": 6168
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7134932405240115,
      "learning_rate": 1.1542025855021815e-06,
      "loss": 0.5491,
      "step": 6169
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6603433432560484,
      "learning_rate": 1.1528843418709623e-06,
      "loss": 0.5253,
      "step": 6170
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8306929964653622,
      "learning_rate": 1.1515667533634607e-06,
      "loss": 0.61,
      "step": 6171
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7710547110438372,
      "learning_rate": 1.150249820204048e-06,
      "loss": 0.58,
      "step": 6172
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9072139986073352,
      "learning_rate": 1.148933542616985e-06,
      "loss": 0.6498,
      "step": 6173
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7795215662855888,
      "learning_rate": 1.1476179208264194e-06,
      "loss": 0.623,
      "step": 6174
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8515107013953397,
      "learning_rate": 1.1463029550563865e-06,
      "loss": 0.662,
      "step": 6175
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7988210288360347,
      "learning_rate": 1.144988645530814e-06,
      "loss": 0.5976,
      "step": 6176
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6430245707432374,
      "learning_rate": 1.1436749924735142e-06,
      "loss": 0.535,
      "step": 6177
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8144922769304165,
      "learning_rate": 1.1423619961081884e-06,
      "loss": 0.621,
      "step": 6178
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7654573309480317,
      "learning_rate": 1.1410496566584261e-06,
      "loss": 0.5927,
      "step": 6179
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8609749546291151,
      "learning_rate": 1.1397379743477049e-06,
      "loss": 0.6713,
      "step": 6180
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7678564312715521,
      "learning_rate": 1.1384269493993904e-06,
      "loss": 0.595,
      "step": 6181
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7057751995240582,
      "learning_rate": 1.1371165820367374e-06,
      "loss": 0.4971,
      "step": 6182
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6928187456092068,
      "learning_rate": 1.1358068724828869e-06,
      "loss": 0.5229,
      "step": 6183
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6949700436378984,
      "learning_rate": 1.1344978209608693e-06,
      "loss": 0.5015,
      "step": 6184
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7633083830399727,
      "learning_rate": 1.1331894276936023e-06,
      "loss": 0.638,
      "step": 6185
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8868990583661809,
      "learning_rate": 1.1318816929038917e-06,
      "loss": 0.6378,
      "step": 6186
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7482377697075585,
      "learning_rate": 1.1305746168144311e-06,
      "loss": 0.5627,
      "step": 6187
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8106571710188469,
      "learning_rate": 1.1292681996478016e-06,
      "loss": 0.6062,
      "step": 6188
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.881086958854107,
      "learning_rate": 1.1279624416264728e-06,
      "loss": 0.6472,
      "step": 6189
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9418688480993389,
      "learning_rate": 1.1266573429728017e-06,
      "loss": 0.5853,
      "step": 6190
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8565746535553717,
      "learning_rate": 1.1253529039090328e-06,
      "loss": 0.6326,
      "step": 6191
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8673629367375326,
      "learning_rate": 1.124049124657297e-06,
      "loss": 0.5691,
      "step": 6192
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7131959255254904,
      "learning_rate": 1.1227460054396177e-06,
      "loss": 0.5023,
      "step": 6193
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6438808413111474,
      "learning_rate": 1.1214435464779006e-06,
      "loss": 0.4696,
      "step": 6194
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.629225317410022,
      "learning_rate": 1.1201417479939409e-06,
      "loss": 0.5351,
      "step": 6195
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9578969302653552,
      "learning_rate": 1.1188406102094212e-06,
      "loss": 0.6284,
      "step": 6196
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6630072380433006,
      "learning_rate": 1.1175401333459125e-06,
      "loss": 0.5373,
      "step": 6197
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9070015214670604,
      "learning_rate": 1.1162403176248715e-06,
      "loss": 0.5891,
      "step": 6198
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5866880757374822,
      "learning_rate": 1.1149411632676438e-06,
      "loss": 0.4916,
      "step": 6199
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.58782424096755,
      "learning_rate": 1.1136426704954622e-06,
      "loss": 0.4622,
      "step": 6200
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8056465797658702,
      "learning_rate": 1.1123448395294456e-06,
      "loss": 0.5852,
      "step": 6201
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8645160049101256,
      "learning_rate": 1.1110476705906015e-06,
      "loss": 0.6357,
      "step": 6202
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.731765507538455,
      "learning_rate": 1.1097511638998242e-06,
      "loss": 0.5594,
      "step": 6203
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7261411816489569,
      "learning_rate": 1.1084553196778958e-06,
      "loss": 0.5415,
      "step": 6204
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.848535599717028,
      "learning_rate": 1.1071601381454845e-06,
      "loss": 0.5527,
      "step": 6205
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8321029980681731,
      "learning_rate": 1.105865619523146e-06,
      "loss": 0.6308,
      "step": 6206
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9830634768572302,
      "learning_rate": 1.1045717640313242e-06,
      "loss": 0.6878,
      "step": 6207
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7405648641021086,
      "learning_rate": 1.1032785718903483e-06,
      "loss": 0.5296,
      "step": 6208
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7601487978703901,
      "learning_rate": 1.101986043320436e-06,
      "loss": 0.6108,
      "step": 6209
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8406239980286435,
      "learning_rate": 1.1006941785416913e-06,
      "loss": 0.6328,
      "step": 6210
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7530948194857948,
      "learning_rate": 1.099402977774105e-06,
      "loss": 0.541,
      "step": 6211
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6766893753807784,
      "learning_rate": 1.098112441237556e-06,
      "loss": 0.5138,
      "step": 6212
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6519095171781057,
      "learning_rate": 1.0968225691518087e-06,
      "loss": 0.513,
      "step": 6213
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8115321958296825,
      "learning_rate": 1.0955333617365143e-06,
      "loss": 0.5079,
      "step": 6214
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6460120675259842,
      "learning_rate": 1.0942448192112126e-06,
      "loss": 0.5183,
      "step": 6215
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6380077886955908,
      "learning_rate": 1.0929569417953279e-06,
      "loss": 0.4961,
      "step": 6216
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6552424049911305,
      "learning_rate": 1.091669729708173e-06,
      "loss": 0.5324,
      "step": 6217
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9788464200653787,
      "learning_rate": 1.090383183168946e-06,
      "loss": 0.6374,
      "step": 6218
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8082605448167149,
      "learning_rate": 1.0890973023967316e-06,
      "loss": 0.5505,
      "step": 6219
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6598846093612981,
      "learning_rate": 1.0878120876105048e-06,
      "loss": 0.448,
      "step": 6220
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9606216679552523,
      "learning_rate": 1.0865275390291224e-06,
      "loss": 0.6129,
      "step": 6221
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7190480441961784,
      "learning_rate": 1.0852436568713298e-06,
      "loss": 0.5212,
      "step": 6222
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6236530283726269,
      "learning_rate": 1.0839604413557586e-06,
      "loss": 0.5166,
      "step": 6223
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8125470559413996,
      "learning_rate": 1.082677892700928e-06,
      "loss": 0.6029,
      "step": 6224
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7001503441112323,
      "learning_rate": 1.0813960111252414e-06,
      "loss": 0.549,
      "step": 6225
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7816738434130006,
      "learning_rate": 1.0801147968469899e-06,
      "loss": 0.635,
      "step": 6226
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7835350219471248,
      "learning_rate": 1.0788342500843523e-06,
      "loss": 0.6075,
      "step": 6227
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6754054734801362,
      "learning_rate": 1.077554371055391e-06,
      "loss": 0.5141,
      "step": 6228
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9462088514474958,
      "learning_rate": 1.0762751599780568e-06,
      "loss": 0.6178,
      "step": 6229
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9549454473405544,
      "learning_rate": 1.0749966170701853e-06,
      "loss": 0.6146,
      "step": 6230
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6648093085411874,
      "learning_rate": 1.0737187425495e-06,
      "loss": 0.4686,
      "step": 6231
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8424240595325632,
      "learning_rate": 1.0724415366336094e-06,
      "loss": 0.6633,
      "step": 6232
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8061080345778929,
      "learning_rate": 1.0711649995400074e-06,
      "loss": 0.5888,
      "step": 6233
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7234579542316615,
      "learning_rate": 1.069889131486076e-06,
      "loss": 0.6015,
      "step": 6234
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.853396978446328,
      "learning_rate": 1.068613932689082e-06,
      "loss": 0.6162,
      "step": 6235
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8172353131822218,
      "learning_rate": 1.0673394033661783e-06,
      "loss": 0.5776,
      "step": 6236
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8205851146633035,
      "learning_rate": 1.0660655437344041e-06,
      "loss": 0.594,
      "step": 6237
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.011073369699019,
      "learning_rate": 1.0647923540106842e-06,
      "loss": 0.6244,
      "step": 6238
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9431651575566694,
      "learning_rate": 1.0635198344118298e-06,
      "loss": 0.563,
      "step": 6239
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7203332072907978,
      "learning_rate": 1.0622479851545376e-06,
      "loss": 0.5172,
      "step": 6240
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5330732846790553,
      "learning_rate": 1.0609768064553906e-06,
      "loss": 0.6169,
      "step": 6241
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7922575471273464,
      "learning_rate": 1.0597062985308565e-06,
      "loss": 0.6055,
      "step": 6242
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8960165388719186,
      "learning_rate": 1.058436461597291e-06,
      "loss": 0.607,
      "step": 6243
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7009826156987234,
      "learning_rate": 1.0571672958709323e-06,
      "loss": 0.4863,
      "step": 6244
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8613762557827476,
      "learning_rate": 1.0558988015679056e-06,
      "loss": 0.5474,
      "step": 6245
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5716277494162099,
      "learning_rate": 1.0546309789042252e-06,
      "loss": 0.4953,
      "step": 6246
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8906396133121085,
      "learning_rate": 1.053363828095787e-06,
      "loss": 0.6278,
      "step": 6247
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.614837720931292,
      "learning_rate": 1.0520973493583725e-06,
      "loss": 0.527,
      "step": 6248
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6737021852045152,
      "learning_rate": 1.0508315429076509e-06,
      "loss": 0.5392,
      "step": 6249
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8202359192341226,
      "learning_rate": 1.049566408959175e-06,
      "loss": 0.6121,
      "step": 6250
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7362648230998704,
      "learning_rate": 1.0483019477283845e-06,
      "loss": 0.596,
      "step": 6251
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8718961809706266,
      "learning_rate": 1.047038159430604e-06,
      "loss": 0.6618,
      "step": 6252
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8422450533487897,
      "learning_rate": 1.0457750442810433e-06,
      "loss": 0.5744,
      "step": 6253
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6586926569265514,
      "learning_rate": 1.0445126024947971e-06,
      "loss": 0.5133,
      "step": 6254
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.684533044519361,
      "learning_rate": 1.0432508342868475e-06,
      "loss": 0.5882,
      "step": 6255
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7911646447510978,
      "learning_rate": 1.041989739872059e-06,
      "loss": 0.5693,
      "step": 6256
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2598379082550029,
      "learning_rate": 1.0407293194651841e-06,
      "loss": 0.6607,
      "step": 6257
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7961612572007144,
      "learning_rate": 1.0394695732808586e-06,
      "loss": 0.5328,
      "step": 6258
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6672435034180854,
      "learning_rate": 1.0382105015336042e-06,
      "loss": 0.5146,
      "step": 6259
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7198639090313526,
      "learning_rate": 1.0369521044378278e-06,
      "loss": 0.5718,
      "step": 6260
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8923439659012693,
      "learning_rate": 1.0356943822078213e-06,
      "loss": 0.5887,
      "step": 6261
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.995128344461852,
      "learning_rate": 1.034437335057762e-06,
      "loss": 0.6101,
      "step": 6262
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6566195268725807,
      "learning_rate": 1.033180963201711e-06,
      "loss": 0.5229,
      "step": 6263
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.820809713397241,
      "learning_rate": 1.031925266853616e-06,
      "loss": 0.6099,
      "step": 6264
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7802241174866805,
      "learning_rate": 1.0306702462273093e-06,
      "loss": 0.5923,
      "step": 6265
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8156660113492312,
      "learning_rate": 1.0294159015365073e-06,
      "loss": 0.5571,
      "step": 6266
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.682182596830582,
      "learning_rate": 1.0281622329948122e-06,
      "loss": 0.5005,
      "step": 6267
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6995287117136341,
      "learning_rate": 1.0269092408157104e-06,
      "loss": 0.502,
      "step": 6268
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8567453344893206,
      "learning_rate": 1.0256569252125736e-06,
      "loss": 0.6122,
      "step": 6269
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7311172055675106,
      "learning_rate": 1.0244052863986575e-06,
      "loss": 0.5522,
      "step": 6270
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8766170310779489,
      "learning_rate": 1.0231543245871028e-06,
      "loss": 0.5918,
      "step": 6271
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.713738796145529,
      "learning_rate": 1.0219040399909375e-06,
      "loss": 0.5271,
      "step": 6272
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9568214224760472,
      "learning_rate": 1.0206544328230705e-06,
      "loss": 0.5951,
      "step": 6273
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8079660719584748,
      "learning_rate": 1.0194055032962973e-06,
      "loss": 0.6152,
      "step": 6274
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.832200353042896,
      "learning_rate": 1.0181572516232973e-06,
      "loss": 0.6097,
      "step": 6275
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7485972629506407,
      "learning_rate": 1.0169096780166348e-06,
      "loss": 0.5347,
      "step": 6276
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7655463644938711,
      "learning_rate": 1.0156627826887572e-06,
      "loss": 0.5326,
      "step": 6277
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6501270495038894,
      "learning_rate": 1.0144165658520006e-06,
      "loss": 0.5254,
      "step": 6278
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6745780888093885,
      "learning_rate": 1.0131710277185814e-06,
      "loss": 0.537,
      "step": 6279
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7763817561059477,
      "learning_rate": 1.0119261685006016e-06,
      "loss": 0.6129,
      "step": 6280
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6481494702559437,
      "learning_rate": 1.0106819884100478e-06,
      "loss": 0.5146,
      "step": 6281
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8744103792126252,
      "learning_rate": 1.0094384876587909e-06,
      "loss": 0.6936,
      "step": 6282
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8081329439939369,
      "learning_rate": 1.0081956664585863e-06,
      "loss": 0.5273,
      "step": 6283
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8218241318554043,
      "learning_rate": 1.0069535250210733e-06,
      "loss": 0.5723,
      "step": 6284
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8323508997339945,
      "learning_rate": 1.005712063557776e-06,
      "loss": 0.5859,
      "step": 6285
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6833483456005509,
      "learning_rate": 1.0044712822801018e-06,
      "loss": 0.5094,
      "step": 6286
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7376026633247432,
      "learning_rate": 1.0032311813993433e-06,
      "loss": 0.5705,
      "step": 6287
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6637827017026487,
      "learning_rate": 1.001991761126676e-06,
      "loss": 0.531,
      "step": 6288
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7060537044332207,
      "learning_rate": 1.0007530216731614e-06,
      "loss": 0.5112,
      "step": 6289
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7486848973427744,
      "learning_rate": 9.995149632497435e-07,
      "loss": 0.5348,
      "step": 6290
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.713976176966678,
      "learning_rate": 9.982775860672507e-07,
      "loss": 0.5323,
      "step": 6291
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9228682757431986,
      "learning_rate": 9.97040890336395e-07,
      "loss": 0.6437,
      "step": 6292
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8681026717342937,
      "learning_rate": 9.95804876267773e-07,
      "loss": 0.6467,
      "step": 6293
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.891720177211184,
      "learning_rate": 9.94569544071865e-07,
      "loss": 0.6566,
      "step": 6294
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7155541886546622,
      "learning_rate": 9.933348939590358e-07,
      "loss": 0.5291,
      "step": 6295
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8106697195823868,
      "learning_rate": 9.921009261395332e-07,
      "loss": 0.702,
      "step": 6296
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8378337582728091,
      "learning_rate": 9.908676408234884e-07,
      "loss": 0.5963,
      "step": 6297
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7856225166543965,
      "learning_rate": 9.896350382209175e-07,
      "loss": 0.5445,
      "step": 6298
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6981942904144177,
      "learning_rate": 9.884031185417204e-07,
      "loss": 0.534,
      "step": 6299
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6213183746973765,
      "learning_rate": 9.871718819956794e-07,
      "loss": 0.6102,
      "step": 6300
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7772848459838978,
      "learning_rate": 9.859413287924613e-07,
      "loss": 0.6649,
      "step": 6301
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9619917126053887,
      "learning_rate": 9.847114591416174e-07,
      "loss": 0.6294,
      "step": 6302
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6182236137916943,
      "learning_rate": 9.834822732525795e-07,
      "loss": 0.4744,
      "step": 6303
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7906434671275144,
      "learning_rate": 9.822537713346674e-07,
      "loss": 0.5943,
      "step": 6304
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6999346251123362,
      "learning_rate": 9.810259535970822e-07,
      "loss": 0.5183,
      "step": 6305
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8120641418406452,
      "learning_rate": 9.797988202489073e-07,
      "loss": 0.6163,
      "step": 6306
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9170177494928352,
      "learning_rate": 9.785723714991113e-07,
      "loss": 0.6439,
      "step": 6307
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7407648585684432,
      "learning_rate": 9.773466075565457e-07,
      "loss": 0.5794,
      "step": 6308
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7570388545399602,
      "learning_rate": 9.761215286299448e-07,
      "loss": 0.5619,
      "step": 6309
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7003609097323054,
      "learning_rate": 9.748971349279273e-07,
      "loss": 0.5164,
      "step": 6310
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7846353207242666,
      "learning_rate": 9.736734266589949e-07,
      "loss": 0.633,
      "step": 6311
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7442210558441857,
      "learning_rate": 9.724504040315313e-07,
      "loss": 0.4941,
      "step": 6312
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7851932250667001,
      "learning_rate": 9.712280672538056e-07,
      "loss": 0.5389,
      "step": 6313
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0628178993798705,
      "learning_rate": 9.700064165339679e-07,
      "loss": 0.6568,
      "step": 6314
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8270627781034178,
      "learning_rate": 9.687854520800538e-07,
      "loss": 0.6764,
      "step": 6315
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8703573663692341,
      "learning_rate": 9.67565174099981e-07,
      "loss": 0.6785,
      "step": 6316
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7432694516472188,
      "learning_rate": 9.663455828015483e-07,
      "loss": 0.5056,
      "step": 6317
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8401794413763661,
      "learning_rate": 9.651266783924413e-07,
      "loss": 0.5762,
      "step": 6318
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6502974963213034,
      "learning_rate": 9.639084610802258e-07,
      "loss": 0.5132,
      "step": 6319
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9041666141911302,
      "learning_rate": 9.626909310723514e-07,
      "loss": 0.6339,
      "step": 6320
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9661907743730896,
      "learning_rate": 9.614740885761514e-07,
      "loss": 0.6125,
      "step": 6321
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6989020740784724,
      "learning_rate": 9.602579337988405e-07,
      "loss": 0.5244,
      "step": 6322
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6942587172255824,
      "learning_rate": 9.590424669475185e-07,
      "loss": 0.5369,
      "step": 6323
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8215775919701693,
      "learning_rate": 9.578276882291654e-07,
      "loss": 0.5938,
      "step": 6324
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7763457187065593,
      "learning_rate": 9.566135978506463e-07,
      "loss": 0.6062,
      "step": 6325
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8439439083412723,
      "learning_rate": 9.554001960187076e-07,
      "loss": 0.5933,
      "step": 6326
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6865102763966442,
      "learning_rate": 9.541874829399794e-07,
      "loss": 0.5598,
      "step": 6327
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.792086123196357,
      "learning_rate": 9.529754588209739e-07,
      "loss": 0.5553,
      "step": 6328
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7952880125025787,
      "learning_rate": 9.517641238680858e-07,
      "loss": 0.5129,
      "step": 6329
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9459410360699414,
      "learning_rate": 9.505534782875919e-07,
      "loss": 0.6355,
      "step": 6330
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0417818703953796,
      "learning_rate": 9.493435222856556e-07,
      "loss": 0.6733,
      "step": 6331
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7685259581956614,
      "learning_rate": 9.481342560683177e-07,
      "loss": 0.5507,
      "step": 6332
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8153557506737833,
      "learning_rate": 9.469256798415039e-07,
      "loss": 0.6318,
      "step": 6333
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.905984596158335,
      "learning_rate": 9.457177938110224e-07,
      "loss": 0.6344,
      "step": 6334
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.713504844088086,
      "learning_rate": 9.445105981825631e-07,
      "loss": 0.5502,
      "step": 6335
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8501016308531509,
      "learning_rate": 9.433040931616994e-07,
      "loss": 0.6587,
      "step": 6336
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8894184030141193,
      "learning_rate": 9.420982789538863e-07,
      "loss": 0.5844,
      "step": 6337
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8014474699975243,
      "learning_rate": 9.408931557644612e-07,
      "loss": 0.5991,
      "step": 6338
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6616724922259084,
      "learning_rate": 9.396887237986446e-07,
      "loss": 0.5244,
      "step": 6339
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.670124248743178,
      "learning_rate": 9.384849832615378e-07,
      "loss": 0.5429,
      "step": 6340
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9663526142345502,
      "learning_rate": 9.372819343581263e-07,
      "loss": 0.6628,
      "step": 6341
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1966347654919072,
      "learning_rate": 9.360795772932757e-07,
      "loss": 0.6262,
      "step": 6342
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8772406751189397,
      "learning_rate": 9.348779122717361e-07,
      "loss": 0.6318,
      "step": 6343
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6198695705419238,
      "learning_rate": 9.336769394981377e-07,
      "loss": 0.4884,
      "step": 6344
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7309703264407184,
      "learning_rate": 9.324766591769935e-07,
      "loss": 0.5811,
      "step": 6345
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7704208564997415,
      "learning_rate": 9.312770715126995e-07,
      "loss": 0.6154,
      "step": 6346
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8255495150700635,
      "learning_rate": 9.300781767095324e-07,
      "loss": 0.6275,
      "step": 6347
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6469598942949039,
      "learning_rate": 9.288799749716515e-07,
      "loss": 0.5169,
      "step": 6348
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7209148556957957,
      "learning_rate": 9.276824665030987e-07,
      "loss": 0.5758,
      "step": 6349
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6903977614344166,
      "learning_rate": 9.264856515077963e-07,
      "loss": 0.5184,
      "step": 6350
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7610622064284739,
      "learning_rate": 9.252895301895498e-07,
      "loss": 0.5348,
      "step": 6351
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6234572038760733,
      "learning_rate": 9.240941027520467e-07,
      "loss": 0.5325,
      "step": 6352
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9358115177267461,
      "learning_rate": 9.228993693988552e-07,
      "loss": 0.6189,
      "step": 6353
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6530056342410644,
      "learning_rate": 9.217053303334261e-07,
      "loss": 0.5526,
      "step": 6354
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9130393891859884,
      "learning_rate": 9.205119857590922e-07,
      "loss": 0.6836,
      "step": 6355
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7188109948012957,
      "learning_rate": 9.193193358790659e-07,
      "loss": 0.5856,
      "step": 6356
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7572151585042509,
      "learning_rate": 9.181273808964458e-07,
      "loss": 0.5072,
      "step": 6357
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6728016940172079,
      "learning_rate": 9.16936121014208e-07,
      "loss": 0.5378,
      "step": 6358
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7294785169408458,
      "learning_rate": 9.157455564352113e-07,
      "loss": 0.5511,
      "step": 6359
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6325989437731061,
      "learning_rate": 9.145556873621975e-07,
      "loss": 0.5211,
      "step": 6360
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6986411665808339,
      "learning_rate": 9.133665139977882e-07,
      "loss": 0.5113,
      "step": 6361
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8420509759587462,
      "learning_rate": 9.121780365444872e-07,
      "loss": 0.6081,
      "step": 6362
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7028574459494599,
      "learning_rate": 9.109902552046801e-07,
      "loss": 0.5869,
      "step": 6363
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7961708199236058,
      "learning_rate": 9.098031701806337e-07,
      "loss": 0.5573,
      "step": 6364
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.927958601548571,
      "learning_rate": 9.086167816744956e-07,
      "loss": 0.6459,
      "step": 6365
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6727780101487731,
      "learning_rate": 9.074310898882959e-07,
      "loss": 0.5791,
      "step": 6366
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6969774042320578,
      "learning_rate": 9.06246095023946e-07,
      "loss": 0.5567,
      "step": 6367
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6832756967576599,
      "learning_rate": 9.050617972832371e-07,
      "loss": 0.5323,
      "step": 6368
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.873415656932259,
      "learning_rate": 9.03878196867844e-07,
      "loss": 0.6124,
      "step": 6369
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7211061479204822,
      "learning_rate": 9.026952939793205e-07,
      "loss": 0.5625,
      "step": 6370
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6046508241330434,
      "learning_rate": 9.015130888191026e-07,
      "loss": 0.4904,
      "step": 6371
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7240367374676969,
      "learning_rate": 9.003315815885083e-07,
      "loss": 0.5416,
      "step": 6372
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7552968507889684,
      "learning_rate": 8.99150772488736e-07,
      "loss": 0.5648,
      "step": 6373
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6444843294447101,
      "learning_rate": 8.979706617208644e-07,
      "loss": 0.4902,
      "step": 6374
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8652895633378936,
      "learning_rate": 8.96791249485855e-07,
      "loss": 0.6235,
      "step": 6375
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7362087482833665,
      "learning_rate": 8.95612535984548e-07,
      "loss": 0.4955,
      "step": 6376
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7442295393655132,
      "learning_rate": 8.944345214176675e-07,
      "loss": 0.5593,
      "step": 6377
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8884109501699472,
      "learning_rate": 8.932572059858152e-07,
      "loss": 0.6404,
      "step": 6378
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.710718206757313,
      "learning_rate": 8.920805898894785e-07,
      "loss": 0.5346,
      "step": 6379
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7705635707168244,
      "learning_rate": 8.909046733290211e-07,
      "loss": 0.606,
      "step": 6380
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9028447540402339,
      "learning_rate": 8.897294565046894e-07,
      "loss": 0.6082,
      "step": 6381
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8651942490622496,
      "learning_rate": 8.885549396166109e-07,
      "loss": 0.5092,
      "step": 6382
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7637105023210538,
      "learning_rate": 8.873811228647933e-07,
      "loss": 0.4897,
      "step": 6383
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8446746908123323,
      "learning_rate": 8.862080064491258e-07,
      "loss": 0.6442,
      "step": 6384
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7506576838512039,
      "learning_rate": 8.850355905693775e-07,
      "loss": 0.5856,
      "step": 6385
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6526432601773813,
      "learning_rate": 8.838638754251988e-07,
      "loss": 0.4714,
      "step": 6386
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2469464106570451,
      "learning_rate": 8.826928612161206e-07,
      "loss": 0.6614,
      "step": 6387
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9555022354121456,
      "learning_rate": 8.815225481415529e-07,
      "loss": 0.65,
      "step": 6388
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8031010869418164,
      "learning_rate": 8.803529364007907e-07,
      "loss": 0.5896,
      "step": 6389
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6506408573927392,
      "learning_rate": 8.791840261930051e-07,
      "loss": 0.5076,
      "step": 6390
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7958142341409334,
      "learning_rate": 8.780158177172493e-07,
      "loss": 0.6211,
      "step": 6391
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6992844880535665,
      "learning_rate": 8.76848311172458e-07,
      "loss": 0.5018,
      "step": 6392
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8464495817244385,
      "learning_rate": 8.756815067574437e-07,
      "loss": 0.5834,
      "step": 6393
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6769978154615174,
      "learning_rate": 8.745154046709025e-07,
      "loss": 0.522,
      "step": 6394
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7282635362494032,
      "learning_rate": 8.733500051114085e-07,
      "loss": 0.5223,
      "step": 6395
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7064724956371685,
      "learning_rate": 8.721853082774179e-07,
      "loss": 0.5072,
      "step": 6396
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6736115477866378,
      "learning_rate": 8.710213143672658e-07,
      "loss": 0.4898,
      "step": 6397
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6872224234036635,
      "learning_rate": 8.698580235791687e-07,
      "loss": 0.4781,
      "step": 6398
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7645808313815857,
      "learning_rate": 8.686954361112226e-07,
      "loss": 0.6034,
      "step": 6399
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8205171128663795,
      "learning_rate": 8.675335521614036e-07,
      "loss": 0.675,
      "step": 6400
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6973765243887022,
      "learning_rate": 8.663723719275696e-07,
      "loss": 0.5552,
      "step": 6401
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8716466453068313,
      "learning_rate": 8.652118956074568e-07,
      "loss": 0.5486,
      "step": 6402
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.910889051339716,
      "learning_rate": 8.640521233986821e-07,
      "loss": 0.6764,
      "step": 6403
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.897066922239198,
      "learning_rate": 8.62893055498743e-07,
      "loss": 0.5139,
      "step": 6404
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9734767889047309,
      "learning_rate": 8.617346921050162e-07,
      "loss": 0.6108,
      "step": 6405
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7916151181150848,
      "learning_rate": 8.60577033414759e-07,
      "loss": 0.6009,
      "step": 6406
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8315242732182044,
      "learning_rate": 8.59420079625109e-07,
      "loss": 0.5479,
      "step": 6407
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2298868688925846,
      "learning_rate": 8.582638309330832e-07,
      "loss": 0.6719,
      "step": 6408
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8750952535243524,
      "learning_rate": 8.57108287535579e-07,
      "loss": 0.6203,
      "step": 6409
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6481336043044734,
      "learning_rate": 8.559534496293731e-07,
      "loss": 0.5055,
      "step": 6410
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7258832671269168,
      "learning_rate": 8.547993174111219e-07,
      "loss": 0.5998,
      "step": 6411
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8904719368762815,
      "learning_rate": 8.536458910773626e-07,
      "loss": 0.6317,
      "step": 6412
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7792863333282625,
      "learning_rate": 8.524931708245121e-07,
      "loss": 0.5018,
      "step": 6413
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0350283676168204,
      "learning_rate": 8.513411568488645e-07,
      "loss": 0.5705,
      "step": 6414
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7300904436997702,
      "learning_rate": 8.501898493465987e-07,
      "loss": 0.5725,
      "step": 6415
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9430342021084578,
      "learning_rate": 8.490392485137694e-07,
      "loss": 0.6605,
      "step": 6416
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8221835254769719,
      "learning_rate": 8.478893545463118e-07,
      "loss": 0.5958,
      "step": 6417
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9080898966147469,
      "learning_rate": 8.467401676400405e-07,
      "loss": 0.6525,
      "step": 6418
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9063293526838265,
      "learning_rate": 8.455916879906506e-07,
      "loss": 0.5777,
      "step": 6419
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6812565107290276,
      "learning_rate": 8.444439157937157e-07,
      "loss": 0.5403,
      "step": 6420
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8449181236722197,
      "learning_rate": 8.432968512446899e-07,
      "loss": 0.6618,
      "step": 6421
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7151088648532885,
      "learning_rate": 8.421504945389058e-07,
      "loss": 0.5615,
      "step": 6422
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8819943013699523,
      "learning_rate": 8.410048458715763e-07,
      "loss": 0.611,
      "step": 6423
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7398960907214395,
      "learning_rate": 8.398599054377938e-07,
      "loss": 0.5817,
      "step": 6424
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6737527039318283,
      "learning_rate": 8.387156734325286e-07,
      "loss": 0.4806,
      "step": 6425
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8898809918516477,
      "learning_rate": 8.375721500506329e-07,
      "loss": 0.6666,
      "step": 6426
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.705381848449586,
      "learning_rate": 8.364293354868364e-07,
      "loss": 0.5413,
      "step": 6427
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9201966811073016,
      "learning_rate": 8.352872299357478e-07,
      "loss": 0.6921,
      "step": 6428
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9764504006838134,
      "learning_rate": 8.341458335918562e-07,
      "loss": 0.6028,
      "step": 6429
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8930701182556481,
      "learning_rate": 8.330051466495298e-07,
      "loss": 0.6304,
      "step": 6430
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8090905722357825,
      "learning_rate": 8.318651693030161e-07,
      "loss": 0.5732,
      "step": 6431
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7990617014192907,
      "learning_rate": 8.307259017464397e-07,
      "loss": 0.5524,
      "step": 6432
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8723261145153438,
      "learning_rate": 8.295873441738078e-07,
      "loss": 0.4804,
      "step": 6433
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9497407108176819,
      "learning_rate": 8.284494967790036e-07,
      "loss": 0.6029,
      "step": 6434
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7487588504033302,
      "learning_rate": 8.27312359755792e-07,
      "loss": 0.5049,
      "step": 6435
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9875815722822804,
      "learning_rate": 8.261759332978142e-07,
      "loss": 0.6082,
      "step": 6436
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.602675938569681,
      "learning_rate": 8.250402175985922e-07,
      "loss": 0.4879,
      "step": 6437
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6970143125962176,
      "learning_rate": 8.239052128515268e-07,
      "loss": 0.5487,
      "step": 6438
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6684524885507424,
      "learning_rate": 8.227709192498979e-07,
      "loss": 0.5187,
      "step": 6439
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7580073133390448,
      "learning_rate": 8.21637336986863e-07,
      "loss": 0.5891,
      "step": 6440
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8224837185440056,
      "learning_rate": 8.205044662554584e-07,
      "loss": 0.565,
      "step": 6441
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8587563777130872,
      "learning_rate": 8.193723072486032e-07,
      "loss": 0.6328,
      "step": 6442
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7415384257076991,
      "learning_rate": 8.182408601590902e-07,
      "loss": 0.5113,
      "step": 6443
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.828380262117733,
      "learning_rate": 8.171101251795937e-07,
      "loss": 0.6017,
      "step": 6444
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6839197399037709,
      "learning_rate": 8.159801025026659e-07,
      "loss": 0.5056,
      "step": 6445
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7660586088529385,
      "learning_rate": 8.148507923207377e-07,
      "loss": 0.5758,
      "step": 6446
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.777553505297608,
      "learning_rate": 8.137221948261187e-07,
      "loss": 0.5946,
      "step": 6447
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6120111706389496,
      "learning_rate": 8.125943102109979e-07,
      "loss": 0.5102,
      "step": 6448
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8154463983584009,
      "learning_rate": 8.114671386674422e-07,
      "loss": 0.5613,
      "step": 6449
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7925579121584788,
      "learning_rate": 8.103406803873969e-07,
      "loss": 0.5276,
      "step": 6450
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8180563185991548,
      "learning_rate": 8.092149355626861e-07,
      "loss": 0.5566,
      "step": 6451
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6520611315975993,
      "learning_rate": 8.080899043850121e-07,
      "loss": 0.5128,
      "step": 6452
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6409915168930695,
      "learning_rate": 8.06965587045957e-07,
      "loss": 0.4664,
      "step": 6453
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7785096698611864,
      "learning_rate": 8.058419837369791e-07,
      "loss": 0.5641,
      "step": 6454
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2059408329842793,
      "learning_rate": 8.047190946494176e-07,
      "loss": 0.5678,
      "step": 6455
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8636108318229886,
      "learning_rate": 8.035969199744875e-07,
      "loss": 0.6432,
      "step": 6456
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6784763867394926,
      "learning_rate": 8.024754599032847e-07,
      "loss": 0.4905,
      "step": 6457
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6312876033399646,
      "learning_rate": 8.013547146267814e-07,
      "loss": 0.5438,
      "step": 6458
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7317142855710564,
      "learning_rate": 8.002346843358289e-07,
      "loss": 0.5523,
      "step": 6459
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.897920036628498,
      "learning_rate": 7.991153692211568e-07,
      "loss": 0.5509,
      "step": 6460
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8568066928443241,
      "learning_rate": 7.979967694733726e-07,
      "loss": 0.6415,
      "step": 6461
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.892723819171723,
      "learning_rate": 7.968788852829629e-07,
      "loss": 0.6203,
      "step": 6462
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7037148952275748,
      "learning_rate": 7.957617168402909e-07,
      "loss": 0.5094,
      "step": 6463
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8011052057685065,
      "learning_rate": 7.946452643355995e-07,
      "loss": 0.6489,
      "step": 6464
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.908649020884252,
      "learning_rate": 7.935295279590082e-07,
      "loss": 0.6019,
      "step": 6465
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8202694790046764,
      "learning_rate": 7.924145079005158e-07,
      "loss": 0.6325,
      "step": 6466
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6918177701681298,
      "learning_rate": 7.913002043499968e-07,
      "loss": 0.5543,
      "step": 6467
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8514098585170228,
      "learning_rate": 7.901866174972084e-07,
      "loss": 0.5362,
      "step": 6468
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9216796790734681,
      "learning_rate": 7.890737475317816e-07,
      "loss": 0.5834,
      "step": 6469
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6654423347433402,
      "learning_rate": 7.879615946432268e-07,
      "loss": 0.5403,
      "step": 6470
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7535850823083603,
      "learning_rate": 7.868501590209315e-07,
      "loss": 0.5056,
      "step": 6471
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6950152086840601,
      "learning_rate": 7.857394408541619e-07,
      "loss": 0.5256,
      "step": 6472
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6618727857489061,
      "learning_rate": 7.846294403320614e-07,
      "loss": 0.5756,
      "step": 6473
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8479021801667956,
      "learning_rate": 7.835201576436518e-07,
      "loss": 0.6183,
      "step": 6474
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7812017074292636,
      "learning_rate": 7.824115929778325e-07,
      "loss": 0.5057,
      "step": 6475
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7723323111542498,
      "learning_rate": 7.813037465233808e-07,
      "loss": 0.4913,
      "step": 6476
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6851069637669087,
      "learning_rate": 7.801966184689502e-07,
      "loss": 0.5185,
      "step": 6477
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8930163809330414,
      "learning_rate": 7.790902090030745e-07,
      "loss": 0.6458,
      "step": 6478
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8950594834668176,
      "learning_rate": 7.779845183141622e-07,
      "loss": 0.6127,
      "step": 6479
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8542798718681037,
      "learning_rate": 7.768795465905021e-07,
      "loss": 0.6416,
      "step": 6480
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9204954413015148,
      "learning_rate": 7.757752940202596e-07,
      "loss": 0.5953,
      "step": 6481
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6560953156263014,
      "learning_rate": 7.746717607914766e-07,
      "loss": 0.478,
      "step": 6482
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6999493642693491,
      "learning_rate": 7.735689470920738e-07,
      "loss": 0.5704,
      "step": 6483
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7103873071471618,
      "learning_rate": 7.724668531098489e-07,
      "loss": 0.5308,
      "step": 6484
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8182183839372367,
      "learning_rate": 7.713654790324765e-07,
      "loss": 0.5327,
      "step": 6485
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6088936193910871,
      "learning_rate": 7.702648250475092e-07,
      "loss": 0.4662,
      "step": 6486
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6945340773217124,
      "learning_rate": 7.691648913423772e-07,
      "loss": 0.5371,
      "step": 6487
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8052921052230595,
      "learning_rate": 7.680656781043882e-07,
      "loss": 0.6224,
      "step": 6488
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6811345915823981,
      "learning_rate": 7.669671855207256e-07,
      "loss": 0.5459,
      "step": 6489
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8879446267811603,
      "learning_rate": 7.658694137784522e-07,
      "loss": 0.626,
      "step": 6490
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6489813321023465,
      "learning_rate": 7.64772363064507e-07,
      "loss": 0.5046,
      "step": 6491
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7594974198505549,
      "learning_rate": 7.636760335657056e-07,
      "loss": 0.5707,
      "step": 6492
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7211531903587437,
      "learning_rate": 7.625804254687425e-07,
      "loss": 0.5205,
      "step": 6493
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8114582002808618,
      "learning_rate": 7.614855389601872e-07,
      "loss": 0.6189,
      "step": 6494
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9914081689024536,
      "learning_rate": 7.603913742264885e-07,
      "loss": 0.6116,
      "step": 6495
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6756113126695669,
      "learning_rate": 7.592979314539711e-07,
      "loss": 0.509,
      "step": 6496
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7881046685308296,
      "learning_rate": 7.582052108288363e-07,
      "loss": 0.5875,
      "step": 6497
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7854487899893072,
      "learning_rate": 7.571132125371633e-07,
      "loss": 0.6147,
      "step": 6498
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8426822431831505,
      "learning_rate": 7.560219367649069e-07,
      "loss": 0.6141,
      "step": 6499
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7097615281401756,
      "learning_rate": 7.549313836979027e-07,
      "loss": 0.5226,
      "step": 6500
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8762150764731828,
      "learning_rate": 7.538415535218591e-07,
      "loss": 0.5852,
      "step": 6501
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9341001814991424,
      "learning_rate": 7.527524464223629e-07,
      "loss": 0.6607,
      "step": 6502
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8474017520442968,
      "learning_rate": 7.51664062584877e-07,
      "loss": 0.702,
      "step": 6503
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8612114281507735,
      "learning_rate": 7.505764021947426e-07,
      "loss": 0.5472,
      "step": 6504
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6497753113001647,
      "learning_rate": 7.494894654371771e-07,
      "loss": 0.6018,
      "step": 6505
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6210852166384915,
      "learning_rate": 7.484032524972739e-07,
      "loss": 0.4705,
      "step": 6506
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9260109906457434,
      "learning_rate": 7.473177635600037e-07,
      "loss": 0.6305,
      "step": 6507
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7667925016785997,
      "learning_rate": 7.462329988102147e-07,
      "loss": 0.6503,
      "step": 6508
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7437470904870618,
      "learning_rate": 7.451489584326304e-07,
      "loss": 0.5631,
      "step": 6509
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7458368296289911,
      "learning_rate": 7.440656426118515e-07,
      "loss": 0.5599,
      "step": 6510
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6960985561781085,
      "learning_rate": 7.42983051532356e-07,
      "loss": 0.5344,
      "step": 6511
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6847519213495826,
      "learning_rate": 7.419011853784974e-07,
      "loss": 0.5393,
      "step": 6512
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6739700716475066,
      "learning_rate": 7.408200443345065e-07,
      "loss": 0.4921,
      "step": 6513
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7677987528044851,
      "learning_rate": 7.397396285844898e-07,
      "loss": 0.5781,
      "step": 6514
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6881564934431853,
      "learning_rate": 7.386599383124321e-07,
      "loss": 0.523,
      "step": 6515
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7891656762827007,
      "learning_rate": 7.375809737021922e-07,
      "loss": 0.6497,
      "step": 6516
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6748799703713955,
      "learning_rate": 7.365027349375065e-07,
      "loss": 0.5322,
      "step": 6517
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8133659434698493,
      "learning_rate": 7.354252222019887e-07,
      "loss": 0.5289,
      "step": 6518
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8721716971664203,
      "learning_rate": 7.343484356791275e-07,
      "loss": 0.5698,
      "step": 6519
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8306903272994864,
      "learning_rate": 7.33272375552288e-07,
      "loss": 0.705,
      "step": 6520
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7301242419997787,
      "learning_rate": 7.321970420047131e-07,
      "loss": 0.5408,
      "step": 6521
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.851566074482624,
      "learning_rate": 7.311224352195201e-07,
      "loss": 0.6491,
      "step": 6522
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8994519614635279,
      "learning_rate": 7.300485553797037e-07,
      "loss": 0.6776,
      "step": 6523
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9604227114442911,
      "learning_rate": 7.289754026681339e-07,
      "loss": 0.6479,
      "step": 6524
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7741664738252255,
      "learning_rate": 7.279029772675572e-07,
      "loss": 0.5746,
      "step": 6525
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9252040164173223,
      "learning_rate": 7.268312793605975e-07,
      "loss": 0.6237,
      "step": 6526
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8382225232199589,
      "learning_rate": 7.257603091297538e-07,
      "loss": 0.5213,
      "step": 6527
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7815582144118317,
      "learning_rate": 7.246900667574009e-07,
      "loss": 0.5466,
      "step": 6528
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.772197561483472,
      "learning_rate": 7.236205524257894e-07,
      "loss": 0.5132,
      "step": 6529
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7428604057593859,
      "learning_rate": 7.225517663170473e-07,
      "loss": 0.5197,
      "step": 6530
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7127185987053908,
      "learning_rate": 7.214837086131771e-07,
      "loss": 0.5333,
      "step": 6531
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.903382842631152,
      "learning_rate": 7.204163794960578e-07,
      "loss": 0.6325,
      "step": 6532
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7398743569942173,
      "learning_rate": 7.193497791474446e-07,
      "loss": 0.5368,
      "step": 6533
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9228566647147254,
      "learning_rate": 7.182839077489689e-07,
      "loss": 0.6272,
      "step": 6534
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7204189569886309,
      "learning_rate": 7.172187654821372e-07,
      "loss": 0.5373,
      "step": 6535
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8046945371198233,
      "learning_rate": 7.161543525283315e-07,
      "loss": 0.6012,
      "step": 6536
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9334020107419905,
      "learning_rate": 7.150906690688114e-07,
      "loss": 0.6378,
      "step": 6537
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8252522540635658,
      "learning_rate": 7.140277152847102e-07,
      "loss": 0.5813,
      "step": 6538
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8441720435695788,
      "learning_rate": 7.129654913570383e-07,
      "loss": 0.6096,
      "step": 6539
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.046191285128024,
      "learning_rate": 7.119039974666809e-07,
      "loss": 0.6358,
      "step": 6540
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7897050072671297,
      "learning_rate": 7.108432337943999e-07,
      "loss": 0.5614,
      "step": 6541
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.925825742956098,
      "learning_rate": 7.097832005208322e-07,
      "loss": 0.6194,
      "step": 6542
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6639325753349582,
      "learning_rate": 7.087238978264899e-07,
      "loss": 0.525,
      "step": 6543
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8458210571181917,
      "learning_rate": 7.076653258917615e-07,
      "loss": 0.5256,
      "step": 6544
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8970400226029809,
      "learning_rate": 7.066074848969112e-07,
      "loss": 0.6703,
      "step": 6545
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7288540836577638,
      "learning_rate": 7.055503750220783e-07,
      "loss": 0.5601,
      "step": 6546
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7880073716352396,
      "learning_rate": 7.044939964472769e-07,
      "loss": 0.6089,
      "step": 6547
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.807433246236488,
      "learning_rate": 7.034383493523978e-07,
      "loss": 0.6235,
      "step": 6548
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8460268456842512,
      "learning_rate": 7.023834339172064e-07,
      "loss": 0.6064,
      "step": 6549
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7127437865258521,
      "learning_rate": 7.013292503213443e-07,
      "loss": 0.564,
      "step": 6550
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8372618732884886,
      "learning_rate": 7.002757987443271e-07,
      "loss": 0.6523,
      "step": 6551
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6510529742660217,
      "learning_rate": 6.992230793655464e-07,
      "loss": 0.4872,
      "step": 6552
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8825271137113972,
      "learning_rate": 6.98171092364271e-07,
      "loss": 0.6358,
      "step": 6553
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9186629204336472,
      "learning_rate": 6.971198379196425e-07,
      "loss": 0.6642,
      "step": 6554
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6899110144642793,
      "learning_rate": 6.960693162106791e-07,
      "loss": 0.5621,
      "step": 6555
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7517082236620101,
      "learning_rate": 6.950195274162725e-07,
      "loss": 0.5229,
      "step": 6556
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.63533970528908,
      "learning_rate": 6.939704717151913e-07,
      "loss": 0.5014,
      "step": 6557
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8511302440395219,
      "learning_rate": 6.929221492860794e-07,
      "loss": 0.6717,
      "step": 6558
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6902038290606284,
      "learning_rate": 6.918745603074545e-07,
      "loss": 0.5023,
      "step": 6559
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6495942208233723,
      "learning_rate": 6.9082770495771e-07,
      "loss": 0.5581,
      "step": 6560
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6920938888084776,
      "learning_rate": 6.89781583415115e-07,
      "loss": 0.5082,
      "step": 6561
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8274598308488191,
      "learning_rate": 6.887361958578126e-07,
      "loss": 0.6074,
      "step": 6562
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8229797456862055,
      "learning_rate": 6.876915424638214e-07,
      "loss": 0.6803,
      "step": 6563
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7435217812970396,
      "learning_rate": 6.866476234110358e-07,
      "loss": 0.6069,
      "step": 6564
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6633564205218362,
      "learning_rate": 6.856044388772232e-07,
      "loss": 0.5188,
      "step": 6565
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9011349377109444,
      "learning_rate": 6.845619890400279e-07,
      "loss": 0.6222,
      "step": 6566
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7242900348331709,
      "learning_rate": 6.83520274076968e-07,
      "loss": 0.5769,
      "step": 6567
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7172579274798075,
      "learning_rate": 6.824792941654367e-07,
      "loss": 0.5016,
      "step": 6568
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7334040085187365,
      "learning_rate": 6.814390494827017e-07,
      "loss": 0.4865,
      "step": 6569
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6922043380499469,
      "learning_rate": 6.803995402059066e-07,
      "loss": 0.6136,
      "step": 6570
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8507336565683287,
      "learning_rate": 6.793607665120683e-07,
      "loss": 0.5847,
      "step": 6571
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8544148798495429,
      "learning_rate": 6.783227285780791e-07,
      "loss": 0.632,
      "step": 6572
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8094824740712867,
      "learning_rate": 6.772854265807067e-07,
      "loss": 0.5908,
      "step": 6573
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7314383141767523,
      "learning_rate": 6.762488606965923e-07,
      "loss": 0.5619,
      "step": 6574
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9592265697085839,
      "learning_rate": 6.752130311022526e-07,
      "loss": 0.6096,
      "step": 6575
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7136288729789788,
      "learning_rate": 6.741779379740782e-07,
      "loss": 0.5345,
      "step": 6576
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6801727801386589,
      "learning_rate": 6.731435814883347e-07,
      "loss": 0.5616,
      "step": 6577
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9150043542854669,
      "learning_rate": 6.721099618211613e-07,
      "loss": 0.5897,
      "step": 6578
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6573435031350073,
      "learning_rate": 6.710770791485749e-07,
      "loss": 0.494,
      "step": 6579
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8458907566928735,
      "learning_rate": 6.700449336464637e-07,
      "loss": 0.6492,
      "step": 6580
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7891506612026545,
      "learning_rate": 6.690135254905916e-07,
      "loss": 0.507,
      "step": 6581
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7363962078087404,
      "learning_rate": 6.679828548565953e-07,
      "loss": 0.5688,
      "step": 6582
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7483263661502465,
      "learning_rate": 6.669529219199877e-07,
      "loss": 0.5266,
      "step": 6583
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6614566801717008,
      "learning_rate": 6.659237268561569e-07,
      "loss": 0.5625,
      "step": 6584
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6997752087560764,
      "learning_rate": 6.648952698403638e-07,
      "loss": 0.5789,
      "step": 6585
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8622418962488968,
      "learning_rate": 6.638675510477433e-07,
      "loss": 0.5844,
      "step": 6586
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6496142281971781,
      "learning_rate": 6.628405706533053e-07,
      "loss": 0.4949,
      "step": 6587
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7752862159838645,
      "learning_rate": 6.618143288319334e-07,
      "loss": 0.62,
      "step": 6588
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6884560532238665,
      "learning_rate": 6.607888257583872e-07,
      "loss": 0.5267,
      "step": 6589
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9632589696241081,
      "learning_rate": 6.597640616072982e-07,
      "loss": 0.6133,
      "step": 6590
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.175798276460779,
      "learning_rate": 6.587400365531727e-07,
      "loss": 0.6573,
      "step": 6591
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6957212247020323,
      "learning_rate": 6.577167507703924e-07,
      "loss": 0.5223,
      "step": 6592
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8352043556627344,
      "learning_rate": 6.56694204433212e-07,
      "loss": 0.6036,
      "step": 6593
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9007777722190855,
      "learning_rate": 6.556723977157603e-07,
      "loss": 0.6608,
      "step": 6594
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2455559278980968,
      "learning_rate": 6.546513307920405e-07,
      "loss": 0.6454,
      "step": 6595
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.895344559397274,
      "learning_rate": 6.536310038359295e-07,
      "loss": 0.5853,
      "step": 6596
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8147482627636214,
      "learning_rate": 6.526114170211784e-07,
      "loss": 0.5642,
      "step": 6597
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7078279820985721,
      "learning_rate": 6.515925705214121e-07,
      "loss": 0.5498,
      "step": 6598
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7088968242192337,
      "learning_rate": 6.505744645101297e-07,
      "loss": 0.6174,
      "step": 6599
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6904135820440285,
      "learning_rate": 6.495570991607037e-07,
      "loss": 0.5551,
      "step": 6600
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7862387716656356,
      "learning_rate": 6.485404746463814e-07,
      "loss": 0.5577,
      "step": 6601
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.864439197272147,
      "learning_rate": 6.475245911402834e-07,
      "loss": 0.6271,
      "step": 6602
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9083952512629536,
      "learning_rate": 6.465094488154033e-07,
      "loss": 0.5907,
      "step": 6603
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.025873277259067,
      "learning_rate": 6.454950478446098e-07,
      "loss": 0.6752,
      "step": 6604
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7064691761211881,
      "learning_rate": 6.444813884006451e-07,
      "loss": 0.5554,
      "step": 6605
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8670253134946831,
      "learning_rate": 6.434684706561245e-07,
      "loss": 0.6097,
      "step": 6606
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.685954258627498,
      "learning_rate": 6.424562947835367e-07,
      "loss": 0.5271,
      "step": 6607
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8960855454330223,
      "learning_rate": 6.41444860955246e-07,
      "loss": 0.6234,
      "step": 6608
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6920440160616677,
      "learning_rate": 6.404341693434879e-07,
      "loss": 0.495,
      "step": 6609
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8065475664869005,
      "learning_rate": 6.394242201203721e-07,
      "loss": 0.6373,
      "step": 6610
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6925968971186459,
      "learning_rate": 6.384150134578848e-07,
      "loss": 0.5267,
      "step": 6611
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.755069660541801,
      "learning_rate": 6.374065495278819e-07,
      "loss": 0.5097,
      "step": 6612
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8191009070321031,
      "learning_rate": 6.363988285020945e-07,
      "loss": 0.5046,
      "step": 6613
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8960369799273593,
      "learning_rate": 6.353918505521273e-07,
      "loss": 0.6315,
      "step": 6614
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6764562784266911,
      "learning_rate": 6.343856158494571e-07,
      "loss": 0.5137,
      "step": 6615
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6454672083155628,
      "learning_rate": 6.333801245654364e-07,
      "loss": 0.4892,
      "step": 6616
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8045098208261016,
      "learning_rate": 6.323753768712892e-07,
      "loss": 0.621,
      "step": 6617
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8393630167875734,
      "learning_rate": 6.313713729381138e-07,
      "loss": 0.6323,
      "step": 6618
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9187812000656891,
      "learning_rate": 6.303681129368811e-07,
      "loss": 0.6311,
      "step": 6619
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7491693734229161,
      "learning_rate": 6.293655970384365e-07,
      "loss": 0.5697,
      "step": 6620
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.808694753028369,
      "learning_rate": 6.283638254134982e-07,
      "loss": 0.6031,
      "step": 6621
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6929546853512615,
      "learning_rate": 6.273627982326569e-07,
      "loss": 0.5125,
      "step": 6622
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.712977412043284,
      "learning_rate": 6.263625156663771e-07,
      "loss": 0.5327,
      "step": 6623
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7497393177219862,
      "learning_rate": 6.253629778849968e-07,
      "loss": 0.5211,
      "step": 6624
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7829935478188985,
      "learning_rate": 6.243641850587262e-07,
      "loss": 0.5412,
      "step": 6625
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8015704685460228,
      "learning_rate": 6.233661373576505e-07,
      "loss": 0.5993,
      "step": 6626
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7311230006799819,
      "learning_rate": 6.22368834951726e-07,
      "loss": 0.5025,
      "step": 6627
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6522736470894342,
      "learning_rate": 6.213722780107829e-07,
      "loss": 0.5278,
      "step": 6628
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9495278766822307,
      "learning_rate": 6.203764667045248e-07,
      "loss": 0.6341,
      "step": 6629
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6600076316837938,
      "learning_rate": 6.193814012025278e-07,
      "loss": 0.5431,
      "step": 6630
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7223962732789876,
      "learning_rate": 6.18387081674241e-07,
      "loss": 0.5033,
      "step": 6631
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0115853671943351,
      "learning_rate": 6.173935082889871e-07,
      "loss": 0.618,
      "step": 6632
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6169789762530803,
      "learning_rate": 6.164006812159612e-07,
      "loss": 0.4767,
      "step": 6633
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6001650551054903,
      "learning_rate": 6.15408600624231e-07,
      "loss": 0.4583,
      "step": 6634
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7520824242428634,
      "learning_rate": 6.14417266682738e-07,
      "loss": 0.5102,
      "step": 6635
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8362768901726602,
      "learning_rate": 6.134266795602944e-07,
      "loss": 0.6298,
      "step": 6636
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7315763207232844,
      "learning_rate": 6.124368394255898e-07,
      "loss": 0.5012,
      "step": 6637
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.852498104191264,
      "learning_rate": 6.114477464471819e-07,
      "loss": 0.6188,
      "step": 6638
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9094063787593737,
      "learning_rate": 6.104594007935033e-07,
      "loss": 0.6702,
      "step": 6639
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8265869620461759,
      "learning_rate": 6.094718026328583e-07,
      "loss": 0.5774,
      "step": 6640
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6946989349757569,
      "learning_rate": 6.084849521334257e-07,
      "loss": 0.5109,
      "step": 6641
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7640435512437086,
      "learning_rate": 6.07498849463255e-07,
      "loss": 0.5647,
      "step": 6642
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9579411548752673,
      "learning_rate": 6.065134947902695e-07,
      "loss": 0.645,
      "step": 6643
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8216598148726845,
      "learning_rate": 6.055288882822652e-07,
      "loss": 0.6178,
      "step": 6644
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9543165896804842,
      "learning_rate": 6.0454503010691e-07,
      "loss": 0.4812,
      "step": 6645
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9689780840641171,
      "learning_rate": 6.035619204317439e-07,
      "loss": 0.5543,
      "step": 6646
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6761672107852109,
      "learning_rate": 6.025795594241817e-07,
      "loss": 0.5438,
      "step": 6647
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6635736510253682,
      "learning_rate": 6.015979472515082e-07,
      "loss": 0.5071,
      "step": 6648
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9002661217661749,
      "learning_rate": 6.006170840808823e-07,
      "loss": 0.6033,
      "step": 6649
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5050216595890071,
      "learning_rate": 5.996369700793342e-07,
      "loss": 0.688,
      "step": 6650
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.630109754315328,
      "learning_rate": 5.986576054137677e-07,
      "loss": 0.5178,
      "step": 6651
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9155447520534753,
      "learning_rate": 5.976789902509583e-07,
      "loss": 0.6613,
      "step": 6652
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9820046030654918,
      "learning_rate": 5.967011247575532e-07,
      "loss": 0.6213,
      "step": 6653
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6558362385868037,
      "learning_rate": 5.957240091000738e-07,
      "loss": 0.5553,
      "step": 6654
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6903335376409027,
      "learning_rate": 5.947476434449123e-07,
      "loss": 0.5505,
      "step": 6655
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.64420493720284,
      "learning_rate": 5.937720279583337e-07,
      "loss": 0.5081,
      "step": 6656
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8145204462810093,
      "learning_rate": 5.927971628064744e-07,
      "loss": 0.5657,
      "step": 6657
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.001511706551358,
      "learning_rate": 5.918230481553444e-07,
      "loss": 0.6998,
      "step": 6658
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8172986374485259,
      "learning_rate": 5.908496841708255e-07,
      "loss": 0.6622,
      "step": 6659
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6178531025454447,
      "learning_rate": 5.898770710186714e-07,
      "loss": 0.4946,
      "step": 6660
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7907835344553366,
      "learning_rate": 5.889052088645075e-07,
      "loss": 0.5728,
      "step": 6661
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8495992604157405,
      "learning_rate": 5.879340978738313e-07,
      "loss": 0.5893,
      "step": 6662
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8712011458641244,
      "learning_rate": 5.869637382120142e-07,
      "loss": 0.5844,
      "step": 6663
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7367028112167259,
      "learning_rate": 5.859941300442984e-07,
      "loss": 0.5627,
      "step": 6664
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6850367441887795,
      "learning_rate": 5.850252735357969e-07,
      "loss": 0.543,
      "step": 6665
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8034512080275277,
      "learning_rate": 5.840571688514968e-07,
      "loss": 0.5555,
      "step": 6666
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0041028325002337,
      "learning_rate": 5.83089816156256e-07,
      "loss": 0.6227,
      "step": 6667
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7423828439667035,
      "learning_rate": 5.821232156148038e-07,
      "loss": 0.5036,
      "step": 6668
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6867344045954835,
      "learning_rate": 5.811573673917437e-07,
      "loss": 0.5458,
      "step": 6669
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7532799713461582,
      "learning_rate": 5.801922716515484e-07,
      "loss": 0.5531,
      "step": 6670
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7337520098242731,
      "learning_rate": 5.792279285585639e-07,
      "loss": 0.5768,
      "step": 6671
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5865350997259149,
      "learning_rate": 5.782643382770081e-07,
      "loss": 0.4887,
      "step": 6672
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6961869592594258,
      "learning_rate": 5.773015009709699e-07,
      "loss": 0.5084,
      "step": 6673
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8443426684067711,
      "learning_rate": 5.763394168044112e-07,
      "loss": 0.5076,
      "step": 6674
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8130278404048547,
      "learning_rate": 5.75378085941164e-07,
      "loss": 0.6074,
      "step": 6675
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8791188186089853,
      "learning_rate": 5.744175085449338e-07,
      "loss": 0.6357,
      "step": 6676
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8281016205869466,
      "learning_rate": 5.734576847792961e-07,
      "loss": 0.6446,
      "step": 6677
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.783230280814367,
      "learning_rate": 5.724986148076995e-07,
      "loss": 0.6188,
      "step": 6678
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7720714095568783,
      "learning_rate": 5.715402987934632e-07,
      "loss": 0.6254,
      "step": 6679
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9144659141459703,
      "learning_rate": 5.70582736899779e-07,
      "loss": 0.5646,
      "step": 6680
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6811205167451434,
      "learning_rate": 5.696259292897094e-07,
      "loss": 0.5407,
      "step": 6681
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7029733085662122,
      "learning_rate": 5.686698761261872e-07,
      "loss": 0.5328,
      "step": 6682
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7271438792959343,
      "learning_rate": 5.677145775720216e-07,
      "loss": 0.5423,
      "step": 6683
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8482020211725333,
      "learning_rate": 5.667600337898876e-07,
      "loss": 0.61,
      "step": 6684
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8934355627262323,
      "learning_rate": 5.658062449423352e-07,
      "loss": 0.5782,
      "step": 6685
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.732886356591622,
      "learning_rate": 5.648532111917837e-07,
      "loss": 0.5205,
      "step": 6686
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7406662787341809,
      "learning_rate": 5.639009327005257e-07,
      "loss": 0.5273,
      "step": 6687
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8336562427239234,
      "learning_rate": 5.629494096307236e-07,
      "loss": 0.5765,
      "step": 6688
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7316241170428661,
      "learning_rate": 5.619986421444124e-07,
      "loss": 0.5216,
      "step": 6689
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8118423680253526,
      "learning_rate": 5.610486304034973e-07,
      "loss": 0.5909,
      "step": 6690
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6992098729325374,
      "learning_rate": 5.600993745697564e-07,
      "loss": 0.5521,
      "step": 6691
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9295298419127371,
      "learning_rate": 5.591508748048369e-07,
      "loss": 0.6369,
      "step": 6692
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.701918935071087,
      "learning_rate": 5.582031312702596e-07,
      "loss": 0.5326,
      "step": 6693
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8317144429426071,
      "learning_rate": 5.572561441274132e-07,
      "loss": 0.6267,
      "step": 6694
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7718526690059179,
      "learning_rate": 5.563099135375622e-07,
      "loss": 0.5816,
      "step": 6695
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7734736046367386,
      "learning_rate": 5.55364439661839e-07,
      "loss": 0.5425,
      "step": 6696
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6256423017637482,
      "learning_rate": 5.544197226612486e-07,
      "loss": 0.5377,
      "step": 6697
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8050142198012837,
      "learning_rate": 5.534757626966647e-07,
      "loss": 0.5776,
      "step": 6698
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9749231815572074,
      "learning_rate": 5.525325599288356e-07,
      "loss": 0.6004,
      "step": 6699
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9218144609161856,
      "learning_rate": 5.515901145183778e-07,
      "loss": 0.6094,
      "step": 6700
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6859431604713325,
      "learning_rate": 5.5064842662578e-07,
      "loss": 0.5202,
      "step": 6701
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6858568460525114,
      "learning_rate": 5.497074964114024e-07,
      "loss": 0.5397,
      "step": 6702
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.680589784458555,
      "learning_rate": 5.487673240354757e-07,
      "loss": 0.5687,
      "step": 6703
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7866750202681904,
      "learning_rate": 5.478279096581008e-07,
      "loss": 0.6138,
      "step": 6704
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8416185289132421,
      "learning_rate": 5.468892534392506e-07,
      "loss": 0.6023,
      "step": 6705
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6983827388076834,
      "learning_rate": 5.45951355538768e-07,
      "loss": 0.4549,
      "step": 6706
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9529964173405592,
      "learning_rate": 5.450142161163679e-07,
      "loss": 0.623,
      "step": 6707
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7934857079047395,
      "learning_rate": 5.440778353316351e-07,
      "loss": 0.5279,
      "step": 6708
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0550014155465883,
      "learning_rate": 5.431422133440256e-07,
      "loss": 0.617,
      "step": 6709
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6292006488676265,
      "learning_rate": 5.422073503128655e-07,
      "loss": 0.5203,
      "step": 6710
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9030709829990416,
      "learning_rate": 5.412732463973531e-07,
      "loss": 0.6435,
      "step": 6711
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7470482344475862,
      "learning_rate": 5.403399017565563e-07,
      "loss": 0.5252,
      "step": 6712
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8329525197801656,
      "learning_rate": 5.394073165494135e-07,
      "loss": 0.6394,
      "step": 6713
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7564476104651021,
      "learning_rate": 5.384754909347345e-07,
      "loss": 0.555,
      "step": 6714
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7557044998004324,
      "learning_rate": 5.375444250711997e-07,
      "loss": 0.5702,
      "step": 6715
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6666910710711206,
      "learning_rate": 5.366141191173602e-07,
      "loss": 0.5073,
      "step": 6716
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8077592649107482,
      "learning_rate": 5.356845732316368e-07,
      "loss": 0.6512,
      "step": 6717
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7297283234079667,
      "learning_rate": 5.347557875723214e-07,
      "loss": 0.5567,
      "step": 6718
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7379214844064317,
      "learning_rate": 5.338277622975769e-07,
      "loss": 0.5573,
      "step": 6719
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7973404610329959,
      "learning_rate": 5.329004975654361e-07,
      "loss": 0.6162,
      "step": 6720
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8953338522024854,
      "learning_rate": 5.31973993533802e-07,
      "loss": 0.6205,
      "step": 6721
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.624479714645648,
      "learning_rate": 5.310482503604497e-07,
      "loss": 0.4784,
      "step": 6722
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6572203102259121,
      "learning_rate": 5.301232682030239e-07,
      "loss": 0.5187,
      "step": 6723
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.746509996698656,
      "learning_rate": 5.291990472190378e-07,
      "loss": 0.5764,
      "step": 6724
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7252402771027143,
      "learning_rate": 5.28275587565878e-07,
      "loss": 0.5205,
      "step": 6725
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.769748151715065,
      "learning_rate": 5.273528894007995e-07,
      "loss": 0.5533,
      "step": 6726
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9482714920919001,
      "learning_rate": 5.264309528809286e-07,
      "loss": 0.6273,
      "step": 6727
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8245267925991313,
      "learning_rate": 5.255097781632606e-07,
      "loss": 0.5199,
      "step": 6728
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7754770650072839,
      "learning_rate": 5.245893654046625e-07,
      "loss": 0.5463,
      "step": 6729
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7053218324500792,
      "learning_rate": 5.236697147618713e-07,
      "loss": 0.4901,
      "step": 6730
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.681869034081981,
      "learning_rate": 5.227508263914933e-07,
      "loss": 0.5462,
      "step": 6731
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6961163291499552,
      "learning_rate": 5.218327004500062e-07,
      "loss": 0.5314,
      "step": 6732
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6538120026732325,
      "learning_rate": 5.209153370937569e-07,
      "loss": 0.5247,
      "step": 6733
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6701074681331374,
      "learning_rate": 5.199987364789627e-07,
      "loss": 0.4855,
      "step": 6734
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8418016593954959,
      "learning_rate": 5.190828987617119e-07,
      "loss": 0.5844,
      "step": 6735
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0377313602386256,
      "learning_rate": 5.18167824097961e-07,
      "loss": 0.6338,
      "step": 6736
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8535672143176243,
      "learning_rate": 5.172535126435385e-07,
      "loss": 0.6308,
      "step": 6737
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6670861902778317,
      "learning_rate": 5.163399645541417e-07,
      "loss": 0.5323,
      "step": 6738
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7787532767284215,
      "learning_rate": 5.154271799853389e-07,
      "loss": 0.5894,
      "step": 6739
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7249869752081559,
      "learning_rate": 5.145151590925673e-07,
      "loss": 0.5289,
      "step": 6740
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6668592754131172,
      "learning_rate": 5.136039020311345e-07,
      "loss": 0.5567,
      "step": 6741
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8118886276253745,
      "learning_rate": 5.126934089562186e-07,
      "loss": 0.6115,
      "step": 6742
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9056117908349427,
      "learning_rate": 5.11783680022866e-07,
      "loss": 0.5527,
      "step": 6743
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7347999822566383,
      "learning_rate": 5.108747153859956e-07,
      "loss": 0.5184,
      "step": 6744
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8355829946032209,
      "learning_rate": 5.099665152003929e-07,
      "loss": 0.5964,
      "step": 6745
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6416412954058793,
      "learning_rate": 5.090590796207168e-07,
      "loss": 0.4566,
      "step": 6746
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8396672141426126,
      "learning_rate": 5.081524088014916e-07,
      "loss": 0.6658,
      "step": 6747
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8610716348412157,
      "learning_rate": 5.072465028971168e-07,
      "loss": 0.6407,
      "step": 6748
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6193157597771248,
      "learning_rate": 5.063413620618568e-07,
      "loss": 0.5334,
      "step": 6749
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8897653668198241,
      "learning_rate": 5.054369864498487e-07,
      "loss": 0.5912,
      "step": 6750
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7508538914643151,
      "learning_rate": 5.045333762150978e-07,
      "loss": 0.5816,
      "step": 6751
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6397178648767955,
      "learning_rate": 5.036305315114792e-07,
      "loss": 0.5154,
      "step": 6752
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7066585449650707,
      "learning_rate": 5.027284524927384e-07,
      "loss": 0.5581,
      "step": 6753
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8739775262682591,
      "learning_rate": 5.018271393124901e-07,
      "loss": 0.6349,
      "step": 6754
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4116853772701352,
      "learning_rate": 5.009265921242185e-07,
      "loss": 0.6274,
      "step": 6755
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8242199581327988,
      "learning_rate": 5.000268110812767e-07,
      "loss": 0.6128,
      "step": 6756
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7677744210600361,
      "learning_rate": 4.99127796336889e-07,
      "loss": 0.6398,
      "step": 6757
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.911019440437457,
      "learning_rate": 4.982295480441473e-07,
      "loss": 0.6458,
      "step": 6758
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6492105159862692,
      "learning_rate": 4.97332066356015e-07,
      "loss": 0.4959,
      "step": 6759
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6192634049603718,
      "learning_rate": 4.964353514253234e-07,
      "loss": 0.517,
      "step": 6760
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8562076206174601,
      "learning_rate": 4.955394034047734e-07,
      "loss": 0.5531,
      "step": 6761
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9545415698057342,
      "learning_rate": 4.946442224469356e-07,
      "loss": 0.5951,
      "step": 6762
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8673426410389913,
      "learning_rate": 4.937498087042508e-07,
      "loss": 0.5969,
      "step": 6763
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8770459392939213,
      "learning_rate": 4.928561623290273e-07,
      "loss": 0.664,
      "step": 6764
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8525983029934454,
      "learning_rate": 4.919632834734451e-07,
      "loss": 0.5981,
      "step": 6765
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8186084441320477,
      "learning_rate": 4.910711722895506e-07,
      "loss": 0.6199,
      "step": 6766
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.711732577811161,
      "learning_rate": 4.901798289292619e-07,
      "loss": 0.5136,
      "step": 6767
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8118998111549554,
      "learning_rate": 4.892892535443655e-07,
      "loss": 0.6057,
      "step": 6768
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7495231699115711,
      "learning_rate": 4.883994462865166e-07,
      "loss": 0.4684,
      "step": 6769
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7256228022868758,
      "learning_rate": 4.875104073072412e-07,
      "loss": 0.566,
      "step": 6770
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8726281530599873,
      "learning_rate": 4.866221367579321e-07,
      "loss": 0.6354,
      "step": 6771
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6618751822917529,
      "learning_rate": 4.857346347898534e-07,
      "loss": 0.5518,
      "step": 6772
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5953706836835683,
      "learning_rate": 4.848479015541357e-07,
      "loss": 0.4961,
      "step": 6773
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7022808077604614,
      "learning_rate": 4.839619372017829e-07,
      "loss": 0.5512,
      "step": 6774
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.838539482973408,
      "learning_rate": 4.830767418836646e-07,
      "loss": 0.521,
      "step": 6775
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8278782733605144,
      "learning_rate": 4.821923157505204e-07,
      "loss": 0.605,
      "step": 6776
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8249562991141697,
      "learning_rate": 4.813086589529587e-07,
      "loss": 0.5902,
      "step": 6777
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9229908607233321,
      "learning_rate": 4.804257716414568e-07,
      "loss": 0.6813,
      "step": 6778
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7503060863534825,
      "learning_rate": 4.795436539663611e-07,
      "loss": 0.6196,
      "step": 6779
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9066843622351747,
      "learning_rate": 4.786623060778878e-07,
      "loss": 0.5739,
      "step": 6780
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6401371007213178,
      "learning_rate": 4.777817281261205e-07,
      "loss": 0.5523,
      "step": 6781
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9687108354645209,
      "learning_rate": 4.769019202610131e-07,
      "loss": 0.5962,
      "step": 6782
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7686738272549012,
      "learning_rate": 4.7602288263238704e-07,
      "loss": 0.5638,
      "step": 6783
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.913255053557828,
      "learning_rate": 4.7514461538993306e-07,
      "loss": 0.6536,
      "step": 6784
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6467988572123885,
      "learning_rate": 4.7426711868321206e-07,
      "loss": 0.5335,
      "step": 6785
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.636413510811312,
      "learning_rate": 4.733903926616523e-07,
      "loss": 0.5066,
      "step": 6786
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7325449776923049,
      "learning_rate": 4.725144374745505e-07,
      "loss": 0.5856,
      "step": 6787
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7791334682511725,
      "learning_rate": 4.716392532710734e-07,
      "loss": 0.5576,
      "step": 6788
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6791214542215358,
      "learning_rate": 4.7076484020025515e-07,
      "loss": 0.5726,
      "step": 6789
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6854480019351055,
      "learning_rate": 4.6989119841099885e-07,
      "loss": 0.4893,
      "step": 6790
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7958724188131521,
      "learning_rate": 4.6901832805207767e-07,
      "loss": 0.5659,
      "step": 6791
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6175377429627246,
      "learning_rate": 4.6814622927213173e-07,
      "loss": 0.5306,
      "step": 6792
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7080386859850634,
      "learning_rate": 4.6727490221967045e-07,
      "loss": 0.4996,
      "step": 6793
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7368830841915116,
      "learning_rate": 4.664043470430718e-07,
      "loss": 0.5,
      "step": 6794
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7737900367302412,
      "learning_rate": 4.655345638905823e-07,
      "loss": 0.594,
      "step": 6795
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6922047848659214,
      "learning_rate": 4.646655529103167e-07,
      "loss": 0.5209,
      "step": 6796
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8008251346769502,
      "learning_rate": 4.6379731425025833e-07,
      "loss": 0.5682,
      "step": 6797
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6835830699019857,
      "learning_rate": 4.6292984805826005e-07,
      "loss": 0.5429,
      "step": 6798
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8126309441794485,
      "learning_rate": 4.620631544820414e-07,
      "loss": 0.5396,
      "step": 6799
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8292290049068698,
      "learning_rate": 4.6119723366919165e-07,
      "loss": 0.6184,
      "step": 6800
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.695727262243584,
      "learning_rate": 4.6033208576716837e-07,
      "loss": 0.5559,
      "step": 6801
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.686152965410741,
      "learning_rate": 4.59467710923297e-07,
      "loss": 0.5584,
      "step": 6802
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.869218102651399,
      "learning_rate": 4.586041092847715e-07,
      "loss": 0.6628,
      "step": 6803
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7033760378930044,
      "learning_rate": 4.5774128099865423e-07,
      "loss": 0.5452,
      "step": 6804
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8817873099132584,
      "learning_rate": 4.5687922621187486e-07,
      "loss": 0.6602,
      "step": 6805
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6778791026065074,
      "learning_rate": 4.560179450712343e-07,
      "loss": 0.5421,
      "step": 6806
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6867619275953373,
      "learning_rate": 4.551574377233986e-07,
      "loss": 0.5645,
      "step": 6807
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6442054288426311,
      "learning_rate": 4.5429770431490384e-07,
      "loss": 0.544,
      "step": 6808
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7733245367956972,
      "learning_rate": 4.534387449921529e-07,
      "loss": 0.5778,
      "step": 6809
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7175828112212366,
      "learning_rate": 4.5258055990141824e-07,
      "loss": 0.5527,
      "step": 6810
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8169946273011858,
      "learning_rate": 4.5172314918883966e-07,
      "loss": 0.6252,
      "step": 6811
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7653382954104186,
      "learning_rate": 4.508665130004247e-07,
      "loss": 0.5589,
      "step": 6812
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7825437685036122,
      "learning_rate": 4.500106514820507e-07,
      "loss": 0.6105,
      "step": 6813
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.703887035643475,
      "learning_rate": 4.49155564779461e-07,
      "loss": 0.5048,
      "step": 6814
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6794530379352032,
      "learning_rate": 4.48301253038268e-07,
      "loss": 0.4855,
      "step": 6815
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7483641478404777,
      "learning_rate": 4.474477164039526e-07,
      "loss": 0.6125,
      "step": 6816
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7479259027806663,
      "learning_rate": 4.465949550218629e-07,
      "loss": 0.5981,
      "step": 6817
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7668018295572554,
      "learning_rate": 4.457429690372156e-07,
      "loss": 0.5848,
      "step": 6818
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7771909454271351,
      "learning_rate": 4.4489175859509457e-07,
      "loss": 0.6349,
      "step": 6819
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8611013648956578,
      "learning_rate": 4.4404132384045175e-07,
      "loss": 0.5848,
      "step": 6820
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7529478721056361,
      "learning_rate": 4.4319166491810837e-07,
      "loss": 0.5377,
      "step": 6821
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7823556060885173,
      "learning_rate": 4.423427819727522e-07,
      "loss": 0.61,
      "step": 6822
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8724258091935903,
      "learning_rate": 4.414946751489385e-07,
      "loss": 0.6478,
      "step": 6823
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7172657941061589,
      "learning_rate": 4.4064734459109196e-07,
      "loss": 0.5273,
      "step": 6824
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8003943794910167,
      "learning_rate": 4.398007904435031e-07,
      "loss": 0.5955,
      "step": 6825
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6323228053391567,
      "learning_rate": 4.389550128503317e-07,
      "loss": 0.5389,
      "step": 6826
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6960791593406801,
      "learning_rate": 4.381100119556053e-07,
      "loss": 0.5472,
      "step": 6827
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7696280108354582,
      "learning_rate": 4.3726578790321883e-07,
      "loss": 0.5116,
      "step": 6828
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7935042196525566,
      "learning_rate": 4.3642234083693394e-07,
      "loss": 0.5808,
      "step": 6829
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9761111474908527,
      "learning_rate": 4.355796709003812e-07,
      "loss": 0.6043,
      "step": 6830
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7815814672097576,
      "learning_rate": 4.347377782370593e-07,
      "loss": 0.5275,
      "step": 6831
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9070702883255608,
      "learning_rate": 4.3389666299033187e-07,
      "loss": 0.5954,
      "step": 6832
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7346600417424236,
      "learning_rate": 4.330563253034348e-07,
      "loss": 0.6451,
      "step": 6833
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7604800538695748,
      "learning_rate": 4.32216765319467e-07,
      "loss": 0.5662,
      "step": 6834
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7100245554364574,
      "learning_rate": 4.3137798318139743e-07,
      "loss": 0.5223,
      "step": 6835
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6917532187602273,
      "learning_rate": 4.305399790320619e-07,
      "loss": 0.5409,
      "step": 6836
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7800980219393966,
      "learning_rate": 4.297027530141634e-07,
      "loss": 0.5641,
      "step": 6837
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4123768643865868,
      "learning_rate": 4.2886630527027296e-07,
      "loss": 0.6374,
      "step": 6838
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7812681699758023,
      "learning_rate": 4.2803063594282933e-07,
      "loss": 0.5542,
      "step": 6839
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7374181134282078,
      "learning_rate": 4.271957451741376e-07,
      "loss": 0.5465,
      "step": 6840
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7627086107356028,
      "learning_rate": 4.2636163310637136e-07,
      "loss": 0.5041,
      "step": 6841
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8017589177021514,
      "learning_rate": 4.255282998815713e-07,
      "loss": 0.559,
      "step": 6842
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.392871557551146,
      "learning_rate": 4.246957456416456e-07,
      "loss": 0.5955,
      "step": 6843
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8788631862402585,
      "learning_rate": 4.2386397052836924e-07,
      "loss": 0.5684,
      "step": 6844
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8175707495151532,
      "learning_rate": 4.230329746833844e-07,
      "loss": 0.6572,
      "step": 6845
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.076979857917219,
      "learning_rate": 4.2220275824820136e-07,
      "loss": 0.6191,
      "step": 6846
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6767655043363255,
      "learning_rate": 4.2137332136419796e-07,
      "loss": 0.5426,
      "step": 6847
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8756275194949769,
      "learning_rate": 4.2054466417261797e-07,
      "loss": 0.5679,
      "step": 6848
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6903222954808087,
      "learning_rate": 4.1971678681457294e-07,
      "loss": 0.5825,
      "step": 6849
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0013682367500893,
      "learning_rate": 4.1888968943104236e-07,
      "loss": 0.6232,
      "step": 6850
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9861620341408214,
      "learning_rate": 4.180633721628724e-07,
      "loss": 0.5604,
      "step": 6851
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6825068009700247,
      "learning_rate": 4.1723783515077555e-07,
      "loss": 0.5647,
      "step": 6852
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.865924622991732,
      "learning_rate": 4.164130785353321e-07,
      "loss": 0.6838,
      "step": 6853
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.846563986788896,
      "learning_rate": 4.1558910245699027e-07,
      "loss": 0.6136,
      "step": 6854
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6599117250326576,
      "learning_rate": 4.1476590705606445e-07,
      "loss": 0.5418,
      "step": 6855
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.738252561715404,
      "learning_rate": 4.139434924727359e-07,
      "loss": 0.5662,
      "step": 6856
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7003117425747266,
      "learning_rate": 4.131218588470531e-07,
      "loss": 0.4961,
      "step": 6857
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6164644232338188,
      "learning_rate": 4.1230100631893147e-07,
      "loss": 0.5168,
      "step": 6858
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7704882681491474,
      "learning_rate": 4.1148093502815456e-07,
      "loss": 0.5307,
      "step": 6859
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8390704993099708,
      "learning_rate": 4.106616451143719e-07,
      "loss": 0.5816,
      "step": 6860
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.878284303444981,
      "learning_rate": 4.0984313671709964e-07,
      "loss": 0.5812,
      "step": 6861
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7511873474742895,
      "learning_rate": 4.0902540997572126e-07,
      "loss": 0.5249,
      "step": 6862
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7069526483100713,
      "learning_rate": 4.082084650294876e-07,
      "loss": 0.5638,
      "step": 6863
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6295548390546648,
      "learning_rate": 4.0739230201751467e-07,
      "loss": 0.5217,
      "step": 6864
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8218383948225362,
      "learning_rate": 4.0657692107878786e-07,
      "loss": 0.5297,
      "step": 6865
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.716755321479113,
      "learning_rate": 4.057623223521573e-07,
      "loss": 0.6145,
      "step": 6866
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7018049628331753,
      "learning_rate": 4.0494850597634085e-07,
      "loss": 0.5328,
      "step": 6867
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6346296975776975,
      "learning_rate": 4.0413547208992266e-07,
      "loss": 0.4875,
      "step": 6868
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6149852180625857,
      "learning_rate": 4.0332322083135423e-07,
      "loss": 0.5551,
      "step": 6869
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7891505455240773,
      "learning_rate": 4.025117523389538e-07,
      "loss": 0.61,
      "step": 6870
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9327963460010891,
      "learning_rate": 4.0170106675090525e-07,
      "loss": 0.6267,
      "step": 6871
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6265477901565917,
      "learning_rate": 4.008911642052604e-07,
      "loss": 0.4838,
      "step": 6872
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9771155422037211,
      "learning_rate": 4.0008204483993727e-07,
      "loss": 0.6863,
      "step": 6873
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7949880738497038,
      "learning_rate": 3.992737087927201e-07,
      "loss": 0.6243,
      "step": 6874
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7942470456138682,
      "learning_rate": 3.984661562012604e-07,
      "loss": 0.5645,
      "step": 6875
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8241258590694551,
      "learning_rate": 3.9765938720307597e-07,
      "loss": 0.622,
      "step": 6876
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7746195041475429,
      "learning_rate": 3.9685340193555134e-07,
      "loss": 0.5296,
      "step": 6877
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7994063499442703,
      "learning_rate": 3.9604820053593674e-07,
      "loss": 0.6392,
      "step": 6878
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.846324670541516,
      "learning_rate": 3.952437831413497e-07,
      "loss": 0.5781,
      "step": 6879
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.837393186171623,
      "learning_rate": 3.944401498887751e-07,
      "loss": 0.5723,
      "step": 6880
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0279746020972238,
      "learning_rate": 3.936373009150618e-07,
      "loss": 0.6141,
      "step": 6881
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7405639040079075,
      "learning_rate": 3.9283523635692824e-07,
      "loss": 0.5521,
      "step": 6882
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7662720425572059,
      "learning_rate": 3.920339563509562e-07,
      "loss": 0.6346,
      "step": 6883
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.661707733285121,
      "learning_rate": 3.912334610335955e-07,
      "loss": 0.5168,
      "step": 6884
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8866194093469825,
      "learning_rate": 3.904337505411632e-07,
      "loss": 0.6181,
      "step": 6885
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6639006598345805,
      "learning_rate": 3.8963482500984084e-07,
      "loss": 0.5233,
      "step": 6886
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7089907406100479,
      "learning_rate": 3.88836684575678e-07,
      "loss": 0.5224,
      "step": 6887
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9709052690857305,
      "learning_rate": 3.880393293745882e-07,
      "loss": 0.6476,
      "step": 6888
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7078290522664243,
      "learning_rate": 3.872427595423545e-07,
      "loss": 0.5246,
      "step": 6889
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7773499366125478,
      "learning_rate": 3.8644697521462217e-07,
      "loss": 0.6141,
      "step": 6890
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7890174550710611,
      "learning_rate": 3.856519765269068e-07,
      "loss": 0.61,
      "step": 6891
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7002798302807485,
      "learning_rate": 3.8485776361458837e-07,
      "loss": 0.5213,
      "step": 6892
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7727571508671105,
      "learning_rate": 3.8406433661291264e-07,
      "loss": 0.6327,
      "step": 6893
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7869232079443725,
      "learning_rate": 3.832716956569915e-07,
      "loss": 0.6688,
      "step": 6894
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.750352468157192,
      "learning_rate": 3.8247984088180425e-07,
      "loss": 0.5299,
      "step": 6895
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6735478717148347,
      "learning_rate": 3.8168877242219517e-07,
      "loss": 0.547,
      "step": 6896
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9261104360253203,
      "learning_rate": 3.808984904128743e-07,
      "loss": 0.6403,
      "step": 6897
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9402187707065872,
      "learning_rate": 3.8010899498841956e-07,
      "loss": 0.5862,
      "step": 6898
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9271254088548732,
      "learning_rate": 3.7932028628327346e-07,
      "loss": 0.6467,
      "step": 6899
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9380125441395577,
      "learning_rate": 3.785323644317446e-07,
      "loss": 0.6566,
      "step": 6900
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7674381775107388,
      "learning_rate": 3.7774522956800786e-07,
      "loss": 0.6122,
      "step": 6901
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9153304736106843,
      "learning_rate": 3.7695888182610385e-07,
      "loss": 0.6076,
      "step": 6902
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7997953495084293,
      "learning_rate": 3.7617332133994046e-07,
      "loss": 0.5367,
      "step": 6903
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6222969877308744,
      "learning_rate": 3.753885482432895e-07,
      "loss": 0.5279,
      "step": 6904
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7001504885983985,
      "learning_rate": 3.746045626697897e-07,
      "loss": 0.5526,
      "step": 6905
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6752696819991477,
      "learning_rate": 3.738213647529459e-07,
      "loss": 0.4499,
      "step": 6906
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.709618406756323,
      "learning_rate": 3.730389546261287e-07,
      "loss": 0.5586,
      "step": 6907
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9123097924192907,
      "learning_rate": 3.7225733242257424e-07,
      "loss": 0.6448,
      "step": 6908
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9488789912478209,
      "learning_rate": 3.714764982753843e-07,
      "loss": 0.6166,
      "step": 6909
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2451797262702748,
      "learning_rate": 3.7069645231752713e-07,
      "loss": 0.6686,
      "step": 6910
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8699991128307344,
      "learning_rate": 3.6991719468183696e-07,
      "loss": 0.6018,
      "step": 6911
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6653320708322982,
      "learning_rate": 3.6913872550101206e-07,
      "loss": 0.5214,
      "step": 6912
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6648400282397972,
      "learning_rate": 3.683610449076186e-07,
      "loss": 0.5315,
      "step": 6913
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8988892255364088,
      "learning_rate": 3.6758415303408677e-07,
      "loss": 0.6633,
      "step": 6914
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8331837088938187,
      "learning_rate": 3.6680805001271413e-07,
      "loss": 0.6502,
      "step": 6915
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8522064293592045,
      "learning_rate": 3.66032735975661e-07,
      "loss": 0.5894,
      "step": 6916
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7267399888318924,
      "learning_rate": 3.652582110549574e-07,
      "loss": 0.556,
      "step": 6917
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6525631118985582,
      "learning_rate": 3.6448447538249607e-07,
      "loss": 0.5178,
      "step": 6918
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6678621811405187,
      "learning_rate": 3.6371152909003617e-07,
      "loss": 0.5311,
      "step": 6919
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7755896507457781,
      "learning_rate": 3.6293937230920286e-07,
      "loss": 0.5775,
      "step": 6920
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7233229254905214,
      "learning_rate": 3.621680051714854e-07,
      "loss": 0.6355,
      "step": 6921
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7517800436879831,
      "learning_rate": 3.6139742780823993e-07,
      "loss": 0.5584,
      "step": 6922
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8555129232744482,
      "learning_rate": 3.606276403506881e-07,
      "loss": 0.6097,
      "step": 6923
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9443308726560262,
      "learning_rate": 3.598586429299167e-07,
      "loss": 0.5809,
      "step": 6924
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.090612087503481,
      "learning_rate": 3.5909043567687764e-07,
      "loss": 0.6676,
      "step": 6925
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.824867716377936,
      "learning_rate": 3.583230187223885e-07,
      "loss": 0.603,
      "step": 6926
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8694607584208969,
      "learning_rate": 3.575563921971331e-07,
      "loss": 0.6721,
      "step": 6927
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7461706664584211,
      "learning_rate": 3.5679055623165924e-07,
      "loss": 0.4813,
      "step": 6928
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8876390230410423,
      "learning_rate": 3.5602551095638094e-07,
      "loss": 0.5108,
      "step": 6929
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6559274253866437,
      "learning_rate": 3.5526125650157725e-07,
      "loss": 0.533,
      "step": 6930
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7457679782258849,
      "learning_rate": 3.544977929973936e-07,
      "loss": 0.5776,
      "step": 6931
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.702490934007531,
      "learning_rate": 3.537351205738393e-07,
      "loss": 0.5473,
      "step": 6932
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6445776896524193,
      "learning_rate": 3.529732393607893e-07,
      "loss": 0.51,
      "step": 6933
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6670299732439671,
      "learning_rate": 3.522121494879843e-07,
      "loss": 0.5552,
      "step": 6934
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7736167503730588,
      "learning_rate": 3.5145185108502954e-07,
      "loss": 0.5255,
      "step": 6935
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7171381500724084,
      "learning_rate": 3.50692344281397e-07,
      "loss": 0.5473,
      "step": 6936
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7235075676515516,
      "learning_rate": 3.4993362920642203e-07,
      "loss": 0.5106,
      "step": 6937
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6993561792287761,
      "learning_rate": 3.491757059893053e-07,
      "loss": 0.554,
      "step": 6938
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9627727313690821,
      "learning_rate": 3.4841857475911454e-07,
      "loss": 0.5873,
      "step": 6939
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9825846628224079,
      "learning_rate": 3.4766223564478063e-07,
      "loss": 0.6347,
      "step": 6940
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7347986735468995,
      "learning_rate": 3.469066887750999e-07,
      "loss": 0.5362,
      "step": 6941
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8234358645824681,
      "learning_rate": 3.461519342787351e-07,
      "loss": 0.6099,
      "step": 6942
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9191586072061311,
      "learning_rate": 3.45397972284211e-07,
      "loss": 0.6638,
      "step": 6943
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6503457217687219,
      "learning_rate": 3.4464480291992177e-07,
      "loss": 0.5246,
      "step": 6944
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.582986678206098,
      "learning_rate": 3.4389242631412413e-07,
      "loss": 0.4604,
      "step": 6945
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.637010140064547,
      "learning_rate": 3.4314084259493897e-07,
      "loss": 0.4951,
      "step": 6946
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7290929898169166,
      "learning_rate": 3.4239005189035335e-07,
      "loss": 0.5715,
      "step": 6947
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.27325381155192,
      "learning_rate": 3.416400543282189e-07,
      "loss": 0.571,
      "step": 6948
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8251831808010429,
      "learning_rate": 3.408908500362534e-07,
      "loss": 0.6635,
      "step": 6949
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8581039923204896,
      "learning_rate": 3.401424391420377e-07,
      "loss": 0.5924,
      "step": 6950
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7975402178987853,
      "learning_rate": 3.393948217730186e-07,
      "loss": 0.6368,
      "step": 6951
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6974268376522019,
      "learning_rate": 3.386479980565077e-07,
      "loss": 0.5624,
      "step": 6952
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6936500310076437,
      "learning_rate": 3.379019681196805e-07,
      "loss": 0.4911,
      "step": 6953
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9422566394040627,
      "learning_rate": 3.371567320895791e-07,
      "loss": 0.6401,
      "step": 6954
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7677258658613064,
      "learning_rate": 3.364122900931094e-07,
      "loss": 0.7253,
      "step": 6955
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6747384078911947,
      "learning_rate": 3.3566864225704156e-07,
      "loss": 0.4613,
      "step": 6956
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7914947855322968,
      "learning_rate": 3.34925788708011e-07,
      "loss": 0.6493,
      "step": 6957
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8611323040500779,
      "learning_rate": 3.3418372957251864e-07,
      "loss": 0.5927,
      "step": 6958
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6892243682656503,
      "learning_rate": 3.3344246497692856e-07,
      "loss": 0.5473,
      "step": 6959
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8071568528194664,
      "learning_rate": 3.3270199504747137e-07,
      "loss": 0.5912,
      "step": 6960
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6489081725854973,
      "learning_rate": 3.319623199102406e-07,
      "loss": 0.5496,
      "step": 6961
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7645147360763048,
      "learning_rate": 3.312234396911956e-07,
      "loss": 0.578,
      "step": 6962
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7636841622102487,
      "learning_rate": 3.304853545161596e-07,
      "loss": 0.5495,
      "step": 6963
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7478663981948215,
      "learning_rate": 3.297480645108214e-07,
      "loss": 0.5565,
      "step": 6964
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7910201536702967,
      "learning_rate": 3.290115698007329e-07,
      "loss": 0.5828,
      "step": 6965
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7601905653887429,
      "learning_rate": 3.282758705113126e-07,
      "loss": 0.5878,
      "step": 6966
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7885189484176697,
      "learning_rate": 3.27540966767842e-07,
      "loss": 0.6145,
      "step": 6967
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8083902611325703,
      "learning_rate": 3.26806858695467e-07,
      "loss": 0.6785,
      "step": 6968
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9553180885231035,
      "learning_rate": 3.260735464191983e-07,
      "loss": 0.5001,
      "step": 6969
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8958323072326773,
      "learning_rate": 3.2534103006391314e-07,
      "loss": 0.6256,
      "step": 6970
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8858237745038109,
      "learning_rate": 3.246093097543501e-07,
      "loss": 0.5929,
      "step": 6971
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8546518752288892,
      "learning_rate": 3.238783856151134e-07,
      "loss": 0.5647,
      "step": 6972
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.732904673898036,
      "learning_rate": 3.2314825777067294e-07,
      "loss": 0.5361,
      "step": 6973
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.742062478261483,
      "learning_rate": 3.2241892634536034e-07,
      "loss": 0.5175,
      "step": 6974
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9138286581049283,
      "learning_rate": 3.2169039146337457e-07,
      "loss": 0.6797,
      "step": 6975
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7558495500258164,
      "learning_rate": 3.209626532487764e-07,
      "loss": 0.4705,
      "step": 6976
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.710825217787473,
      "learning_rate": 3.202357118254923e-07,
      "loss": 0.5359,
      "step": 6977
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.605834096758279,
      "learning_rate": 3.195095673173132e-07,
      "loss": 0.5336,
      "step": 6978
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6864356020580893,
      "learning_rate": 3.187842198478941e-07,
      "loss": 0.5513,
      "step": 6979
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6759669168563152,
      "learning_rate": 3.18059669540754e-07,
      "loss": 0.5288,
      "step": 6980
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7045823353040241,
      "learning_rate": 3.173359165192752e-07,
      "loss": 0.549,
      "step": 6981
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7323056226383078,
      "learning_rate": 3.16612960906707e-07,
      "loss": 0.5864,
      "step": 6982
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7001829028875254,
      "learning_rate": 3.158908028261598e-07,
      "loss": 0.5756,
      "step": 6983
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4567425588892098,
      "learning_rate": 3.1516944240061077e-07,
      "loss": 0.5913,
      "step": 6984
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8593649698855941,
      "learning_rate": 3.1444887975289886e-07,
      "loss": 0.6559,
      "step": 6985
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8825695459196331,
      "learning_rate": 3.1372911500572923e-07,
      "loss": 0.5615,
      "step": 6986
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6486826066326216,
      "learning_rate": 3.130101482816694e-07,
      "loss": 0.5393,
      "step": 6987
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7771684236717392,
      "learning_rate": 3.1229197970315374e-07,
      "loss": 0.5436,
      "step": 6988
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8971593602959002,
      "learning_rate": 3.115746093924771e-07,
      "loss": 0.6208,
      "step": 6989
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6243847288722311,
      "learning_rate": 3.108580374718012e-07,
      "loss": 0.4787,
      "step": 6990
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6895107679319626,
      "learning_rate": 3.1014226406315076e-07,
      "loss": 0.5202,
      "step": 6991
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6869952310809828,
      "learning_rate": 3.094272892884137e-07,
      "loss": 0.5258,
      "step": 6992
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7981868981637653,
      "learning_rate": 3.087131132693433e-07,
      "loss": 0.5602,
      "step": 6993
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8797551499257745,
      "learning_rate": 3.0799973612755607e-07,
      "loss": 0.5667,
      "step": 6994
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7329088767484716,
      "learning_rate": 3.0728715798453266e-07,
      "loss": 0.5361,
      "step": 6995
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6915315346563154,
      "learning_rate": 3.0657537896161824e-07,
      "loss": 0.5534,
      "step": 6996
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8777997273576092,
      "learning_rate": 3.0586439918002143e-07,
      "loss": 0.6639,
      "step": 6997
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8835129072842285,
      "learning_rate": 3.051542187608136e-07,
      "loss": 0.5706,
      "step": 6998
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8822879692217039,
      "learning_rate": 3.0444483782493207e-07,
      "loss": 0.6302,
      "step": 6999
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7144881672330305,
      "learning_rate": 3.0373625649317674e-07,
      "loss": 0.5518,
      "step": 7000
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.712558302776765,
      "learning_rate": 3.0302847488621114e-07,
      "loss": 0.5903,
      "step": 7001
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7100044367656192,
      "learning_rate": 3.023214931245638e-07,
      "loss": 0.5521,
      "step": 7002
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6749515165421132,
      "learning_rate": 3.016153113286263e-07,
      "loss": 0.5523,
      "step": 7003
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7636478222843134,
      "learning_rate": 3.0090992961865404e-07,
      "loss": 0.5154,
      "step": 7004
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8319166412829024,
      "learning_rate": 3.00205348114766e-07,
      "loss": 0.6102,
      "step": 7005
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8699956819100007,
      "learning_rate": 2.99501566936945e-07,
      "loss": 0.6081,
      "step": 7006
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9130595218705513,
      "learning_rate": 2.987985862050374e-07,
      "loss": 0.6893,
      "step": 7007
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8666271146032645,
      "learning_rate": 2.9809640603875423e-07,
      "loss": 0.5186,
      "step": 7008
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6086223377233909,
      "learning_rate": 2.973950265576686e-07,
      "loss": 0.5689,
      "step": 7009
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6261903218404566,
      "learning_rate": 2.966944478812189e-07,
      "loss": 0.5089,
      "step": 7010
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7795110203080939,
      "learning_rate": 2.9599467012870585e-07,
      "loss": 0.6468,
      "step": 7011
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7562381923264567,
      "learning_rate": 2.9529569341929465e-07,
      "loss": 0.5411,
      "step": 7012
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6648073384045954,
      "learning_rate": 2.9459751787201396e-07,
      "loss": 0.5316,
      "step": 7013
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7704531930552118,
      "learning_rate": 2.939001436057548e-07,
      "loss": 0.5426,
      "step": 7014
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7559707076164255,
      "learning_rate": 2.932035707392739e-07,
      "loss": 0.4867,
      "step": 7015
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8009764462752249,
      "learning_rate": 2.9250779939118954e-07,
      "loss": 0.6485,
      "step": 7016
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6360209916378563,
      "learning_rate": 2.9181282967998493e-07,
      "loss": 0.525,
      "step": 7017
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.216844001749365,
      "learning_rate": 2.9111866172400527e-07,
      "loss": 0.5708,
      "step": 7018
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6724771473420179,
      "learning_rate": 2.9042529564146104e-07,
      "loss": 0.5188,
      "step": 7019
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6107138625887601,
      "learning_rate": 2.8973273155042556e-07,
      "loss": 0.4381,
      "step": 7020
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8667014730760305,
      "learning_rate": 2.89040969568834e-07,
      "loss": 0.6348,
      "step": 7021
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8609136351307967,
      "learning_rate": 2.88350009814487e-07,
      "loss": 0.5861,
      "step": 7022
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7231364731422838,
      "learning_rate": 2.876598524050478e-07,
      "loss": 0.5114,
      "step": 7023
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7918572201065193,
      "learning_rate": 2.869704974580434e-07,
      "loss": 0.64,
      "step": 7024
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.677159760913107,
      "learning_rate": 2.862819450908627e-07,
      "loss": 0.5171,
      "step": 7025
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8175295099283835,
      "learning_rate": 2.8559419542075973e-07,
      "loss": 0.5664,
      "step": 7026
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8599328960190116,
      "learning_rate": 2.8490724856485017e-07,
      "loss": 0.6107,
      "step": 7027
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6916106104323896,
      "learning_rate": 2.842211046401155e-07,
      "loss": 0.4957,
      "step": 7028
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6372831000949382,
      "learning_rate": 2.8353576376339786e-07,
      "loss": 0.4869,
      "step": 7029
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8402321887312569,
      "learning_rate": 2.8285122605140433e-07,
      "loss": 0.6356,
      "step": 7030
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.709162788295609,
      "learning_rate": 2.82167491620704e-07,
      "loss": 0.4921,
      "step": 7031
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2340712363566009,
      "learning_rate": 2.8148456058772977e-07,
      "loss": 0.6281,
      "step": 7032
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0184218311929112,
      "learning_rate": 2.808024330687775e-07,
      "loss": 0.5297,
      "step": 7033
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6485840936874373,
      "learning_rate": 2.8012110918000657e-07,
      "loss": 0.5183,
      "step": 7034
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9654053535424876,
      "learning_rate": 2.7944058903743964e-07,
      "loss": 0.6208,
      "step": 7035
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7202838090490248,
      "learning_rate": 2.787608727569624e-07,
      "loss": 0.5151,
      "step": 7036
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6263902949244659,
      "learning_rate": 2.7808196045432224e-07,
      "loss": 0.4849,
      "step": 7037
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8266628551698149,
      "learning_rate": 2.774038522451322e-07,
      "loss": 0.6024,
      "step": 7038
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7833067891524704,
      "learning_rate": 2.767265482448661e-07,
      "loss": 0.5607,
      "step": 7039
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7353334391585161,
      "learning_rate": 2.760500485688622e-07,
      "loss": 0.5339,
      "step": 7040
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6841793443135431,
      "learning_rate": 2.7537435333232165e-07,
      "loss": 0.5078,
      "step": 7041
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7670098693714952,
      "learning_rate": 2.746994626503074e-07,
      "loss": 0.5406,
      "step": 7042
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7998903518140569,
      "learning_rate": 2.740253766377471e-07,
      "loss": 0.5527,
      "step": 7043
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6328764137402102,
      "learning_rate": 2.733520954094304e-07,
      "loss": 0.4937,
      "step": 7044
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7104193587876197,
      "learning_rate": 2.726796190800096e-07,
      "loss": 0.5558,
      "step": 7045
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6689463628416755,
      "learning_rate": 2.720079477640009e-07,
      "loss": 0.5306,
      "step": 7046
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9270457056764949,
      "learning_rate": 2.713370815757832e-07,
      "loss": 0.5915,
      "step": 7047
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7134306407832558,
      "learning_rate": 2.706670206295975e-07,
      "loss": 0.5354,
      "step": 7048
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.968517769346105,
      "learning_rate": 2.69997765039548e-07,
      "loss": 0.6725,
      "step": 7049
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7462057343588644,
      "learning_rate": 2.69329314919603e-07,
      "loss": 0.5452,
      "step": 7050
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6084785648605658,
      "learning_rate": 2.6866167038359146e-07,
      "loss": 0.5083,
      "step": 7051
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6947343598459573,
      "learning_rate": 2.679948315452069e-07,
      "loss": 0.5775,
      "step": 7052
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8446069298115492,
      "learning_rate": 2.673287985180045e-07,
      "loss": 0.6017,
      "step": 7053
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6175663214902423,
      "learning_rate": 2.6666357141540255e-07,
      "loss": 0.4735,
      "step": 7054
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8468003612838216,
      "learning_rate": 2.6599915035068367e-07,
      "loss": 0.5983,
      "step": 7055
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9358853973868989,
      "learning_rate": 2.653355354369908e-07,
      "loss": 0.6184,
      "step": 7056
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7350697162970625,
      "learning_rate": 2.6467272678733114e-07,
      "loss": 0.5699,
      "step": 7057
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.757707184325397,
      "learning_rate": 2.6401072451457397e-07,
      "loss": 0.5448,
      "step": 7058
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8562357487747339,
      "learning_rate": 2.633495287314508e-07,
      "loss": 0.6307,
      "step": 7059
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7088049314602112,
      "learning_rate": 2.626891395505571e-07,
      "loss": 0.5397,
      "step": 7060
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7305906116858653,
      "learning_rate": 2.620295570843501e-07,
      "loss": 0.5357,
      "step": 7061
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8256628776120437,
      "learning_rate": 2.6137078144514937e-07,
      "loss": 0.5285,
      "step": 7062
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.857129435342946,
      "learning_rate": 2.6071281274513805e-07,
      "loss": 0.6252,
      "step": 7063
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6269493699640977,
      "learning_rate": 2.6005565109636086e-07,
      "loss": 0.4871,
      "step": 7064
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7914612364642583,
      "learning_rate": 2.5939929661072615e-07,
      "loss": 0.6056,
      "step": 7065
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0079455349570206,
      "learning_rate": 2.587437494000045e-07,
      "loss": 0.6326,
      "step": 7066
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3023054964119716,
      "learning_rate": 2.580890095758276e-07,
      "loss": 0.5742,
      "step": 7067
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7769674375256881,
      "learning_rate": 2.5743507724969196e-07,
      "loss": 0.6068,
      "step": 7068
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7533057660642877,
      "learning_rate": 2.5678195253295455e-07,
      "loss": 0.5454,
      "step": 7069
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6712511631210043,
      "learning_rate": 2.5612963553683646e-07,
      "loss": 0.5452,
      "step": 7070
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6758664301975796,
      "learning_rate": 2.554781263724199e-07,
      "loss": 0.5027,
      "step": 7071
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7390186762902936,
      "learning_rate": 2.548274251506505e-07,
      "loss": 0.5291,
      "step": 7072
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6643365682069159,
      "learning_rate": 2.541775319823359e-07,
      "loss": 0.5136,
      "step": 7073
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6939284519542831,
      "learning_rate": 2.535284469781457e-07,
      "loss": 0.4921,
      "step": 7074
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7192443709846023,
      "learning_rate": 2.5288017024861275e-07,
      "loss": 0.495,
      "step": 7075
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7236574478596427,
      "learning_rate": 2.522327019041315e-07,
      "loss": 0.592,
      "step": 7076
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7095519864309726,
      "learning_rate": 2.515860420549593e-07,
      "loss": 0.5323,
      "step": 7077
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.664579149964256,
      "learning_rate": 2.509401908112158e-07,
      "loss": 0.5595,
      "step": 7078
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9287731271909274,
      "learning_rate": 2.5029514828288194e-07,
      "loss": 0.6237,
      "step": 7079
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9682793347658527,
      "learning_rate": 2.4965091457980204e-07,
      "loss": 0.6218,
      "step": 7080
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6187793924499251,
      "learning_rate": 2.490074898116829e-07,
      "loss": 0.4864,
      "step": 7081
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.752378765205412,
      "learning_rate": 2.4836487408809284e-07,
      "loss": 0.6115,
      "step": 7082
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9326199565186433,
      "learning_rate": 2.4772306751846273e-07,
      "loss": 0.6125,
      "step": 7083
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.767021366413531,
      "learning_rate": 2.470820702120857e-07,
      "loss": 0.535,
      "step": 7084
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6465448790357634,
      "learning_rate": 2.4644188227811615e-07,
      "loss": 0.4647,
      "step": 7085
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8689702530254806,
      "learning_rate": 2.458025038255718e-07,
      "loss": 0.6542,
      "step": 7086
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0249396177769077,
      "learning_rate": 2.4516393496333225e-07,
      "loss": 0.6297,
      "step": 7087
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8459240902913864,
      "learning_rate": 2.4452617580013827e-07,
      "loss": 0.6601,
      "step": 7088
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7201875641647064,
      "learning_rate": 2.4388922644459525e-07,
      "loss": 0.5253,
      "step": 7089
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7910108479186113,
      "learning_rate": 2.4325308700516805e-07,
      "loss": 0.6053,
      "step": 7090
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8791270265381143,
      "learning_rate": 2.4261775759018445e-07,
      "loss": 0.6523,
      "step": 7091
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8256320099237336,
      "learning_rate": 2.419832383078352e-07,
      "loss": 0.6083,
      "step": 7092
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7394375310809276,
      "learning_rate": 2.413495292661716e-07,
      "loss": 0.5249,
      "step": 7093
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9538122054240401,
      "learning_rate": 2.407166305731079e-07,
      "loss": 0.5523,
      "step": 7094
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8663068288282182,
      "learning_rate": 2.400845423364201e-07,
      "loss": 0.5097,
      "step": 7095
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8748138919881248,
      "learning_rate": 2.3945326466374653e-07,
      "loss": 0.5417,
      "step": 7096
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8982500360953076,
      "learning_rate": 2.3882279766258676e-07,
      "loss": 0.6382,
      "step": 7097
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.795692760464459,
      "learning_rate": 2.381931414403027e-07,
      "loss": 0.6448,
      "step": 7098
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1017615614465852,
      "learning_rate": 2.3756429610411913e-07,
      "loss": 0.6318,
      "step": 7099
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8856021951495591,
      "learning_rate": 2.3693626176112096e-07,
      "loss": 0.5187,
      "step": 7100
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9442964311471541,
      "learning_rate": 2.3630903851825593e-07,
      "loss": 0.6505,
      "step": 7101
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7271985109025816,
      "learning_rate": 2.356826264823342e-07,
      "loss": 0.5709,
      "step": 7102
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6359329365587,
      "learning_rate": 2.3505702576002654e-07,
      "loss": 0.4809,
      "step": 7103
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7253713559731307,
      "learning_rate": 2.3443223645786662e-07,
      "loss": 0.5283,
      "step": 7104
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8485637511844898,
      "learning_rate": 2.3380825868224987e-07,
      "loss": 0.628,
      "step": 7105
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6863163524885713,
      "learning_rate": 2.3318509253943245e-07,
      "loss": 0.5776,
      "step": 7106
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6164989318007248,
      "learning_rate": 2.3256273813553387e-07,
      "loss": 0.5022,
      "step": 7107
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.713443878747001,
      "learning_rate": 2.3194119557653384e-07,
      "loss": 0.5374,
      "step": 7108
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7591216189472731,
      "learning_rate": 2.3132046496827486e-07,
      "loss": 0.5584,
      "step": 7109
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.747974880105595,
      "learning_rate": 2.3070054641646134e-07,
      "loss": 0.5796,
      "step": 7110
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8951868454865017,
      "learning_rate": 2.3008144002665823e-07,
      "loss": 0.6346,
      "step": 7111
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8349950816077631,
      "learning_rate": 2.2946314590429287e-07,
      "loss": 0.6016,
      "step": 7112
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6399477910833015,
      "learning_rate": 2.288456641546549e-07,
      "loss": 0.5322,
      "step": 7113
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6932096174340461,
      "learning_rate": 2.2822899488289474e-07,
      "loss": 0.6073,
      "step": 7114
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6732670674436677,
      "learning_rate": 2.2761313819402553e-07,
      "loss": 0.5779,
      "step": 7115
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2724381653920382,
      "learning_rate": 2.2699809419292007e-07,
      "loss": 0.6472,
      "step": 7116
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9303540914468582,
      "learning_rate": 2.2638386298431404e-07,
      "loss": 0.6572,
      "step": 7117
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7968642434156843,
      "learning_rate": 2.257704446728054e-07,
      "loss": 0.5871,
      "step": 7118
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7161511395365222,
      "learning_rate": 2.251578393628523e-07,
      "loss": 0.5825,
      "step": 7119
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8062276010121023,
      "learning_rate": 2.245460471587757e-07,
      "loss": 0.6411,
      "step": 7120
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6448369603234269,
      "learning_rate": 2.2393506816475618e-07,
      "loss": 0.5076,
      "step": 7121
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7975509266431865,
      "learning_rate": 2.2332490248483828e-07,
      "loss": 0.5272,
      "step": 7122
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8450583049842941,
      "learning_rate": 2.2271555022292669e-07,
      "loss": 0.6058,
      "step": 7123
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8186238837891533,
      "learning_rate": 2.2210701148278724e-07,
      "loss": 0.5575,
      "step": 7124
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8552212947485776,
      "learning_rate": 2.2149928636804817e-07,
      "loss": 0.5459,
      "step": 7125
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7051197562685936,
      "learning_rate": 2.2089237498219839e-07,
      "loss": 0.5758,
      "step": 7126
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.235319009836637,
      "learning_rate": 2.2028627742858855e-07,
      "loss": 0.5888,
      "step": 7127
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8552803463757278,
      "learning_rate": 2.1968099381043106e-07,
      "loss": 0.6369,
      "step": 7128
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6312595287701854,
      "learning_rate": 2.190765242307996e-07,
      "loss": 0.4955,
      "step": 7129
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6625882948581738,
      "learning_rate": 2.1847286879262852e-07,
      "loss": 0.5401,
      "step": 7130
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8906231153802382,
      "learning_rate": 2.1787002759871446e-07,
      "loss": 0.6107,
      "step": 7131
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6984324581123562,
      "learning_rate": 2.1726800075171473e-07,
      "loss": 0.4947,
      "step": 7132
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8061261657145004,
      "learning_rate": 2.1666678835414846e-07,
      "loss": 0.629,
      "step": 7133
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6426059207406561,
      "learning_rate": 2.1606639050839596e-07,
      "loss": 0.5209,
      "step": 7134
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.734655949442228,
      "learning_rate": 2.1546680731669767e-07,
      "loss": 0.5614,
      "step": 7135
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.028804723195199,
      "learning_rate": 2.1486803888115805e-07,
      "loss": 0.626,
      "step": 7136
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8773926003625261,
      "learning_rate": 2.1427008530373994e-07,
      "loss": 0.6251,
      "step": 7137
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9811830901944624,
      "learning_rate": 2.1367294668626803e-07,
      "loss": 0.7057,
      "step": 7138
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8689753685764708,
      "learning_rate": 2.1307662313043043e-07,
      "loss": 0.5909,
      "step": 7139
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6305367710103641,
      "learning_rate": 2.1248111473777422e-07,
      "loss": 0.4956,
      "step": 7140
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6734045563812018,
      "learning_rate": 2.118864216097083e-07,
      "loss": 0.5158,
      "step": 7141
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7333230002556563,
      "learning_rate": 2.1129254384750275e-07,
      "loss": 0.563,
      "step": 7142
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7174059188948292,
      "learning_rate": 2.1069948155228838e-07,
      "loss": 0.5815,
      "step": 7143
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7389695046596599,
      "learning_rate": 2.1010723482505712e-07,
      "loss": 0.6053,
      "step": 7144
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.829117219036659,
      "learning_rate": 2.0951580376666393e-07,
      "loss": 0.5968,
      "step": 7145
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6973702716805741,
      "learning_rate": 2.0892518847782205e-07,
      "loss": 0.5013,
      "step": 7146
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7223751531504962,
      "learning_rate": 2.0833538905910723e-07,
      "loss": 0.5644,
      "step": 7147
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8771362530749554,
      "learning_rate": 2.0774640561095682e-07,
      "loss": 0.5349,
      "step": 7148
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8448018995415694,
      "learning_rate": 2.0715823823366844e-07,
      "loss": 0.6504,
      "step": 7149
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8098850175375887,
      "learning_rate": 2.0657088702740024e-07,
      "loss": 0.5978,
      "step": 7150
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9019805800373341,
      "learning_rate": 2.0598435209217228e-07,
      "loss": 0.6332,
      "step": 7151
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8902668900601771,
      "learning_rate": 2.053986335278657e-07,
      "loss": 0.6241,
      "step": 7152
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7405242224551115,
      "learning_rate": 2.0481373143422244e-07,
      "loss": 0.4857,
      "step": 7153
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8218848980735411,
      "learning_rate": 2.0422964591084394e-07,
      "loss": 0.6267,
      "step": 7154
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8089294191267027,
      "learning_rate": 2.036463770571956e-07,
      "loss": 0.6374,
      "step": 7155
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7234180045107353,
      "learning_rate": 2.0306392497260075e-07,
      "loss": 0.5363,
      "step": 7156
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8849831781816686,
      "learning_rate": 2.0248228975624506e-07,
      "loss": 0.622,
      "step": 7157
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8085358743725332,
      "learning_rate": 2.0190147150717598e-07,
      "loss": 0.5461,
      "step": 7158
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.752844305249223,
      "learning_rate": 2.013214703242994e-07,
      "loss": 0.5222,
      "step": 7159
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7678663736220976,
      "learning_rate": 2.007422863063846e-07,
      "loss": 0.5523,
      "step": 7160
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6080486702035511,
      "learning_rate": 2.0016391955205993e-07,
      "loss": 0.5184,
      "step": 7161
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7833129578078779,
      "learning_rate": 1.9958637015981552e-07,
      "loss": 0.5631,
      "step": 7162
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.803368956510164,
      "learning_rate": 1.9900963822800212e-07,
      "loss": 0.5957,
      "step": 7163
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9315691973525265,
      "learning_rate": 1.9843372385483062e-07,
      "loss": 0.6144,
      "step": 7164
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6501987968813367,
      "learning_rate": 1.9785862713837312e-07,
      "loss": 0.557,
      "step": 7165
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6644622950023554,
      "learning_rate": 1.972843481765635e-07,
      "loss": 0.5413,
      "step": 7166
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7703229252873747,
      "learning_rate": 1.967108870671952e-07,
      "loss": 0.5275,
      "step": 7167
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.845235080616675,
      "learning_rate": 1.961382439079229e-07,
      "loss": 0.5838,
      "step": 7168
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9493000787658042,
      "learning_rate": 1.955664187962608e-07,
      "loss": 0.6042,
      "step": 7169
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7990135664535877,
      "learning_rate": 1.9499541182958604e-07,
      "loss": 0.6013,
      "step": 7170
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9409366215620388,
      "learning_rate": 1.9442522310513355e-07,
      "loss": 0.5653,
      "step": 7171
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7415705644774591,
      "learning_rate": 1.9385585272000241e-07,
      "loss": 0.5289,
      "step": 7172
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8663803473444577,
      "learning_rate": 1.9328730077114889e-07,
      "loss": 0.6092,
      "step": 7173
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6466471897218875,
      "learning_rate": 1.9271956735539222e-07,
      "loss": 0.491,
      "step": 7174
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9657893941012731,
      "learning_rate": 1.9215265256941117e-07,
      "loss": 0.6575,
      "step": 7175
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7102076254729801,
      "learning_rate": 1.915865565097458e-07,
      "loss": 0.528,
      "step": 7176
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6270302751772797,
      "learning_rate": 1.9102127927279613e-07,
      "loss": 0.5478,
      "step": 7177
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.721118094924588,
      "learning_rate": 1.9045682095482298e-07,
      "loss": 0.5566,
      "step": 7178
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9338382777522265,
      "learning_rate": 1.898931816519478e-07,
      "loss": 0.6155,
      "step": 7179
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.828588403972346,
      "learning_rate": 1.8933036146015271e-07,
      "loss": 0.6461,
      "step": 7180
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8703385692153343,
      "learning_rate": 1.8876836047527937e-07,
      "loss": 0.5794,
      "step": 7181
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8499247307989124,
      "learning_rate": 1.8820717879303173e-07,
      "loss": 0.5769,
      "step": 7182
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8346166661850465,
      "learning_rate": 1.876468165089723e-07,
      "loss": 0.5915,
      "step": 7183
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9179178226621876,
      "learning_rate": 1.8708727371852531e-07,
      "loss": 0.6033,
      "step": 7184
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6874965214042009,
      "learning_rate": 1.8652855051697504e-07,
      "loss": 0.5139,
      "step": 7185
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8641984261272758,
      "learning_rate": 1.8597064699946654e-07,
      "loss": 0.6015,
      "step": 7186
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6971071038458619,
      "learning_rate": 1.8541356326100436e-07,
      "loss": 0.5064,
      "step": 7187
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7535879680846994,
      "learning_rate": 1.8485729939645425e-07,
      "loss": 0.523,
      "step": 7188
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7943957796459852,
      "learning_rate": 1.8430185550054214e-07,
      "loss": 0.5439,
      "step": 7189
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7762325154146982,
      "learning_rate": 1.8374723166785456e-07,
      "loss": 0.6019,
      "step": 7190
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7455010974633227,
      "learning_rate": 1.8319342799283824e-07,
      "loss": 0.5829,
      "step": 7191
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7681783232511509,
      "learning_rate": 1.826404445697999e-07,
      "loss": 0.4493,
      "step": 7192
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7840365994342315,
      "learning_rate": 1.8208828149290702e-07,
      "loss": 0.5468,
      "step": 7193
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7118399932943732,
      "learning_rate": 1.8153693885618718e-07,
      "loss": 0.5628,
      "step": 7194
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6103364414898095,
      "learning_rate": 1.80986416753528e-07,
      "loss": 0.5268,
      "step": 7195
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7428047166651373,
      "learning_rate": 1.8043671527867846e-07,
      "loss": 0.5388,
      "step": 7196
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6431049238974033,
      "learning_rate": 1.7988783452524583e-07,
      "loss": 0.5188,
      "step": 7197
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8226918770684055,
      "learning_rate": 1.7933977458669982e-07,
      "loss": 0.6034,
      "step": 7198
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9584408885605662,
      "learning_rate": 1.7879253555636855e-07,
      "loss": 0.6481,
      "step": 7199
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8108181407093012,
      "learning_rate": 1.7824611752744193e-07,
      "loss": 0.6093,
      "step": 7200
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7789346752929458,
      "learning_rate": 1.7770052059296882e-07,
      "loss": 0.5973,
      "step": 7201
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8033982393375146,
      "learning_rate": 1.7715574484585885e-07,
      "loss": 0.6144,
      "step": 7202
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6650795973550269,
      "learning_rate": 1.766117903788811e-07,
      "loss": 0.549,
      "step": 7203
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9048984414670657,
      "learning_rate": 1.7606865728466592e-07,
      "loss": 0.5862,
      "step": 7204
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6989273872268915,
      "learning_rate": 1.7552634565570325e-07,
      "loss": 0.544,
      "step": 7205
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.800429379818265,
      "learning_rate": 1.7498485558434252e-07,
      "loss": 0.5694,
      "step": 7206
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.733207102092401,
      "learning_rate": 1.7444418716279443e-07,
      "loss": 0.5328,
      "step": 7207
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8065312252122538,
      "learning_rate": 1.739043404831292e-07,
      "loss": 0.5699,
      "step": 7208
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7543913815098466,
      "learning_rate": 1.7336531563727664e-07,
      "loss": 0.6131,
      "step": 7209
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8908361515944642,
      "learning_rate": 1.7282711271702723e-07,
      "loss": 0.6604,
      "step": 7210
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9352180956734087,
      "learning_rate": 1.7228973181403153e-07,
      "loss": 0.6345,
      "step": 7211
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8337364430948481,
      "learning_rate": 1.7175317301979967e-07,
      "loss": 0.6209,
      "step": 7212
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7236231476194895,
      "learning_rate": 1.7121743642570244e-07,
      "loss": 0.5255,
      "step": 7213
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6804012606244086,
      "learning_rate": 1.7068252212296966e-07,
      "loss": 0.5429,
      "step": 7214
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7443964473025415,
      "learning_rate": 1.7014843020269178e-07,
      "loss": 0.5419,
      "step": 7215
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7612930878922145,
      "learning_rate": 1.696151607558194e-07,
      "loss": 0.6044,
      "step": 7216
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0337458171805556,
      "learning_rate": 1.6908271387316267e-07,
      "loss": 0.6532,
      "step": 7217
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6108009146553584,
      "learning_rate": 1.6855108964539124e-07,
      "loss": 0.4885,
      "step": 7218
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7919708868274475,
      "learning_rate": 1.6802028816303606e-07,
      "loss": 0.5667,
      "step": 7219
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9418347945434669,
      "learning_rate": 1.6749030951648647e-07,
      "loss": 0.6393,
      "step": 7220
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9298620630681734,
      "learning_rate": 1.6696115379599255e-07,
      "loss": 0.635,
      "step": 7221
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7577334959722177,
      "learning_rate": 1.6643282109166326e-07,
      "loss": 0.5552,
      "step": 7222
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8401159965497546,
      "learning_rate": 1.659053114934689e-07,
      "loss": 0.643,
      "step": 7223
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7632485871117627,
      "learning_rate": 1.6537862509123924e-07,
      "loss": 0.5098,
      "step": 7224
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6836376357237847,
      "learning_rate": 1.648527619746626e-07,
      "loss": 0.5736,
      "step": 7225
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.856336635320358,
      "learning_rate": 1.6432772223328898e-07,
      "loss": 0.6342,
      "step": 7226
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8174785267813208,
      "learning_rate": 1.6380350595652628e-07,
      "loss": 0.582,
      "step": 7227
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8619754409539745,
      "learning_rate": 1.632801132336431e-07,
      "loss": 0.5851,
      "step": 7228
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8209396215900154,
      "learning_rate": 1.627575441537682e-07,
      "loss": 0.6033,
      "step": 7229
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7839055606645394,
      "learning_rate": 1.622357988058898e-07,
      "loss": 0.5837,
      "step": 7230
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7393864397438376,
      "learning_rate": 1.6171487727885515e-07,
      "loss": 0.473,
      "step": 7231
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6820447291478476,
      "learning_rate": 1.611947796613722e-07,
      "loss": 0.5325,
      "step": 7232
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8375674756471989,
      "learning_rate": 1.6067550604200787e-07,
      "loss": 0.5889,
      "step": 7233
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6278770645329156,
      "learning_rate": 1.601570565091892e-07,
      "loss": 0.4811,
      "step": 7234
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9216177944482743,
      "learning_rate": 1.5963943115120284e-07,
      "loss": 0.6323,
      "step": 7235
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.662379481099137,
      "learning_rate": 1.5912263005619433e-07,
      "loss": 0.5101,
      "step": 7236
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0434824488140055,
      "learning_rate": 1.5860665331217052e-07,
      "loss": 0.6077,
      "step": 7237
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6463294416645726,
      "learning_rate": 1.580915010069961e-07,
      "loss": 0.5412,
      "step": 7238
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.79385198066356,
      "learning_rate": 1.575771732283965e-07,
      "loss": 0.66,
      "step": 7239
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.811795868922921,
      "learning_rate": 1.5706367006395607e-07,
      "loss": 0.5595,
      "step": 7240
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9237649889918583,
      "learning_rate": 1.5655099160111986e-07,
      "loss": 0.6342,
      "step": 7241
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9513765092354893,
      "learning_rate": 1.5603913792719083e-07,
      "loss": 0.6577,
      "step": 7242
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8941049062976215,
      "learning_rate": 1.5552810912933202e-07,
      "loss": 0.6472,
      "step": 7243
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.762418050132673,
      "learning_rate": 1.5501790529456772e-07,
      "loss": 0.6102,
      "step": 7244
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7413108397970571,
      "learning_rate": 1.54508526509779e-07,
      "loss": 0.5173,
      "step": 7245
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9122060565943907,
      "learning_rate": 1.5399997286170865e-07,
      "loss": 0.5982,
      "step": 7246
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7603507605809376,
      "learning_rate": 1.5349224443695742e-07,
      "loss": 0.5359,
      "step": 7247
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8170087182992468,
      "learning_rate": 1.5298534132198616e-07,
      "loss": 0.5981,
      "step": 7248
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7791241193311133,
      "learning_rate": 1.5247926360311528e-07,
      "loss": 0.6004,
      "step": 7249
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7312295761256361,
      "learning_rate": 1.5197401136652523e-07,
      "loss": 0.5328,
      "step": 7250
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6092954745272051,
      "learning_rate": 1.5146958469825446e-07,
      "loss": 0.4666,
      "step": 7251
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8649372402444326,
      "learning_rate": 1.50965983684202e-07,
      "loss": 0.5055,
      "step": 7252
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8306144127054994,
      "learning_rate": 1.5046320841012541e-07,
      "loss": 0.6647,
      "step": 7253
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7027455000400481,
      "learning_rate": 1.4996125896164226e-07,
      "loss": 0.5335,
      "step": 7254
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8171187402406849,
      "learning_rate": 1.4946013542422976e-07,
      "loss": 0.5794,
      "step": 7255
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7163602134643401,
      "learning_rate": 1.4895983788322298e-07,
      "loss": 0.5958,
      "step": 7256
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6363625728689332,
      "learning_rate": 1.4846036642381878e-07,
      "loss": 0.5228,
      "step": 7257
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7820200790663094,
      "learning_rate": 1.4796172113107076e-07,
      "loss": 0.5387,
      "step": 7258
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8966288364832229,
      "learning_rate": 1.474639020898938e-07,
      "loss": 0.6284,
      "step": 7259
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7253105896826562,
      "learning_rate": 1.4696690938506063e-07,
      "loss": 0.663,
      "step": 7260
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8053078621884595,
      "learning_rate": 1.4647074310120467e-07,
      "loss": 0.5073,
      "step": 7261
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7558488047770283,
      "learning_rate": 1.4597540332281724e-07,
      "loss": 0.5786,
      "step": 7262
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8031110169382346,
      "learning_rate": 1.454808901342497e-07,
      "loss": 0.5034,
      "step": 7263
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7999272606864025,
      "learning_rate": 1.4498720361971251e-07,
      "loss": 0.6249,
      "step": 7264
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7622721677428819,
      "learning_rate": 1.444943438632751e-07,
      "loss": 0.5436,
      "step": 7265
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7103629644539494,
      "learning_rate": 1.4400231094886696e-07,
      "loss": 0.5383,
      "step": 7266
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7517678667871269,
      "learning_rate": 1.435111049602761e-07,
      "loss": 0.5413,
      "step": 7267
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7716418733229863,
      "learning_rate": 1.4302072598114947e-07,
      "loss": 0.5301,
      "step": 7268
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7210693409852775,
      "learning_rate": 1.4253117409499307e-07,
      "loss": 0.5437,
      "step": 7269
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6810966492044397,
      "learning_rate": 1.4204244938517353e-07,
      "loss": 0.5174,
      "step": 7270
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7882773009113102,
      "learning_rate": 1.4155455193491485e-07,
      "loss": 0.5773,
      "step": 7271
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7625913996140953,
      "learning_rate": 1.4106748182730057e-07,
      "loss": 0.5178,
      "step": 7272
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7765444837546138,
      "learning_rate": 1.4058123914527489e-07,
      "loss": 0.5614,
      "step": 7273
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8589502591680025,
      "learning_rate": 1.4009582397163878e-07,
      "loss": 0.6607,
      "step": 7274
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6871440011206116,
      "learning_rate": 1.3961123638905338e-07,
      "loss": 0.5279,
      "step": 7275
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6566957303391037,
      "learning_rate": 1.3912747648003933e-07,
      "loss": 0.5171,
      "step": 7276
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8791288500475686,
      "learning_rate": 1.3864454432697626e-07,
      "loss": 0.6833,
      "step": 7277
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6856361885371615,
      "learning_rate": 1.3816244001210177e-07,
      "loss": 0.4997,
      "step": 7278
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8135724637012066,
      "learning_rate": 1.376811636175135e-07,
      "loss": 0.6011,
      "step": 7279
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.695346015847785,
      "learning_rate": 1.372007152251681e-07,
      "loss": 0.5035,
      "step": 7280
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7426620018324094,
      "learning_rate": 1.3672109491688068e-07,
      "loss": 0.5772,
      "step": 7281
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.874240425056211,
      "learning_rate": 1.3624230277432538e-07,
      "loss": 0.5638,
      "step": 7282
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8266477060745181,
      "learning_rate": 1.357643388790353e-07,
      "loss": 0.6329,
      "step": 7283
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8310711632642739,
      "learning_rate": 1.3528720331240363e-07,
      "loss": 0.6437,
      "step": 7284
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7857312126715882,
      "learning_rate": 1.3481089615568044e-07,
      "loss": 0.542,
      "step": 7285
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6357937854081518,
      "learning_rate": 1.3433541748997692e-07,
      "loss": 0.4875,
      "step": 7286
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8116090325592704,
      "learning_rate": 1.3386076739626164e-07,
      "loss": 0.5983,
      "step": 7287
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7126544222392629,
      "learning_rate": 1.3338694595536218e-07,
      "loss": 0.5082,
      "step": 7288
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8862799074073311,
      "learning_rate": 1.3291395324796618e-07,
      "loss": 0.6758,
      "step": 7289
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8850685983174424,
      "learning_rate": 1.3244178935461926e-07,
      "loss": 0.5197,
      "step": 7290
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6416868278447686,
      "learning_rate": 1.3197045435572487e-07,
      "loss": 0.4744,
      "step": 7291
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6065229160659318,
      "learning_rate": 1.3149994833154768e-07,
      "loss": 0.515,
      "step": 7292
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6255487609177446,
      "learning_rate": 1.310302713622097e-07,
      "loss": 0.4644,
      "step": 7293
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7671393331462449,
      "learning_rate": 1.3056142352769253e-07,
      "loss": 0.5709,
      "step": 7294
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8288261003774176,
      "learning_rate": 1.3009340490783507e-07,
      "loss": 0.5772,
      "step": 7295
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8965150342414727,
      "learning_rate": 1.2962621558233635e-07,
      "loss": 0.6201,
      "step": 7296
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7358651322956123,
      "learning_rate": 1.2915985563075383e-07,
      "loss": 0.5704,
      "step": 7297
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5996505589400662,
      "learning_rate": 1.2869432513250403e-07,
      "loss": 0.4781,
      "step": 7298
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8826346651961483,
      "learning_rate": 1.282296241668618e-07,
      "loss": 0.6141,
      "step": 7299
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6917842156335661,
      "learning_rate": 1.2776575281296055e-07,
      "loss": 0.5725,
      "step": 7300
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8483391325350825,
      "learning_rate": 1.2730271114979375e-07,
      "loss": 0.5803,
      "step": 7301
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.662984599520145,
      "learning_rate": 1.268404992562111e-07,
      "loss": 0.4692,
      "step": 7302
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.716657251695465,
      "learning_rate": 1.2637911721092356e-07,
      "loss": 0.5637,
      "step": 7303
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0023802468834073,
      "learning_rate": 1.2591856509249935e-07,
      "loss": 0.6214,
      "step": 7304
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6850418100402297,
      "learning_rate": 1.2545884297936572e-07,
      "loss": 0.5286,
      "step": 7305
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7705106271550255,
      "learning_rate": 1.2499995094980843e-07,
      "loss": 0.5944,
      "step": 7306
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7177916983823144,
      "learning_rate": 1.245418890819716e-07,
      "loss": 0.5682,
      "step": 7307
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8325556302563488,
      "learning_rate": 1.24084657453859e-07,
      "loss": 0.6101,
      "step": 7308
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6735129378054524,
      "learning_rate": 1.2362825614333275e-07,
      "loss": 0.5494,
      "step": 7309
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6406439998324431,
      "learning_rate": 1.2317268522811286e-07,
      "loss": 0.5456,
      "step": 7310
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8626057098563875,
      "learning_rate": 1.2271794478577847e-07,
      "loss": 0.6458,
      "step": 7311
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7293757717691276,
      "learning_rate": 1.222640348937665e-07,
      "loss": 0.543,
      "step": 7312
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7205390701467687,
      "learning_rate": 1.218109556293734e-07,
      "loss": 0.5554,
      "step": 7313
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6394006074449649,
      "learning_rate": 1.2135870706975416e-07,
      "loss": 0.488,
      "step": 7314
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6875498888448462,
      "learning_rate": 1.209072892919222e-07,
      "loss": 0.5286,
      "step": 7315
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8722385122980011,
      "learning_rate": 1.204567023727482e-07,
      "loss": 0.556,
      "step": 7316
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8141119333855233,
      "learning_rate": 1.200069463889636e-07,
      "loss": 0.6031,
      "step": 7317
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8181392856626931,
      "learning_rate": 1.19558021417156e-07,
      "loss": 0.6232,
      "step": 7318
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8190091678953136,
      "learning_rate": 1.191099275337737e-07,
      "loss": 0.579,
      "step": 7319
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0702031458544996,
      "learning_rate": 1.1866266481512234e-07,
      "loss": 0.6566,
      "step": 7320
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9393952041689086,
      "learning_rate": 1.1821623333736487e-07,
      "loss": 0.6161,
      "step": 7321
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.892329138021765,
      "learning_rate": 1.1777063317652549e-07,
      "loss": 0.7195,
      "step": 7322
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6428168154385282,
      "learning_rate": 1.1732586440848459e-07,
      "loss": 0.5451,
      "step": 7323
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7467473564336984,
      "learning_rate": 1.168819271089816e-07,
      "loss": 0.4558,
      "step": 7324
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6996962877786095,
      "learning_rate": 1.1643882135361384e-07,
      "loss": 0.5489,
      "step": 7325
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.729384449939817,
      "learning_rate": 1.159965472178387e-07,
      "loss": 0.4909,
      "step": 7326
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9210812763635579,
      "learning_rate": 1.155551047769704e-07,
      "loss": 0.6422,
      "step": 7327
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7522939144579935,
      "learning_rate": 1.1511449410618214e-07,
      "loss": 0.547,
      "step": 7328
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9392851914349472,
      "learning_rate": 1.14674715280505e-07,
      "loss": 0.6285,
      "step": 7329
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8490034260807524,
      "learning_rate": 1.1423576837482908e-07,
      "loss": 0.6993,
      "step": 7330
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7692444173967874,
      "learning_rate": 1.1379765346390181e-07,
      "loss": 0.5675,
      "step": 7331
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7618326757462582,
      "learning_rate": 1.1336037062233075e-07,
      "loss": 0.5529,
      "step": 7332
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9470098883084807,
      "learning_rate": 1.1292391992457907e-07,
      "loss": 0.6862,
      "step": 7333
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9390875430241059,
      "learning_rate": 1.124883014449707e-07,
      "loss": 0.6755,
      "step": 7334
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8053956607457149,
      "learning_rate": 1.1205351525768737e-07,
      "loss": 0.5763,
      "step": 7335
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8040419476669075,
      "learning_rate": 1.1161956143676822e-07,
      "loss": 0.6063,
      "step": 7336
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8922416387459771,
      "learning_rate": 1.1118644005611079e-07,
      "loss": 0.62,
      "step": 7337
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6470085687473525,
      "learning_rate": 1.1075415118947108e-07,
      "loss": 0.4567,
      "step": 7338
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6174593204703737,
      "learning_rate": 1.103226949104641e-07,
      "loss": 0.4962,
      "step": 7339
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7269615687151015,
      "learning_rate": 1.0989207129256163e-07,
      "loss": 0.5989,
      "step": 7340
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8502106075474133,
      "learning_rate": 1.0946228040909445e-07,
      "loss": 0.6283,
      "step": 7341
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6486284023748357,
      "learning_rate": 1.0903332233325181e-07,
      "loss": 0.4695,
      "step": 7342
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7905882462156122,
      "learning_rate": 1.0860519713808082e-07,
      "loss": 0.601,
      "step": 7343
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7528406797193351,
      "learning_rate": 1.081779048964865e-07,
      "loss": 0.5193,
      "step": 7344
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7405114275518584,
      "learning_rate": 1.077514456812323e-07,
      "loss": 0.532,
      "step": 7345
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.73006527424087,
      "learning_rate": 1.0732581956494015e-07,
      "loss": 0.5076,
      "step": 7346
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7338725281136042,
      "learning_rate": 1.0690102662008928e-07,
      "loss": 0.5611,
      "step": 7347
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5958310173568875,
      "learning_rate": 1.0647706691901738e-07,
      "loss": 0.5405,
      "step": 7348
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7392212911350581,
      "learning_rate": 1.060539405339206e-07,
      "loss": 0.5911,
      "step": 7349
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9405858773290489,
      "learning_rate": 1.0563164753685296e-07,
      "loss": 0.6476,
      "step": 7350
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7958212301674082,
      "learning_rate": 1.0521018799972693e-07,
      "loss": 0.5629,
      "step": 7351
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7013041877092918,
      "learning_rate": 1.0478956199431233e-07,
      "loss": 0.5491,
      "step": 7352
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7769767823289179,
      "learning_rate": 1.0436976959223688e-07,
      "loss": 0.5288,
      "step": 7353
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7095681781276898,
      "learning_rate": 1.0395081086498781e-07,
      "loss": 0.5226,
      "step": 7354
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8854674966142587,
      "learning_rate": 1.0353268588390863e-07,
      "loss": 0.6602,
      "step": 7355
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6555574370439939,
      "learning_rate": 1.0311539472020238e-07,
      "loss": 0.4829,
      "step": 7356
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7159034068352127,
      "learning_rate": 1.0269893744492887e-07,
      "loss": 0.5749,
      "step": 7357
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.666760197261771,
      "learning_rate": 1.0228331412900693e-07,
      "loss": 0.4956,
      "step": 7358
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9035166141226554,
      "learning_rate": 1.0186852484321218e-07,
      "loss": 0.5535,
      "step": 7359
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7517025558418778,
      "learning_rate": 1.0145456965817868e-07,
      "loss": 0.5361,
      "step": 7360
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7588075554988031,
      "learning_rate": 1.0104144864440002e-07,
      "loss": 0.5532,
      "step": 7361
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6815012408779785,
      "learning_rate": 1.0062916187222549e-07,
      "loss": 0.488,
      "step": 7362
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0222338230754746,
      "learning_rate": 1.0021770941186393e-07,
      "loss": 0.6169,
      "step": 7363
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8091819969509825,
      "learning_rate": 9.98070913333804e-08,
      "loss": 0.6313,
      "step": 7364
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7173543941400221,
      "learning_rate": 9.939730770669953e-08,
      "loss": 0.5647,
      "step": 7365
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7284321572404366,
      "learning_rate": 9.898835860160272e-08,
      "loss": 0.5762,
      "step": 7366
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6400188578213188,
      "learning_rate": 9.858024408773036e-08,
      "loss": 0.5446,
      "step": 7367
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7736441661010852,
      "learning_rate": 9.817296423457968e-08,
      "loss": 0.5464,
      "step": 7368
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8063318126842265,
      "learning_rate": 9.776651911150681e-08,
      "loss": 0.6151,
      "step": 7369
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9714943737295673,
      "learning_rate": 9.73609087877242e-08,
      "loss": 0.635,
      "step": 7370
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9055985251724772,
      "learning_rate": 9.69561333323038e-08,
      "loss": 0.6246,
      "step": 7371
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7802595122283081,
      "learning_rate": 9.655219281417382e-08,
      "loss": 0.5992,
      "step": 7372
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7642877634701126,
      "learning_rate": 9.614908730212202e-08,
      "loss": 0.6061,
      "step": 7373
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7614338289147581,
      "learning_rate": 9.574681686479237e-08,
      "loss": 0.5357,
      "step": 7374
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7084461885573448,
      "learning_rate": 9.53453815706884e-08,
      "loss": 0.5143,
      "step": 7375
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7294685552871213,
      "learning_rate": 9.494478148816877e-08,
      "loss": 0.5369,
      "step": 7376
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6528965384956722,
      "learning_rate": 9.454501668545279e-08,
      "loss": 0.4736,
      "step": 7377
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6911121271333714,
      "learning_rate": 9.414608723061602e-08,
      "loss": 0.5119,
      "step": 7378
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6698477467275437,
      "learning_rate": 9.374799319159134e-08,
      "loss": 0.5089,
      "step": 7379
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.858908076645722,
      "learning_rate": 9.335073463617062e-08,
      "loss": 0.5925,
      "step": 7380
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9747074207088203,
      "learning_rate": 9.295431163200308e-08,
      "loss": 0.6454,
      "step": 7381
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9200028770155853,
      "learning_rate": 9.255872424659418e-08,
      "loss": 0.6253,
      "step": 7382
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7470864329520069,
      "learning_rate": 9.216397254731002e-08,
      "loss": 0.581,
      "step": 7383
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.623530278497163,
      "learning_rate": 9.177005660137129e-08,
      "loss": 0.5032,
      "step": 7384
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7311882987243897,
      "learning_rate": 9.137697647585875e-08,
      "loss": 0.5648,
      "step": 7385
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7830188672111648,
      "learning_rate": 9.098473223770943e-08,
      "loss": 0.6198,
      "step": 7386
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6719440031138225,
      "learning_rate": 9.059332395371768e-08,
      "loss": 0.4882,
      "step": 7387
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.736446488086922,
      "learning_rate": 9.020275169053793e-08,
      "loss": 0.4992,
      "step": 7388
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8108198488966872,
      "learning_rate": 8.981301551467924e-08,
      "loss": 0.5367,
      "step": 7389
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6705747796265014,
      "learning_rate": 8.942411549251018e-08,
      "loss": 0.4735,
      "step": 7390
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9262665393606445,
      "learning_rate": 8.903605169025609e-08,
      "loss": 0.5458,
      "step": 7391
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6807210799599605,
      "learning_rate": 8.864882417400078e-08,
      "loss": 0.5708,
      "step": 7392
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7040951620549809,
      "learning_rate": 8.826243300968374e-08,
      "loss": 0.5729,
      "step": 7393
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7551867210440104,
      "learning_rate": 8.787687826310509e-08,
      "loss": 0.6155,
      "step": 7394
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7681511535557327,
      "learning_rate": 8.749215999991956e-08,
      "loss": 0.5192,
      "step": 7395
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8799553300687178,
      "learning_rate": 8.710827828564139e-08,
      "loss": 0.5438,
      "step": 7396
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8819019453552943,
      "learning_rate": 8.672523318564108e-08,
      "loss": 0.5513,
      "step": 7397
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6555108477083655,
      "learning_rate": 8.634302476514755e-08,
      "loss": 0.4952,
      "step": 7398
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8049771846196506,
      "learning_rate": 8.596165308924708e-08,
      "loss": 0.4966,
      "step": 7399
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6445621320653898,
      "learning_rate": 8.558111822288217e-08,
      "loss": 0.5245,
      "step": 7400
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8071006753641993,
      "learning_rate": 8.520142023085543e-08,
      "loss": 0.6203,
      "step": 7401
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7892589034145806,
      "learning_rate": 8.482255917782455e-08,
      "loss": 0.557,
      "step": 7402
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6910387541917993,
      "learning_rate": 8.444453512830631e-08,
      "loss": 0.5416,
      "step": 7403
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.659051561439889,
      "learning_rate": 8.406734814667361e-08,
      "loss": 0.5465,
      "step": 7404
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6302135302637807,
      "learning_rate": 8.369099829715733e-08,
      "loss": 0.5138,
      "step": 7405
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.021620722476716,
      "learning_rate": 8.331548564384672e-08,
      "loss": 0.6353,
      "step": 7406
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6548546798751134,
      "learning_rate": 8.294081025068734e-08,
      "loss": 0.4992,
      "step": 7407
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7814362816415003,
      "learning_rate": 8.256697218148202e-08,
      "loss": 0.5288,
      "step": 7408
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6080749573520178,
      "learning_rate": 8.219397149989206e-08,
      "loss": 0.5044,
      "step": 7409
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8875353285515969,
      "learning_rate": 8.182180826943554e-08,
      "loss": 0.6534,
      "step": 7410
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9955848566920733,
      "learning_rate": 8.145048255348786e-08,
      "loss": 0.6272,
      "step": 7411
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2900601951864894,
      "learning_rate": 8.10799944152818e-08,
      "loss": 0.6536,
      "step": 7412
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6691976740484893,
      "learning_rate": 8.071034391790799e-08,
      "loss": 0.5216,
      "step": 7413
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6992770355368774,
      "learning_rate": 8.034153112431331e-08,
      "loss": 0.5573,
      "step": 7414
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7573448448022497,
      "learning_rate": 7.997355609730361e-08,
      "loss": 0.5036,
      "step": 7415
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7895226861238228,
      "learning_rate": 7.960641889954102e-08,
      "loss": 0.5915,
      "step": 7416
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8569780606468798,
      "learning_rate": 7.924011959354494e-08,
      "loss": 0.6104,
      "step": 7417
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8948860917878033,
      "learning_rate": 7.887465824169216e-08,
      "loss": 0.609,
      "step": 7418
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8587982446316301,
      "learning_rate": 7.851003490621678e-08,
      "loss": 0.6538,
      "step": 7419
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8061055937651449,
      "learning_rate": 7.814624964921136e-08,
      "loss": 0.592,
      "step": 7420
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8501248495231092,
      "learning_rate": 7.778330253262357e-08,
      "loss": 0.6031,
      "step": 7421
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7508365902515983,
      "learning_rate": 7.742119361826061e-08,
      "loss": 0.5953,
      "step": 7422
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.769294198884973,
      "learning_rate": 7.705992296778542e-08,
      "loss": 0.5824,
      "step": 7423
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7137904186482988,
      "learning_rate": 7.66994906427182e-08,
      "loss": 0.4806,
      "step": 7424
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8894916387203268,
      "learning_rate": 7.63398967044371e-08,
      "loss": 0.6612,
      "step": 7425
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7751909187222735,
      "learning_rate": 7.598114121417754e-08,
      "loss": 0.6082,
      "step": 7426
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8879137173213574,
      "learning_rate": 7.562322423303125e-08,
      "loss": 0.6166,
      "step": 7427
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8549700924191678,
      "learning_rate": 7.526614582194836e-08,
      "loss": 0.5266,
      "step": 7428
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9274097809705091,
      "learning_rate": 7.490990604173521e-08,
      "loss": 0.6383,
      "step": 7429
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8741603042536286,
      "learning_rate": 7.455450495305606e-08,
      "loss": 0.6504,
      "step": 7430
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7201206310120831,
      "learning_rate": 7.419994261643192e-08,
      "loss": 0.5486,
      "step": 7431
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9767964118496452,
      "learning_rate": 7.384621909224066e-08,
      "loss": 0.6272,
      "step": 7432
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7223315650826927,
      "learning_rate": 7.349333444071793e-08,
      "loss": 0.6089,
      "step": 7433
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6780399729649443,
      "learning_rate": 7.314128872195626e-08,
      "loss": 0.5385,
      "step": 7434
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.707306368940346,
      "learning_rate": 7.279008199590543e-08,
      "loss": 0.486,
      "step": 7435
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.714139965985434,
      "learning_rate": 7.243971432237263e-08,
      "loss": 0.5739,
      "step": 7436
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6162192494164707,
      "learning_rate": 7.209018576102178e-08,
      "loss": 0.4752,
      "step": 7437
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8631658155881962,
      "learning_rate": 7.174149637137307e-08,
      "loss": 0.5679,
      "step": 7438
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7819558887318622,
      "learning_rate": 7.139364621280564e-08,
      "loss": 0.6359,
      "step": 7439
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6684400294528868,
      "learning_rate": 7.104663534455436e-08,
      "loss": 0.5222,
      "step": 7440
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7666833899696217,
      "learning_rate": 7.070046382571194e-08,
      "loss": 0.5476,
      "step": 7441
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6993628436058983,
      "learning_rate": 7.035513171522678e-08,
      "loss": 0.5163,
      "step": 7442
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8518681412246044,
      "learning_rate": 7.00106390719063e-08,
      "loss": 0.626,
      "step": 7443
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8753522895777665,
      "learning_rate": 6.966698595441413e-08,
      "loss": 0.6539,
      "step": 7444
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7745927228490971,
      "learning_rate": 6.932417242126954e-08,
      "loss": 0.5003,
      "step": 7445
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.859209627688157,
      "learning_rate": 6.898219853085142e-08,
      "loss": 0.6418,
      "step": 7446
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.648242136691597,
      "learning_rate": 6.864106434139428e-08,
      "loss": 0.5473,
      "step": 7447
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6810104864839329,
      "learning_rate": 6.830076991098889e-08,
      "loss": 0.4944,
      "step": 7448
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8989056436038333,
      "learning_rate": 6.796131529758443e-08,
      "loss": 0.6665,
      "step": 7449
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6133260543909741,
      "learning_rate": 6.762270055898689e-08,
      "loss": 0.5133,
      "step": 7450
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.895883922095325,
      "learning_rate": 6.728492575285795e-08,
      "loss": 0.6544,
      "step": 7451
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7175767020902932,
      "learning_rate": 6.694799093671712e-08,
      "loss": 0.5446,
      "step": 7452
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8772505057973679,
      "learning_rate": 6.661189616794184e-08,
      "loss": 0.5742,
      "step": 7453
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8528062763836372,
      "learning_rate": 6.627664150376467e-08,
      "loss": 0.605,
      "step": 7454
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9298121551861412,
      "learning_rate": 6.594222700127605e-08,
      "loss": 0.5195,
      "step": 7455
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0489493131058054,
      "learning_rate": 6.560865271742434e-08,
      "loss": 0.6109,
      "step": 7456
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8531231142070393,
      "learning_rate": 6.52759187090124e-08,
      "loss": 0.6511,
      "step": 7457
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7094782016869042,
      "learning_rate": 6.494402503270159e-08,
      "loss": 0.5327,
      "step": 7458
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.778488280364918,
      "learning_rate": 6.461297174501058e-08,
      "loss": 0.6375,
      "step": 7459
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9649809611221075,
      "learning_rate": 6.428275890231428e-08,
      "loss": 0.6085,
      "step": 7460
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8384867714665719,
      "learning_rate": 6.395338656084382e-08,
      "loss": 0.5417,
      "step": 7461
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7172016219462297,
      "learning_rate": 6.362485477668878e-08,
      "loss": 0.5185,
      "step": 7462
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.734621564244927,
      "learning_rate": 6.329716360579386e-08,
      "loss": 0.5063,
      "step": 7463
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8585544568767859,
      "learning_rate": 6.297031310396162e-08,
      "loss": 0.5736,
      "step": 7464
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6924314831354269,
      "learning_rate": 6.264430332685145e-08,
      "loss": 0.5083,
      "step": 7465
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.631501622783633,
      "learning_rate": 6.231913432997949e-08,
      "loss": 0.5091,
      "step": 7466
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9919905000100189,
      "learning_rate": 6.199480616871867e-08,
      "loss": 0.6368,
      "step": 7467
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6427426645437417,
      "learning_rate": 6.167131889829869e-08,
      "loss": 0.5226,
      "step": 7468
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9290370498472447,
      "learning_rate": 6.134867257380606e-08,
      "loss": 0.6453,
      "step": 7469
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8110301863562172,
      "learning_rate": 6.102686725018348e-08,
      "loss": 0.6369,
      "step": 7470
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6910169488205844,
      "learning_rate": 6.070590298223156e-08,
      "loss": 0.5709,
      "step": 7471
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6731087264860027,
      "learning_rate": 6.03857798246077e-08,
      "loss": 0.5665,
      "step": 7472
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8943346689620795,
      "learning_rate": 6.006649783182494e-08,
      "loss": 0.6289,
      "step": 7473
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8167250978026973,
      "learning_rate": 5.974805705825315e-08,
      "loss": 0.5844,
      "step": 7474
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.984246334537608,
      "learning_rate": 5.9430457558120604e-08,
      "loss": 0.5991,
      "step": 7475
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6799249381001408,
      "learning_rate": 5.9113699385510703e-08,
      "loss": 0.5534,
      "step": 7476
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7261401602783657,
      "learning_rate": 5.879778259436364e-08,
      "loss": 0.5101,
      "step": 7477
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8532129506866355,
      "learning_rate": 5.8482707238477485e-08,
      "loss": 0.5765,
      "step": 7478
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.806207622017008,
      "learning_rate": 5.8168473371505994e-08,
      "loss": 0.5542,
      "step": 7479
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6545822991323784,
      "learning_rate": 5.785508104695969e-08,
      "loss": 0.5447,
      "step": 7480
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9957514847587225,
      "learning_rate": 5.754253031820589e-08,
      "loss": 0.6748,
      "step": 7481
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7380907981824266,
      "learning_rate": 5.723082123846924e-08,
      "loss": 0.5567,
      "step": 7482
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9386289161624838,
      "learning_rate": 5.69199538608306e-08,
      "loss": 0.6698,
      "step": 7483
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9054781923651454,
      "learning_rate": 5.660992823822708e-08,
      "loss": 0.6854,
      "step": 7484
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6393259310632982,
      "learning_rate": 5.630074442345257e-08,
      "loss": 0.5401,
      "step": 7485
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7125136364076405,
      "learning_rate": 5.599240246915882e-08,
      "loss": 0.5059,
      "step": 7486
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9879095274100037,
      "learning_rate": 5.5684902427852184e-08,
      "loss": 0.5688,
      "step": 7487
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8698599708981681,
      "learning_rate": 5.537824435189742e-08,
      "loss": 0.6757,
      "step": 7488
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.055570003412716,
      "learning_rate": 5.507242829351445e-08,
      "loss": 0.6763,
      "step": 7489
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8004758143381725,
      "learning_rate": 5.476745430478159e-08,
      "loss": 0.6024,
      "step": 7490
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1545071889606615,
      "learning_rate": 5.446332243763175e-08,
      "loss": 0.6503,
      "step": 7491
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8329510466577515,
      "learning_rate": 5.416003274385628e-08,
      "loss": 0.5604,
      "step": 7492
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7050925062683882,
      "learning_rate": 5.385758527510165e-08,
      "loss": 0.4843,
      "step": 7493
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.797023627805345,
      "learning_rate": 5.355598008287166e-08,
      "loss": 0.5367,
      "step": 7494
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6770177519989335,
      "learning_rate": 5.325521721852634e-08,
      "loss": 0.5337,
      "step": 7495
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7650629487914867,
      "learning_rate": 5.2955296733283035e-08,
      "loss": 0.5682,
      "step": 7496
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6991932344743291,
      "learning_rate": 5.2656218678214796e-08,
      "loss": 0.5234,
      "step": 7497
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7406186254989756,
      "learning_rate": 5.235798310425144e-08,
      "loss": 0.567,
      "step": 7498
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6333883574602763,
      "learning_rate": 5.2060590062179564e-08,
      "loss": 0.5034,
      "step": 7499
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8411842986062696,
      "learning_rate": 5.1764039602641425e-08,
      "loss": 0.6154,
      "step": 7500
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8398522524183845,
      "learning_rate": 5.1468331776137194e-08,
      "loss": 0.5931,
      "step": 7501
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6295971104189083,
      "learning_rate": 5.117346663302325e-08,
      "loss": 0.5134,
      "step": 7502
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7133737735344948,
      "learning_rate": 5.0879444223510545e-08,
      "loss": 0.5133,
      "step": 7503
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.785016914148015,
      "learning_rate": 5.058626459766902e-08,
      "loss": 0.5233,
      "step": 7504
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6561288306975454,
      "learning_rate": 5.02939278054243e-08,
      "loss": 0.5598,
      "step": 7505
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7467025471001423,
      "learning_rate": 5.000243389655823e-08,
      "loss": 0.5059,
      "step": 7506
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6969993780799142,
      "learning_rate": 4.9711782920708884e-08,
      "loss": 0.5592,
      "step": 7507
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7316437964232073,
      "learning_rate": 4.9421974927371106e-08,
      "loss": 0.5455,
      "step": 7508
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8792334846283368,
      "learning_rate": 4.913300996589598e-08,
      "loss": 0.6052,
      "step": 7509
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8550110046030751,
      "learning_rate": 4.8844888085491924e-08,
      "loss": 0.6383,
      "step": 7510
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9171351622488054,
      "learning_rate": 4.855760933522247e-08,
      "loss": 0.6374,
      "step": 7511
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8955492983908454,
      "learning_rate": 4.827117376400847e-08,
      "loss": 0.5689,
      "step": 7512
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7459772028562747,
      "learning_rate": 4.798558142062648e-08,
      "loss": 0.5588,
      "step": 7513
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8590116947538903,
      "learning_rate": 4.770083235371037e-08,
      "loss": 0.6369,
      "step": 7514
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8198050199769966,
      "learning_rate": 4.741692661175023e-08,
      "loss": 0.6195,
      "step": 7515
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7873992295593397,
      "learning_rate": 4.7133864243091296e-08,
      "loss": 0.5965,
      "step": 7516
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9160868531555808,
      "learning_rate": 4.685164529593722e-08,
      "loss": 0.5691,
      "step": 7517
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8000447445083542,
      "learning_rate": 4.657026981834623e-08,
      "loss": 0.6108,
      "step": 7518
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8085491423626913,
      "learning_rate": 4.628973785823387e-08,
      "loss": 0.6165,
      "step": 7519
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6795877964477249,
      "learning_rate": 4.6010049463371953e-08,
      "loss": 0.512,
      "step": 7520
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.975881642312962,
      "learning_rate": 4.573120468138792e-08,
      "loss": 0.6768,
      "step": 7521
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.751163012352079,
      "learning_rate": 4.5453203559766566e-08,
      "loss": 0.5635,
      "step": 7522
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7118732092834387,
      "learning_rate": 4.517604614584892e-08,
      "loss": 0.5145,
      "step": 7523
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6823053069108216,
      "learning_rate": 4.489973248683111e-08,
      "loss": 0.4954,
      "step": 7524
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7182483845788249,
      "learning_rate": 4.462426262976716e-08,
      "loss": 0.5542,
      "step": 7525
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8170515846596358,
      "learning_rate": 4.434963662156677e-08,
      "loss": 0.6004,
      "step": 7526
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.3205693102806864,
      "learning_rate": 4.407585450899587e-08,
      "loss": 0.6224,
      "step": 7527
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8456537587567948,
      "learning_rate": 4.380291633867606e-08,
      "loss": 0.6312,
      "step": 7528
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9061800139308304,
      "learning_rate": 4.353082215708626e-08,
      "loss": 0.6636,
      "step": 7529
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6291401034723934,
      "learning_rate": 4.3259572010561635e-08,
      "loss": 0.5266,
      "step": 7530
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9078034208417431,
      "learning_rate": 4.2989165945292453e-08,
      "loss": 0.6589,
      "step": 7531
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8683726750442915,
      "learning_rate": 4.2719604007327445e-08,
      "loss": 0.6134,
      "step": 7532
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8111394987920763,
      "learning_rate": 4.2450886242568767e-08,
      "loss": 0.5968,
      "step": 7533
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7948536402146265,
      "learning_rate": 4.2183012696776494e-08,
      "loss": 0.5738,
      "step": 7534
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8439442302535923,
      "learning_rate": 4.191598341556746e-08,
      "loss": 0.6332,
      "step": 7535
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8414604968394939,
      "learning_rate": 4.164979844441308e-08,
      "loss": 0.6055,
      "step": 7536
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6850448926497814,
      "learning_rate": 4.138445782864264e-08,
      "loss": 0.5203,
      "step": 7537
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7083976371231931,
      "learning_rate": 4.111996161344056e-08,
      "loss": 0.6204,
      "step": 7538
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8068246758263687,
      "learning_rate": 4.085630984384803e-08,
      "loss": 0.5136,
      "step": 7539
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7421138418611758,
      "learning_rate": 4.0593502564761354e-08,
      "loss": 0.5004,
      "step": 7540
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.680298252757217,
      "learning_rate": 4.033153982093474e-08,
      "loss": 0.5057,
      "step": 7541
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.892761612727089,
      "learning_rate": 4.00704216569775e-08,
      "loss": 0.5793,
      "step": 7542
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5991013086495907,
      "learning_rate": 3.9810148117355175e-08,
      "loss": 0.4995,
      "step": 7543
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7813304815649467,
      "learning_rate": 3.955071924639009e-08,
      "loss": 0.535,
      "step": 7544
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7147271161143031,
      "learning_rate": 3.929213508825913e-08,
      "loss": 0.528,
      "step": 7545
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7271019639566279,
      "learning_rate": 3.903439568699818e-08,
      "loss": 0.5357,
      "step": 7546
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6924238607904464,
      "learning_rate": 3.877750108649603e-08,
      "loss": 0.5599,
      "step": 7547
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8227709696525372,
      "learning_rate": 3.8521451330499894e-08,
      "loss": 0.6816,
      "step": 7548
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8139320579366447,
      "learning_rate": 3.8266246462612675e-08,
      "loss": 0.5432,
      "step": 7549
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8977256524792877,
      "learning_rate": 3.8011886526292394e-08,
      "loss": 0.6683,
      "step": 7550
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7313993538405529,
      "learning_rate": 3.77583715648544e-08,
      "loss": 0.5152,
      "step": 7551
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1248106578866355,
      "learning_rate": 3.7505701621469714e-08,
      "loss": 0.582,
      "step": 7552
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7232042933147221,
      "learning_rate": 3.725387673916448e-08,
      "loss": 0.5303,
      "step": 7553
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8607725374247168,
      "learning_rate": 3.700289696082271e-08,
      "loss": 0.622,
      "step": 7554
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6810540978016121,
      "learning_rate": 3.675276232918357e-08,
      "loss": 0.554,
      "step": 7555
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7691704879145688,
      "learning_rate": 3.6503472886842415e-08,
      "loss": 0.5098,
      "step": 7556
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7989962994326033,
      "learning_rate": 3.6255028676250305e-08,
      "loss": 0.6368,
      "step": 7557
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.954583259642225,
      "learning_rate": 3.600742973971505e-08,
      "loss": 0.6488,
      "step": 7558
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7879537975219921,
      "learning_rate": 3.57606761193996e-08,
      "loss": 0.5595,
      "step": 7559
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7481630461900315,
      "learning_rate": 3.5514767857324774e-08,
      "loss": 0.5969,
      "step": 7560
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2515041888048102,
      "learning_rate": 3.526970499536486e-08,
      "loss": 0.6275,
      "step": 7561
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7774979304075383,
      "learning_rate": 3.5025487575251485e-08,
      "loss": 0.5216,
      "step": 7562
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6939490686347481,
      "learning_rate": 3.4782115638573586e-08,
      "loss": 0.54,
      "step": 7563
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7695443028938527,
      "learning_rate": 3.453958922677414e-08,
      "loss": 0.621,
      "step": 7564
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6020808565150131,
      "learning_rate": 3.4297908381152324e-08,
      "loss": 0.4413,
      "step": 7565
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.76437625223097,
      "learning_rate": 3.405707314286466e-08,
      "loss": 0.5897,
      "step": 7566
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6480458169495762,
      "learning_rate": 3.3817083552922793e-08,
      "loss": 0.5276,
      "step": 7567
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7104413272447241,
      "learning_rate": 3.357793965219402e-08,
      "loss": 0.5646,
      "step": 7568
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6836036129920416,
      "learning_rate": 3.3339641481402455e-08,
      "loss": 0.5546,
      "step": 7569
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7126474655394591,
      "learning_rate": 3.3102189081127835e-08,
      "loss": 0.5127,
      "step": 7570
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7887757503717319,
      "learning_rate": 3.286558249180505e-08,
      "loss": 0.4984,
      "step": 7571
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6364014173268464,
      "learning_rate": 3.262982175372686e-08,
      "loss": 0.5045,
      "step": 7572
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9592387600886884,
      "learning_rate": 3.239490690704006e-08,
      "loss": 0.6247,
      "step": 7573
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6965653049718208,
      "learning_rate": 3.216083799174874e-08,
      "loss": 0.5304,
      "step": 7574
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6726101160073129,
      "learning_rate": 3.1927615047712135e-08,
      "loss": 0.5497,
      "step": 7575
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7061453200551265,
      "learning_rate": 3.16952381146457e-08,
      "loss": 0.5149,
      "step": 7576
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6775519593995715,
      "learning_rate": 3.146370723212056e-08,
      "loss": 0.5449,
      "step": 7577
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6943310331760596,
      "learning_rate": 3.1233022439564055e-08,
      "loss": 0.5371,
      "step": 7578
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8425427798288586,
      "learning_rate": 3.100318377625977e-08,
      "loss": 0.6013,
      "step": 7579
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9160785143760598,
      "learning_rate": 3.0774191281346956e-08,
      "loss": 0.6379,
      "step": 7580
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.703069754176784,
      "learning_rate": 3.054604499381997e-08,
      "loss": 0.5836,
      "step": 7581
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7052626811005301,
      "learning_rate": 3.0318744952529954e-08,
      "loss": 0.5668,
      "step": 7582
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6703817169944095,
      "learning_rate": 3.0092291196184285e-08,
      "loss": 0.5498,
      "step": 7583
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7093961310763612,
      "learning_rate": 2.986668376334545e-08,
      "loss": 0.4832,
      "step": 7584
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8194420737638893,
      "learning_rate": 2.96419226924316e-08,
      "loss": 0.5692,
      "step": 7585
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6302911122407827,
      "learning_rate": 2.941800802171768e-08,
      "loss": 0.4973,
      "step": 7586
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7758949579106059,
      "learning_rate": 2.9194939789333743e-08,
      "loss": 0.5273,
      "step": 7587
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8939951709116091,
      "learning_rate": 2.8972718033266068e-08,
      "loss": 0.5628,
      "step": 7588
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7439240101071243,
      "learning_rate": 2.875134279135716e-08,
      "loss": 0.5666,
      "step": 7589
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6641650024767523,
      "learning_rate": 2.8530814101304094e-08,
      "loss": 0.5479,
      "step": 7590
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8909326656246606,
      "learning_rate": 2.8311132000660712e-08,
      "loss": 0.6051,
      "step": 7591
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9336998038701715,
      "learning_rate": 2.8092296526837092e-08,
      "loss": 0.6334,
      "step": 7592
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9265661366575673,
      "learning_rate": 2.7874307717098425e-08,
      "loss": 0.6137,
      "step": 7593
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.57553316797106,
      "learning_rate": 2.7657165608566128e-08,
      "loss": 0.4506,
      "step": 7594
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8949490207016785,
      "learning_rate": 2.7440870238216733e-08,
      "loss": 0.5264,
      "step": 7595
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6224418763795442,
      "learning_rate": 2.7225421642883554e-08,
      "loss": 0.5402,
      "step": 7596
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9481257507679741,
      "learning_rate": 2.7010819859255022e-08,
      "loss": 0.6386,
      "step": 7597
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8198017356079316,
      "learning_rate": 2.6797064923875792e-08,
      "loss": 0.654,
      "step": 7598
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9115867113271215,
      "learning_rate": 2.6584156873145638e-08,
      "loss": 0.583,
      "step": 7599
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8080440274525793,
      "learning_rate": 2.6372095743321115e-08,
      "loss": 0.5628,
      "step": 7600
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6894831506206119,
      "learning_rate": 2.6160881570513884e-08,
      "loss": 0.5348,
      "step": 7601
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9214827202214521,
      "learning_rate": 2.5950514390691296e-08,
      "loss": 0.6196,
      "step": 7602
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4383952745109176,
      "learning_rate": 2.574099423967691e-08,
      "loss": 0.5653,
      "step": 7603
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9832497162835971,
      "learning_rate": 2.5532321153149962e-08,
      "loss": 0.5451,
      "step": 7604
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.760822492437758,
      "learning_rate": 2.5324495166644813e-08,
      "loss": 0.5494,
      "step": 7605
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6849650968792051,
      "learning_rate": 2.5117516315552038e-08,
      "loss": 0.4845,
      "step": 7606
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7551381448446217,
      "learning_rate": 2.4911384635118996e-08,
      "loss": 0.5166,
      "step": 7607
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6942683914485269,
      "learning_rate": 2.47061001604465e-08,
      "loss": 0.5681,
      "step": 7608
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6164606785817738,
      "learning_rate": 2.4501662926493252e-08,
      "loss": 0.5015,
      "step": 7609
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.933629322704433,
      "learning_rate": 2.429807296807196e-08,
      "loss": 0.6731,
      "step": 7610
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7075975356508541,
      "learning_rate": 2.409533031985267e-08,
      "loss": 0.528,
      "step": 7611
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9408293318238553,
      "learning_rate": 2.389343501635999e-08,
      "loss": 0.6494,
      "step": 7612
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.739327088547856,
      "learning_rate": 2.3692387091974744e-08,
      "loss": 0.512,
      "step": 7613
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6672513485985363,
      "learning_rate": 2.3492186580932884e-08,
      "loss": 0.5183,
      "step": 7614
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6385530075485687,
      "learning_rate": 2.329283351732714e-08,
      "loss": 0.4924,
      "step": 7615
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6117564201872002,
      "learning_rate": 2.30943279351048e-08,
      "loss": 0.5451,
      "step": 7616
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7053936823477176,
      "learning_rate": 2.2896669868069378e-08,
      "loss": 0.5679,
      "step": 7617
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6139386673899543,
      "learning_rate": 2.269985934988006e-08,
      "loss": 0.5171,
      "step": 7618
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6924980335519657,
      "learning_rate": 2.2503896414051153e-08,
      "loss": 0.5184,
      "step": 7619
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7836532609520706,
      "learning_rate": 2.2308781093953736e-08,
      "loss": 0.5199,
      "step": 7620
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6822585771271885,
      "learning_rate": 2.211451342281401e-08,
      "loss": 0.5106,
      "step": 7621
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6498117269238174,
      "learning_rate": 2.1921093433712737e-08,
      "loss": 0.527,
      "step": 7622
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8263831589742552,
      "learning_rate": 2.1728521159588566e-08,
      "loss": 0.6286,
      "step": 7623
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9478364554102605,
      "learning_rate": 2.1536796633233603e-08,
      "loss": 0.6193,
      "step": 7624
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8074256039605504,
      "learning_rate": 2.1345919887297282e-08,
      "loss": 0.5908,
      "step": 7625
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7687298793058737,
      "learning_rate": 2.115589095428361e-08,
      "loss": 0.5159,
      "step": 7626
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8953012025285586,
      "learning_rate": 2.0966709866552247e-08,
      "loss": 0.6637,
      "step": 7627
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6510354904056129,
      "learning_rate": 2.07783766563191e-08,
      "loss": 0.485,
      "step": 7628
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6771934372010611,
      "learning_rate": 2.0590891355655172e-08,
      "loss": 0.5039,
      "step": 7629
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8862268120628601,
      "learning_rate": 2.0404253996487156e-08,
      "loss": 0.6675,
      "step": 7630
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6549563332423868,
      "learning_rate": 2.0218464610597956e-08,
      "loss": 0.5177,
      "step": 7631
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.815536908391313,
      "learning_rate": 2.003352322962504e-08,
      "loss": 0.6283,
      "step": 7632
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8024763610746151,
      "learning_rate": 1.9849429885062664e-08,
      "loss": 0.5837,
      "step": 7633
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6538272674063264,
      "learning_rate": 1.9666184608259086e-08,
      "loss": 0.512,
      "step": 7634
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9747424145626216,
      "learning_rate": 1.9483787430419345e-08,
      "loss": 0.5918,
      "step": 7635
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7269490009815838,
      "learning_rate": 1.9302238382604143e-08,
      "loss": 0.5391,
      "step": 7636
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7955981522535058,
      "learning_rate": 1.9121537495729313e-08,
      "loss": 0.553,
      "step": 7637
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7532457359610292,
      "learning_rate": 1.894168480056635e-08,
      "loss": 0.5258,
      "step": 7638
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7786838948230345,
      "learning_rate": 1.8762680327741867e-08,
      "loss": 0.5158,
      "step": 7639
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7705090147624136,
      "learning_rate": 1.8584524107739254e-08,
      "loss": 0.6039,
      "step": 7640
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6271619679119295,
      "learning_rate": 1.840721617089536e-08,
      "loss": 0.4755,
      "step": 7641
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8242831530757934,
      "learning_rate": 1.823075654740547e-08,
      "loss": 0.5875,
      "step": 7642
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9118442549462102,
      "learning_rate": 1.8055145267317777e-08,
      "loss": 0.6527,
      "step": 7643
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8089574031526846,
      "learning_rate": 1.788038236053724e-08,
      "loss": 0.5842,
      "step": 7644
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0375731049153643,
      "learning_rate": 1.7706467856824506e-08,
      "loss": 0.5502,
      "step": 7645
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8947846294822129,
      "learning_rate": 1.753340178579477e-08,
      "loss": 0.6349,
      "step": 7646
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7154147934546784,
      "learning_rate": 1.7361184176920008e-08,
      "loss": 0.5444,
      "step": 7647
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7088735427201656,
      "learning_rate": 1.718981505952677e-08,
      "loss": 0.5615,
      "step": 7648
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9218429625809412,
      "learning_rate": 1.7019294462797265e-08,
      "loss": 0.6155,
      "step": 7649
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8195366290720348,
      "learning_rate": 1.6849622415769928e-08,
      "loss": 0.6005,
      "step": 7650
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8740285290283316,
      "learning_rate": 1.6680798947337762e-08,
      "loss": 0.6114,
      "step": 7651
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8197519517165034,
      "learning_rate": 1.6512824086249435e-08,
      "loss": 0.6723,
      "step": 7652
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.848332027468816,
      "learning_rate": 1.6345697861109843e-08,
      "loss": 0.6257,
      "step": 7653
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7602282230177202,
      "learning_rate": 1.6179420300378447e-08,
      "loss": 0.5368,
      "step": 7654
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.523336022225611,
      "learning_rate": 1.6013991432370368e-08,
      "loss": 0.6458,
      "step": 7655
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7357873236779066,
      "learning_rate": 1.584941128525752e-08,
      "loss": 0.6184,
      "step": 7656
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6648318920546201,
      "learning_rate": 1.568567988706471e-08,
      "loss": 0.4788,
      "step": 7657
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7455860455840642,
      "learning_rate": 1.5522797265674628e-08,
      "loss": 0.5713,
      "step": 7658
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.767895085522298,
      "learning_rate": 1.5360763448824533e-08,
      "loss": 0.5014,
      "step": 7659
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6436561928192716,
      "learning_rate": 1.51995784641068e-08,
      "loss": 0.5185,
      "step": 7660
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.946774318535707,
      "learning_rate": 1.5039242338969474e-08,
      "loss": 0.5827,
      "step": 7661
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8146163446823751,
      "learning_rate": 1.4879755100716265e-08,
      "loss": 0.6285,
      "step": 7662
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.963849630989354,
      "learning_rate": 1.4721116776506005e-08,
      "loss": 0.6304,
      "step": 7663
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.050497809462164,
      "learning_rate": 1.45633273933532e-08,
      "loss": 0.6008,
      "step": 7664
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8619927286062997,
      "learning_rate": 1.4406386978128017e-08,
      "loss": 0.6373,
      "step": 7665
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.754761807069326,
      "learning_rate": 1.4250295557555748e-08,
      "loss": 0.553,
      "step": 7666
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8762840585867455,
      "learning_rate": 1.4095053158216244e-08,
      "loss": 0.578,
      "step": 7667
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6853056973897615,
      "learning_rate": 1.3940659806547242e-08,
      "loss": 0.4753,
      "step": 7668
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7746630447024893,
      "learning_rate": 1.3787115528839379e-08,
      "loss": 0.5713,
      "step": 7669
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6237274074845542,
      "learning_rate": 1.363442035123952e-08,
      "loss": 0.495,
      "step": 7670
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8055965862715443,
      "learning_rate": 1.3482574299750195e-08,
      "loss": 0.6074,
      "step": 7671
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7166271483017809,
      "learning_rate": 1.3331577400229611e-08,
      "loss": 0.5492,
      "step": 7672
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9617096777816452,
      "learning_rate": 1.3181429678390534e-08,
      "loss": 0.6534,
      "step": 7673
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6914267583377445,
      "learning_rate": 1.3032131159801397e-08,
      "loss": 0.5591,
      "step": 7674
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9030150608544938,
      "learning_rate": 1.2883681869886866e-08,
      "loss": 0.6313,
      "step": 7675
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6728825040994538,
      "learning_rate": 1.2736081833925607e-08,
      "loss": 0.5605,
      "step": 7676
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8455372774306043,
      "learning_rate": 1.258933107705307e-08,
      "loss": 0.6027,
      "step": 7677
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8132978841711667,
      "learning_rate": 1.2443429624258707e-08,
      "loss": 0.6431,
      "step": 7678
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7470156413722059,
      "learning_rate": 1.2298377500388204e-08,
      "loss": 0.6046,
      "step": 7679
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6282495803286285,
      "learning_rate": 1.2154174730142908e-08,
      "loss": 0.4967,
      "step": 7680
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7577861601460566,
      "learning_rate": 1.2010821338078183e-08,
      "loss": 0.5325,
      "step": 7681
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8544789700296355,
      "learning_rate": 1.1868317348606163e-08,
      "loss": 0.6396,
      "step": 7682
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9000443677134143,
      "learning_rate": 1.1726662785994103e-08,
      "loss": 0.6668,
      "step": 7683
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7819674662950078,
      "learning_rate": 1.1585857674363821e-08,
      "loss": 0.6322,
      "step": 7684
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8640397434417426,
      "learning_rate": 1.1445902037692801e-08,
      "loss": 0.5418,
      "step": 7685
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6942189142401901,
      "learning_rate": 1.1306795899814205e-08,
      "loss": 0.5229,
      "step": 7686
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8252149382811961,
      "learning_rate": 1.1168539284416858e-08,
      "loss": 0.5897,
      "step": 7687
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9370495691978186,
      "learning_rate": 1.1031132215043593e-08,
      "loss": 0.6481,
      "step": 7688
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9468573208554496,
      "learning_rate": 1.089457471509403e-08,
      "loss": 0.6029,
      "step": 7689
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7799486706748221,
      "learning_rate": 1.0758866807822344e-08,
      "loss": 0.6144,
      "step": 7690
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9564609198009918,
      "learning_rate": 1.0624008516337825e-08,
      "loss": 0.6432,
      "step": 7691
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9950670412142393,
      "learning_rate": 1.0489999863605994e-08,
      "loss": 0.6406,
      "step": 7692
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9867067801806185,
      "learning_rate": 1.0356840872446373e-08,
      "loss": 0.6618,
      "step": 7693
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.691692246098567,
      "learning_rate": 1.0224531565535267e-08,
      "loss": 0.5272,
      "step": 7694
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9529959200558118,
      "learning_rate": 1.0093071965403544e-08,
      "loss": 0.6626,
      "step": 7695
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8631900570623116,
      "learning_rate": 9.962462094437186e-09,
      "loss": 0.6504,
      "step": 7696
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7051312279628598,
      "learning_rate": 9.83270197487729e-09,
      "loss": 0.5292,
      "step": 7697
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.624240652680998,
      "learning_rate": 9.703791628821735e-09,
      "loss": 0.5128,
      "step": 7698
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0862893000426475,
      "learning_rate": 9.575731078221295e-09,
      "loss": 0.628,
      "step": 7699
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6151464782986507,
      "learning_rate": 9.448520344884082e-09,
      "loss": 0.5222,
      "step": 7700
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8302844615144636,
      "learning_rate": 9.322159450472768e-09,
      "loss": 0.618,
      "step": 7701
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9629715401901966,
      "learning_rate": 9.196648416505694e-09,
      "loss": 0.638,
      "step": 7702
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6284057979685642,
      "learning_rate": 9.071987264355209e-09,
      "loss": 0.4622,
      "step": 7703
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7946363867564209,
      "learning_rate": 8.948176015249887e-09,
      "loss": 0.5813,
      "step": 7704
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8240984944842134,
      "learning_rate": 8.825214690273976e-09,
      "loss": 0.5794,
      "step": 7705
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7708128742040896,
      "learning_rate": 8.703103310366279e-09,
      "loss": 0.53,
      "step": 7706
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7930533939687292,
      "learning_rate": 8.581841896321274e-09,
      "loss": 0.5901,
      "step": 7707
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8868068286654145,
      "learning_rate": 8.461430468788557e-09,
      "loss": 0.588,
      "step": 7708
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8641175805955916,
      "learning_rate": 8.341869048272278e-09,
      "loss": 0.6371,
      "step": 7709
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8669576713465519,
      "learning_rate": 8.223157655133374e-09,
      "loss": 0.5611,
      "step": 7710
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6462270456572182,
      "learning_rate": 8.105296309586786e-09,
      "loss": 0.5056,
      "step": 7711
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6507443854245548,
      "learning_rate": 7.988285031702569e-09,
      "loss": 0.5629,
      "step": 7712
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9160757491823722,
      "learning_rate": 7.872123841407564e-09,
      "loss": 0.6535,
      "step": 7713
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7848714663431117,
      "learning_rate": 7.756812758482058e-09,
      "loss": 0.5844,
      "step": 7714
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9968209939293037,
      "learning_rate": 7.642351802562564e-09,
      "loss": 0.6688,
      "step": 7715
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7743367133853505,
      "learning_rate": 7.52874099314016e-09,
      "loss": 0.5677,
      "step": 7716
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.985855860243906,
      "learning_rate": 7.4159803495627014e-09,
      "loss": 0.5944,
      "step": 7717
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7769300759622173,
      "learning_rate": 7.3040698910309405e-09,
      "loss": 0.5758,
      "step": 7718
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8238070914861223,
      "learning_rate": 7.1930096366024104e-09,
      "loss": 0.6125,
      "step": 7719
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8968157418227001,
      "learning_rate": 7.08279960518976e-09,
      "loss": 0.6793,
      "step": 7720
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7954210920795329,
      "learning_rate": 6.9734398155607565e-09,
      "loss": 0.5517,
      "step": 7721
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9231282029029846,
      "learning_rate": 6.864930286337723e-09,
      "loss": 0.5853,
      "step": 7722
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6722337777132225,
      "learning_rate": 6.757271035999213e-09,
      "loss": 0.5089,
      "step": 7723
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7967805112616456,
      "learning_rate": 6.650462082878339e-09,
      "loss": 0.6144,
      "step": 7724
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8618017600310541,
      "learning_rate": 6.544503445162776e-09,
      "loss": 0.641,
      "step": 7725
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8572291298679268,
      "learning_rate": 6.439395140896976e-09,
      "loss": 0.6132,
      "step": 7726
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8126804757304016,
      "learning_rate": 6.3351371879799565e-09,
      "loss": 0.5717,
      "step": 7727
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8912996518047204,
      "learning_rate": 6.2317296041658485e-09,
      "loss": 0.6713,
      "step": 7728
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7494857653277093,
      "learning_rate": 6.129172407062789e-09,
      "loss": 0.5811,
      "step": 7729
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6505078175920673,
      "learning_rate": 6.027465614136252e-09,
      "loss": 0.5485,
      "step": 7730
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7161590675565519,
      "learning_rate": 5.9266092427051614e-09,
      "loss": 0.5381,
      "step": 7731
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.727954679948721,
      "learning_rate": 5.826603309944667e-09,
      "loss": 0.5307,
      "step": 7732
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7296040349339669,
      "learning_rate": 5.727447832885036e-09,
      "loss": 0.4967,
      "step": 7733
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8287125089496792,
      "learning_rate": 5.6291428284110934e-09,
      "loss": 0.6687,
      "step": 7734
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7386894344806016,
      "learning_rate": 5.531688313262784e-09,
      "loss": 0.5609,
      "step": 7735
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6992615364351703,
      "learning_rate": 5.435084304036276e-09,
      "loss": 0.521,
      "step": 7736
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.645566685331598,
      "learning_rate": 5.339330817181743e-09,
      "loss": 0.5036,
      "step": 7737
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7415998513341338,
      "learning_rate": 5.244427869005586e-09,
      "loss": 0.5239,
      "step": 7738
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8696586769269208,
      "learning_rate": 5.150375475667657e-09,
      "loss": 0.6374,
      "step": 7739
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6585700511068008,
      "learning_rate": 5.0571736531851435e-09,
      "loss": 0.4942,
      "step": 7740
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6818694657527492,
      "learning_rate": 4.964822417429238e-09,
      "loss": 0.5017,
      "step": 7741
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9738124329608369,
      "learning_rate": 4.87332178412625e-09,
      "loss": 0.6612,
      "step": 7742
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5989502745504444,
      "learning_rate": 4.7826717688570504e-09,
      "loss": 0.465,
      "step": 7743
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7610246981313702,
      "learning_rate": 4.692872387059844e-09,
      "loss": 0.5602,
      "step": 7744
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7948919264223075,
      "learning_rate": 4.603923654025177e-09,
      "loss": 0.524,
      "step": 7745
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9546635423186289,
      "learning_rate": 4.515825584900935e-09,
      "loss": 0.6557,
      "step": 7746
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6425175540783005,
      "learning_rate": 4.428578194689004e-09,
      "loss": 0.4951,
      "step": 7747
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7739844154182344,
      "learning_rate": 4.3421814982463896e-09,
      "loss": 0.5958,
      "step": 7748
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7354734804398161,
      "learning_rate": 4.256635510286322e-09,
      "loss": 0.5468,
      "step": 7749
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7094949902187434,
      "learning_rate": 4.171940245375483e-09,
      "loss": 0.5637,
      "step": 7750
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9152832736516887,
      "learning_rate": 4.0880957179373345e-09,
      "loss": 0.6787,
      "step": 7751
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8337972076499247,
      "learning_rate": 4.005101942249345e-09,
      "loss": 0.6361,
      "step": 7752
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7116564328879181,
      "learning_rate": 3.922958932444654e-09,
      "loss": 0.5385,
      "step": 7753
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7469676186804468,
      "learning_rate": 3.841666702511515e-09,
      "loss": 0.5242,
      "step": 7754
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6889087565562085,
      "learning_rate": 3.761225266292745e-09,
      "loss": 0.5099,
      "step": 7755
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8503511104789901,
      "learning_rate": 3.6816346374868307e-09,
      "loss": 0.5719,
      "step": 7756
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8357174587071159,
      "learning_rate": 3.602894829647374e-09,
      "loss": 0.5505,
      "step": 7757
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7224718547138523,
      "learning_rate": 3.5250058561825397e-09,
      "loss": 0.5524,
      "step": 7758
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6978070506328778,
      "learning_rate": 3.447967730356716e-09,
      "loss": 0.4938,
      "step": 7759
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.88907344584442,
      "learning_rate": 3.371780465288299e-09,
      "loss": 0.5532,
      "step": 7760
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8917303071815923,
      "learning_rate": 3.2964440739508e-09,
      "loss": 0.6157,
      "step": 7761
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6840176788264855,
      "learning_rate": 3.221958569173955e-09,
      "loss": 0.5724,
      "step": 7762
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7945173049507899,
      "learning_rate": 3.148323963641509e-09,
      "loss": 0.5738,
      "step": 7763
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6977984037923312,
      "learning_rate": 3.0755402698923185e-09,
      "loss": 0.5739,
      "step": 7764
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8102152513572651,
      "learning_rate": 3.0036075003209152e-09,
      "loss": 0.5876,
      "step": 7765
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8342865402711346,
      "learning_rate": 2.9325256671774992e-09,
      "loss": 0.5828,
      "step": 7766
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8904558550414632,
      "learning_rate": 2.862294782565167e-09,
      "loss": 0.6963,
      "step": 7767
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8758157724870405,
      "learning_rate": 2.792914858444906e-09,
      "loss": 0.5964,
      "step": 7768
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7425124539882405,
      "learning_rate": 2.7243859066306e-09,
      "loss": 0.5852,
      "step": 7769
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7460310516973931,
      "learning_rate": 2.6567079387918026e-09,
      "loss": 0.5309,
      "step": 7770
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7889244025296429,
      "learning_rate": 2.5898809664537395e-09,
      "loss": 0.5286,
      "step": 7771
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8218427560324548,
      "learning_rate": 2.5239050009961962e-09,
      "loss": 0.5818,
      "step": 7772
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7004383419013205,
      "learning_rate": 2.4587800536546304e-09,
      "loss": 0.5161,
      "step": 7773
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8498689634209067,
      "learning_rate": 2.3945061355185038e-09,
      "loss": 0.626,
      "step": 7774
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.794196644507347,
      "learning_rate": 2.3310832575335065e-09,
      "loss": 0.5921,
      "step": 7775
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7973033067988463,
      "learning_rate": 2.2685114304998868e-09,
      "loss": 0.637,
      "step": 7776
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8177026760959344,
      "learning_rate": 2.2067906650724556e-09,
      "loss": 0.6068,
      "step": 7777
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8371703718276913,
      "learning_rate": 2.14592097176225e-09,
      "loss": 0.5878,
      "step": 7778
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7940912155992044,
      "learning_rate": 2.085902360934311e-09,
      "loss": 0.5465,
      "step": 7779
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6542199991423644,
      "learning_rate": 2.0267348428087974e-09,
      "loss": 0.5269,
      "step": 7780
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8336999965186498,
      "learning_rate": 1.9684184274626482e-09,
      "loss": 0.6512,
      "step": 7781
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7203921560481334,
      "learning_rate": 1.910953124825143e-09,
      "loss": 0.5767,
      "step": 7782
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7152695576658792,
      "learning_rate": 1.8543389446828986e-09,
      "loss": 0.5201,
      "step": 7783
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6492815330947064,
      "learning_rate": 1.7985758966759802e-09,
      "loss": 0.49,
      "step": 7784
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.672969385685014,
      "learning_rate": 1.7436639903012364e-09,
      "loss": 0.5268,
      "step": 7785
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.870950256196372,
      "learning_rate": 1.6896032349089653e-09,
      "loss": 0.6033,
      "step": 7786
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8307803805180253,
      "learning_rate": 1.636393639705136e-09,
      "loss": 0.6131,
      "step": 7787
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.685915269680486,
      "learning_rate": 1.5840352137508342e-09,
      "loss": 0.5046,
      "step": 7788
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2242661160829826,
      "learning_rate": 1.5325279659622605e-09,
      "loss": 0.6561,
      "step": 7789
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8246040025098534,
      "learning_rate": 1.4818719051107323e-09,
      "loss": 0.5787,
      "step": 7790
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8654962246155131,
      "learning_rate": 1.432067039822127e-09,
      "loss": 0.6085,
      "step": 7791
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7440648543329675,
      "learning_rate": 1.3831133785774387e-09,
      "loss": 0.5854,
      "step": 7792
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6533064685289787,
      "learning_rate": 1.3350109297138868e-09,
      "loss": 0.5154,
      "step": 7793
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6605468557127613,
      "learning_rate": 1.287759701421587e-09,
      "loss": 0.5663,
      "step": 7794
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6640633822597499,
      "learning_rate": 1.2413597017479907e-09,
      "loss": 0.4827,
      "step": 7795
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9687083555099592,
      "learning_rate": 1.1958109385945549e-09,
      "loss": 0.6545,
      "step": 7796
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9359954040840036,
      "learning_rate": 1.1511134197167429e-09,
      "loss": 0.6436,
      "step": 7797
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8085712408662021,
      "learning_rate": 1.1072671527267986e-09,
      "loss": 0.5153,
      "step": 7798
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7900497162608829,
      "learning_rate": 1.0642721450915272e-09,
      "loss": 0.57,
      "step": 7799
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0073835613094655,
      "learning_rate": 1.0221284041317392e-09,
      "loss": 0.6231,
      "step": 7800
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7447281939135997,
      "learning_rate": 9.808359370244713e-10,
      "loss": 0.5375,
      "step": 7801
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2691562437176505,
      "learning_rate": 9.403947508018763e-10,
      "loss": 0.5765,
      "step": 7802
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.603492045493172,
      "learning_rate": 9.008048523501123e-10,
      "loss": 0.4676,
      "step": 7803
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8399561608212126,
      "learning_rate": 8.620662484110087e-10,
      "loss": 0.6406,
      "step": 7804
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6660470559642359,
      "learning_rate": 8.241789455809557e-10,
      "loss": 0.487,
      "step": 7805
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7526997649275916,
      "learning_rate": 7.871429503125694e-10,
      "loss": 0.5727,
      "step": 7806
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8386335849922518,
      "learning_rate": 7.509582689124717e-10,
      "loss": 0.5359,
      "step": 7807
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8791408182547873,
      "learning_rate": 7.156249075424004e-10,
      "loss": 0.5908,
      "step": 7808
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8814804310839345,
      "learning_rate": 6.811428722186542e-10,
      "loss": 0.5871,
      "step": 7809
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8069524659368271,
      "learning_rate": 6.475121688143127e-10,
      "loss": 0.5866,
      "step": 7810
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0796992262489185,
      "learning_rate": 6.147328030553512e-10,
      "loss": 0.6439,
      "step": 7811
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6471025574360484,
      "learning_rate": 5.828047805245263e-10,
      "loss": 0.5278,
      "step": 7812
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7213080347264894,
      "learning_rate": 5.517281066586e-10,
      "loss": 0.5464,
      "step": 7813
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7544696732992185,
      "learning_rate": 5.215027867488954e-10,
      "loss": 0.5973,
      "step": 7814
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7980019980413136,
      "learning_rate": 4.921288259435164e-10,
      "loss": 0.5612,
      "step": 7815
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7209293917073583,
      "learning_rate": 4.63606229244018e-10,
      "loss": 0.5556,
      "step": 7816
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6936144025753888,
      "learning_rate": 4.359350015076258e-10,
      "loss": 0.5407,
      "step": 7817
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8459246286115272,
      "learning_rate": 4.0911514744668146e-10,
      "loss": 0.6508,
      "step": 7818
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.728279364160396,
      "learning_rate": 3.8314667162753226e-10,
      "loss": 0.5827,
      "step": 7819
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6240523422549289,
      "learning_rate": 3.5802957847330675e-10,
      "loss": 0.5268,
      "step": 7820
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1505071053576004,
      "learning_rate": 3.33763872260584e-10,
      "loss": 0.6181,
      "step": 7821
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.781740776424288,
      "learning_rate": 3.103495571216142e-10,
      "loss": 0.6263,
      "step": 7822
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7154723188236493,
      "learning_rate": 2.877866370443183e-10,
      "loss": 0.5604,
      "step": 7823
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7804506360670698,
      "learning_rate": 2.6607511586951295e-10,
      "loss": 0.531,
      "step": 7824
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6245231106333156,
      "learning_rate": 2.4521499729590617e-10,
      "loss": 0.4899,
      "step": 7825
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8060527603216245,
      "learning_rate": 2.252062848745462e-10,
      "loss": 0.6094,
      "step": 7826
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7712624981343347,
      "learning_rate": 2.0604898201381784e-10,
      "loss": 0.5505,
      "step": 7827
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8325203915473648,
      "learning_rate": 1.8774309197500117e-10,
      "loss": 0.5425,
      "step": 7828
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6362454048618623,
      "learning_rate": 1.7028861787615757e-10,
      "loss": 0.5784,
      "step": 7829
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9404219665387779,
      "learning_rate": 1.5368556268935407e-10,
      "loss": 0.6056,
      "step": 7830
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.65911909680477,
      "learning_rate": 1.3793392924177362e-10,
      "loss": 0.5173,
      "step": 7831
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0428669071958223,
      "learning_rate": 1.2303372021571502e-10,
      "loss": 0.6511,
      "step": 7832
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7835975398135873,
      "learning_rate": 1.0898493814859301e-10,
      "loss": 0.5284,
      "step": 7833
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6754899777119601,
      "learning_rate": 9.57875854329382e-11,
      "loss": 0.5642,
      "step": 7834
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7817867229811551,
      "learning_rate": 8.34416643163971e-11,
      "loss": 0.5251,
      "step": 7835
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7628512553475276,
      "learning_rate": 7.194717690062191e-11,
      "loss": 0.5022,
      "step": 7836
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8942812694349074,
      "learning_rate": 6.130412514349094e-11,
      "loss": 0.6451,
      "step": 7837
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7026924685199204,
      "learning_rate": 5.151251085744324e-11,
      "loss": 0.5829,
      "step": 7838
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8938957575622153,
      "learning_rate": 4.2572335709478717e-11,
      "loss": 0.6132,
      "step": 7839
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6940844492508118,
      "learning_rate": 3.4483601222268237e-11,
      "loss": 0.5458,
      "step": 7840
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7017763326458094,
      "learning_rate": 2.724630877304346e-11,
      "loss": 0.538,
      "step": 7841
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7365141654995742,
      "learning_rate": 2.0860459594707062e-11,
      "loss": 0.5385,
      "step": 7842
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7958613214386577,
      "learning_rate": 1.532605477472249e-11,
      "loss": 0.5561,
      "step": 7843
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7231877128825722,
      "learning_rate": 1.0643095255113978e-11,
      "loss": 0.5354,
      "step": 7844
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9451208547386291,
      "learning_rate": 6.811581833021663e-12,
      "loss": 0.6171,
      "step": 7845
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9173825027155778,
      "learning_rate": 3.831515161811794e-12,
      "loss": 0.5922,
      "step": 7846
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.748525232596623,
      "learning_rate": 1.702895748301181e-12,
      "loss": 0.5628,
      "step": 7847
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6750411804468949,
      "learning_rate": 4.257239549776415e-13,
      "loss": 0.4978,
      "step": 7848
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8171339809221412,
      "learning_rate": 0.0,
      "loss": 0.5818,
      "step": 7849
    },
    {
      "epoch": 1.0,
      "step": 7849,
      "total_flos": 4047659296358400.0,
      "train_loss": 0.5952112748412087,
      "train_runtime": 40810.2508,
      "train_samples_per_second": 24.62,
      "train_steps_per_second": 0.192
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 7849,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2000,
  "total_flos": 4047659296358400.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}