{
  "best_metric": 1.5972942113876343,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.12101043715020421,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000605052185751021,
      "grad_norm": 8.824080467224121,
      "learning_rate": 1e-05,
      "loss": 3.857,
      "step": 1
    },
    {
      "epoch": 0.000605052185751021,
      "eval_loss": 3.967219829559326,
      "eval_runtime": 131.6153,
      "eval_samples_per_second": 21.153,
      "eval_steps_per_second": 5.288,
      "step": 1
    },
    {
      "epoch": 0.001210104371502042,
      "grad_norm": 9.042195320129395,
      "learning_rate": 2e-05,
      "loss": 3.4688,
      "step": 2
    },
    {
      "epoch": 0.001815156557253063,
      "grad_norm": 9.631072998046875,
      "learning_rate": 3e-05,
      "loss": 3.6481,
      "step": 3
    },
    {
      "epoch": 0.002420208743004084,
      "grad_norm": 10.320423126220703,
      "learning_rate": 4e-05,
      "loss": 3.9826,
      "step": 4
    },
    {
      "epoch": 0.003025260928755105,
      "grad_norm": 8.81252670288086,
      "learning_rate": 5e-05,
      "loss": 4.0196,
      "step": 5
    },
    {
      "epoch": 0.003630313114506126,
      "grad_norm": 7.749195575714111,
      "learning_rate": 6e-05,
      "loss": 3.2208,
      "step": 6
    },
    {
      "epoch": 0.004235365300257147,
      "grad_norm": 8.090449333190918,
      "learning_rate": 7e-05,
      "loss": 3.27,
      "step": 7
    },
    {
      "epoch": 0.004840417486008168,
      "grad_norm": 7.524991035461426,
      "learning_rate": 8e-05,
      "loss": 2.7439,
      "step": 8
    },
    {
      "epoch": 0.00544546967175919,
      "grad_norm": 7.981743812561035,
      "learning_rate": 9e-05,
      "loss": 2.9816,
      "step": 9
    },
    {
      "epoch": 0.00605052185751021,
      "grad_norm": 5.805087089538574,
      "learning_rate": 0.0001,
      "loss": 2.9419,
      "step": 10
    },
    {
      "epoch": 0.0066555740432612314,
      "grad_norm": 7.2685136795043945,
      "learning_rate": 9.999316524962345e-05,
      "loss": 3.5016,
      "step": 11
    },
    {
      "epoch": 0.007260626229012252,
      "grad_norm": 6.021042823791504,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.5152,
      "step": 12
    },
    {
      "epoch": 0.007865678414763273,
      "grad_norm": 6.932441711425781,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.6719,
      "step": 13
    },
    {
      "epoch": 0.008470730600514294,
      "grad_norm": 7.189040184020996,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.9583,
      "step": 14
    },
    {
      "epoch": 0.009075782786265316,
      "grad_norm": 6.571204662322998,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.6229,
      "step": 15
    },
    {
      "epoch": 0.009680834972016336,
      "grad_norm": 7.477680206298828,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.8054,
      "step": 16
    },
    {
      "epoch": 0.010285887157767357,
      "grad_norm": 7.641745090484619,
      "learning_rate": 9.966546331768191e-05,
      "loss": 3.1001,
      "step": 17
    },
    {
      "epoch": 0.01089093934351838,
      "grad_norm": 9.802349090576172,
      "learning_rate": 9.956320346634876e-05,
      "loss": 3.0516,
      "step": 18
    },
    {
      "epoch": 0.0114959915292694,
      "grad_norm": 7.617214202880859,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.9733,
      "step": 19
    },
    {
      "epoch": 0.01210104371502042,
      "grad_norm": 7.413455009460449,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.3252,
      "step": 20
    },
    {
      "epoch": 0.012706095900771442,
      "grad_norm": 5.545858860015869,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.0681,
      "step": 21
    },
    {
      "epoch": 0.013311148086522463,
      "grad_norm": 5.678832054138184,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.074,
      "step": 22
    },
    {
      "epoch": 0.013916200272273483,
      "grad_norm": 5.938708782196045,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.7884,
      "step": 23
    },
    {
      "epoch": 0.014521252458024504,
      "grad_norm": 5.408363342285156,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.874,
      "step": 24
    },
    {
      "epoch": 0.015126304643775526,
      "grad_norm": 5.655430793762207,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.1017,
      "step": 25
    },
    {
      "epoch": 0.015731356829526547,
      "grad_norm": 4.935225963592529,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.5395,
      "step": 26
    },
    {
      "epoch": 0.016336409015277567,
      "grad_norm": 6.2746782302856445,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.0652,
      "step": 27
    },
    {
      "epoch": 0.016941461201028588,
      "grad_norm": 4.640069961547852,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.5109,
      "step": 28
    },
    {
      "epoch": 0.017546513386779608,
      "grad_norm": 5.904468536376953,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.5498,
      "step": 29
    },
    {
      "epoch": 0.018151565572530632,
      "grad_norm": 4.789588451385498,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.8035,
      "step": 30
    },
    {
      "epoch": 0.018756617758281652,
      "grad_norm": 5.2518229484558105,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.2286,
      "step": 31
    },
    {
      "epoch": 0.019361669944032673,
      "grad_norm": 4.67473030090332,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.8263,
      "step": 32
    },
    {
      "epoch": 0.019966722129783693,
      "grad_norm": 5.545779705047607,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.6349,
      "step": 33
    },
    {
      "epoch": 0.020571774315534714,
      "grad_norm": 4.631152629852295,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.3875,
      "step": 34
    },
    {
      "epoch": 0.021176826501285734,
      "grad_norm": 4.845266342163086,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.366,
      "step": 35
    },
    {
      "epoch": 0.02178187868703676,
      "grad_norm": 5.808229446411133,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.593,
      "step": 36
    },
    {
      "epoch": 0.02238693087278778,
      "grad_norm": 4.492132186889648,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.1577,
      "step": 37
    },
    {
      "epoch": 0.0229919830585388,
      "grad_norm": 5.4371747970581055,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.7733,
      "step": 38
    },
    {
      "epoch": 0.02359703524428982,
      "grad_norm": 5.477941036224365,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.6819,
      "step": 39
    },
    {
      "epoch": 0.02420208743004084,
      "grad_norm": 5.033182621002197,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.7058,
      "step": 40
    },
    {
      "epoch": 0.02480713961579186,
      "grad_norm": 5.059697151184082,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.7372,
      "step": 41
    },
    {
      "epoch": 0.025412191801542885,
      "grad_norm": 4.732717990875244,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.3914,
      "step": 42
    },
    {
      "epoch": 0.026017243987293905,
      "grad_norm": 5.424059867858887,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.6854,
      "step": 43
    },
    {
      "epoch": 0.026622296173044926,
      "grad_norm": 7.047700881958008,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.2959,
      "step": 44
    },
    {
      "epoch": 0.027227348358795946,
      "grad_norm": 6.0506815910339355,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.1825,
      "step": 45
    },
    {
      "epoch": 0.027832400544546967,
      "grad_norm": 4.7166595458984375,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.5374,
      "step": 46
    },
    {
      "epoch": 0.028437452730297987,
      "grad_norm": 5.0626678466796875,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.7514,
      "step": 47
    },
    {
      "epoch": 0.029042504916049008,
      "grad_norm": 5.607767581939697,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.0604,
      "step": 48
    },
    {
      "epoch": 0.02964755710180003,
      "grad_norm": 4.763421058654785,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.7194,
      "step": 49
    },
    {
      "epoch": 0.030252609287551052,
      "grad_norm": 5.836249828338623,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.4789,
      "step": 50
    },
    {
      "epoch": 0.030252609287551052,
      "eval_loss": 1.9229539632797241,
      "eval_runtime": 133.0651,
      "eval_samples_per_second": 20.922,
      "eval_steps_per_second": 5.231,
      "step": 50
    },
    {
      "epoch": 0.030857661473302073,
      "grad_norm": 4.283012390136719,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.3281,
      "step": 51
    },
    {
      "epoch": 0.03146271365905309,
      "grad_norm": 4.356929779052734,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.9618,
      "step": 52
    },
    {
      "epoch": 0.032067765844804114,
      "grad_norm": 4.299388885498047,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.3413,
      "step": 53
    },
    {
      "epoch": 0.032672818030555134,
      "grad_norm": 4.3752360343933105,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.363,
      "step": 54
    },
    {
      "epoch": 0.033277870216306155,
      "grad_norm": 5.005733966827393,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.5032,
      "step": 55
    },
    {
      "epoch": 0.033882922402057175,
      "grad_norm": 4.872354030609131,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.3465,
      "step": 56
    },
    {
      "epoch": 0.034487974587808196,
      "grad_norm": 4.530215740203857,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.2999,
      "step": 57
    },
    {
      "epoch": 0.035093026773559216,
      "grad_norm": 4.4950032234191895,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.2542,
      "step": 58
    },
    {
      "epoch": 0.035698078959310243,
      "grad_norm": 6.2269511222839355,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.401,
      "step": 59
    },
    {
      "epoch": 0.036303131145061264,
      "grad_norm": 5.563892364501953,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.3522,
      "step": 60
    },
    {
      "epoch": 0.036908183330812284,
      "grad_norm": 5.401533603668213,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.0331,
      "step": 61
    },
    {
      "epoch": 0.037513235516563305,
      "grad_norm": 5.7106194496154785,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.1083,
      "step": 62
    },
    {
      "epoch": 0.038118287702314325,
      "grad_norm": 8.203621864318848,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.4249,
      "step": 63
    },
    {
      "epoch": 0.038723339888065346,
      "grad_norm": 5.082009792327881,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.31,
      "step": 64
    },
    {
      "epoch": 0.039328392073816366,
      "grad_norm": 6.3080644607543945,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.8694,
      "step": 65
    },
    {
      "epoch": 0.03993344425956739,
      "grad_norm": 6.578169822692871,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.2567,
      "step": 66
    },
    {
      "epoch": 0.04053849644531841,
      "grad_norm": 15.368583679199219,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.6325,
      "step": 67
    },
    {
      "epoch": 0.04114354863106943,
      "grad_norm": 7.130876064300537,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.3104,
      "step": 68
    },
    {
      "epoch": 0.04174860081682045,
      "grad_norm": 7.857542514801025,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.45,
      "step": 69
    },
    {
      "epoch": 0.04235365300257147,
      "grad_norm": 5.319846153259277,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.6475,
      "step": 70
    },
    {
      "epoch": 0.04295870518832249,
      "grad_norm": 5.050972938537598,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.5534,
      "step": 71
    },
    {
      "epoch": 0.04356375737407352,
      "grad_norm": 4.796183109283447,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.6907,
      "step": 72
    },
    {
      "epoch": 0.04416880955982454,
      "grad_norm": 5.4603753089904785,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.468,
      "step": 73
    },
    {
      "epoch": 0.04477386174557556,
      "grad_norm": 4.92110013961792,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.8974,
      "step": 74
    },
    {
      "epoch": 0.04537891393132658,
      "grad_norm": 4.946081638336182,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.7965,
      "step": 75
    },
    {
      "epoch": 0.0459839661170776,
      "grad_norm": 4.532042980194092,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.4453,
      "step": 76
    },
    {
      "epoch": 0.04658901830282862,
      "grad_norm": 4.3761725425720215,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.4019,
      "step": 77
    },
    {
      "epoch": 0.04719407048857964,
      "grad_norm": 4.8957295417785645,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.7838,
      "step": 78
    },
    {
      "epoch": 0.04779912267433066,
      "grad_norm": 3.9672493934631348,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.3741,
      "step": 79
    },
    {
      "epoch": 0.04840417486008168,
      "grad_norm": 4.684700012207031,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.7226,
      "step": 80
    },
    {
      "epoch": 0.0490092270458327,
      "grad_norm": 6.024520397186279,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.8488,
      "step": 81
    },
    {
      "epoch": 0.04961427923158372,
      "grad_norm": 5.137774467468262,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.1449,
      "step": 82
    },
    {
      "epoch": 0.05021933141733474,
      "grad_norm": 4.6780290603637695,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.6421,
      "step": 83
    },
    {
      "epoch": 0.05082438360308577,
      "grad_norm": 3.9340310096740723,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.3248,
      "step": 84
    },
    {
      "epoch": 0.05142943578883679,
      "grad_norm": 4.935190200805664,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.0658,
      "step": 85
    },
    {
      "epoch": 0.05203448797458781,
      "grad_norm": 3.860861301422119,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.0084,
      "step": 86
    },
    {
      "epoch": 0.05263954016033883,
      "grad_norm": 4.6443891525268555,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.5877,
      "step": 87
    },
    {
      "epoch": 0.05324459234608985,
      "grad_norm": 5.341885566711426,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.39,
      "step": 88
    },
    {
      "epoch": 0.05384964453184087,
      "grad_norm": 4.4416422843933105,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.5717,
      "step": 89
    },
    {
      "epoch": 0.05445469671759189,
      "grad_norm": 4.4941325187683105,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.6604,
      "step": 90
    },
    {
      "epoch": 0.05505974890334291,
      "grad_norm": 5.161505699157715,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.7155,
      "step": 91
    },
    {
      "epoch": 0.055664801089093933,
      "grad_norm": 5.349496841430664,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.7318,
      "step": 92
    },
    {
      "epoch": 0.056269853274844954,
      "grad_norm": 5.1708149909973145,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.5856,
      "step": 93
    },
    {
      "epoch": 0.056874905460595974,
      "grad_norm": 4.384572505950928,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.2994,
      "step": 94
    },
    {
      "epoch": 0.057479957646346995,
      "grad_norm": 4.298950672149658,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.4626,
      "step": 95
    },
    {
      "epoch": 0.058085009832098015,
      "grad_norm": 5.496049404144287,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.0517,
      "step": 96
    },
    {
      "epoch": 0.05869006201784904,
      "grad_norm": 4.826742649078369,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.442,
      "step": 97
    },
    {
      "epoch": 0.05929511420360006,
      "grad_norm": 5.374789237976074,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.3767,
      "step": 98
    },
    {
      "epoch": 0.059900166389351084,
      "grad_norm": 4.710012435913086,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.5977,
      "step": 99
    },
    {
      "epoch": 0.060505218575102104,
      "grad_norm": 5.694589614868164,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.0857,
      "step": 100
    },
    {
      "epoch": 0.060505218575102104,
      "eval_loss": 1.7223000526428223,
      "eval_runtime": 132.8888,
      "eval_samples_per_second": 20.95,
      "eval_steps_per_second": 5.237,
      "step": 100
    },
    {
      "epoch": 0.061110270760853125,
      "grad_norm": 4.256014347076416,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.8564,
      "step": 101
    },
    {
      "epoch": 0.061715322946604145,
      "grad_norm": 4.559025287628174,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.2042,
      "step": 102
    },
    {
      "epoch": 0.062320375132355166,
      "grad_norm": 4.099578857421875,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.0992,
      "step": 103
    },
    {
      "epoch": 0.06292542731810619,
      "grad_norm": 4.300604820251465,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.1802,
      "step": 104
    },
    {
      "epoch": 0.0635304795038572,
      "grad_norm": 4.9460296630859375,
      "learning_rate": 5e-05,
      "loss": 2.6528,
      "step": 105
    },
    {
      "epoch": 0.06413553168960823,
      "grad_norm": 4.572192668914795,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.1446,
      "step": 106
    },
    {
      "epoch": 0.06474058387535925,
      "grad_norm": 4.888028621673584,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.1126,
      "step": 107
    },
    {
      "epoch": 0.06534563606111027,
      "grad_norm": 4.279664039611816,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.838,
      "step": 108
    },
    {
      "epoch": 0.06595068824686129,
      "grad_norm": 4.849969387054443,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.0839,
      "step": 109
    },
    {
      "epoch": 0.06655574043261231,
      "grad_norm": 5.360795974731445,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.6644,
      "step": 110
    },
    {
      "epoch": 0.06716079261836333,
      "grad_norm": 4.785168647766113,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.9794,
      "step": 111
    },
    {
      "epoch": 0.06776584480411435,
      "grad_norm": 4.891386032104492,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.619,
      "step": 112
    },
    {
      "epoch": 0.06837089698986537,
      "grad_norm": 6.79011344909668,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.9895,
      "step": 113
    },
    {
      "epoch": 0.06897594917561639,
      "grad_norm": 5.493412971496582,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.6992,
      "step": 114
    },
    {
      "epoch": 0.06958100136136741,
      "grad_norm": 5.8443450927734375,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.9605,
      "step": 115
    },
    {
      "epoch": 0.07018605354711843,
      "grad_norm": 5.376211166381836,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.5237,
      "step": 116
    },
    {
      "epoch": 0.07079110573286947,
      "grad_norm": 6.854538440704346,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.1044,
      "step": 117
    },
    {
      "epoch": 0.07139615791862049,
      "grad_norm": 12.249556541442871,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.9499,
      "step": 118
    },
    {
      "epoch": 0.07200121010437151,
      "grad_norm": 4.763735294342041,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.6537,
      "step": 119
    },
    {
      "epoch": 0.07260626229012253,
      "grad_norm": 5.7027363777160645,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.2904,
      "step": 120
    },
    {
      "epoch": 0.07321131447587355,
      "grad_norm": 5.195602893829346,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.6341,
      "step": 121
    },
    {
      "epoch": 0.07381636666162457,
      "grad_norm": 4.989786624908447,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.6039,
      "step": 122
    },
    {
      "epoch": 0.07442141884737559,
      "grad_norm": 4.652569770812988,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.6898,
      "step": 123
    },
    {
      "epoch": 0.07502647103312661,
      "grad_norm": 4.230212211608887,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.27,
      "step": 124
    },
    {
      "epoch": 0.07563152321887763,
      "grad_norm": 4.758596897125244,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.6071,
      "step": 125
    },
    {
      "epoch": 0.07623657540462865,
      "grad_norm": 4.950014114379883,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.8124,
      "step": 126
    },
    {
      "epoch": 0.07684162759037967,
      "grad_norm": 4.507136344909668,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.232,
      "step": 127
    },
    {
      "epoch": 0.07744667977613069,
      "grad_norm": 4.617379665374756,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.6234,
      "step": 128
    },
    {
      "epoch": 0.07805173196188171,
      "grad_norm": 4.021543979644775,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.2495,
      "step": 129
    },
    {
      "epoch": 0.07865678414763273,
      "grad_norm": 3.2041749954223633,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.9134,
      "step": 130
    },
    {
      "epoch": 0.07926183633338375,
      "grad_norm": 4.642094135284424,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.5549,
      "step": 131
    },
    {
      "epoch": 0.07986688851913477,
      "grad_norm": 4.646106243133545,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.372,
      "step": 132
    },
    {
      "epoch": 0.0804719407048858,
      "grad_norm": 4.656652450561523,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.8913,
      "step": 133
    },
    {
      "epoch": 0.08107699289063681,
      "grad_norm": 4.523721694946289,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.6731,
      "step": 134
    },
    {
      "epoch": 0.08168204507638784,
      "grad_norm": 4.131584167480469,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.5144,
      "step": 135
    },
    {
      "epoch": 0.08228709726213886,
      "grad_norm": 4.05930233001709,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.3557,
      "step": 136
    },
    {
      "epoch": 0.08289214944788988,
      "grad_norm": 4.66646671295166,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.3276,
      "step": 137
    },
    {
      "epoch": 0.0834972016336409,
      "grad_norm": 4.368364334106445,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.554,
      "step": 138
    },
    {
      "epoch": 0.08410225381939192,
      "grad_norm": 4.241589546203613,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.4698,
      "step": 139
    },
    {
      "epoch": 0.08470730600514294,
      "grad_norm": 4.435495376586914,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.572,
      "step": 140
    },
    {
      "epoch": 0.08531235819089396,
      "grad_norm": 4.079944610595703,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.0826,
      "step": 141
    },
    {
      "epoch": 0.08591741037664498,
      "grad_norm": 4.599911689758301,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.4314,
      "step": 142
    },
    {
      "epoch": 0.08652246256239601,
      "grad_norm": 3.965014934539795,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.1389,
      "step": 143
    },
    {
      "epoch": 0.08712751474814703,
      "grad_norm": 4.101932048797607,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.2636,
      "step": 144
    },
    {
      "epoch": 0.08773256693389805,
      "grad_norm": 4.412632465362549,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.4291,
      "step": 145
    },
    {
      "epoch": 0.08833761911964907,
      "grad_norm": 4.265484809875488,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.4033,
      "step": 146
    },
    {
      "epoch": 0.0889426713054001,
      "grad_norm": 4.544005393981934,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.4107,
      "step": 147
    },
    {
      "epoch": 0.08954772349115112,
      "grad_norm": 4.222890853881836,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.3296,
      "step": 148
    },
    {
      "epoch": 0.09015277567690214,
      "grad_norm": 5.043741226196289,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.7521,
      "step": 149
    },
    {
      "epoch": 0.09075782786265316,
      "grad_norm": 5.090908050537109,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.8676,
      "step": 150
    },
    {
      "epoch": 0.09075782786265316,
      "eval_loss": 1.6341633796691895,
      "eval_runtime": 132.9165,
      "eval_samples_per_second": 20.945,
      "eval_steps_per_second": 5.236,
      "step": 150
    },
    {
      "epoch": 0.09136288004840418,
      "grad_norm": 4.010870456695557,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.015,
      "step": 151
    },
    {
      "epoch": 0.0919679322341552,
      "grad_norm": 4.418432712554932,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.9752,
      "step": 152
    },
    {
      "epoch": 0.09257298441990622,
      "grad_norm": 3.9958014488220215,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.0753,
      "step": 153
    },
    {
      "epoch": 0.09317803660565724,
      "grad_norm": 4.166998386383057,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.7546,
      "step": 154
    },
    {
      "epoch": 0.09378308879140826,
      "grad_norm": 3.985368251800537,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.6191,
      "step": 155
    },
    {
      "epoch": 0.09438814097715928,
      "grad_norm": 4.3636369705200195,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.0277,
      "step": 156
    },
    {
      "epoch": 0.0949931931629103,
      "grad_norm": 4.95756721496582,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.3159,
      "step": 157
    },
    {
      "epoch": 0.09559824534866132,
      "grad_norm": 4.428719997406006,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.7416,
      "step": 158
    },
    {
      "epoch": 0.09620329753441234,
      "grad_norm": 4.474020957946777,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.7884,
      "step": 159
    },
    {
      "epoch": 0.09680834972016336,
      "grad_norm": 4.405423164367676,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.6668,
      "step": 160
    },
    {
      "epoch": 0.09741340190591438,
      "grad_norm": 6.775367259979248,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.8023,
      "step": 161
    },
    {
      "epoch": 0.0980184540916654,
      "grad_norm": 5.818444728851318,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.1367,
      "step": 162
    },
    {
      "epoch": 0.09862350627741642,
      "grad_norm": 4.062469005584717,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.2394,
      "step": 163
    },
    {
      "epoch": 0.09922855846316744,
      "grad_norm": 5.237635612487793,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.8316,
      "step": 164
    },
    {
      "epoch": 0.09983361064891846,
      "grad_norm": 5.004766941070557,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.7498,
      "step": 165
    },
    {
      "epoch": 0.10043866283466948,
      "grad_norm": 6.062614440917969,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.2138,
      "step": 166
    },
    {
      "epoch": 0.1010437150204205,
      "grad_norm": 6.577429294586182,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.4752,
      "step": 167
    },
    {
      "epoch": 0.10164876720617154,
      "grad_norm": 4.324187755584717,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.4536,
      "step": 168
    },
    {
      "epoch": 0.10225381939192256,
      "grad_norm": 6.81287145614624,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.7088,
      "step": 169
    },
    {
      "epoch": 0.10285887157767358,
      "grad_norm": 4.584249019622803,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.629,
      "step": 170
    },
    {
      "epoch": 0.1034639237634246,
      "grad_norm": 3.824648141860962,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.2652,
      "step": 171
    },
    {
      "epoch": 0.10406897594917562,
      "grad_norm": 4.415689945220947,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.8574,
      "step": 172
    },
    {
      "epoch": 0.10467402813492664,
      "grad_norm": 6.334531784057617,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.3935,
      "step": 173
    },
    {
      "epoch": 0.10527908032067766,
      "grad_norm": 3.9672675132751465,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.3739,
      "step": 174
    },
    {
      "epoch": 0.10588413250642868,
      "grad_norm": 4.651885032653809,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.7055,
      "step": 175
    },
    {
      "epoch": 0.1064891846921797,
      "grad_norm": 4.223604202270508,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.3369,
      "step": 176
    },
    {
      "epoch": 0.10709423687793072,
      "grad_norm": 4.983737945556641,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.7146,
      "step": 177
    },
    {
      "epoch": 0.10769928906368174,
      "grad_norm": 5.22819185256958,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.7389,
      "step": 178
    },
    {
      "epoch": 0.10830434124943276,
      "grad_norm": 3.9994022846221924,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.35,
      "step": 179
    },
    {
      "epoch": 0.10890939343518379,
      "grad_norm": 4.338124752044678,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.4418,
      "step": 180
    },
    {
      "epoch": 0.1095144456209348,
      "grad_norm": 4.586472511291504,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.6367,
      "step": 181
    },
    {
      "epoch": 0.11011949780668583,
      "grad_norm": 4.578819751739502,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.6142,
      "step": 182
    },
    {
      "epoch": 0.11072454999243685,
      "grad_norm": 3.8915982246398926,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.3661,
      "step": 183
    },
    {
      "epoch": 0.11132960217818787,
      "grad_norm": 5.164275646209717,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.8204,
      "step": 184
    },
    {
      "epoch": 0.11193465436393889,
      "grad_norm": 4.32537841796875,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.5363,
      "step": 185
    },
    {
      "epoch": 0.11253970654968991,
      "grad_norm": 4.528438568115234,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.1212,
      "step": 186
    },
    {
      "epoch": 0.11314475873544093,
      "grad_norm": 4.616623401641846,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.6208,
      "step": 187
    },
    {
      "epoch": 0.11374981092119195,
      "grad_norm": 3.87123966217041,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.9631,
      "step": 188
    },
    {
      "epoch": 0.11435486310694297,
      "grad_norm": 4.51935338973999,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.4062,
      "step": 189
    },
    {
      "epoch": 0.11495991529269399,
      "grad_norm": 4.743103981018066,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.2416,
      "step": 190
    },
    {
      "epoch": 0.11556496747844501,
      "grad_norm": 4.292260646820068,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.4508,
      "step": 191
    },
    {
      "epoch": 0.11617001966419603,
      "grad_norm": 4.7620673179626465,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.3951,
      "step": 192
    },
    {
      "epoch": 0.11677507184994705,
      "grad_norm": 5.560548782348633,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.7636,
      "step": 193
    },
    {
      "epoch": 0.11738012403569809,
      "grad_norm": 5.016361713409424,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.8073,
      "step": 194
    },
    {
      "epoch": 0.1179851762214491,
      "grad_norm": 5.063380241394043,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.7441,
      "step": 195
    },
    {
      "epoch": 0.11859022840720013,
      "grad_norm": 4.034510135650635,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.2482,
      "step": 196
    },
    {
      "epoch": 0.11919528059295115,
      "grad_norm": 5.894397735595703,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.3427,
      "step": 197
    },
    {
      "epoch": 0.11980033277870217,
      "grad_norm": 5.2676310539245605,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.5469,
      "step": 198
    },
    {
      "epoch": 0.12040538496445319,
      "grad_norm": 4.946696758270264,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.1969,
      "step": 199
    },
    {
      "epoch": 0.12101043715020421,
      "grad_norm": 5.685734748840332,
      "learning_rate": 0.0,
      "loss": 1.8717,
      "step": 200
    },
    {
      "epoch": 0.12101043715020421,
      "eval_loss": 1.5972942113876343,
      "eval_runtime": 133.4035,
      "eval_samples_per_second": 20.869,
      "eval_steps_per_second": 5.217,
      "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": 1.1068744294780109e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}