{
  "best_metric": 2.245424270629883,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.012011470954761797,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 6.005735477380899e-05,
      "grad_norm": 0.770524263381958,
      "learning_rate": 1e-05,
      "loss": 1.3234,
      "step": 1
    },
    {
      "epoch": 6.005735477380899e-05,
      "eval_loss": 3.1906468868255615,
      "eval_runtime": 2137.5414,
      "eval_samples_per_second": 13.12,
      "eval_steps_per_second": 3.28,
      "step": 1
    },
    {
      "epoch": 0.00012011470954761798,
      "grad_norm": 0.8618220686912537,
      "learning_rate": 2e-05,
      "loss": 1.4146,
      "step": 2
    },
    {
      "epoch": 0.00018017206432142696,
      "grad_norm": 0.8978565335273743,
      "learning_rate": 3e-05,
      "loss": 1.5296,
      "step": 3
    },
    {
      "epoch": 0.00024022941909523595,
      "grad_norm": 0.9844889640808105,
      "learning_rate": 4e-05,
      "loss": 1.6595,
      "step": 4
    },
    {
      "epoch": 0.00030028677386904494,
      "grad_norm": 0.8339329957962036,
      "learning_rate": 5e-05,
      "loss": 1.6757,
      "step": 5
    },
    {
      "epoch": 0.00036034412864285393,
      "grad_norm": 0.7451542615890503,
      "learning_rate": 6e-05,
      "loss": 1.6839,
      "step": 6
    },
    {
      "epoch": 0.0004204014834166629,
      "grad_norm": 0.9352735877037048,
      "learning_rate": 7e-05,
      "loss": 1.6442,
      "step": 7
    },
    {
      "epoch": 0.0004804588381904719,
      "grad_norm": 0.9434142708778381,
      "learning_rate": 8e-05,
      "loss": 1.5872,
      "step": 8
    },
    {
      "epoch": 0.0005405161929642808,
      "grad_norm": 0.9863334894180298,
      "learning_rate": 9e-05,
      "loss": 1.7174,
      "step": 9
    },
    {
      "epoch": 0.0006005735477380899,
      "grad_norm": 0.6982995867729187,
      "learning_rate": 0.0001,
      "loss": 1.442,
      "step": 10
    },
    {
      "epoch": 0.0006606309025118988,
      "grad_norm": 0.7425613403320312,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.5336,
      "step": 11
    },
    {
      "epoch": 0.0007206882572857079,
      "grad_norm": 0.862773060798645,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.47,
      "step": 12
    },
    {
      "epoch": 0.0007807456120595168,
      "grad_norm": 0.8265248537063599,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.6426,
      "step": 13
    },
    {
      "epoch": 0.0008408029668333258,
      "grad_norm": 0.7517850995063782,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.7018,
      "step": 14
    },
    {
      "epoch": 0.0009008603216071348,
      "grad_norm": 0.7315523028373718,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.4801,
      "step": 15
    },
    {
      "epoch": 0.0009609176763809438,
      "grad_norm": 0.7213664054870605,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.6447,
      "step": 16
    },
    {
      "epoch": 0.0010209750311547527,
      "grad_norm": 0.8049549460411072,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.3661,
      "step": 17
    },
    {
      "epoch": 0.0010810323859285617,
      "grad_norm": 0.8196991682052612,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.4959,
      "step": 18
    },
    {
      "epoch": 0.0011410897407023708,
      "grad_norm": 0.84516841173172,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.6066,
      "step": 19
    },
    {
      "epoch": 0.0012011470954761798,
      "grad_norm": 0.8220654726028442,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.5342,
      "step": 20
    },
    {
      "epoch": 0.0012612044502499887,
      "grad_norm": 0.9309256076812744,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.6366,
      "step": 21
    },
    {
      "epoch": 0.0013212618050237976,
      "grad_norm": 1.6053969860076904,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.5904,
      "step": 22
    },
    {
      "epoch": 0.0013813191597976068,
      "grad_norm": 0.9796047210693359,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.5951,
      "step": 23
    },
    {
      "epoch": 0.0014413765145714157,
      "grad_norm": 1.05988347530365,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.7388,
      "step": 24
    },
    {
      "epoch": 0.0015014338693452246,
      "grad_norm": 1.0321321487426758,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.9195,
      "step": 25
    },
    {
      "epoch": 0.0015614912241190336,
      "grad_norm": 1.396426796913147,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.8129,
      "step": 26
    },
    {
      "epoch": 0.0016215485788928427,
      "grad_norm": 1.1280184984207153,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.8458,
      "step": 27
    },
    {
      "epoch": 0.0016816059336666517,
      "grad_norm": 1.1668058633804321,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.7866,
      "step": 28
    },
    {
      "epoch": 0.0017416632884404606,
      "grad_norm": 1.2741973400115967,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.8008,
      "step": 29
    },
    {
      "epoch": 0.0018017206432142695,
      "grad_norm": 1.2448976039886475,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.8068,
      "step": 30
    },
    {
      "epoch": 0.0018617779979880787,
      "grad_norm": 1.45051109790802,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.0329,
      "step": 31
    },
    {
      "epoch": 0.0019218353527618876,
      "grad_norm": 2.0939557552337646,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.7654,
      "step": 32
    },
    {
      "epoch": 0.0019818927075356966,
      "grad_norm": 1.4269686937332153,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.0806,
      "step": 33
    },
    {
      "epoch": 0.0020419500623095055,
      "grad_norm": 2.7957756519317627,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.2719,
      "step": 34
    },
    {
      "epoch": 0.0021020074170833144,
      "grad_norm": 2.373943328857422,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.3571,
      "step": 35
    },
    {
      "epoch": 0.0021620647718571234,
      "grad_norm": 4.211935520172119,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.6789,
      "step": 36
    },
    {
      "epoch": 0.0022221221266309327,
      "grad_norm": 3.9513766765594482,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.7947,
      "step": 37
    },
    {
      "epoch": 0.0022821794814047417,
      "grad_norm": 2.819199562072754,
      "learning_rate": 9.473646649103818e-05,
      "loss": 3.1013,
      "step": 38
    },
    {
      "epoch": 0.0023422368361785506,
      "grad_norm": 3.6481869220733643,
      "learning_rate": 9.43611409721806e-05,
      "loss": 3.0809,
      "step": 39
    },
    {
      "epoch": 0.0024022941909523595,
      "grad_norm": 3.8346762657165527,
      "learning_rate": 9.397368756032445e-05,
      "loss": 3.4865,
      "step": 40
    },
    {
      "epoch": 0.0024623515457261685,
      "grad_norm": 4.402808666229248,
      "learning_rate": 9.357421218136386e-05,
      "loss": 3.7069,
      "step": 41
    },
    {
      "epoch": 0.0025224089004999774,
      "grad_norm": 4.706476211547852,
      "learning_rate": 9.316282404787871e-05,
      "loss": 3.5655,
      "step": 42
    },
    {
      "epoch": 0.0025824662552737863,
      "grad_norm": 8.770739555358887,
      "learning_rate": 9.273963562927695e-05,
      "loss": 4.0911,
      "step": 43
    },
    {
      "epoch": 0.0026425236100475953,
      "grad_norm": 13.75476360321045,
      "learning_rate": 9.230476262104677e-05,
      "loss": 3.6738,
      "step": 44
    },
    {
      "epoch": 0.0027025809648214046,
      "grad_norm": 10.422087669372559,
      "learning_rate": 9.185832391312644e-05,
      "loss": 4.1666,
      "step": 45
    },
    {
      "epoch": 0.0027626383195952136,
      "grad_norm": 5.070408821105957,
      "learning_rate": 9.140044155740101e-05,
      "loss": 3.6621,
      "step": 46
    },
    {
      "epoch": 0.0028226956743690225,
      "grad_norm": 16.344045639038086,
      "learning_rate": 9.093124073433463e-05,
      "loss": 4.2733,
      "step": 47
    },
    {
      "epoch": 0.0028827530291428314,
      "grad_norm": 8.721118927001953,
      "learning_rate": 9.045084971874738e-05,
      "loss": 4.7064,
      "step": 48
    },
    {
      "epoch": 0.0029428103839166404,
      "grad_norm": 12.818589210510254,
      "learning_rate": 8.995939984474624e-05,
      "loss": 4.5642,
      "step": 49
    },
    {
      "epoch": 0.0030028677386904493,
      "grad_norm": 11.10518741607666,
      "learning_rate": 8.945702546981969e-05,
      "loss": 4.632,
      "step": 50
    },
    {
      "epoch": 0.0030028677386904493,
      "eval_loss": 2.88491153717041,
      "eval_runtime": 2144.0816,
      "eval_samples_per_second": 13.08,
      "eval_steps_per_second": 3.27,
      "step": 50
    },
    {
      "epoch": 0.0030629250934642582,
      "grad_norm": 4.857649803161621,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.881,
      "step": 51
    },
    {
      "epoch": 0.003122982448238067,
      "grad_norm": 2.592421293258667,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.7319,
      "step": 52
    },
    {
      "epoch": 0.0031830398030118765,
      "grad_norm": 1.252772331237793,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.5248,
      "step": 53
    },
    {
      "epoch": 0.0032430971577856855,
      "grad_norm": 1.4710232019424438,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.4511,
      "step": 54
    },
    {
      "epoch": 0.0033031545125594944,
      "grad_norm": 0.6959516406059265,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.3836,
      "step": 55
    },
    {
      "epoch": 0.0033632118673333033,
      "grad_norm": 0.67060387134552,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.4728,
      "step": 56
    },
    {
      "epoch": 0.0034232692221071123,
      "grad_norm": 0.7281978130340576,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.506,
      "step": 57
    },
    {
      "epoch": 0.003483326576880921,
      "grad_norm": 0.5831403732299805,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.4249,
      "step": 58
    },
    {
      "epoch": 0.00354338393165473,
      "grad_norm": 0.6770080924034119,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.574,
      "step": 59
    },
    {
      "epoch": 0.003603441286428539,
      "grad_norm": 0.6479463577270508,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.4924,
      "step": 60
    },
    {
      "epoch": 0.0036634986412023484,
      "grad_norm": 0.7681052684783936,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.6914,
      "step": 61
    },
    {
      "epoch": 0.0037235559959761574,
      "grad_norm": 0.6489607095718384,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.5407,
      "step": 62
    },
    {
      "epoch": 0.0037836133507499663,
      "grad_norm": 0.6738668084144592,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.4574,
      "step": 63
    },
    {
      "epoch": 0.0038436707055237752,
      "grad_norm": 0.6503859758377075,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.4774,
      "step": 64
    },
    {
      "epoch": 0.003903728060297584,
      "grad_norm": 0.7435093522071838,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.7024,
      "step": 65
    },
    {
      "epoch": 0.003963785415071393,
      "grad_norm": 0.6974945068359375,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.5928,
      "step": 66
    },
    {
      "epoch": 0.004023842769845202,
      "grad_norm": 1.033766746520996,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.5197,
      "step": 67
    },
    {
      "epoch": 0.004083900124619011,
      "grad_norm": 0.8381729125976562,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.6606,
      "step": 68
    },
    {
      "epoch": 0.00414395747939282,
      "grad_norm": 0.8566480278968811,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.6861,
      "step": 69
    },
    {
      "epoch": 0.004204014834166629,
      "grad_norm": 0.799872875213623,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.5304,
      "step": 70
    },
    {
      "epoch": 0.004264072188940438,
      "grad_norm": 0.938618540763855,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.802,
      "step": 71
    },
    {
      "epoch": 0.004324129543714247,
      "grad_norm": 0.7965691685676575,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.6406,
      "step": 72
    },
    {
      "epoch": 0.0043841868984880565,
      "grad_norm": 0.851714015007019,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.8181,
      "step": 73
    },
    {
      "epoch": 0.0044442442532618654,
      "grad_norm": 0.8853086829185486,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.7945,
      "step": 74
    },
    {
      "epoch": 0.004504301608035674,
      "grad_norm": 1.0283277034759521,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.6908,
      "step": 75
    },
    {
      "epoch": 0.004564358962809483,
      "grad_norm": 0.885177731513977,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.6584,
      "step": 76
    },
    {
      "epoch": 0.004624416317583292,
      "grad_norm": 1.063603162765503,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.7969,
      "step": 77
    },
    {
      "epoch": 0.004684473672357101,
      "grad_norm": 1.1493542194366455,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.7811,
      "step": 78
    },
    {
      "epoch": 0.00474453102713091,
      "grad_norm": 1.0672365427017212,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.8951,
      "step": 79
    },
    {
      "epoch": 0.004804588381904719,
      "grad_norm": 1.2150990962982178,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.8018,
      "step": 80
    },
    {
      "epoch": 0.004864645736678528,
      "grad_norm": 1.3201502561569214,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.9971,
      "step": 81
    },
    {
      "epoch": 0.004924703091452337,
      "grad_norm": 1.1746817827224731,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.912,
      "step": 82
    },
    {
      "epoch": 0.004984760446226146,
      "grad_norm": 1.5162415504455566,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.3503,
      "step": 83
    },
    {
      "epoch": 0.005044817800999955,
      "grad_norm": 2.4259698390960693,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.2487,
      "step": 84
    },
    {
      "epoch": 0.005104875155773764,
      "grad_norm": 2.470768451690674,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.4274,
      "step": 85
    },
    {
      "epoch": 0.005164932510547573,
      "grad_norm": 2.2791121006011963,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.5619,
      "step": 86
    },
    {
      "epoch": 0.005224989865321382,
      "grad_norm": 2.2025554180145264,
      "learning_rate": 6.466250186922325e-05,
      "loss": 3.1082,
      "step": 87
    },
    {
      "epoch": 0.0052850472200951905,
      "grad_norm": 2.408917188644409,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.9163,
      "step": 88
    },
    {
      "epoch": 0.005345104574869,
      "grad_norm": 3.0242130756378174,
      "learning_rate": 6.307399704769099e-05,
      "loss": 3.2261,
      "step": 89
    },
    {
      "epoch": 0.005405161929642809,
      "grad_norm": 3.487112045288086,
      "learning_rate": 6.227427435703997e-05,
      "loss": 3.6184,
      "step": 90
    },
    {
      "epoch": 0.005465219284416618,
      "grad_norm": 5.258666515350342,
      "learning_rate": 6.147119600233758e-05,
      "loss": 3.8135,
      "step": 91
    },
    {
      "epoch": 0.005525276639190427,
      "grad_norm": 4.33889627456665,
      "learning_rate": 6.066498153718735e-05,
      "loss": 3.8951,
      "step": 92
    },
    {
      "epoch": 0.005585333993964236,
      "grad_norm": 4.518934726715088,
      "learning_rate": 5.985585137257401e-05,
      "loss": 3.699,
      "step": 93
    },
    {
      "epoch": 0.005645391348738045,
      "grad_norm": 5.626092910766602,
      "learning_rate": 5.90440267166055e-05,
      "loss": 3.9828,
      "step": 94
    },
    {
      "epoch": 0.005705448703511854,
      "grad_norm": 7.326063632965088,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 3.9095,
      "step": 95
    },
    {
      "epoch": 0.005765506058285663,
      "grad_norm": 5.146079063415527,
      "learning_rate": 5.74131823855921e-05,
      "loss": 3.9802,
      "step": 96
    },
    {
      "epoch": 0.005825563413059472,
      "grad_norm": 5.464537143707275,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 4.2323,
      "step": 97
    },
    {
      "epoch": 0.005885620767833281,
      "grad_norm": 10.174283027648926,
      "learning_rate": 5.577423184847932e-05,
      "loss": 4.355,
      "step": 98
    },
    {
      "epoch": 0.00594567812260709,
      "grad_norm": 6.405921459197998,
      "learning_rate": 5.495227651252315e-05,
      "loss": 4.3089,
      "step": 99
    },
    {
      "epoch": 0.006005735477380899,
      "grad_norm": 20.77309226989746,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 5.1427,
      "step": 100
    },
    {
      "epoch": 0.006005735477380899,
      "eval_loss": 2.8183083534240723,
      "eval_runtime": 2144.036,
      "eval_samples_per_second": 13.08,
      "eval_steps_per_second": 3.27,
      "step": 100
    },
    {
      "epoch": 0.0060657928321547075,
      "grad_norm": 4.833406925201416,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.9641,
      "step": 101
    },
    {
      "epoch": 0.0061258501869285165,
      "grad_norm": 3.734908103942871,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.8651,
      "step": 102
    },
    {
      "epoch": 0.006185907541702325,
      "grad_norm": 2.424708604812622,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.7192,
      "step": 103
    },
    {
      "epoch": 0.006245964896476134,
      "grad_norm": 1.2214038372039795,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.5439,
      "step": 104
    },
    {
      "epoch": 0.006306022251249943,
      "grad_norm": 1.0090330839157104,
      "learning_rate": 5e-05,
      "loss": 1.7109,
      "step": 105
    },
    {
      "epoch": 0.006366079606023753,
      "grad_norm": 0.6321302652359009,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.5019,
      "step": 106
    },
    {
      "epoch": 0.006426136960797562,
      "grad_norm": 0.6775772571563721,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.5734,
      "step": 107
    },
    {
      "epoch": 0.006486194315571371,
      "grad_norm": 0.6290256381034851,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.4641,
      "step": 108
    },
    {
      "epoch": 0.00654625167034518,
      "grad_norm": 0.6630206108093262,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.4423,
      "step": 109
    },
    {
      "epoch": 0.006606309025118989,
      "grad_norm": 0.5929522514343262,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.5018,
      "step": 110
    },
    {
      "epoch": 0.006666366379892798,
      "grad_norm": 0.7440500259399414,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.719,
      "step": 111
    },
    {
      "epoch": 0.006726423734666607,
      "grad_norm": 0.7187187075614929,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.4171,
      "step": 112
    },
    {
      "epoch": 0.006786481089440416,
      "grad_norm": 0.6664174199104309,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.4871,
      "step": 113
    },
    {
      "epoch": 0.0068465384442142245,
      "grad_norm": 0.8212423920631409,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.6519,
      "step": 114
    },
    {
      "epoch": 0.0069065957989880335,
      "grad_norm": 0.7325797080993652,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.5503,
      "step": 115
    },
    {
      "epoch": 0.006966653153761842,
      "grad_norm": 0.7924835681915283,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.5493,
      "step": 116
    },
    {
      "epoch": 0.007026710508535651,
      "grad_norm": 0.7400676608085632,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.4353,
      "step": 117
    },
    {
      "epoch": 0.00708676786330946,
      "grad_norm": 0.7383843660354614,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.607,
      "step": 118
    },
    {
      "epoch": 0.007146825218083269,
      "grad_norm": 0.8036738634109497,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.7235,
      "step": 119
    },
    {
      "epoch": 0.007206882572857078,
      "grad_norm": 0.9643393754959106,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.6605,
      "step": 120
    },
    {
      "epoch": 0.007266939927630887,
      "grad_norm": 0.7977961897850037,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.5874,
      "step": 121
    },
    {
      "epoch": 0.007326997282404697,
      "grad_norm": 1.0039657354354858,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.7547,
      "step": 122
    },
    {
      "epoch": 0.007387054637178506,
      "grad_norm": 0.8533593416213989,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.5897,
      "step": 123
    },
    {
      "epoch": 0.007447111991952315,
      "grad_norm": 0.9100041389465332,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.8671,
      "step": 124
    },
    {
      "epoch": 0.007507169346726124,
      "grad_norm": 0.9276086688041687,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.823,
      "step": 125
    },
    {
      "epoch": 0.007567226701499933,
      "grad_norm": 0.9275705218315125,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.7562,
      "step": 126
    },
    {
      "epoch": 0.0076272840562737415,
      "grad_norm": 1.480225682258606,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.6622,
      "step": 127
    },
    {
      "epoch": 0.0076873414110475505,
      "grad_norm": 0.9616846442222595,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.8093,
      "step": 128
    },
    {
      "epoch": 0.007747398765821359,
      "grad_norm": 1.14011812210083,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.8223,
      "step": 129
    },
    {
      "epoch": 0.007807456120595168,
      "grad_norm": 1.2321470975875854,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.8567,
      "step": 130
    },
    {
      "epoch": 0.007867513475368978,
      "grad_norm": 1.0687763690948486,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.8913,
      "step": 131
    },
    {
      "epoch": 0.007927570830142786,
      "grad_norm": 1.063493013381958,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.0323,
      "step": 132
    },
    {
      "epoch": 0.007987628184916596,
      "grad_norm": 1.641831636428833,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.4082,
      "step": 133
    },
    {
      "epoch": 0.008047685539690404,
      "grad_norm": 1.429152488708496,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.2811,
      "step": 134
    },
    {
      "epoch": 0.008107742894464214,
      "grad_norm": 1.641626238822937,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.6107,
      "step": 135
    },
    {
      "epoch": 0.008167800249238022,
      "grad_norm": 2.325291395187378,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.9367,
      "step": 136
    },
    {
      "epoch": 0.008227857604011832,
      "grad_norm": 2.3526947498321533,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.8885,
      "step": 137
    },
    {
      "epoch": 0.00828791495878564,
      "grad_norm": 2.213717222213745,
      "learning_rate": 2.405152131093926e-05,
      "loss": 3.1196,
      "step": 138
    },
    {
      "epoch": 0.00834797231355945,
      "grad_norm": 2.8324809074401855,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 3.4588,
      "step": 139
    },
    {
      "epoch": 0.008408029668333258,
      "grad_norm": 3.3284666538238525,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 3.7411,
      "step": 140
    },
    {
      "epoch": 0.008468087023107067,
      "grad_norm": 4.056039333343506,
      "learning_rate": 2.196424713241637e-05,
      "loss": 3.9056,
      "step": 141
    },
    {
      "epoch": 0.008528144377880876,
      "grad_norm": 3.690812110900879,
      "learning_rate": 2.128356686585282e-05,
      "loss": 3.9602,
      "step": 142
    },
    {
      "epoch": 0.008588201732654685,
      "grad_norm": 4.101503372192383,
      "learning_rate": 2.061073738537635e-05,
      "loss": 4.0565,
      "step": 143
    },
    {
      "epoch": 0.008648259087428493,
      "grad_norm": 4.115358352661133,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 3.9999,
      "step": 144
    },
    {
      "epoch": 0.008708316442202303,
      "grad_norm": 4.536535263061523,
      "learning_rate": 1.928936436551661e-05,
      "loss": 3.9276,
      "step": 145
    },
    {
      "epoch": 0.008768373796976113,
      "grad_norm": 5.131693363189697,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 3.9758,
      "step": 146
    },
    {
      "epoch": 0.008828431151749921,
      "grad_norm": 6.091833114624023,
      "learning_rate": 1.800157297483417e-05,
      "loss": 4.0936,
      "step": 147
    },
    {
      "epoch": 0.008888488506523731,
      "grad_norm": 11.448600769042969,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 4.6619,
      "step": 148
    },
    {
      "epoch": 0.008948545861297539,
      "grad_norm": 8.706692695617676,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 3.9178,
      "step": 149
    },
    {
      "epoch": 0.009008603216071349,
      "grad_norm": 17.97637176513672,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 4.8413,
      "step": 150
    },
    {
      "epoch": 0.009008603216071349,
      "eval_loss": 2.2935338020324707,
      "eval_runtime": 2144.7077,
      "eval_samples_per_second": 13.076,
      "eval_steps_per_second": 3.269,
      "step": 150
    },
    {
      "epoch": 0.009068660570845157,
      "grad_norm": 0.4741351008415222,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.2606,
      "step": 151
    },
    {
      "epoch": 0.009128717925618967,
      "grad_norm": 0.5311182737350464,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.2989,
      "step": 152
    },
    {
      "epoch": 0.009188775280392775,
      "grad_norm": 0.5777656435966492,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.3516,
      "step": 153
    },
    {
      "epoch": 0.009248832635166584,
      "grad_norm": 0.5261973142623901,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.4222,
      "step": 154
    },
    {
      "epoch": 0.009308889989940393,
      "grad_norm": 0.565487265586853,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.4194,
      "step": 155
    },
    {
      "epoch": 0.009368947344714202,
      "grad_norm": 0.594511866569519,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.5465,
      "step": 156
    },
    {
      "epoch": 0.00942900469948801,
      "grad_norm": 0.6569897532463074,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.3701,
      "step": 157
    },
    {
      "epoch": 0.00948906205426182,
      "grad_norm": 0.6133348941802979,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.6571,
      "step": 158
    },
    {
      "epoch": 0.009549119409035628,
      "grad_norm": 0.658849835395813,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.563,
      "step": 159
    },
    {
      "epoch": 0.009609176763809438,
      "grad_norm": 0.7560611367225647,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.3805,
      "step": 160
    },
    {
      "epoch": 0.009669234118583246,
      "grad_norm": 0.6569478511810303,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.5825,
      "step": 161
    },
    {
      "epoch": 0.009729291473357056,
      "grad_norm": 0.6861472725868225,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.4875,
      "step": 162
    },
    {
      "epoch": 0.009789348828130866,
      "grad_norm": 0.77292799949646,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.5709,
      "step": 163
    },
    {
      "epoch": 0.009849406182904674,
      "grad_norm": 0.6188532710075378,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.5094,
      "step": 164
    },
    {
      "epoch": 0.009909463537678484,
      "grad_norm": 0.7523466348648071,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.6228,
      "step": 165
    },
    {
      "epoch": 0.009969520892452292,
      "grad_norm": 0.7730385661125183,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.6141,
      "step": 166
    },
    {
      "epoch": 0.010029578247226102,
      "grad_norm": 0.6758378744125366,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.4716,
      "step": 167
    },
    {
      "epoch": 0.01008963560199991,
      "grad_norm": 0.7794915437698364,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.5349,
      "step": 168
    },
    {
      "epoch": 0.01014969295677372,
      "grad_norm": 0.8888735175132751,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.5984,
      "step": 169
    },
    {
      "epoch": 0.010209750311547527,
      "grad_norm": 0.8209103345870972,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.5371,
      "step": 170
    },
    {
      "epoch": 0.010269807666321337,
      "grad_norm": 0.8117617964744568,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.6463,
      "step": 171
    },
    {
      "epoch": 0.010329865021095145,
      "grad_norm": 0.9295036792755127,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.7469,
      "step": 172
    },
    {
      "epoch": 0.010389922375868955,
      "grad_norm": 0.7867856621742249,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.6214,
      "step": 173
    },
    {
      "epoch": 0.010449979730642763,
      "grad_norm": 0.7696926593780518,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.6501,
      "step": 174
    },
    {
      "epoch": 0.010510037085416573,
      "grad_norm": 1.0101313591003418,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.6691,
      "step": 175
    },
    {
      "epoch": 0.010570094440190381,
      "grad_norm": 0.9581689238548279,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.7672,
      "step": 176
    },
    {
      "epoch": 0.01063015179496419,
      "grad_norm": 0.9825965166091919,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.7463,
      "step": 177
    },
    {
      "epoch": 0.010690209149738,
      "grad_norm": 1.0118306875228882,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.765,
      "step": 178
    },
    {
      "epoch": 0.010750266504511809,
      "grad_norm": 1.1941312551498413,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.005,
      "step": 179
    },
    {
      "epoch": 0.010810323859285619,
      "grad_norm": 1.0240905284881592,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.8371,
      "step": 180
    },
    {
      "epoch": 0.010870381214059427,
      "grad_norm": 1.2060580253601074,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.9201,
      "step": 181
    },
    {
      "epoch": 0.010930438568833236,
      "grad_norm": 1.159066915512085,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.0884,
      "step": 182
    },
    {
      "epoch": 0.010990495923607044,
      "grad_norm": 1.2375589609146118,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.0756,
      "step": 183
    },
    {
      "epoch": 0.011050553278380854,
      "grad_norm": 1.4156990051269531,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.9746,
      "step": 184
    },
    {
      "epoch": 0.011110610633154662,
      "grad_norm": 1.5409784317016602,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.577,
      "step": 185
    },
    {
      "epoch": 0.011170667987928472,
      "grad_norm": 1.7621561288833618,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.7555,
      "step": 186
    },
    {
      "epoch": 0.01123072534270228,
      "grad_norm": 1.8674893379211426,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.8826,
      "step": 187
    },
    {
      "epoch": 0.01129078269747609,
      "grad_norm": 2.479443073272705,
      "learning_rate": 9.810017062595322e-07,
      "loss": 3.0932,
      "step": 188
    },
    {
      "epoch": 0.011350840052249898,
      "grad_norm": 2.1900508403778076,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.9696,
      "step": 189
    },
    {
      "epoch": 0.011410897407023708,
      "grad_norm": 2.491612434387207,
      "learning_rate": 6.819348298638839e-07,
      "loss": 3.5814,
      "step": 190
    },
    {
      "epoch": 0.011470954761797516,
      "grad_norm": 3.3416852951049805,
      "learning_rate": 5.526064699265753e-07,
      "loss": 3.6031,
      "step": 191
    },
    {
      "epoch": 0.011531012116571326,
      "grad_norm": 3.326714515686035,
      "learning_rate": 4.367965336512403e-07,
      "loss": 3.7905,
      "step": 192
    },
    {
      "epoch": 0.011591069471345134,
      "grad_norm": 4.072110176086426,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 4.0016,
      "step": 193
    },
    {
      "epoch": 0.011651126826118944,
      "grad_norm": 3.6170566082000732,
      "learning_rate": 2.458548727494292e-07,
      "loss": 3.7446,
      "step": 194
    },
    {
      "epoch": 0.011711184180892753,
      "grad_norm": 4.770792007446289,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 3.8111,
      "step": 195
    },
    {
      "epoch": 0.011771241535666561,
      "grad_norm": 4.090066432952881,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 4.0008,
      "step": 196
    },
    {
      "epoch": 0.011831298890440371,
      "grad_norm": 5.364345550537109,
      "learning_rate": 6.150154258476315e-08,
      "loss": 3.9344,
      "step": 197
    },
    {
      "epoch": 0.01189135624521418,
      "grad_norm": 5.561614990234375,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 4.2718,
      "step": 198
    },
    {
      "epoch": 0.011951413599987989,
      "grad_norm": 6.925398826599121,
      "learning_rate": 6.834750376549792e-09,
      "loss": 4.6472,
      "step": 199
    },
    {
      "epoch": 0.012011470954761797,
      "grad_norm": 11.039876937866211,
      "learning_rate": 0.0,
      "loss": 4.8495,
      "step": 200
    },
    {
      "epoch": 0.012011470954761797,
      "eval_loss": 2.245424270629883,
      "eval_runtime": 2145.3013,
      "eval_samples_per_second": 13.072,
      "eval_steps_per_second": 3.268,
      "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": 3.0321122932791706e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}