{
  "best_metric": 0.2604656517505646,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.7881773399014779,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003940886699507389,
      "grad_norm": 4.605724811553955,
      "learning_rate": 1e-05,
      "loss": 0.8683,
      "step": 1
    },
    {
      "epoch": 0.003940886699507389,
      "eval_loss": 2.55507493019104,
      "eval_runtime": 31.8336,
      "eval_samples_per_second": 13.445,
      "eval_steps_per_second": 3.361,
      "step": 1
    },
    {
      "epoch": 0.007881773399014778,
      "grad_norm": 5.66981315612793,
      "learning_rate": 2e-05,
      "loss": 1.1386,
      "step": 2
    },
    {
      "epoch": 0.011822660098522168,
      "grad_norm": 5.069392204284668,
      "learning_rate": 3e-05,
      "loss": 1.2048,
      "step": 3
    },
    {
      "epoch": 0.015763546798029555,
      "grad_norm": 5.734151363372803,
      "learning_rate": 4e-05,
      "loss": 1.277,
      "step": 4
    },
    {
      "epoch": 0.019704433497536946,
      "grad_norm": 3.5209782123565674,
      "learning_rate": 5e-05,
      "loss": 1.1337,
      "step": 5
    },
    {
      "epoch": 0.023645320197044337,
      "grad_norm": 3.9932868480682373,
      "learning_rate": 6e-05,
      "loss": 1.1903,
      "step": 6
    },
    {
      "epoch": 0.027586206896551724,
      "grad_norm": 4.947808265686035,
      "learning_rate": 7e-05,
      "loss": 1.1634,
      "step": 7
    },
    {
      "epoch": 0.03152709359605911,
      "grad_norm": 3.9849352836608887,
      "learning_rate": 8e-05,
      "loss": 0.8613,
      "step": 8
    },
    {
      "epoch": 0.035467980295566505,
      "grad_norm": 3.9003443717956543,
      "learning_rate": 9e-05,
      "loss": 0.8209,
      "step": 9
    },
    {
      "epoch": 0.03940886699507389,
      "grad_norm": 3.5782792568206787,
      "learning_rate": 0.0001,
      "loss": 0.8184,
      "step": 10
    },
    {
      "epoch": 0.04334975369458128,
      "grad_norm": 2.893052101135254,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.7108,
      "step": 11
    },
    {
      "epoch": 0.04729064039408867,
      "grad_norm": 3.7223823070526123,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.7857,
      "step": 12
    },
    {
      "epoch": 0.05123152709359606,
      "grad_norm": 3.174710273742676,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.5848,
      "step": 13
    },
    {
      "epoch": 0.05517241379310345,
      "grad_norm": 2.620410442352295,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.4895,
      "step": 14
    },
    {
      "epoch": 0.059113300492610835,
      "grad_norm": 2.8039543628692627,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.3098,
      "step": 15
    },
    {
      "epoch": 0.06305418719211822,
      "grad_norm": 2.8827335834503174,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.3726,
      "step": 16
    },
    {
      "epoch": 0.06699507389162561,
      "grad_norm": 3.061081886291504,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.4516,
      "step": 17
    },
    {
      "epoch": 0.07093596059113301,
      "grad_norm": 3.5195717811584473,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.5393,
      "step": 18
    },
    {
      "epoch": 0.0748768472906404,
      "grad_norm": 3.7049005031585693,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.4818,
      "step": 19
    },
    {
      "epoch": 0.07881773399014778,
      "grad_norm": 2.5747106075286865,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.2806,
      "step": 20
    },
    {
      "epoch": 0.08275862068965517,
      "grad_norm": 2.381589889526367,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.3042,
      "step": 21
    },
    {
      "epoch": 0.08669950738916256,
      "grad_norm": 2.599273920059204,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.2739,
      "step": 22
    },
    {
      "epoch": 0.09064039408866995,
      "grad_norm": 3.358445644378662,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.5456,
      "step": 23
    },
    {
      "epoch": 0.09458128078817735,
      "grad_norm": 2.716853380203247,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.3143,
      "step": 24
    },
    {
      "epoch": 0.09852216748768473,
      "grad_norm": 3.0183498859405518,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.3835,
      "step": 25
    },
    {
      "epoch": 0.10246305418719212,
      "grad_norm": 4.4335761070251465,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.5713,
      "step": 26
    },
    {
      "epoch": 0.10640394088669951,
      "grad_norm": 3.446942090988159,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.3276,
      "step": 27
    },
    {
      "epoch": 0.1103448275862069,
      "grad_norm": 2.479736328125,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.2167,
      "step": 28
    },
    {
      "epoch": 0.11428571428571428,
      "grad_norm": 3.836677074432373,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.4218,
      "step": 29
    },
    {
      "epoch": 0.11822660098522167,
      "grad_norm": 2.7648138999938965,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.2113,
      "step": 30
    },
    {
      "epoch": 0.12216748768472907,
      "grad_norm": 3.2018415927886963,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.4064,
      "step": 31
    },
    {
      "epoch": 0.12610837438423644,
      "grad_norm": 3.6798958778381348,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.3362,
      "step": 32
    },
    {
      "epoch": 0.13004926108374384,
      "grad_norm": 2.7453086376190186,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.2485,
      "step": 33
    },
    {
      "epoch": 0.13399014778325122,
      "grad_norm": 3.7431647777557373,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.417,
      "step": 34
    },
    {
      "epoch": 0.13793103448275862,
      "grad_norm": 5.276279449462891,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.6466,
      "step": 35
    },
    {
      "epoch": 0.14187192118226602,
      "grad_norm": 4.7364325523376465,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.4518,
      "step": 36
    },
    {
      "epoch": 0.1458128078817734,
      "grad_norm": 16.259431838989258,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.0207,
      "step": 37
    },
    {
      "epoch": 0.1497536945812808,
      "grad_norm": 10.356752395629883,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.1488,
      "step": 38
    },
    {
      "epoch": 0.15369458128078817,
      "grad_norm": 6.155438423156738,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.9367,
      "step": 39
    },
    {
      "epoch": 0.15763546798029557,
      "grad_norm": 2.4262003898620605,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.3892,
      "step": 40
    },
    {
      "epoch": 0.16157635467980297,
      "grad_norm": 2.5450832843780518,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.5764,
      "step": 41
    },
    {
      "epoch": 0.16551724137931034,
      "grad_norm": 3.6339566707611084,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.5184,
      "step": 42
    },
    {
      "epoch": 0.16945812807881774,
      "grad_norm": 2.639420509338379,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.4674,
      "step": 43
    },
    {
      "epoch": 0.17339901477832512,
      "grad_norm": 1.4921849966049194,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.3817,
      "step": 44
    },
    {
      "epoch": 0.17733990147783252,
      "grad_norm": 1.7085474729537964,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.4652,
      "step": 45
    },
    {
      "epoch": 0.1812807881773399,
      "grad_norm": 1.2810380458831787,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.3,
      "step": 46
    },
    {
      "epoch": 0.1852216748768473,
      "grad_norm": 1.4004738330841064,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.3166,
      "step": 47
    },
    {
      "epoch": 0.1891625615763547,
      "grad_norm": 1.2935832738876343,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.309,
      "step": 48
    },
    {
      "epoch": 0.19310344827586207,
      "grad_norm": 1.9144483804702759,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.6953,
      "step": 49
    },
    {
      "epoch": 0.19704433497536947,
      "grad_norm": 3.1915669441223145,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.1621,
      "step": 50
    },
    {
      "epoch": 0.19704433497536947,
      "eval_loss": 0.475941926240921,
      "eval_runtime": 31.8814,
      "eval_samples_per_second": 13.425,
      "eval_steps_per_second": 3.356,
      "step": 50
    },
    {
      "epoch": 0.20098522167487684,
      "grad_norm": 0.93958979845047,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.2102,
      "step": 51
    },
    {
      "epoch": 0.20492610837438424,
      "grad_norm": 1.406104564666748,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.3783,
      "step": 52
    },
    {
      "epoch": 0.20886699507389161,
      "grad_norm": 1.484674334526062,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.3237,
      "step": 53
    },
    {
      "epoch": 0.21280788177339902,
      "grad_norm": 1.52517831325531,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.3155,
      "step": 54
    },
    {
      "epoch": 0.21674876847290642,
      "grad_norm": 1.9273359775543213,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.3627,
      "step": 55
    },
    {
      "epoch": 0.2206896551724138,
      "grad_norm": 1.719754934310913,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.456,
      "step": 56
    },
    {
      "epoch": 0.2246305418719212,
      "grad_norm": 1.8616478443145752,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.4314,
      "step": 57
    },
    {
      "epoch": 0.22857142857142856,
      "grad_norm": 1.5278351306915283,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.2611,
      "step": 58
    },
    {
      "epoch": 0.23251231527093597,
      "grad_norm": 1.606553077697754,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.3329,
      "step": 59
    },
    {
      "epoch": 0.23645320197044334,
      "grad_norm": 1.8797303438186646,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.4514,
      "step": 60
    },
    {
      "epoch": 0.24039408866995074,
      "grad_norm": 1.6210929155349731,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.4083,
      "step": 61
    },
    {
      "epoch": 0.24433497536945814,
      "grad_norm": 1.8047562837600708,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.4039,
      "step": 62
    },
    {
      "epoch": 0.2482758620689655,
      "grad_norm": 2.107999563217163,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.4446,
      "step": 63
    },
    {
      "epoch": 0.2522167487684729,
      "grad_norm": 2.1669254302978516,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.3722,
      "step": 64
    },
    {
      "epoch": 0.2561576354679803,
      "grad_norm": 2.6170012950897217,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.5331,
      "step": 65
    },
    {
      "epoch": 0.2600985221674877,
      "grad_norm": 1.4068994522094727,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.1781,
      "step": 66
    },
    {
      "epoch": 0.26403940886699506,
      "grad_norm": 2.1255016326904297,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.314,
      "step": 67
    },
    {
      "epoch": 0.26798029556650244,
      "grad_norm": 1.9393290281295776,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.3736,
      "step": 68
    },
    {
      "epoch": 0.27192118226600986,
      "grad_norm": 1.9570231437683105,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.3736,
      "step": 69
    },
    {
      "epoch": 0.27586206896551724,
      "grad_norm": 1.8715764284133911,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.2336,
      "step": 70
    },
    {
      "epoch": 0.2798029556650246,
      "grad_norm": 2.1200549602508545,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.3482,
      "step": 71
    },
    {
      "epoch": 0.28374384236453204,
      "grad_norm": 1.9308780431747437,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.3058,
      "step": 72
    },
    {
      "epoch": 0.2876847290640394,
      "grad_norm": 1.8106579780578613,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.2779,
      "step": 73
    },
    {
      "epoch": 0.2916256157635468,
      "grad_norm": 2.214245557785034,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.3715,
      "step": 74
    },
    {
      "epoch": 0.2955665024630542,
      "grad_norm": 2.6384432315826416,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.336,
      "step": 75
    },
    {
      "epoch": 0.2995073891625616,
      "grad_norm": 2.050410509109497,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.273,
      "step": 76
    },
    {
      "epoch": 0.30344827586206896,
      "grad_norm": 3.054478645324707,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.3908,
      "step": 77
    },
    {
      "epoch": 0.30738916256157633,
      "grad_norm": 2.5260796546936035,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.2106,
      "step": 78
    },
    {
      "epoch": 0.31133004926108376,
      "grad_norm": 2.7985095977783203,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.4464,
      "step": 79
    },
    {
      "epoch": 0.31527093596059114,
      "grad_norm": 1.4435943365097046,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.1433,
      "step": 80
    },
    {
      "epoch": 0.3192118226600985,
      "grad_norm": 2.0289454460144043,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.1928,
      "step": 81
    },
    {
      "epoch": 0.32315270935960594,
      "grad_norm": 2.266352891921997,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.2891,
      "step": 82
    },
    {
      "epoch": 0.3270935960591133,
      "grad_norm": 2.526263475418091,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.3348,
      "step": 83
    },
    {
      "epoch": 0.3310344827586207,
      "grad_norm": 3.377316951751709,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.2571,
      "step": 84
    },
    {
      "epoch": 0.33497536945812806,
      "grad_norm": 4.159484386444092,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.3926,
      "step": 85
    },
    {
      "epoch": 0.3389162561576355,
      "grad_norm": 4.5916900634765625,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.6609,
      "step": 86
    },
    {
      "epoch": 0.34285714285714286,
      "grad_norm": 7.815732955932617,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.8648,
      "step": 87
    },
    {
      "epoch": 0.34679802955665023,
      "grad_norm": 7.154582500457764,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.6757,
      "step": 88
    },
    {
      "epoch": 0.35073891625615766,
      "grad_norm": 2.3498964309692383,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.3888,
      "step": 89
    },
    {
      "epoch": 0.35467980295566504,
      "grad_norm": 2.000180721282959,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.4885,
      "step": 90
    },
    {
      "epoch": 0.3586206896551724,
      "grad_norm": 1.2019802331924438,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.2243,
      "step": 91
    },
    {
      "epoch": 0.3625615763546798,
      "grad_norm": 1.1074315309524536,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.1624,
      "step": 92
    },
    {
      "epoch": 0.3665024630541872,
      "grad_norm": 1.6011695861816406,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.4666,
      "step": 93
    },
    {
      "epoch": 0.3704433497536946,
      "grad_norm": 0.920390248298645,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.2325,
      "step": 94
    },
    {
      "epoch": 0.37438423645320196,
      "grad_norm": 1.4502712488174438,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.4634,
      "step": 95
    },
    {
      "epoch": 0.3783251231527094,
      "grad_norm": 1.074723243713379,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.2225,
      "step": 96
    },
    {
      "epoch": 0.38226600985221676,
      "grad_norm": 0.7659882307052612,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.1655,
      "step": 97
    },
    {
      "epoch": 0.38620689655172413,
      "grad_norm": 0.860362708568573,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.1621,
      "step": 98
    },
    {
      "epoch": 0.3901477832512315,
      "grad_norm": 1.1806848049163818,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.372,
      "step": 99
    },
    {
      "epoch": 0.39408866995073893,
      "grad_norm": 1.4738397598266602,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.4585,
      "step": 100
    },
    {
      "epoch": 0.39408866995073893,
      "eval_loss": 0.434915691614151,
      "eval_runtime": 32.0682,
      "eval_samples_per_second": 13.347,
      "eval_steps_per_second": 3.337,
      "step": 100
    },
    {
      "epoch": 0.3980295566502463,
      "grad_norm": 0.9950315952301025,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.3496,
      "step": 101
    },
    {
      "epoch": 0.4019704433497537,
      "grad_norm": 1.4049850702285767,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.4795,
      "step": 102
    },
    {
      "epoch": 0.4059113300492611,
      "grad_norm": 1.1271802186965942,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.2861,
      "step": 103
    },
    {
      "epoch": 0.4098522167487685,
      "grad_norm": 1.2322853803634644,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.3483,
      "step": 104
    },
    {
      "epoch": 0.41379310344827586,
      "grad_norm": 1.181979775428772,
      "learning_rate": 5e-05,
      "loss": 0.2755,
      "step": 105
    },
    {
      "epoch": 0.41773399014778323,
      "grad_norm": 1.2165412902832031,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.3267,
      "step": 106
    },
    {
      "epoch": 0.42167487684729066,
      "grad_norm": 1.3426710367202759,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.3285,
      "step": 107
    },
    {
      "epoch": 0.42561576354679803,
      "grad_norm": 1.3296332359313965,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.2895,
      "step": 108
    },
    {
      "epoch": 0.4295566502463054,
      "grad_norm": 1.6037664413452148,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.4341,
      "step": 109
    },
    {
      "epoch": 0.43349753694581283,
      "grad_norm": 1.3613358736038208,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.2708,
      "step": 110
    },
    {
      "epoch": 0.4374384236453202,
      "grad_norm": 1.8670945167541504,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.4552,
      "step": 111
    },
    {
      "epoch": 0.4413793103448276,
      "grad_norm": 1.343572974205017,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.2287,
      "step": 112
    },
    {
      "epoch": 0.44532019704433495,
      "grad_norm": 1.9313302040100098,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.4153,
      "step": 113
    },
    {
      "epoch": 0.4492610837438424,
      "grad_norm": 1.4682947397232056,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.3328,
      "step": 114
    },
    {
      "epoch": 0.45320197044334976,
      "grad_norm": 1.764954924583435,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.3722,
      "step": 115
    },
    {
      "epoch": 0.45714285714285713,
      "grad_norm": 1.8122596740722656,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.3311,
      "step": 116
    },
    {
      "epoch": 0.46108374384236456,
      "grad_norm": 1.6849801540374756,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.3091,
      "step": 117
    },
    {
      "epoch": 0.46502463054187193,
      "grad_norm": 0.9165311455726624,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.101,
      "step": 118
    },
    {
      "epoch": 0.4689655172413793,
      "grad_norm": 2.010204553604126,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.3682,
      "step": 119
    },
    {
      "epoch": 0.4729064039408867,
      "grad_norm": 1.6442277431488037,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.2343,
      "step": 120
    },
    {
      "epoch": 0.4768472906403941,
      "grad_norm": 1.557530164718628,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.1875,
      "step": 121
    },
    {
      "epoch": 0.4807881773399015,
      "grad_norm": 1.488112211227417,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.1714,
      "step": 122
    },
    {
      "epoch": 0.48472906403940885,
      "grad_norm": 1.8102607727050781,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.1938,
      "step": 123
    },
    {
      "epoch": 0.4886699507389163,
      "grad_norm": 1.8396623134613037,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.1607,
      "step": 124
    },
    {
      "epoch": 0.49261083743842365,
      "grad_norm": 2.4509646892547607,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.2625,
      "step": 125
    },
    {
      "epoch": 0.496551724137931,
      "grad_norm": 1.8777393102645874,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.2728,
      "step": 126
    },
    {
      "epoch": 0.5004926108374385,
      "grad_norm": 2.6126911640167236,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.2017,
      "step": 127
    },
    {
      "epoch": 0.5044334975369458,
      "grad_norm": 2.2730720043182373,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.2649,
      "step": 128
    },
    {
      "epoch": 0.5083743842364532,
      "grad_norm": 1.7744430303573608,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.1378,
      "step": 129
    },
    {
      "epoch": 0.5123152709359606,
      "grad_norm": 2.1662611961364746,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.1538,
      "step": 130
    },
    {
      "epoch": 0.516256157635468,
      "grad_norm": 2.4438276290893555,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.237,
      "step": 131
    },
    {
      "epoch": 0.5201970443349754,
      "grad_norm": 1.7531158924102783,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.1383,
      "step": 132
    },
    {
      "epoch": 0.5241379310344828,
      "grad_norm": 2.0699450969696045,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.1534,
      "step": 133
    },
    {
      "epoch": 0.5280788177339901,
      "grad_norm": 3.246652126312256,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.2998,
      "step": 134
    },
    {
      "epoch": 0.5320197044334976,
      "grad_norm": 6.806807518005371,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.6264,
      "step": 135
    },
    {
      "epoch": 0.5359605911330049,
      "grad_norm": 3.666675567626953,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.4792,
      "step": 136
    },
    {
      "epoch": 0.5399014778325123,
      "grad_norm": 2.5924410820007324,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.525,
      "step": 137
    },
    {
      "epoch": 0.5438423645320197,
      "grad_norm": 1.7887628078460693,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.2162,
      "step": 138
    },
    {
      "epoch": 0.547783251231527,
      "grad_norm": 2.560127019882202,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.4434,
      "step": 139
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 1.782257080078125,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.2569,
      "step": 140
    },
    {
      "epoch": 0.5556650246305419,
      "grad_norm": 1.762956976890564,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.277,
      "step": 141
    },
    {
      "epoch": 0.5596059113300492,
      "grad_norm": 1.4035093784332275,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.167,
      "step": 142
    },
    {
      "epoch": 0.5635467980295567,
      "grad_norm": 1.0826499462127686,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.0938,
      "step": 143
    },
    {
      "epoch": 0.5674876847290641,
      "grad_norm": 1.010370135307312,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.2034,
      "step": 144
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 1.454880952835083,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.3517,
      "step": 145
    },
    {
      "epoch": 0.5753694581280788,
      "grad_norm": 1.653377652168274,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.4822,
      "step": 146
    },
    {
      "epoch": 0.5793103448275863,
      "grad_norm": 1.2127748727798462,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.2397,
      "step": 147
    },
    {
      "epoch": 0.5832512315270936,
      "grad_norm": 1.2672176361083984,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.3219,
      "step": 148
    },
    {
      "epoch": 0.587192118226601,
      "grad_norm": 1.0319257974624634,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.2282,
      "step": 149
    },
    {
      "epoch": 0.5911330049261084,
      "grad_norm": 2.236492395401001,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.7111,
      "step": 150
    },
    {
      "epoch": 0.5911330049261084,
      "eval_loss": 0.2746962308883667,
      "eval_runtime": 31.9121,
      "eval_samples_per_second": 13.412,
      "eval_steps_per_second": 3.353,
      "step": 150
    },
    {
      "epoch": 0.5950738916256157,
      "grad_norm": 1.3970447778701782,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.3817,
      "step": 151
    },
    {
      "epoch": 0.5990147783251232,
      "grad_norm": 1.8816144466400146,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.2996,
      "step": 152
    },
    {
      "epoch": 0.6029556650246305,
      "grad_norm": 1.0152655839920044,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.2239,
      "step": 153
    },
    {
      "epoch": 0.6068965517241379,
      "grad_norm": 1.2198946475982666,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.3124,
      "step": 154
    },
    {
      "epoch": 0.6108374384236454,
      "grad_norm": 1.2820860147476196,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.3085,
      "step": 155
    },
    {
      "epoch": 0.6147783251231527,
      "grad_norm": 1.4625478982925415,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.4283,
      "step": 156
    },
    {
      "epoch": 0.6187192118226601,
      "grad_norm": 1.2759764194488525,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.3072,
      "step": 157
    },
    {
      "epoch": 0.6226600985221675,
      "grad_norm": 1.1684573888778687,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.2927,
      "step": 158
    },
    {
      "epoch": 0.6266009852216748,
      "grad_norm": 1.052973985671997,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.2021,
      "step": 159
    },
    {
      "epoch": 0.6305418719211823,
      "grad_norm": 1.206365704536438,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.196,
      "step": 160
    },
    {
      "epoch": 0.6344827586206897,
      "grad_norm": 1.438977837562561,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.3488,
      "step": 161
    },
    {
      "epoch": 0.638423645320197,
      "grad_norm": 1.5919150114059448,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.3316,
      "step": 162
    },
    {
      "epoch": 0.6423645320197044,
      "grad_norm": 1.5834957361221313,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.3044,
      "step": 163
    },
    {
      "epoch": 0.6463054187192119,
      "grad_norm": 1.3561503887176514,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.2188,
      "step": 164
    },
    {
      "epoch": 0.6502463054187192,
      "grad_norm": 1.728492259979248,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.3912,
      "step": 165
    },
    {
      "epoch": 0.6541871921182266,
      "grad_norm": 1.7723878622055054,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.2262,
      "step": 166
    },
    {
      "epoch": 0.6581280788177339,
      "grad_norm": 2.1955647468566895,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.2562,
      "step": 167
    },
    {
      "epoch": 0.6620689655172414,
      "grad_norm": 1.6165755987167358,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.3041,
      "step": 168
    },
    {
      "epoch": 0.6660098522167488,
      "grad_norm": 1.8477954864501953,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.2558,
      "step": 169
    },
    {
      "epoch": 0.6699507389162561,
      "grad_norm": 1.8062127828598022,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.3622,
      "step": 170
    },
    {
      "epoch": 0.6738916256157635,
      "grad_norm": 1.1588163375854492,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.1784,
      "step": 171
    },
    {
      "epoch": 0.677832512315271,
      "grad_norm": 2.1964800357818604,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.3646,
      "step": 172
    },
    {
      "epoch": 0.6817733990147783,
      "grad_norm": 1.7287336587905884,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.2115,
      "step": 173
    },
    {
      "epoch": 0.6857142857142857,
      "grad_norm": 1.1784448623657227,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.1527,
      "step": 174
    },
    {
      "epoch": 0.6896551724137931,
      "grad_norm": 2.3815059661865234,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.3073,
      "step": 175
    },
    {
      "epoch": 0.6935960591133005,
      "grad_norm": 1.8565568923950195,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.3362,
      "step": 176
    },
    {
      "epoch": 0.6975369458128079,
      "grad_norm": 1.7195888757705688,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.1579,
      "step": 177
    },
    {
      "epoch": 0.7014778325123153,
      "grad_norm": 2.112945318222046,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.2386,
      "step": 178
    },
    {
      "epoch": 0.7054187192118226,
      "grad_norm": 2.121370553970337,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.1858,
      "step": 179
    },
    {
      "epoch": 0.7093596059113301,
      "grad_norm": 2.0235986709594727,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.2022,
      "step": 180
    },
    {
      "epoch": 0.7133004926108374,
      "grad_norm": 3.0182576179504395,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.2456,
      "step": 181
    },
    {
      "epoch": 0.7172413793103448,
      "grad_norm": 2.526031494140625,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.2251,
      "step": 182
    },
    {
      "epoch": 0.7211822660098522,
      "grad_norm": 3.9601526260375977,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.3355,
      "step": 183
    },
    {
      "epoch": 0.7251231527093596,
      "grad_norm": 3.9802451133728027,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.6555,
      "step": 184
    },
    {
      "epoch": 0.729064039408867,
      "grad_norm": 4.811775207519531,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.5472,
      "step": 185
    },
    {
      "epoch": 0.7330049261083744,
      "grad_norm": 5.148604869842529,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.5411,
      "step": 186
    },
    {
      "epoch": 0.7369458128078817,
      "grad_norm": 1.52059805393219,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.337,
      "step": 187
    },
    {
      "epoch": 0.7408866995073892,
      "grad_norm": 1.2510405778884888,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.2873,
      "step": 188
    },
    {
      "epoch": 0.7448275862068966,
      "grad_norm": 0.9079121351242065,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.1653,
      "step": 189
    },
    {
      "epoch": 0.7487684729064039,
      "grad_norm": 0.7605425715446472,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.1099,
      "step": 190
    },
    {
      "epoch": 0.7527093596059113,
      "grad_norm": 1.1136564016342163,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.2461,
      "step": 191
    },
    {
      "epoch": 0.7566502463054188,
      "grad_norm": 1.448406457901001,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.3553,
      "step": 192
    },
    {
      "epoch": 0.7605911330049261,
      "grad_norm": 1.0065648555755615,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.1981,
      "step": 193
    },
    {
      "epoch": 0.7645320197044335,
      "grad_norm": 1.0435380935668945,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.2067,
      "step": 194
    },
    {
      "epoch": 0.7684729064039408,
      "grad_norm": 1.0339488983154297,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.2395,
      "step": 195
    },
    {
      "epoch": 0.7724137931034483,
      "grad_norm": 0.9154281616210938,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.2264,
      "step": 196
    },
    {
      "epoch": 0.7763546798029557,
      "grad_norm": 0.9020112752914429,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.1694,
      "step": 197
    },
    {
      "epoch": 0.780295566502463,
      "grad_norm": 1.000407099723816,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.1575,
      "step": 198
    },
    {
      "epoch": 0.7842364532019704,
      "grad_norm": 1.004696011543274,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.2374,
      "step": 199
    },
    {
      "epoch": 0.7881773399014779,
      "grad_norm": 1.763048529624939,
      "learning_rate": 0.0,
      "loss": 0.5752,
      "step": 200
    },
    {
      "epoch": 0.7881773399014779,
      "eval_loss": 0.2604656517505646,
      "eval_runtime": 31.8962,
      "eval_samples_per_second": 13.419,
      "eval_steps_per_second": 3.355,
      "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": 1.42439008960512e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}