{
  "best_metric": 1.38406240940094,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.5934718100890207,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002967359050445104,
      "grad_norm": 0.9390130639076233,
      "learning_rate": 1e-05,
      "loss": 1.7936,
      "step": 1
    },
    {
      "epoch": 0.002967359050445104,
      "eval_loss": 1.9112251996994019,
      "eval_runtime": 23.7876,
      "eval_samples_per_second": 23.878,
      "eval_steps_per_second": 5.97,
      "step": 1
    },
    {
      "epoch": 0.005934718100890208,
      "grad_norm": 1.0461841821670532,
      "learning_rate": 2e-05,
      "loss": 1.7158,
      "step": 2
    },
    {
      "epoch": 0.008902077151335312,
      "grad_norm": 0.9577460885047913,
      "learning_rate": 3e-05,
      "loss": 1.7936,
      "step": 3
    },
    {
      "epoch": 0.011869436201780416,
      "grad_norm": 1.0381689071655273,
      "learning_rate": 4e-05,
      "loss": 1.721,
      "step": 4
    },
    {
      "epoch": 0.01483679525222552,
      "grad_norm": 0.9465681314468384,
      "learning_rate": 5e-05,
      "loss": 1.6318,
      "step": 5
    },
    {
      "epoch": 0.017804154302670624,
      "grad_norm": 0.8674979209899902,
      "learning_rate": 6e-05,
      "loss": 1.7419,
      "step": 6
    },
    {
      "epoch": 0.020771513353115726,
      "grad_norm": 0.7945682406425476,
      "learning_rate": 7e-05,
      "loss": 1.5909,
      "step": 7
    },
    {
      "epoch": 0.02373887240356083,
      "grad_norm": 0.8964577913284302,
      "learning_rate": 8e-05,
      "loss": 1.7075,
      "step": 8
    },
    {
      "epoch": 0.026706231454005934,
      "grad_norm": 0.9153485894203186,
      "learning_rate": 9e-05,
      "loss": 1.5796,
      "step": 9
    },
    {
      "epoch": 0.02967359050445104,
      "grad_norm": 0.8851525783538818,
      "learning_rate": 0.0001,
      "loss": 1.6185,
      "step": 10
    },
    {
      "epoch": 0.032640949554896145,
      "grad_norm": 0.9287815690040588,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.6709,
      "step": 11
    },
    {
      "epoch": 0.03560830860534125,
      "grad_norm": 0.8513839244842529,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.6026,
      "step": 12
    },
    {
      "epoch": 0.03857566765578635,
      "grad_norm": 0.7979993224143982,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.5212,
      "step": 13
    },
    {
      "epoch": 0.04154302670623145,
      "grad_norm": 0.7577090859413147,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.5349,
      "step": 14
    },
    {
      "epoch": 0.04451038575667656,
      "grad_norm": 0.7057633996009827,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.5037,
      "step": 15
    },
    {
      "epoch": 0.04747774480712166,
      "grad_norm": 0.6886647343635559,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.4208,
      "step": 16
    },
    {
      "epoch": 0.050445103857566766,
      "grad_norm": 0.7384382486343384,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.5444,
      "step": 17
    },
    {
      "epoch": 0.05341246290801187,
      "grad_norm": 0.7319883704185486,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.5301,
      "step": 18
    },
    {
      "epoch": 0.05637982195845697,
      "grad_norm": 0.6773971319198608,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.3967,
      "step": 19
    },
    {
      "epoch": 0.05934718100890208,
      "grad_norm": 0.7057583332061768,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.4312,
      "step": 20
    },
    {
      "epoch": 0.06231454005934718,
      "grad_norm": 0.6787562370300293,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.4796,
      "step": 21
    },
    {
      "epoch": 0.06528189910979229,
      "grad_norm": 0.677161693572998,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.4709,
      "step": 22
    },
    {
      "epoch": 0.06824925816023739,
      "grad_norm": 0.6980742812156677,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.4326,
      "step": 23
    },
    {
      "epoch": 0.0712166172106825,
      "grad_norm": 0.7324345111846924,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.4748,
      "step": 24
    },
    {
      "epoch": 0.07418397626112759,
      "grad_norm": 0.6989796757698059,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.399,
      "step": 25
    },
    {
      "epoch": 0.0771513353115727,
      "grad_norm": 0.6616961359977722,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.526,
      "step": 26
    },
    {
      "epoch": 0.08011869436201781,
      "grad_norm": 0.7103090882301331,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.4837,
      "step": 27
    },
    {
      "epoch": 0.0830860534124629,
      "grad_norm": 0.7382745146751404,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.5351,
      "step": 28
    },
    {
      "epoch": 0.08605341246290801,
      "grad_norm": 0.7211149334907532,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.5275,
      "step": 29
    },
    {
      "epoch": 0.08902077151335312,
      "grad_norm": 0.7221695780754089,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.466,
      "step": 30
    },
    {
      "epoch": 0.09198813056379822,
      "grad_norm": 0.7047070264816284,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.4064,
      "step": 31
    },
    {
      "epoch": 0.09495548961424333,
      "grad_norm": 0.6703099608421326,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.4269,
      "step": 32
    },
    {
      "epoch": 0.09792284866468842,
      "grad_norm": 0.7262135148048401,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.4023,
      "step": 33
    },
    {
      "epoch": 0.10089020771513353,
      "grad_norm": 0.7264920473098755,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.4648,
      "step": 34
    },
    {
      "epoch": 0.10385756676557864,
      "grad_norm": 0.7304619550704956,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.4788,
      "step": 35
    },
    {
      "epoch": 0.10682492581602374,
      "grad_norm": 0.7077768445014954,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.3955,
      "step": 36
    },
    {
      "epoch": 0.10979228486646884,
      "grad_norm": 0.7239018678665161,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.4926,
      "step": 37
    },
    {
      "epoch": 0.11275964391691394,
      "grad_norm": 0.7896670699119568,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.5151,
      "step": 38
    },
    {
      "epoch": 0.11572700296735905,
      "grad_norm": 0.7284670472145081,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.3684,
      "step": 39
    },
    {
      "epoch": 0.11869436201780416,
      "grad_norm": 0.7199960947036743,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.459,
      "step": 40
    },
    {
      "epoch": 0.12166172106824925,
      "grad_norm": 0.7604296803474426,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.4668,
      "step": 41
    },
    {
      "epoch": 0.12462908011869436,
      "grad_norm": 0.7315712571144104,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.4429,
      "step": 42
    },
    {
      "epoch": 0.12759643916913946,
      "grad_norm": 0.8152048587799072,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.5412,
      "step": 43
    },
    {
      "epoch": 0.13056379821958458,
      "grad_norm": 0.7243455052375793,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.3548,
      "step": 44
    },
    {
      "epoch": 0.13353115727002968,
      "grad_norm": 0.7603928446769714,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.4403,
      "step": 45
    },
    {
      "epoch": 0.13649851632047477,
      "grad_norm": 0.709568440914154,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.433,
      "step": 46
    },
    {
      "epoch": 0.1394658753709199,
      "grad_norm": 0.8047810196876526,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.4614,
      "step": 47
    },
    {
      "epoch": 0.142433234421365,
      "grad_norm": 0.8328258395195007,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.417,
      "step": 48
    },
    {
      "epoch": 0.14540059347181009,
      "grad_norm": 0.8486977815628052,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.559,
      "step": 49
    },
    {
      "epoch": 0.14836795252225518,
      "grad_norm": 0.86100834608078,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.4383,
      "step": 50
    },
    {
      "epoch": 0.14836795252225518,
      "eval_loss": 1.4926310777664185,
      "eval_runtime": 24.1203,
      "eval_samples_per_second": 23.549,
      "eval_steps_per_second": 5.887,
      "step": 50
    },
    {
      "epoch": 0.1513353115727003,
      "grad_norm": 1.1528016328811646,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.656,
      "step": 51
    },
    {
      "epoch": 0.1543026706231454,
      "grad_norm": 0.950527012348175,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.5417,
      "step": 52
    },
    {
      "epoch": 0.1572700296735905,
      "grad_norm": 0.7951736450195312,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.5035,
      "step": 53
    },
    {
      "epoch": 0.16023738872403562,
      "grad_norm": 0.6353037357330322,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.4688,
      "step": 54
    },
    {
      "epoch": 0.1632047477744807,
      "grad_norm": 0.5609595775604248,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.5518,
      "step": 55
    },
    {
      "epoch": 0.1661721068249258,
      "grad_norm": 0.5508681535720825,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.4936,
      "step": 56
    },
    {
      "epoch": 0.16913946587537093,
      "grad_norm": 0.6088594794273376,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.4831,
      "step": 57
    },
    {
      "epoch": 0.17210682492581603,
      "grad_norm": 0.6220605969429016,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.4087,
      "step": 58
    },
    {
      "epoch": 0.17507418397626112,
      "grad_norm": 0.6172888875007629,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.4525,
      "step": 59
    },
    {
      "epoch": 0.17804154302670624,
      "grad_norm": 0.6296764016151428,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.3922,
      "step": 60
    },
    {
      "epoch": 0.18100890207715134,
      "grad_norm": 0.5883337259292603,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.4456,
      "step": 61
    },
    {
      "epoch": 0.18397626112759644,
      "grad_norm": 0.5990262627601624,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.4809,
      "step": 62
    },
    {
      "epoch": 0.18694362017804153,
      "grad_norm": 0.5967974066734314,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.4113,
      "step": 63
    },
    {
      "epoch": 0.18991097922848665,
      "grad_norm": 0.6558731198310852,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.4933,
      "step": 64
    },
    {
      "epoch": 0.19287833827893175,
      "grad_norm": 0.6280772686004639,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.4585,
      "step": 65
    },
    {
      "epoch": 0.19584569732937684,
      "grad_norm": 0.5887389779090881,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.3416,
      "step": 66
    },
    {
      "epoch": 0.19881305637982197,
      "grad_norm": 0.6191444396972656,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.368,
      "step": 67
    },
    {
      "epoch": 0.20178041543026706,
      "grad_norm": 0.6135171055793762,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.4333,
      "step": 68
    },
    {
      "epoch": 0.20474777448071216,
      "grad_norm": 0.6104258298873901,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.5187,
      "step": 69
    },
    {
      "epoch": 0.20771513353115728,
      "grad_norm": 0.6439035534858704,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.3439,
      "step": 70
    },
    {
      "epoch": 0.21068249258160238,
      "grad_norm": 0.6498430967330933,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.4518,
      "step": 71
    },
    {
      "epoch": 0.21364985163204747,
      "grad_norm": 0.6102389097213745,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.4113,
      "step": 72
    },
    {
      "epoch": 0.2166172106824926,
      "grad_norm": 0.6316553354263306,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.4384,
      "step": 73
    },
    {
      "epoch": 0.2195845697329377,
      "grad_norm": 0.6329929232597351,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.336,
      "step": 74
    },
    {
      "epoch": 0.22255192878338279,
      "grad_norm": 0.6220859289169312,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.4376,
      "step": 75
    },
    {
      "epoch": 0.22551928783382788,
      "grad_norm": 0.6214754581451416,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.4113,
      "step": 76
    },
    {
      "epoch": 0.228486646884273,
      "grad_norm": 0.7271621227264404,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.3901,
      "step": 77
    },
    {
      "epoch": 0.2314540059347181,
      "grad_norm": 0.6692347526550293,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.3627,
      "step": 78
    },
    {
      "epoch": 0.2344213649851632,
      "grad_norm": 0.6471211910247803,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.4716,
      "step": 79
    },
    {
      "epoch": 0.23738872403560832,
      "grad_norm": 0.6558648943901062,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.4501,
      "step": 80
    },
    {
      "epoch": 0.2403560830860534,
      "grad_norm": 0.6556353569030762,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.3693,
      "step": 81
    },
    {
      "epoch": 0.2433234421364985,
      "grad_norm": 0.6138564348220825,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.3153,
      "step": 82
    },
    {
      "epoch": 0.24629080118694363,
      "grad_norm": 0.6149715781211853,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.3472,
      "step": 83
    },
    {
      "epoch": 0.24925816023738873,
      "grad_norm": 0.6439250111579895,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.4507,
      "step": 84
    },
    {
      "epoch": 0.2522255192878338,
      "grad_norm": 0.6602970361709595,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.3618,
      "step": 85
    },
    {
      "epoch": 0.2551928783382789,
      "grad_norm": 0.6500831246376038,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.3503,
      "step": 86
    },
    {
      "epoch": 0.258160237388724,
      "grad_norm": 0.6519495844841003,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.2981,
      "step": 87
    },
    {
      "epoch": 0.26112759643916916,
      "grad_norm": 0.6412690877914429,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.4225,
      "step": 88
    },
    {
      "epoch": 0.26409495548961426,
      "grad_norm": 0.6644392013549805,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.4254,
      "step": 89
    },
    {
      "epoch": 0.26706231454005935,
      "grad_norm": 0.8173342347145081,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.4293,
      "step": 90
    },
    {
      "epoch": 0.27002967359050445,
      "grad_norm": 0.6676755547523499,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.3785,
      "step": 91
    },
    {
      "epoch": 0.27299703264094954,
      "grad_norm": 0.716145396232605,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.5122,
      "step": 92
    },
    {
      "epoch": 0.27596439169139464,
      "grad_norm": 0.6954559087753296,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.333,
      "step": 93
    },
    {
      "epoch": 0.2789317507418398,
      "grad_norm": 0.7437573075294495,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.4149,
      "step": 94
    },
    {
      "epoch": 0.2818991097922849,
      "grad_norm": 0.757072925567627,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.4235,
      "step": 95
    },
    {
      "epoch": 0.28486646884273,
      "grad_norm": 0.7700419425964355,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.4789,
      "step": 96
    },
    {
      "epoch": 0.2878338278931751,
      "grad_norm": 0.767932116985321,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.4574,
      "step": 97
    },
    {
      "epoch": 0.29080118694362017,
      "grad_norm": 0.7921847701072693,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.4761,
      "step": 98
    },
    {
      "epoch": 0.29376854599406527,
      "grad_norm": 0.7436829805374146,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.4238,
      "step": 99
    },
    {
      "epoch": 0.29673590504451036,
      "grad_norm": 0.9294240474700928,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.3764,
      "step": 100
    },
    {
      "epoch": 0.29673590504451036,
      "eval_loss": 1.4243048429489136,
      "eval_runtime": 24.1568,
      "eval_samples_per_second": 23.513,
      "eval_steps_per_second": 5.878,
      "step": 100
    },
    {
      "epoch": 0.2997032640949555,
      "grad_norm": 0.7108097672462463,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.4564,
      "step": 101
    },
    {
      "epoch": 0.3026706231454006,
      "grad_norm": 0.7227379083633423,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.4768,
      "step": 102
    },
    {
      "epoch": 0.3056379821958457,
      "grad_norm": 0.6515297889709473,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.5221,
      "step": 103
    },
    {
      "epoch": 0.3086053412462908,
      "grad_norm": 0.6635713577270508,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.3973,
      "step": 104
    },
    {
      "epoch": 0.3115727002967359,
      "grad_norm": 0.5767274498939514,
      "learning_rate": 5e-05,
      "loss": 1.3611,
      "step": 105
    },
    {
      "epoch": 0.314540059347181,
      "grad_norm": 0.5633724927902222,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.3804,
      "step": 106
    },
    {
      "epoch": 0.31750741839762614,
      "grad_norm": 0.5423171520233154,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.3908,
      "step": 107
    },
    {
      "epoch": 0.32047477744807124,
      "grad_norm": 0.5702859163284302,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.3531,
      "step": 108
    },
    {
      "epoch": 0.32344213649851633,
      "grad_norm": 0.5615251660346985,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.4314,
      "step": 109
    },
    {
      "epoch": 0.3264094955489614,
      "grad_norm": 0.5738047957420349,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.4177,
      "step": 110
    },
    {
      "epoch": 0.3293768545994065,
      "grad_norm": 0.5917168259620667,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.3784,
      "step": 111
    },
    {
      "epoch": 0.3323442136498516,
      "grad_norm": 0.6052302122116089,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.4513,
      "step": 112
    },
    {
      "epoch": 0.3353115727002967,
      "grad_norm": 0.5679163336753845,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.3371,
      "step": 113
    },
    {
      "epoch": 0.33827893175074186,
      "grad_norm": 0.5959806442260742,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.3729,
      "step": 114
    },
    {
      "epoch": 0.34124629080118696,
      "grad_norm": 0.6050249934196472,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.3484,
      "step": 115
    },
    {
      "epoch": 0.34421364985163205,
      "grad_norm": 0.569413959980011,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.4277,
      "step": 116
    },
    {
      "epoch": 0.34718100890207715,
      "grad_norm": 0.6120415925979614,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.454,
      "step": 117
    },
    {
      "epoch": 0.35014836795252224,
      "grad_norm": 0.5939725041389465,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.3871,
      "step": 118
    },
    {
      "epoch": 0.35311572700296734,
      "grad_norm": 0.6111284494400024,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.2666,
      "step": 119
    },
    {
      "epoch": 0.3560830860534125,
      "grad_norm": 0.6026867628097534,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.3433,
      "step": 120
    },
    {
      "epoch": 0.3590504451038576,
      "grad_norm": 0.621806800365448,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.2591,
      "step": 121
    },
    {
      "epoch": 0.3620178041543027,
      "grad_norm": 0.6404162049293518,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.3787,
      "step": 122
    },
    {
      "epoch": 0.3649851632047478,
      "grad_norm": 0.6309717893600464,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.361,
      "step": 123
    },
    {
      "epoch": 0.36795252225519287,
      "grad_norm": 0.621569037437439,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.3831,
      "step": 124
    },
    {
      "epoch": 0.37091988130563797,
      "grad_norm": 0.6342136263847351,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.412,
      "step": 125
    },
    {
      "epoch": 0.37388724035608306,
      "grad_norm": 0.618765115737915,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.3994,
      "step": 126
    },
    {
      "epoch": 0.3768545994065282,
      "grad_norm": 0.6354244947433472,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.4568,
      "step": 127
    },
    {
      "epoch": 0.3798219584569733,
      "grad_norm": 0.6155465245246887,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.3748,
      "step": 128
    },
    {
      "epoch": 0.3827893175074184,
      "grad_norm": 0.6294691562652588,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.3385,
      "step": 129
    },
    {
      "epoch": 0.3857566765578635,
      "grad_norm": 0.6291654109954834,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.3601,
      "step": 130
    },
    {
      "epoch": 0.3887240356083086,
      "grad_norm": 0.6313578486442566,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.3743,
      "step": 131
    },
    {
      "epoch": 0.3916913946587537,
      "grad_norm": 0.6349329352378845,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.359,
      "step": 132
    },
    {
      "epoch": 0.39465875370919884,
      "grad_norm": 0.6437543034553528,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.3214,
      "step": 133
    },
    {
      "epoch": 0.39762611275964393,
      "grad_norm": 0.6678977012634277,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.4971,
      "step": 134
    },
    {
      "epoch": 0.40059347181008903,
      "grad_norm": 0.6663174629211426,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.4844,
      "step": 135
    },
    {
      "epoch": 0.4035608308605341,
      "grad_norm": 0.6650763750076294,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.469,
      "step": 136
    },
    {
      "epoch": 0.4065281899109792,
      "grad_norm": 0.69032222032547,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.436,
      "step": 137
    },
    {
      "epoch": 0.4094955489614243,
      "grad_norm": 0.6544485092163086,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.395,
      "step": 138
    },
    {
      "epoch": 0.4124629080118694,
      "grad_norm": 0.6686225533485413,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.3662,
      "step": 139
    },
    {
      "epoch": 0.41543026706231456,
      "grad_norm": 0.6382662057876587,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.332,
      "step": 140
    },
    {
      "epoch": 0.41839762611275966,
      "grad_norm": 0.7008192539215088,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.3399,
      "step": 141
    },
    {
      "epoch": 0.42136498516320475,
      "grad_norm": 0.6895390152931213,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.351,
      "step": 142
    },
    {
      "epoch": 0.42433234421364985,
      "grad_norm": 0.6697501540184021,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.3511,
      "step": 143
    },
    {
      "epoch": 0.42729970326409494,
      "grad_norm": 0.7038697004318237,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.4292,
      "step": 144
    },
    {
      "epoch": 0.43026706231454004,
      "grad_norm": 0.6954337954521179,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.3809,
      "step": 145
    },
    {
      "epoch": 0.4332344213649852,
      "grad_norm": 0.7463260293006897,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.4764,
      "step": 146
    },
    {
      "epoch": 0.4362017804154303,
      "grad_norm": 0.7155658602714539,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.4374,
      "step": 147
    },
    {
      "epoch": 0.4391691394658754,
      "grad_norm": 0.7358605265617371,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.428,
      "step": 148
    },
    {
      "epoch": 0.4421364985163205,
      "grad_norm": 0.7639449834823608,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.3996,
      "step": 149
    },
    {
      "epoch": 0.44510385756676557,
      "grad_norm": 0.9065217971801758,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.4056,
      "step": 150
    },
    {
      "epoch": 0.44510385756676557,
      "eval_loss": 1.3901383876800537,
      "eval_runtime": 24.1581,
      "eval_samples_per_second": 23.512,
      "eval_steps_per_second": 5.878,
      "step": 150
    },
    {
      "epoch": 0.44807121661721067,
      "grad_norm": 0.5718062520027161,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.4718,
      "step": 151
    },
    {
      "epoch": 0.45103857566765576,
      "grad_norm": 0.6048449277877808,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.2878,
      "step": 152
    },
    {
      "epoch": 0.4540059347181009,
      "grad_norm": 0.5899658799171448,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.3643,
      "step": 153
    },
    {
      "epoch": 0.456973293768546,
      "grad_norm": 0.5835247039794922,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.3515,
      "step": 154
    },
    {
      "epoch": 0.4599406528189911,
      "grad_norm": 0.620940089225769,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.3727,
      "step": 155
    },
    {
      "epoch": 0.4629080118694362,
      "grad_norm": 0.5727185010910034,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.4007,
      "step": 156
    },
    {
      "epoch": 0.4658753709198813,
      "grad_norm": 0.5707109570503235,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.3363,
      "step": 157
    },
    {
      "epoch": 0.4688427299703264,
      "grad_norm": 0.588900625705719,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.3328,
      "step": 158
    },
    {
      "epoch": 0.47181008902077154,
      "grad_norm": 0.55113685131073,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.3329,
      "step": 159
    },
    {
      "epoch": 0.47477744807121663,
      "grad_norm": 0.577501654624939,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.3885,
      "step": 160
    },
    {
      "epoch": 0.47774480712166173,
      "grad_norm": 0.6127510666847229,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.4212,
      "step": 161
    },
    {
      "epoch": 0.4807121661721068,
      "grad_norm": 0.5858021378517151,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.4567,
      "step": 162
    },
    {
      "epoch": 0.4836795252225519,
      "grad_norm": 0.5853855609893799,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.3353,
      "step": 163
    },
    {
      "epoch": 0.486646884272997,
      "grad_norm": 0.5995965003967285,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.4138,
      "step": 164
    },
    {
      "epoch": 0.4896142433234421,
      "grad_norm": 0.6032690405845642,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.4121,
      "step": 165
    },
    {
      "epoch": 0.49258160237388726,
      "grad_norm": 0.5802157521247864,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.3769,
      "step": 166
    },
    {
      "epoch": 0.49554896142433236,
      "grad_norm": 0.5848171710968018,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.3435,
      "step": 167
    },
    {
      "epoch": 0.49851632047477745,
      "grad_norm": 0.5824593901634216,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.4433,
      "step": 168
    },
    {
      "epoch": 0.5014836795252225,
      "grad_norm": 0.58831787109375,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.4107,
      "step": 169
    },
    {
      "epoch": 0.5044510385756676,
      "grad_norm": 0.5733941197395325,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.3252,
      "step": 170
    },
    {
      "epoch": 0.5074183976261127,
      "grad_norm": 0.5888526439666748,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.3818,
      "step": 171
    },
    {
      "epoch": 0.5103857566765578,
      "grad_norm": 0.6103017330169678,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.3278,
      "step": 172
    },
    {
      "epoch": 0.5133531157270029,
      "grad_norm": 0.6007798910140991,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.2266,
      "step": 173
    },
    {
      "epoch": 0.516320474777448,
      "grad_norm": 0.5885071754455566,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.4017,
      "step": 174
    },
    {
      "epoch": 0.5192878338278932,
      "grad_norm": 0.6321878433227539,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.3121,
      "step": 175
    },
    {
      "epoch": 0.5222551928783383,
      "grad_norm": 0.6000406742095947,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.2604,
      "step": 176
    },
    {
      "epoch": 0.5252225519287834,
      "grad_norm": 0.6312501430511475,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.3747,
      "step": 177
    },
    {
      "epoch": 0.5281899109792285,
      "grad_norm": 0.6489883661270142,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.3086,
      "step": 178
    },
    {
      "epoch": 0.5311572700296736,
      "grad_norm": 0.621934175491333,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.3274,
      "step": 179
    },
    {
      "epoch": 0.5341246290801187,
      "grad_norm": 0.6494255065917969,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.3698,
      "step": 180
    },
    {
      "epoch": 0.5370919881305638,
      "grad_norm": 0.6653171181678772,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.3594,
      "step": 181
    },
    {
      "epoch": 0.5400593471810089,
      "grad_norm": 0.6071606874465942,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.2813,
      "step": 182
    },
    {
      "epoch": 0.543026706231454,
      "grad_norm": 0.6710329055786133,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.3311,
      "step": 183
    },
    {
      "epoch": 0.5459940652818991,
      "grad_norm": 0.6964633464813232,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.4293,
      "step": 184
    },
    {
      "epoch": 0.5489614243323442,
      "grad_norm": 0.6795589923858643,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.3489,
      "step": 185
    },
    {
      "epoch": 0.5519287833827893,
      "grad_norm": 0.654394268989563,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.4716,
      "step": 186
    },
    {
      "epoch": 0.5548961424332344,
      "grad_norm": 0.6674098968505859,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.3962,
      "step": 187
    },
    {
      "epoch": 0.5578635014836796,
      "grad_norm": 0.7132275104522705,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.4051,
      "step": 188
    },
    {
      "epoch": 0.5608308605341247,
      "grad_norm": 0.6848230957984924,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.3937,
      "step": 189
    },
    {
      "epoch": 0.5637982195845698,
      "grad_norm": 0.6619846224784851,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.3382,
      "step": 190
    },
    {
      "epoch": 0.5667655786350149,
      "grad_norm": 0.7117395401000977,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.3722,
      "step": 191
    },
    {
      "epoch": 0.56973293768546,
      "grad_norm": 0.6936861872673035,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.3315,
      "step": 192
    },
    {
      "epoch": 0.5727002967359051,
      "grad_norm": 0.7397587299346924,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.4391,
      "step": 193
    },
    {
      "epoch": 0.5756676557863502,
      "grad_norm": 0.7653889656066895,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.5046,
      "step": 194
    },
    {
      "epoch": 0.5786350148367952,
      "grad_norm": 0.7716618180274963,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.4364,
      "step": 195
    },
    {
      "epoch": 0.5816023738872403,
      "grad_norm": 0.7638105750083923,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.437,
      "step": 196
    },
    {
      "epoch": 0.5845697329376854,
      "grad_norm": 0.7750841379165649,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.4018,
      "step": 197
    },
    {
      "epoch": 0.5875370919881305,
      "grad_norm": 0.8630430698394775,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.4002,
      "step": 198
    },
    {
      "epoch": 0.5905044510385756,
      "grad_norm": 0.8892988562583923,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.4058,
      "step": 199
    },
    {
      "epoch": 0.5934718100890207,
      "grad_norm": 0.9203798770904541,
      "learning_rate": 0.0,
      "loss": 1.4318,
      "step": 200
    },
    {
      "epoch": 0.5934718100890207,
      "eval_loss": 1.38406240940094,
      "eval_runtime": 24.1664,
      "eval_samples_per_second": 23.504,
      "eval_steps_per_second": 5.876,
      "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": 1.146618963296256e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}