{
  "best_metric": 2.751559019088745,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.9422850412249706,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004711425206124852,
      "grad_norm": 1.4224753379821777,
      "learning_rate": 1e-05,
      "loss": 4.1193,
      "step": 1
    },
    {
      "epoch": 0.004711425206124852,
      "eval_loss": 4.429855823516846,
      "eval_runtime": 5.6207,
      "eval_samples_per_second": 63.693,
      "eval_steps_per_second": 16.012,
      "step": 1
    },
    {
      "epoch": 0.009422850412249705,
      "grad_norm": 1.7880243062973022,
      "learning_rate": 2e-05,
      "loss": 3.5333,
      "step": 2
    },
    {
      "epoch": 0.014134275618374558,
      "grad_norm": 1.3067415952682495,
      "learning_rate": 3e-05,
      "loss": 3.8069,
      "step": 3
    },
    {
      "epoch": 0.01884570082449941,
      "grad_norm": 1.0573233366012573,
      "learning_rate": 4e-05,
      "loss": 4.228,
      "step": 4
    },
    {
      "epoch": 0.023557126030624265,
      "grad_norm": 1.6455657482147217,
      "learning_rate": 5e-05,
      "loss": 4.0449,
      "step": 5
    },
    {
      "epoch": 0.028268551236749116,
      "grad_norm": 1.360145092010498,
      "learning_rate": 6e-05,
      "loss": 4.1741,
      "step": 6
    },
    {
      "epoch": 0.03297997644287397,
      "grad_norm": 1.3340444564819336,
      "learning_rate": 7e-05,
      "loss": 4.1187,
      "step": 7
    },
    {
      "epoch": 0.03769140164899882,
      "grad_norm": 1.1054108142852783,
      "learning_rate": 8e-05,
      "loss": 4.4052,
      "step": 8
    },
    {
      "epoch": 0.04240282685512368,
      "grad_norm": 1.2251219749450684,
      "learning_rate": 9e-05,
      "loss": 4.0603,
      "step": 9
    },
    {
      "epoch": 0.04711425206124853,
      "grad_norm": 1.1435296535491943,
      "learning_rate": 0.0001,
      "loss": 4.2137,
      "step": 10
    },
    {
      "epoch": 0.05182567726737338,
      "grad_norm": 0.9666030406951904,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.1106,
      "step": 11
    },
    {
      "epoch": 0.05653710247349823,
      "grad_norm": 1.037857174873352,
      "learning_rate": 9.997266286704631e-05,
      "loss": 4.0571,
      "step": 12
    },
    {
      "epoch": 0.061248527679623084,
      "grad_norm": 1.2075278759002686,
      "learning_rate": 9.993849845741524e-05,
      "loss": 4.0131,
      "step": 13
    },
    {
      "epoch": 0.06595995288574794,
      "grad_norm": 1.3902502059936523,
      "learning_rate": 9.989068136093873e-05,
      "loss": 3.8902,
      "step": 14
    },
    {
      "epoch": 0.0706713780918728,
      "grad_norm": 1.3553705215454102,
      "learning_rate": 9.98292246503335e-05,
      "loss": 4.1146,
      "step": 15
    },
    {
      "epoch": 0.07538280329799764,
      "grad_norm": 1.2916676998138428,
      "learning_rate": 9.975414512725057e-05,
      "loss": 3.8055,
      "step": 16
    },
    {
      "epoch": 0.0800942285041225,
      "grad_norm": 1.173845648765564,
      "learning_rate": 9.966546331768191e-05,
      "loss": 3.9704,
      "step": 17
    },
    {
      "epoch": 0.08480565371024736,
      "grad_norm": 1.3278011083602905,
      "learning_rate": 9.956320346634876e-05,
      "loss": 3.9821,
      "step": 18
    },
    {
      "epoch": 0.0895170789163722,
      "grad_norm": 1.1074657440185547,
      "learning_rate": 9.944739353007344e-05,
      "loss": 3.6206,
      "step": 19
    },
    {
      "epoch": 0.09422850412249706,
      "grad_norm": 1.2321492433547974,
      "learning_rate": 9.931806517013612e-05,
      "loss": 3.6249,
      "step": 20
    },
    {
      "epoch": 0.0989399293286219,
      "grad_norm": 1.512702465057373,
      "learning_rate": 9.917525374361912e-05,
      "loss": 3.4217,
      "step": 21
    },
    {
      "epoch": 0.10365135453474676,
      "grad_norm": 1.2779844999313354,
      "learning_rate": 9.901899829374047e-05,
      "loss": 3.9731,
      "step": 22
    },
    {
      "epoch": 0.10836277974087162,
      "grad_norm": 1.5148011445999146,
      "learning_rate": 9.884934153917997e-05,
      "loss": 3.7585,
      "step": 23
    },
    {
      "epoch": 0.11307420494699646,
      "grad_norm": 1.6608439683914185,
      "learning_rate": 9.86663298624003e-05,
      "loss": 3.9828,
      "step": 24
    },
    {
      "epoch": 0.11778563015312132,
      "grad_norm": 2.0406711101531982,
      "learning_rate": 9.847001329696653e-05,
      "loss": 3.8848,
      "step": 25
    },
    {
      "epoch": 0.12249705535924617,
      "grad_norm": 2.0955231189727783,
      "learning_rate": 9.826044551386744e-05,
      "loss": 3.1387,
      "step": 26
    },
    {
      "epoch": 0.127208480565371,
      "grad_norm": 1.6862400770187378,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.9899,
      "step": 27
    },
    {
      "epoch": 0.13191990577149587,
      "grad_norm": 1.7947214841842651,
      "learning_rate": 9.780178907671789e-05,
      "loss": 3.4965,
      "step": 28
    },
    {
      "epoch": 0.13663133097762073,
      "grad_norm": 1.3594914674758911,
      "learning_rate": 9.755282581475769e-05,
      "loss": 3.2996,
      "step": 29
    },
    {
      "epoch": 0.1413427561837456,
      "grad_norm": 1.220553994178772,
      "learning_rate": 9.729086208503174e-05,
      "loss": 3.0813,
      "step": 30
    },
    {
      "epoch": 0.14605418138987045,
      "grad_norm": 1.1600159406661987,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.742,
      "step": 31
    },
    {
      "epoch": 0.15076560659599528,
      "grad_norm": 1.1100362539291382,
      "learning_rate": 9.672822322997305e-05,
      "loss": 3.2874,
      "step": 32
    },
    {
      "epoch": 0.15547703180212014,
      "grad_norm": 1.2839323282241821,
      "learning_rate": 9.642770192448536e-05,
      "loss": 3.1803,
      "step": 33
    },
    {
      "epoch": 0.160188457008245,
      "grad_norm": 1.168896198272705,
      "learning_rate": 9.611448774886924e-05,
      "loss": 3.3228,
      "step": 34
    },
    {
      "epoch": 0.16489988221436985,
      "grad_norm": 1.098732590675354,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.9145,
      "step": 35
    },
    {
      "epoch": 0.1696113074204947,
      "grad_norm": 1.1237868070602417,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.9384,
      "step": 36
    },
    {
      "epoch": 0.17432273262661954,
      "grad_norm": 0.9336548447608948,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.7387,
      "step": 37
    },
    {
      "epoch": 0.1790341578327444,
      "grad_norm": 0.9725769758224487,
      "learning_rate": 9.473646649103818e-05,
      "loss": 3.3192,
      "step": 38
    },
    {
      "epoch": 0.18374558303886926,
      "grad_norm": 0.9544655680656433,
      "learning_rate": 9.43611409721806e-05,
      "loss": 3.037,
      "step": 39
    },
    {
      "epoch": 0.18845700824499412,
      "grad_norm": 1.0939885377883911,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.7841,
      "step": 40
    },
    {
      "epoch": 0.19316843345111898,
      "grad_norm": 0.8907297849655151,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.7532,
      "step": 41
    },
    {
      "epoch": 0.1978798586572438,
      "grad_norm": 0.9159559011459351,
      "learning_rate": 9.316282404787871e-05,
      "loss": 3.0008,
      "step": 42
    },
    {
      "epoch": 0.20259128386336867,
      "grad_norm": 0.9595715403556824,
      "learning_rate": 9.273963562927695e-05,
      "loss": 3.0729,
      "step": 43
    },
    {
      "epoch": 0.20730270906949352,
      "grad_norm": 0.8981564044952393,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.9887,
      "step": 44
    },
    {
      "epoch": 0.21201413427561838,
      "grad_norm": 0.86142498254776,
      "learning_rate": 9.185832391312644e-05,
      "loss": 3.1152,
      "step": 45
    },
    {
      "epoch": 0.21672555948174324,
      "grad_norm": 0.9479924440383911,
      "learning_rate": 9.140044155740101e-05,
      "loss": 3.3846,
      "step": 46
    },
    {
      "epoch": 0.22143698468786807,
      "grad_norm": 0.9610684514045715,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.8478,
      "step": 47
    },
    {
      "epoch": 0.22614840989399293,
      "grad_norm": 0.9329556822776794,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.8681,
      "step": 48
    },
    {
      "epoch": 0.2308598351001178,
      "grad_norm": 0.8114438056945801,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.9582,
      "step": 49
    },
    {
      "epoch": 0.23557126030624265,
      "grad_norm": 0.9317813515663147,
      "learning_rate": 8.945702546981969e-05,
      "loss": 3.1592,
      "step": 50
    },
    {
      "epoch": 0.23557126030624265,
      "eval_loss": 3.092801809310913,
      "eval_runtime": 5.6346,
      "eval_samples_per_second": 63.536,
      "eval_steps_per_second": 15.973,
      "step": 50
    },
    {
      "epoch": 0.24028268551236748,
      "grad_norm": 1.184509038925171,
      "learning_rate": 8.894386393810563e-05,
      "loss": 3.2605,
      "step": 51
    },
    {
      "epoch": 0.24499411071849234,
      "grad_norm": 1.4136345386505127,
      "learning_rate": 8.842005554284296e-05,
      "loss": 3.2183,
      "step": 52
    },
    {
      "epoch": 0.2497055359246172,
      "grad_norm": 1.0584381818771362,
      "learning_rate": 8.788574348801675e-05,
      "loss": 3.4899,
      "step": 53
    },
    {
      "epoch": 0.254416961130742,
      "grad_norm": 0.9841482043266296,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.9919,
      "step": 54
    },
    {
      "epoch": 0.2591283863368669,
      "grad_norm": 1.4427697658538818,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.9883,
      "step": 55
    },
    {
      "epoch": 0.26383981154299174,
      "grad_norm": 0.8978828191757202,
      "learning_rate": 8.622126023955446e-05,
      "loss": 3.0033,
      "step": 56
    },
    {
      "epoch": 0.26855123674911663,
      "grad_norm": 0.9529774785041809,
      "learning_rate": 8.564642241456986e-05,
      "loss": 3.2577,
      "step": 57
    },
    {
      "epoch": 0.27326266195524146,
      "grad_norm": 0.8481725454330444,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.9624,
      "step": 58
    },
    {
      "epoch": 0.2779740871613663,
      "grad_norm": 0.927348256111145,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.822,
      "step": 59
    },
    {
      "epoch": 0.2826855123674912,
      "grad_norm": 0.9241753220558167,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.8834,
      "step": 60
    },
    {
      "epoch": 0.287396937573616,
      "grad_norm": 0.9131187200546265,
      "learning_rate": 8.32512286056924e-05,
      "loss": 3.0752,
      "step": 61
    },
    {
      "epoch": 0.2921083627797409,
      "grad_norm": 0.92178875207901,
      "learning_rate": 8.262928807620843e-05,
      "loss": 3.1421,
      "step": 62
    },
    {
      "epoch": 0.2968197879858657,
      "grad_norm": 1.1212458610534668,
      "learning_rate": 8.199842702516583e-05,
      "loss": 3.6917,
      "step": 63
    },
    {
      "epoch": 0.30153121319199055,
      "grad_norm": 0.9064971804618835,
      "learning_rate": 8.135881792367686e-05,
      "loss": 3.3719,
      "step": 64
    },
    {
      "epoch": 0.30624263839811544,
      "grad_norm": 0.9023184180259705,
      "learning_rate": 8.07106356344834e-05,
      "loss": 3.3009,
      "step": 65
    },
    {
      "epoch": 0.31095406360424027,
      "grad_norm": 0.9435430765151978,
      "learning_rate": 8.005405736415126e-05,
      "loss": 3.109,
      "step": 66
    },
    {
      "epoch": 0.31566548881036516,
      "grad_norm": 0.9783146977424622,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.8671,
      "step": 67
    },
    {
      "epoch": 0.32037691401649,
      "grad_norm": 0.9934388399124146,
      "learning_rate": 7.871643313414718e-05,
      "loss": 3.2788,
      "step": 68
    },
    {
      "epoch": 0.3250883392226148,
      "grad_norm": 0.9181382060050964,
      "learning_rate": 7.803575286758364e-05,
      "loss": 3.1027,
      "step": 69
    },
    {
      "epoch": 0.3297997644287397,
      "grad_norm": 1.0301085710525513,
      "learning_rate": 7.734740790612136e-05,
      "loss": 3.4456,
      "step": 70
    },
    {
      "epoch": 0.33451118963486454,
      "grad_norm": 1.007025122642517,
      "learning_rate": 7.66515864363997e-05,
      "loss": 3.0618,
      "step": 71
    },
    {
      "epoch": 0.3392226148409894,
      "grad_norm": 1.0853885412216187,
      "learning_rate": 7.594847868906076e-05,
      "loss": 3.0318,
      "step": 72
    },
    {
      "epoch": 0.34393404004711425,
      "grad_norm": 1.1132615804672241,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.9215,
      "step": 73
    },
    {
      "epoch": 0.3486454652532391,
      "grad_norm": 1.3701266050338745,
      "learning_rate": 7.452117519152542e-05,
      "loss": 3.1332,
      "step": 74
    },
    {
      "epoch": 0.35335689045936397,
      "grad_norm": 1.3136780261993408,
      "learning_rate": 7.379736965185368e-05,
      "loss": 3.2356,
      "step": 75
    },
    {
      "epoch": 0.3580683156654888,
      "grad_norm": 1.1880289316177368,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.7093,
      "step": 76
    },
    {
      "epoch": 0.3627797408716137,
      "grad_norm": 1.2439075708389282,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.561,
      "step": 77
    },
    {
      "epoch": 0.3674911660777385,
      "grad_norm": 1.152225136756897,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.8727,
      "step": 78
    },
    {
      "epoch": 0.37220259128386335,
      "grad_norm": 1.2609004974365234,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.6404,
      "step": 79
    },
    {
      "epoch": 0.37691401648998824,
      "grad_norm": 1.1906847953796387,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.7368,
      "step": 80
    },
    {
      "epoch": 0.38162544169611307,
      "grad_norm": 1.1014139652252197,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.5298,
      "step": 81
    },
    {
      "epoch": 0.38633686690223795,
      "grad_norm": 1.131485104560852,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.3592,
      "step": 82
    },
    {
      "epoch": 0.3910482921083628,
      "grad_norm": 1.014001727104187,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.5501,
      "step": 83
    },
    {
      "epoch": 0.3957597173144876,
      "grad_norm": 1.2307207584381104,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.9471,
      "step": 84
    },
    {
      "epoch": 0.4004711425206125,
      "grad_norm": 1.0490418672561646,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.9623,
      "step": 85
    },
    {
      "epoch": 0.40518256772673733,
      "grad_norm": 0.9485599994659424,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.8153,
      "step": 86
    },
    {
      "epoch": 0.4098939929328622,
      "grad_norm": 0.9144803881645203,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.1913,
      "step": 87
    },
    {
      "epoch": 0.41460541813898705,
      "grad_norm": 0.9584115147590637,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.3089,
      "step": 88
    },
    {
      "epoch": 0.4193168433451119,
      "grad_norm": 0.8980095982551575,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.7153,
      "step": 89
    },
    {
      "epoch": 0.42402826855123676,
      "grad_norm": 0.914086639881134,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.7107,
      "step": 90
    },
    {
      "epoch": 0.4287396937573616,
      "grad_norm": 0.8021414279937744,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.5325,
      "step": 91
    },
    {
      "epoch": 0.4334511189634865,
      "grad_norm": 0.7937134504318237,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.6215,
      "step": 92
    },
    {
      "epoch": 0.4381625441696113,
      "grad_norm": 0.9508938193321228,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.8083,
      "step": 93
    },
    {
      "epoch": 0.44287396937573614,
      "grad_norm": 0.8266021609306335,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.7794,
      "step": 94
    },
    {
      "epoch": 0.44758539458186103,
      "grad_norm": 0.9294748306274414,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.6317,
      "step": 95
    },
    {
      "epoch": 0.45229681978798586,
      "grad_norm": 0.9131935238838196,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.9648,
      "step": 96
    },
    {
      "epoch": 0.45700824499411075,
      "grad_norm": 0.8720795512199402,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 3.0526,
      "step": 97
    },
    {
      "epoch": 0.4617196702002356,
      "grad_norm": 0.9298259615898132,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.1693,
      "step": 98
    },
    {
      "epoch": 0.4664310954063604,
      "grad_norm": 0.9132508039474487,
      "learning_rate": 5.495227651252315e-05,
      "loss": 3.257,
      "step": 99
    },
    {
      "epoch": 0.4711425206124853,
      "grad_norm": 0.854914665222168,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.7248,
      "step": 100
    },
    {
      "epoch": 0.4711425206124853,
      "eval_loss": 2.875974416732788,
      "eval_runtime": 5.6146,
      "eval_samples_per_second": 63.762,
      "eval_steps_per_second": 16.03,
      "step": 100
    },
    {
      "epoch": 0.4758539458186101,
      "grad_norm": 1.2511489391326904,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.8721,
      "step": 101
    },
    {
      "epoch": 0.48056537102473496,
      "grad_norm": 1.1843417882919312,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.9974,
      "step": 102
    },
    {
      "epoch": 0.48527679623085984,
      "grad_norm": 1.0908939838409424,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.9349,
      "step": 103
    },
    {
      "epoch": 0.48998822143698467,
      "grad_norm": 0.9236489534378052,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 3.1013,
      "step": 104
    },
    {
      "epoch": 0.49469964664310956,
      "grad_norm": 0.8705795407295227,
      "learning_rate": 5e-05,
      "loss": 2.8818,
      "step": 105
    },
    {
      "epoch": 0.4994110718492344,
      "grad_norm": 0.8545961380004883,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.5349,
      "step": 106
    },
    {
      "epoch": 0.5041224970553593,
      "grad_norm": 0.97505784034729,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.7821,
      "step": 107
    },
    {
      "epoch": 0.508833922261484,
      "grad_norm": 1.054458737373352,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.5602,
      "step": 108
    },
    {
      "epoch": 0.5135453474676089,
      "grad_norm": 0.9230842590332031,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.9219,
      "step": 109
    },
    {
      "epoch": 0.5182567726737338,
      "grad_norm": 0.8952105641365051,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.8559,
      "step": 110
    },
    {
      "epoch": 0.5229681978798587,
      "grad_norm": 0.8986338376998901,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.7647,
      "step": 111
    },
    {
      "epoch": 0.5276796230859835,
      "grad_norm": 0.9591684937477112,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.9539,
      "step": 112
    },
    {
      "epoch": 0.5323910482921084,
      "grad_norm": 1.0719199180603027,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 3.2859,
      "step": 113
    },
    {
      "epoch": 0.5371024734982333,
      "grad_norm": 1.1169675588607788,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 3.0344,
      "step": 114
    },
    {
      "epoch": 0.541813898704358,
      "grad_norm": 1.1175668239593506,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.8623,
      "step": 115
    },
    {
      "epoch": 0.5465253239104829,
      "grad_norm": 0.9061518907546997,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.8227,
      "step": 116
    },
    {
      "epoch": 0.5512367491166078,
      "grad_norm": 1.098776936531067,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.7589,
      "step": 117
    },
    {
      "epoch": 0.5559481743227326,
      "grad_norm": 1.1611759662628174,
      "learning_rate": 3.933501846281267e-05,
      "loss": 3.0366,
      "step": 118
    },
    {
      "epoch": 0.5606595995288575,
      "grad_norm": 1.1494617462158203,
      "learning_rate": 3.852880399766243e-05,
      "loss": 3.2172,
      "step": 119
    },
    {
      "epoch": 0.5653710247349824,
      "grad_norm": 1.063607931137085,
      "learning_rate": 3.772572564296005e-05,
      "loss": 3.0673,
      "step": 120
    },
    {
      "epoch": 0.5700824499411072,
      "grad_norm": 1.181087613105774,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 3.0755,
      "step": 121
    },
    {
      "epoch": 0.574793875147232,
      "grad_norm": 1.2603424787521362,
      "learning_rate": 3.612985456190778e-05,
      "loss": 3.1809,
      "step": 122
    },
    {
      "epoch": 0.5795053003533569,
      "grad_norm": 1.3007928133010864,
      "learning_rate": 3.533749813077677e-05,
      "loss": 3.361,
      "step": 123
    },
    {
      "epoch": 0.5842167255594818,
      "grad_norm": 1.364485502243042,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 3.0517,
      "step": 124
    },
    {
      "epoch": 0.5889281507656066,
      "grad_norm": 1.0864237546920776,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 3.2002,
      "step": 125
    },
    {
      "epoch": 0.5936395759717314,
      "grad_norm": 1.2961516380310059,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.2536,
      "step": 126
    },
    {
      "epoch": 0.5983510011778563,
      "grad_norm": 1.274408221244812,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.3761,
      "step": 127
    },
    {
      "epoch": 0.6030624263839811,
      "grad_norm": 1.350342869758606,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.8916,
      "step": 128
    },
    {
      "epoch": 0.607773851590106,
      "grad_norm": 1.1344505548477173,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.5405,
      "step": 129
    },
    {
      "epoch": 0.6124852767962309,
      "grad_norm": 1.3487037420272827,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.9219,
      "step": 130
    },
    {
      "epoch": 0.6171967020023557,
      "grad_norm": 1.1737746000289917,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.4622,
      "step": 131
    },
    {
      "epoch": 0.6219081272084805,
      "grad_norm": 1.2921515703201294,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.7027,
      "step": 132
    },
    {
      "epoch": 0.6266195524146054,
      "grad_norm": 1.0100276470184326,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.502,
      "step": 133
    },
    {
      "epoch": 0.6313309776207303,
      "grad_norm": 1.0417330265045166,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.5498,
      "step": 134
    },
    {
      "epoch": 0.6360424028268551,
      "grad_norm": 1.1279492378234863,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.6156,
      "step": 135
    },
    {
      "epoch": 0.64075382803298,
      "grad_norm": 1.0481832027435303,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.122,
      "step": 136
    },
    {
      "epoch": 0.6454652532391049,
      "grad_norm": 1.0499223470687866,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.5268,
      "step": 137
    },
    {
      "epoch": 0.6501766784452296,
      "grad_norm": 1.1105601787567139,
      "learning_rate": 2.405152131093926e-05,
      "loss": 3.0091,
      "step": 138
    },
    {
      "epoch": 0.6548881036513545,
      "grad_norm": 0.9249874353408813,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.3455,
      "step": 139
    },
    {
      "epoch": 0.6595995288574794,
      "grad_norm": 1.0379689931869507,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.9859,
      "step": 140
    },
    {
      "epoch": 0.6643109540636042,
      "grad_norm": 0.8748549222946167,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.7734,
      "step": 141
    },
    {
      "epoch": 0.6690223792697291,
      "grad_norm": 1.0770591497421265,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.5072,
      "step": 142
    },
    {
      "epoch": 0.673733804475854,
      "grad_norm": 0.9733577966690063,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.5342,
      "step": 143
    },
    {
      "epoch": 0.6784452296819788,
      "grad_norm": 0.9558206796646118,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.1541,
      "step": 144
    },
    {
      "epoch": 0.6831566548881036,
      "grad_norm": 0.9518805146217346,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.7274,
      "step": 145
    },
    {
      "epoch": 0.6878680800942285,
      "grad_norm": 0.9268019199371338,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.3332,
      "step": 146
    },
    {
      "epoch": 0.6925795053003534,
      "grad_norm": 1.0130606889724731,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.7223,
      "step": 147
    },
    {
      "epoch": 0.6972909305064782,
      "grad_norm": 0.8776386380195618,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.61,
      "step": 148
    },
    {
      "epoch": 0.702002355712603,
      "grad_norm": 0.9899702072143555,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 3.0125,
      "step": 149
    },
    {
      "epoch": 0.7067137809187279,
      "grad_norm": 1.0338678359985352,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.7777,
      "step": 150
    },
    {
      "epoch": 0.7067137809187279,
      "eval_loss": 2.7707886695861816,
      "eval_runtime": 5.6127,
      "eval_samples_per_second": 63.784,
      "eval_steps_per_second": 16.035,
      "step": 150
    },
    {
      "epoch": 0.7114252061248527,
      "grad_norm": 0.9890086054801941,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.7659,
      "step": 151
    },
    {
      "epoch": 0.7161366313309776,
      "grad_norm": 0.9277468323707581,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.413,
      "step": 152
    },
    {
      "epoch": 0.7208480565371025,
      "grad_norm": 0.9539352059364319,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.9859,
      "step": 153
    },
    {
      "epoch": 0.7255594817432274,
      "grad_norm": 0.8375805616378784,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.4491,
      "step": 154
    },
    {
      "epoch": 0.7302709069493521,
      "grad_norm": 1.01817786693573,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.9795,
      "step": 155
    },
    {
      "epoch": 0.734982332155477,
      "grad_norm": 0.9246503710746765,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.8541,
      "step": 156
    },
    {
      "epoch": 0.7396937573616019,
      "grad_norm": 0.9075868129730225,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.8676,
      "step": 157
    },
    {
      "epoch": 0.7444051825677267,
      "grad_norm": 0.952269434928894,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.6862,
      "step": 158
    },
    {
      "epoch": 0.7491166077738516,
      "grad_norm": 1.0832068920135498,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 3.1037,
      "step": 159
    },
    {
      "epoch": 0.7538280329799765,
      "grad_norm": 0.954878568649292,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.8786,
      "step": 160
    },
    {
      "epoch": 0.7585394581861012,
      "grad_norm": 0.9489601850509644,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.8139,
      "step": 161
    },
    {
      "epoch": 0.7632508833922261,
      "grad_norm": 1.0556660890579224,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.7036,
      "step": 162
    },
    {
      "epoch": 0.767962308598351,
      "grad_norm": 0.9307450652122498,
      "learning_rate": 9.068759265665384e-06,
      "loss": 3.1366,
      "step": 163
    },
    {
      "epoch": 0.7726737338044759,
      "grad_norm": 1.0892910957336426,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.904,
      "step": 164
    },
    {
      "epoch": 0.7773851590106007,
      "grad_norm": 1.0821411609649658,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.8916,
      "step": 165
    },
    {
      "epoch": 0.7820965842167256,
      "grad_norm": 1.1507179737091064,
      "learning_rate": 7.695237378953223e-06,
      "loss": 3.1046,
      "step": 166
    },
    {
      "epoch": 0.7868080094228505,
      "grad_norm": 1.0595306158065796,
      "learning_rate": 7.260364370723044e-06,
      "loss": 3.088,
      "step": 167
    },
    {
      "epoch": 0.7915194346289752,
      "grad_norm": 1.3114372491836548,
      "learning_rate": 6.837175952121306e-06,
      "loss": 3.2394,
      "step": 168
    },
    {
      "epoch": 0.7962308598351001,
      "grad_norm": 1.2368985414505005,
      "learning_rate": 6.425787818636131e-06,
      "loss": 3.3431,
      "step": 169
    },
    {
      "epoch": 0.800942285041225,
      "grad_norm": 1.3268978595733643,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.9463,
      "step": 170
    },
    {
      "epoch": 0.8056537102473498,
      "grad_norm": 1.1967649459838867,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 3.0073,
      "step": 171
    },
    {
      "epoch": 0.8103651354534747,
      "grad_norm": 1.4040762186050415,
      "learning_rate": 5.263533508961827e-06,
      "loss": 3.494,
      "step": 172
    },
    {
      "epoch": 0.8150765606595995,
      "grad_norm": 1.2713409662246704,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.9515,
      "step": 173
    },
    {
      "epoch": 0.8197879858657244,
      "grad_norm": 1.6765280961990356,
      "learning_rate": 4.549673247541875e-06,
      "loss": 3.4473,
      "step": 174
    },
    {
      "epoch": 0.8244994110718492,
      "grad_norm": 1.618679404258728,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.8997,
      "step": 175
    },
    {
      "epoch": 0.8292108362779741,
      "grad_norm": 0.8593343496322632,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.2641,
      "step": 176
    },
    {
      "epoch": 0.833922261484099,
      "grad_norm": 0.8876267671585083,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.1949,
      "step": 177
    },
    {
      "epoch": 0.8386336866902238,
      "grad_norm": 0.8852311372756958,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.5769,
      "step": 178
    },
    {
      "epoch": 0.8433451118963486,
      "grad_norm": 1.0501874685287476,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.608,
      "step": 179
    },
    {
      "epoch": 0.8480565371024735,
      "grad_norm": 0.9037719964981079,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.5141,
      "step": 180
    },
    {
      "epoch": 0.8527679623085983,
      "grad_norm": 0.9682790040969849,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.5461,
      "step": 181
    },
    {
      "epoch": 0.8574793875147232,
      "grad_norm": 0.8853744864463806,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.6797,
      "step": 182
    },
    {
      "epoch": 0.8621908127208481,
      "grad_norm": 0.9467681050300598,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.3389,
      "step": 183
    },
    {
      "epoch": 0.866902237926973,
      "grad_norm": 0.8484463691711426,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.4208,
      "step": 184
    },
    {
      "epoch": 0.8716136631330977,
      "grad_norm": 0.9523892998695374,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.6641,
      "step": 185
    },
    {
      "epoch": 0.8763250883392226,
      "grad_norm": 0.9082520604133606,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.2748,
      "step": 186
    },
    {
      "epoch": 0.8810365135453475,
      "grad_norm": 1.0969310998916626,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.4917,
      "step": 187
    },
    {
      "epoch": 0.8857479387514723,
      "grad_norm": 0.8878000974655151,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.2523,
      "step": 188
    },
    {
      "epoch": 0.8904593639575972,
      "grad_norm": 0.9266201257705688,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.4628,
      "step": 189
    },
    {
      "epoch": 0.8951707891637221,
      "grad_norm": 0.9008762836456299,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.6504,
      "step": 190
    },
    {
      "epoch": 0.8998822143698468,
      "grad_norm": 0.9340478777885437,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.4853,
      "step": 191
    },
    {
      "epoch": 0.9045936395759717,
      "grad_norm": 0.8499078750610352,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.59,
      "step": 192
    },
    {
      "epoch": 0.9093050647820966,
      "grad_norm": 0.8930243253707886,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.4191,
      "step": 193
    },
    {
      "epoch": 0.9140164899882215,
      "grad_norm": 0.9720921516418457,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.5463,
      "step": 194
    },
    {
      "epoch": 0.9187279151943463,
      "grad_norm": 0.921265721321106,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.2674,
      "step": 195
    },
    {
      "epoch": 0.9234393404004712,
      "grad_norm": 0.8183570504188538,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.4436,
      "step": 196
    },
    {
      "epoch": 0.928150765606596,
      "grad_norm": 1.0162264108657837,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.3652,
      "step": 197
    },
    {
      "epoch": 0.9328621908127208,
      "grad_norm": 0.8739523887634277,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.2952,
      "step": 198
    },
    {
      "epoch": 0.9375736160188457,
      "grad_norm": 0.8950299024581909,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.7807,
      "step": 199
    },
    {
      "epoch": 0.9422850412249706,
      "grad_norm": 0.9458916783332825,
      "learning_rate": 0.0,
      "loss": 2.8684,
      "step": 200
    },
    {
      "epoch": 0.9422850412249706,
      "eval_loss": 2.751559019088745,
      "eval_runtime": 5.5974,
      "eval_samples_per_second": 63.959,
      "eval_steps_per_second": 16.079,
      "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": 7441249896235008.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}