{
  "best_metric": 1.960377812385559,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.47904191616766467,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0023952095808383233,
      "grad_norm": 2.7156052589416504,
      "learning_rate": 1e-05,
      "loss": 2.685,
      "step": 1
    },
    {
      "epoch": 0.0023952095808383233,
      "eval_loss": 4.920567035675049,
      "eval_runtime": 61.5443,
      "eval_samples_per_second": 11.423,
      "eval_steps_per_second": 2.86,
      "step": 1
    },
    {
      "epoch": 0.004790419161676647,
      "grad_norm": 2.7930190563201904,
      "learning_rate": 2e-05,
      "loss": 2.3645,
      "step": 2
    },
    {
      "epoch": 0.00718562874251497,
      "grad_norm": 3.0486397743225098,
      "learning_rate": 3e-05,
      "loss": 2.6402,
      "step": 3
    },
    {
      "epoch": 0.009580838323353293,
      "grad_norm": 3.499586582183838,
      "learning_rate": 4e-05,
      "loss": 2.7066,
      "step": 4
    },
    {
      "epoch": 0.011976047904191617,
      "grad_norm": 3.0709779262542725,
      "learning_rate": 5e-05,
      "loss": 2.38,
      "step": 5
    },
    {
      "epoch": 0.01437125748502994,
      "grad_norm": 3.09730863571167,
      "learning_rate": 6e-05,
      "loss": 2.4444,
      "step": 6
    },
    {
      "epoch": 0.016766467065868262,
      "grad_norm": 2.7251760959625244,
      "learning_rate": 7e-05,
      "loss": 2.5312,
      "step": 7
    },
    {
      "epoch": 0.019161676646706587,
      "grad_norm": 2.6668753623962402,
      "learning_rate": 8e-05,
      "loss": 1.8488,
      "step": 8
    },
    {
      "epoch": 0.02155688622754491,
      "grad_norm": 3.2054877281188965,
      "learning_rate": 9e-05,
      "loss": 2.266,
      "step": 9
    },
    {
      "epoch": 0.023952095808383235,
      "grad_norm": 3.0021796226501465,
      "learning_rate": 0.0001,
      "loss": 2.2668,
      "step": 10
    },
    {
      "epoch": 0.026347305389221556,
      "grad_norm": 3.2809689044952393,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.7756,
      "step": 11
    },
    {
      "epoch": 0.02874251497005988,
      "grad_norm": 3.2694637775421143,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.0423,
      "step": 12
    },
    {
      "epoch": 0.031137724550898204,
      "grad_norm": 2.802170991897583,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.168,
      "step": 13
    },
    {
      "epoch": 0.033532934131736525,
      "grad_norm": 3.2233448028564453,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.2676,
      "step": 14
    },
    {
      "epoch": 0.03592814371257485,
      "grad_norm": 3.124181032180786,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.3027,
      "step": 15
    },
    {
      "epoch": 0.03832335329341317,
      "grad_norm": 4.025321960449219,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.5852,
      "step": 16
    },
    {
      "epoch": 0.0407185628742515,
      "grad_norm": 3.461560010910034,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.0413,
      "step": 17
    },
    {
      "epoch": 0.04311377245508982,
      "grad_norm": 4.176211833953857,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.5048,
      "step": 18
    },
    {
      "epoch": 0.045508982035928146,
      "grad_norm": 2.9072725772857666,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.2905,
      "step": 19
    },
    {
      "epoch": 0.04790419161676647,
      "grad_norm": 3.2296340465545654,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.3121,
      "step": 20
    },
    {
      "epoch": 0.05029940119760479,
      "grad_norm": 3.0049846172332764,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.3655,
      "step": 21
    },
    {
      "epoch": 0.05269461077844311,
      "grad_norm": 3.4838130474090576,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.2905,
      "step": 22
    },
    {
      "epoch": 0.055089820359281436,
      "grad_norm": 3.136869192123413,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.0476,
      "step": 23
    },
    {
      "epoch": 0.05748502994011976,
      "grad_norm": 4.026731967926025,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.6264,
      "step": 24
    },
    {
      "epoch": 0.059880239520958084,
      "grad_norm": 3.9684910774230957,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.1909,
      "step": 25
    },
    {
      "epoch": 0.06227544910179641,
      "grad_norm": 4.399679183959961,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.986,
      "step": 26
    },
    {
      "epoch": 0.06467065868263473,
      "grad_norm": 4.570807933807373,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.0546,
      "step": 27
    },
    {
      "epoch": 0.06706586826347305,
      "grad_norm": 3.9964303970336914,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.4816,
      "step": 28
    },
    {
      "epoch": 0.06946107784431138,
      "grad_norm": 3.470351219177246,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.3921,
      "step": 29
    },
    {
      "epoch": 0.0718562874251497,
      "grad_norm": 4.955052852630615,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.8424,
      "step": 30
    },
    {
      "epoch": 0.07425149700598803,
      "grad_norm": 3.660054922103882,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.0488,
      "step": 31
    },
    {
      "epoch": 0.07664670658682635,
      "grad_norm": 5.011128902435303,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.452,
      "step": 32
    },
    {
      "epoch": 0.07904191616766468,
      "grad_norm": 3.9986367225646973,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.283,
      "step": 33
    },
    {
      "epoch": 0.081437125748503,
      "grad_norm": 4.050436019897461,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.5617,
      "step": 34
    },
    {
      "epoch": 0.08383233532934131,
      "grad_norm": 4.242917537689209,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.0287,
      "step": 35
    },
    {
      "epoch": 0.08622754491017964,
      "grad_norm": 3.5234057903289795,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.2036,
      "step": 36
    },
    {
      "epoch": 0.08862275449101796,
      "grad_norm": 3.7889249324798584,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.4001,
      "step": 37
    },
    {
      "epoch": 0.09101796407185629,
      "grad_norm": 3.3245418071746826,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.119,
      "step": 38
    },
    {
      "epoch": 0.09341317365269461,
      "grad_norm": 4.3238911628723145,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.271,
      "step": 39
    },
    {
      "epoch": 0.09580838323353294,
      "grad_norm": 4.324810981750488,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.0038,
      "step": 40
    },
    {
      "epoch": 0.09820359281437126,
      "grad_norm": 4.097146987915039,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.4418,
      "step": 41
    },
    {
      "epoch": 0.10059880239520957,
      "grad_norm": 3.926846742630005,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.2773,
      "step": 42
    },
    {
      "epoch": 0.1029940119760479,
      "grad_norm": 3.716679334640503,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.3006,
      "step": 43
    },
    {
      "epoch": 0.10538922155688622,
      "grad_norm": 3.246781826019287,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.1566,
      "step": 44
    },
    {
      "epoch": 0.10778443113772455,
      "grad_norm": 4.06755256652832,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.8382,
      "step": 45
    },
    {
      "epoch": 0.11017964071856287,
      "grad_norm": 3.3109564781188965,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.7281,
      "step": 46
    },
    {
      "epoch": 0.1125748502994012,
      "grad_norm": 3.8482367992401123,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.8236,
      "step": 47
    },
    {
      "epoch": 0.11497005988023952,
      "grad_norm": 3.687272071838379,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.6467,
      "step": 48
    },
    {
      "epoch": 0.11736526946107785,
      "grad_norm": 3.618502378463745,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.9025,
      "step": 49
    },
    {
      "epoch": 0.11976047904191617,
      "grad_norm": 6.180907726287842,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.9689,
      "step": 50
    },
    {
      "epoch": 0.11976047904191617,
      "eval_loss": 2.235183000564575,
      "eval_runtime": 62.2788,
      "eval_samples_per_second": 11.288,
      "eval_steps_per_second": 2.826,
      "step": 50
    },
    {
      "epoch": 0.12215568862275449,
      "grad_norm": 2.9579267501831055,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.4739,
      "step": 51
    },
    {
      "epoch": 0.12455089820359282,
      "grad_norm": 2.1343748569488525,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.232,
      "step": 52
    },
    {
      "epoch": 0.12694610778443113,
      "grad_norm": 1.8336989879608154,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.1554,
      "step": 53
    },
    {
      "epoch": 0.12934131736526946,
      "grad_norm": 1.6920663118362427,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.7704,
      "step": 54
    },
    {
      "epoch": 0.1317365269461078,
      "grad_norm": 1.568556785583496,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.7935,
      "step": 55
    },
    {
      "epoch": 0.1341317365269461,
      "grad_norm": 1.6336424350738525,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.8238,
      "step": 56
    },
    {
      "epoch": 0.13652694610778443,
      "grad_norm": 1.8044352531433105,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.9166,
      "step": 57
    },
    {
      "epoch": 0.13892215568862276,
      "grad_norm": 2.1575310230255127,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.1897,
      "step": 58
    },
    {
      "epoch": 0.14131736526946106,
      "grad_norm": 1.9996836185455322,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.2134,
      "step": 59
    },
    {
      "epoch": 0.1437125748502994,
      "grad_norm": 2.1102702617645264,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.2023,
      "step": 60
    },
    {
      "epoch": 0.14610778443113773,
      "grad_norm": 2.0401759147644043,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.0676,
      "step": 61
    },
    {
      "epoch": 0.14850299401197606,
      "grad_norm": 2.1752941608428955,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.4776,
      "step": 62
    },
    {
      "epoch": 0.15089820359281436,
      "grad_norm": 2.2318758964538574,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.234,
      "step": 63
    },
    {
      "epoch": 0.1532934131736527,
      "grad_norm": 2.3220291137695312,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.4716,
      "step": 64
    },
    {
      "epoch": 0.15568862275449102,
      "grad_norm": 2.5199949741363525,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.1721,
      "step": 65
    },
    {
      "epoch": 0.15808383233532936,
      "grad_norm": 2.9603450298309326,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.2214,
      "step": 66
    },
    {
      "epoch": 0.16047904191616766,
      "grad_norm": 3.301903247833252,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.0999,
      "step": 67
    },
    {
      "epoch": 0.162874251497006,
      "grad_norm": 2.8160386085510254,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.8893,
      "step": 68
    },
    {
      "epoch": 0.16526946107784432,
      "grad_norm": 2.4310691356658936,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.0635,
      "step": 69
    },
    {
      "epoch": 0.16766467065868262,
      "grad_norm": 2.6897270679473877,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.0636,
      "step": 70
    },
    {
      "epoch": 0.17005988023952096,
      "grad_norm": 3.205235719680786,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.3447,
      "step": 71
    },
    {
      "epoch": 0.1724550898203593,
      "grad_norm": 2.7721445560455322,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.1925,
      "step": 72
    },
    {
      "epoch": 0.17485029940119762,
      "grad_norm": 2.9733493328094482,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.2444,
      "step": 73
    },
    {
      "epoch": 0.17724550898203592,
      "grad_norm": 2.523561477661133,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.7005,
      "step": 74
    },
    {
      "epoch": 0.17964071856287425,
      "grad_norm": 2.866666316986084,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.1345,
      "step": 75
    },
    {
      "epoch": 0.18203592814371258,
      "grad_norm": 3.354912042617798,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.251,
      "step": 76
    },
    {
      "epoch": 0.1844311377245509,
      "grad_norm": 3.114210605621338,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.9312,
      "step": 77
    },
    {
      "epoch": 0.18682634730538922,
      "grad_norm": 4.299700736999512,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.11,
      "step": 78
    },
    {
      "epoch": 0.18922155688622755,
      "grad_norm": 3.205491781234741,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.1276,
      "step": 79
    },
    {
      "epoch": 0.19161676646706588,
      "grad_norm": 3.380143165588379,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.5675,
      "step": 80
    },
    {
      "epoch": 0.19401197604790418,
      "grad_norm": 3.2885549068450928,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.0711,
      "step": 81
    },
    {
      "epoch": 0.19640718562874251,
      "grad_norm": 3.2571210861206055,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.2092,
      "step": 82
    },
    {
      "epoch": 0.19880239520958085,
      "grad_norm": 3.121706247329712,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.0448,
      "step": 83
    },
    {
      "epoch": 0.20119760479041915,
      "grad_norm": 4.103664398193359,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.9461,
      "step": 84
    },
    {
      "epoch": 0.20359281437125748,
      "grad_norm": 2.8355600833892822,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.9167,
      "step": 85
    },
    {
      "epoch": 0.2059880239520958,
      "grad_norm": 3.132627487182617,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.2493,
      "step": 86
    },
    {
      "epoch": 0.20838323353293414,
      "grad_norm": 4.598633289337158,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.2171,
      "step": 87
    },
    {
      "epoch": 0.21077844311377245,
      "grad_norm": 3.929288864135742,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.5116,
      "step": 88
    },
    {
      "epoch": 0.21317365269461078,
      "grad_norm": 3.933358907699585,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.2981,
      "step": 89
    },
    {
      "epoch": 0.2155688622754491,
      "grad_norm": 4.908536911010742,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.9429,
      "step": 90
    },
    {
      "epoch": 0.2179640718562874,
      "grad_norm": 4.369875431060791,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.879,
      "step": 91
    },
    {
      "epoch": 0.22035928143712574,
      "grad_norm": 3.9446401596069336,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.1076,
      "step": 92
    },
    {
      "epoch": 0.22275449101796407,
      "grad_norm": 4.404725074768066,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.2458,
      "step": 93
    },
    {
      "epoch": 0.2251497005988024,
      "grad_norm": 2.8969099521636963,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.5444,
      "step": 94
    },
    {
      "epoch": 0.2275449101796407,
      "grad_norm": 3.3707683086395264,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.8561,
      "step": 95
    },
    {
      "epoch": 0.22994011976047904,
      "grad_norm": 2.848008394241333,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.6621,
      "step": 96
    },
    {
      "epoch": 0.23233532934131737,
      "grad_norm": 4.26475715637207,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.9782,
      "step": 97
    },
    {
      "epoch": 0.2347305389221557,
      "grad_norm": 3.2886276245117188,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.9189,
      "step": 98
    },
    {
      "epoch": 0.237125748502994,
      "grad_norm": 2.835848331451416,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.6805,
      "step": 99
    },
    {
      "epoch": 0.23952095808383234,
      "grad_norm": 3.6112892627716064,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.2854,
      "step": 100
    },
    {
      "epoch": 0.23952095808383234,
      "eval_loss": 2.1215388774871826,
      "eval_runtime": 62.2757,
      "eval_samples_per_second": 11.289,
      "eval_steps_per_second": 2.826,
      "step": 100
    },
    {
      "epoch": 0.24191616766467067,
      "grad_norm": 1.86229407787323,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.1337,
      "step": 101
    },
    {
      "epoch": 0.24431137724550897,
      "grad_norm": 1.9307291507720947,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.1401,
      "step": 102
    },
    {
      "epoch": 0.2467065868263473,
      "grad_norm": 1.7766854763031006,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.901,
      "step": 103
    },
    {
      "epoch": 0.24910179640718563,
      "grad_norm": 1.6351048946380615,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.2837,
      "step": 104
    },
    {
      "epoch": 0.25149700598802394,
      "grad_norm": 1.786969542503357,
      "learning_rate": 5e-05,
      "loss": 2.0734,
      "step": 105
    },
    {
      "epoch": 0.25389221556886227,
      "grad_norm": 1.5748064517974854,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.8093,
      "step": 106
    },
    {
      "epoch": 0.2562874251497006,
      "grad_norm": 1.5207183361053467,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.7466,
      "step": 107
    },
    {
      "epoch": 0.25868263473053893,
      "grad_norm": 1.506446361541748,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.7074,
      "step": 108
    },
    {
      "epoch": 0.26107784431137726,
      "grad_norm": 1.7950866222381592,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.8384,
      "step": 109
    },
    {
      "epoch": 0.2634730538922156,
      "grad_norm": 1.9788652658462524,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.0645,
      "step": 110
    },
    {
      "epoch": 0.26586826347305387,
      "grad_norm": 1.9470276832580566,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.898,
      "step": 111
    },
    {
      "epoch": 0.2682634730538922,
      "grad_norm": 1.9417413473129272,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.874,
      "step": 112
    },
    {
      "epoch": 0.27065868263473053,
      "grad_norm": 2.201859712600708,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.9008,
      "step": 113
    },
    {
      "epoch": 0.27305389221556886,
      "grad_norm": 2.4181673526763916,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.3725,
      "step": 114
    },
    {
      "epoch": 0.2754491017964072,
      "grad_norm": 1.8668200969696045,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.778,
      "step": 115
    },
    {
      "epoch": 0.2778443113772455,
      "grad_norm": 2.1837172508239746,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.1325,
      "step": 116
    },
    {
      "epoch": 0.28023952095808385,
      "grad_norm": 2.0208983421325684,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.128,
      "step": 117
    },
    {
      "epoch": 0.28263473053892213,
      "grad_norm": 2.2847790718078613,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.7249,
      "step": 118
    },
    {
      "epoch": 0.28502994011976046,
      "grad_norm": 2.183171272277832,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.9463,
      "step": 119
    },
    {
      "epoch": 0.2874251497005988,
      "grad_norm": 2.005955457687378,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.819,
      "step": 120
    },
    {
      "epoch": 0.2898203592814371,
      "grad_norm": 2.585301399230957,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.2503,
      "step": 121
    },
    {
      "epoch": 0.29221556886227545,
      "grad_norm": 3.229210615158081,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.5465,
      "step": 122
    },
    {
      "epoch": 0.2946107784431138,
      "grad_norm": 2.338641881942749,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.1291,
      "step": 123
    },
    {
      "epoch": 0.2970059880239521,
      "grad_norm": 2.5149643421173096,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.0291,
      "step": 124
    },
    {
      "epoch": 0.2994011976047904,
      "grad_norm": 2.461775302886963,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.9684,
      "step": 125
    },
    {
      "epoch": 0.3017964071856287,
      "grad_norm": 2.7158823013305664,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.1838,
      "step": 126
    },
    {
      "epoch": 0.30419161676646705,
      "grad_norm": 2.3903651237487793,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.8388,
      "step": 127
    },
    {
      "epoch": 0.3065868263473054,
      "grad_norm": 2.6435768604278564,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.6648,
      "step": 128
    },
    {
      "epoch": 0.3089820359281437,
      "grad_norm": 2.457359790802002,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.6639,
      "step": 129
    },
    {
      "epoch": 0.31137724550898205,
      "grad_norm": 3.02121901512146,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.8875,
      "step": 130
    },
    {
      "epoch": 0.3137724550898204,
      "grad_norm": 3.076138734817505,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.1737,
      "step": 131
    },
    {
      "epoch": 0.3161676646706587,
      "grad_norm": 4.147399425506592,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.2988,
      "step": 132
    },
    {
      "epoch": 0.318562874251497,
      "grad_norm": 3.4094369411468506,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.7292,
      "step": 133
    },
    {
      "epoch": 0.3209580838323353,
      "grad_norm": 3.789897918701172,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.5084,
      "step": 134
    },
    {
      "epoch": 0.32335329341317365,
      "grad_norm": 3.943809747695923,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.0878,
      "step": 135
    },
    {
      "epoch": 0.325748502994012,
      "grad_norm": 3.7343647480010986,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.9811,
      "step": 136
    },
    {
      "epoch": 0.3281437125748503,
      "grad_norm": 3.721630573272705,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.9745,
      "step": 137
    },
    {
      "epoch": 0.33053892215568864,
      "grad_norm": 4.511366844177246,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.099,
      "step": 138
    },
    {
      "epoch": 0.332934131736527,
      "grad_norm": 4.0060715675354,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.299,
      "step": 139
    },
    {
      "epoch": 0.33532934131736525,
      "grad_norm": 3.0129570960998535,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.7111,
      "step": 140
    },
    {
      "epoch": 0.3377245508982036,
      "grad_norm": 3.2002575397491455,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.8447,
      "step": 141
    },
    {
      "epoch": 0.3401197604790419,
      "grad_norm": 4.177738189697266,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.9311,
      "step": 142
    },
    {
      "epoch": 0.34251497005988024,
      "grad_norm": 4.329373836517334,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.9981,
      "step": 143
    },
    {
      "epoch": 0.3449101796407186,
      "grad_norm": 3.984931707382202,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.3383,
      "step": 144
    },
    {
      "epoch": 0.3473053892215569,
      "grad_norm": 3.767347812652588,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.1531,
      "step": 145
    },
    {
      "epoch": 0.34970059880239523,
      "grad_norm": 3.5382652282714844,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.1704,
      "step": 146
    },
    {
      "epoch": 0.3520958083832335,
      "grad_norm": 3.0679774284362793,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.9053,
      "step": 147
    },
    {
      "epoch": 0.35449101796407184,
      "grad_norm": 4.261399745941162,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.0227,
      "step": 148
    },
    {
      "epoch": 0.3568862275449102,
      "grad_norm": 4.266621112823486,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.8277,
      "step": 149
    },
    {
      "epoch": 0.3592814371257485,
      "grad_norm": 3.6732771396636963,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.6548,
      "step": 150
    },
    {
      "epoch": 0.3592814371257485,
      "eval_loss": 2.024937868118286,
      "eval_runtime": 62.3349,
      "eval_samples_per_second": 11.278,
      "eval_steps_per_second": 2.823,
      "step": 150
    },
    {
      "epoch": 0.36167664670658684,
      "grad_norm": 1.1668760776519775,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.9979,
      "step": 151
    },
    {
      "epoch": 0.36407185628742517,
      "grad_norm": 1.4932260513305664,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.9607,
      "step": 152
    },
    {
      "epoch": 0.3664670658682635,
      "grad_norm": 1.5939315557479858,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.9935,
      "step": 153
    },
    {
      "epoch": 0.3688622754491018,
      "grad_norm": 1.6748764514923096,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.7071,
      "step": 154
    },
    {
      "epoch": 0.3712574850299401,
      "grad_norm": 1.5659757852554321,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.966,
      "step": 155
    },
    {
      "epoch": 0.37365269461077844,
      "grad_norm": 1.7457515001296997,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.8878,
      "step": 156
    },
    {
      "epoch": 0.37604790419161677,
      "grad_norm": 2.206986427307129,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.9995,
      "step": 157
    },
    {
      "epoch": 0.3784431137724551,
      "grad_norm": 1.7352710962295532,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.9208,
      "step": 158
    },
    {
      "epoch": 0.38083832335329343,
      "grad_norm": 1.7520524263381958,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.8047,
      "step": 159
    },
    {
      "epoch": 0.38323353293413176,
      "grad_norm": 2.0070838928222656,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.9983,
      "step": 160
    },
    {
      "epoch": 0.38562874251497004,
      "grad_norm": 1.9782376289367676,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.7591,
      "step": 161
    },
    {
      "epoch": 0.38802395209580837,
      "grad_norm": 1.8567336797714233,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.9367,
      "step": 162
    },
    {
      "epoch": 0.3904191616766467,
      "grad_norm": 1.876726746559143,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.9405,
      "step": 163
    },
    {
      "epoch": 0.39281437125748503,
      "grad_norm": 1.8759047985076904,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.6671,
      "step": 164
    },
    {
      "epoch": 0.39520958083832336,
      "grad_norm": 2.3147873878479004,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.0846,
      "step": 165
    },
    {
      "epoch": 0.3976047904191617,
      "grad_norm": 2.3103091716766357,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.1204,
      "step": 166
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.453749656677246,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.0115,
      "step": 167
    },
    {
      "epoch": 0.4023952095808383,
      "grad_norm": 2.0524468421936035,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.8474,
      "step": 168
    },
    {
      "epoch": 0.40479041916167663,
      "grad_norm": 2.697420597076416,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.4182,
      "step": 169
    },
    {
      "epoch": 0.40718562874251496,
      "grad_norm": 2.5188112258911133,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.1481,
      "step": 170
    },
    {
      "epoch": 0.4095808383233533,
      "grad_norm": 2.564915180206299,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.7769,
      "step": 171
    },
    {
      "epoch": 0.4119760479041916,
      "grad_norm": 2.445690155029297,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.864,
      "step": 172
    },
    {
      "epoch": 0.41437125748502995,
      "grad_norm": 2.7091946601867676,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.488,
      "step": 173
    },
    {
      "epoch": 0.4167664670658683,
      "grad_norm": 3.008167266845703,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.3174,
      "step": 174
    },
    {
      "epoch": 0.41916167664670656,
      "grad_norm": 3.191629409790039,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.1238,
      "step": 175
    },
    {
      "epoch": 0.4215568862275449,
      "grad_norm": 2.5828981399536133,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.0492,
      "step": 176
    },
    {
      "epoch": 0.4239520958083832,
      "grad_norm": 2.6535017490386963,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.0304,
      "step": 177
    },
    {
      "epoch": 0.42634730538922155,
      "grad_norm": 2.6480066776275635,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.0657,
      "step": 178
    },
    {
      "epoch": 0.4287425149700599,
      "grad_norm": 2.4269888401031494,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.8115,
      "step": 179
    },
    {
      "epoch": 0.4311377245508982,
      "grad_norm": 2.877483367919922,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.1785,
      "step": 180
    },
    {
      "epoch": 0.43353293413173655,
      "grad_norm": 3.805612325668335,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.156,
      "step": 181
    },
    {
      "epoch": 0.4359281437125748,
      "grad_norm": 3.093027353286743,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.1996,
      "step": 182
    },
    {
      "epoch": 0.43832335329341315,
      "grad_norm": 3.468073606491089,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.71,
      "step": 183
    },
    {
      "epoch": 0.4407185628742515,
      "grad_norm": 2.9488377571105957,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.2707,
      "step": 184
    },
    {
      "epoch": 0.4431137724550898,
      "grad_norm": 3.6791625022888184,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.2156,
      "step": 185
    },
    {
      "epoch": 0.44550898203592815,
      "grad_norm": 7.463198184967041,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.2354,
      "step": 186
    },
    {
      "epoch": 0.4479041916167665,
      "grad_norm": 2.7715961933135986,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.8003,
      "step": 187
    },
    {
      "epoch": 0.4502994011976048,
      "grad_norm": 3.995654821395874,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.1323,
      "step": 188
    },
    {
      "epoch": 0.45269461077844314,
      "grad_norm": 3.7480030059814453,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.8828,
      "step": 189
    },
    {
      "epoch": 0.4550898203592814,
      "grad_norm": 2.666130542755127,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.1815,
      "step": 190
    },
    {
      "epoch": 0.45748502994011975,
      "grad_norm": 3.5462491512298584,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.818,
      "step": 191
    },
    {
      "epoch": 0.4598802395209581,
      "grad_norm": 4.025123596191406,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.4431,
      "step": 192
    },
    {
      "epoch": 0.4622754491017964,
      "grad_norm": 2.741682767868042,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.4509,
      "step": 193
    },
    {
      "epoch": 0.46467065868263474,
      "grad_norm": 3.475533962249756,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.8404,
      "step": 194
    },
    {
      "epoch": 0.46706586826347307,
      "grad_norm": 3.529165029525757,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.8149,
      "step": 195
    },
    {
      "epoch": 0.4694610778443114,
      "grad_norm": 3.902186632156372,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.1688,
      "step": 196
    },
    {
      "epoch": 0.4718562874251497,
      "grad_norm": 3.5201449394226074,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.3164,
      "step": 197
    },
    {
      "epoch": 0.474251497005988,
      "grad_norm": 3.4739627838134766,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.7835,
      "step": 198
    },
    {
      "epoch": 0.47664670658682634,
      "grad_norm": 4.725226402282715,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.1762,
      "step": 199
    },
    {
      "epoch": 0.47904191616766467,
      "grad_norm": 5.299903869628906,
      "learning_rate": 0.0,
      "loss": 1.785,
      "step": 200
    },
    {
      "epoch": 0.47904191616766467,
      "eval_loss": 1.960377812385559,
      "eval_runtime": 62.2919,
      "eval_samples_per_second": 11.286,
      "eval_steps_per_second": 2.825,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.017028768104448e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}