{
  "best_metric": 0.06954215466976166,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 2.867383512544803,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.014336917562724014,
      "grad_norm": 4.352110862731934,
      "learning_rate": 1e-05,
      "loss": 1.4088,
      "step": 1
    },
    {
      "epoch": 0.014336917562724014,
      "eval_loss": 2.468245267868042,
      "eval_runtime": 8.8557,
      "eval_samples_per_second": 13.325,
      "eval_steps_per_second": 3.388,
      "step": 1
    },
    {
      "epoch": 0.02867383512544803,
      "grad_norm": 4.485035419464111,
      "learning_rate": 2e-05,
      "loss": 1.561,
      "step": 2
    },
    {
      "epoch": 0.043010752688172046,
      "grad_norm": 4.266824245452881,
      "learning_rate": 3e-05,
      "loss": 1.4887,
      "step": 3
    },
    {
      "epoch": 0.05734767025089606,
      "grad_norm": 4.427830696105957,
      "learning_rate": 4e-05,
      "loss": 1.513,
      "step": 4
    },
    {
      "epoch": 0.07168458781362007,
      "grad_norm": 3.4474122524261475,
      "learning_rate": 5e-05,
      "loss": 1.2289,
      "step": 5
    },
    {
      "epoch": 0.08602150537634409,
      "grad_norm": 4.336211204528809,
      "learning_rate": 6e-05,
      "loss": 1.2069,
      "step": 6
    },
    {
      "epoch": 0.1003584229390681,
      "grad_norm": 2.924722909927368,
      "learning_rate": 7e-05,
      "loss": 1.2443,
      "step": 7
    },
    {
      "epoch": 0.11469534050179211,
      "grad_norm": 2.8629510402679443,
      "learning_rate": 8e-05,
      "loss": 0.8519,
      "step": 8
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 2.833808422088623,
      "learning_rate": 9e-05,
      "loss": 1.0116,
      "step": 9
    },
    {
      "epoch": 0.14336917562724014,
      "grad_norm": 3.079848289489746,
      "learning_rate": 0.0001,
      "loss": 1.0873,
      "step": 10
    },
    {
      "epoch": 0.15770609318996415,
      "grad_norm": 3.329118251800537,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.1452,
      "step": 11
    },
    {
      "epoch": 0.17204301075268819,
      "grad_norm": 2.5007457733154297,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.9358,
      "step": 12
    },
    {
      "epoch": 0.1863799283154122,
      "grad_norm": 3.5425992012023926,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.7756,
      "step": 13
    },
    {
      "epoch": 0.2007168458781362,
      "grad_norm": 4.428173542022705,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.7461,
      "step": 14
    },
    {
      "epoch": 0.21505376344086022,
      "grad_norm": 4.069889545440674,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.8013,
      "step": 15
    },
    {
      "epoch": 0.22939068100358423,
      "grad_norm": 6.825600624084473,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.2648,
      "step": 16
    },
    {
      "epoch": 0.24372759856630824,
      "grad_norm": 5.172757148742676,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.1637,
      "step": 17
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 4.992447853088379,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.2698,
      "step": 18
    },
    {
      "epoch": 0.2724014336917563,
      "grad_norm": 2.369986057281494,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.2446,
      "step": 19
    },
    {
      "epoch": 0.2867383512544803,
      "grad_norm": 2.5273282527923584,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.1342,
      "step": 20
    },
    {
      "epoch": 0.3010752688172043,
      "grad_norm": 2.5581085681915283,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.0049,
      "step": 21
    },
    {
      "epoch": 0.3154121863799283,
      "grad_norm": 1.9571079015731812,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.0806,
      "step": 22
    },
    {
      "epoch": 0.32974910394265233,
      "grad_norm": 1.7763464450836182,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.7699,
      "step": 23
    },
    {
      "epoch": 0.34408602150537637,
      "grad_norm": 1.835694432258606,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.6863,
      "step": 24
    },
    {
      "epoch": 0.35842293906810035,
      "grad_norm": 2.2447690963745117,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.7966,
      "step": 25
    },
    {
      "epoch": 0.3727598566308244,
      "grad_norm": 2.7908217906951904,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.2259,
      "step": 26
    },
    {
      "epoch": 0.3870967741935484,
      "grad_norm": 2.485395908355713,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.9482,
      "step": 27
    },
    {
      "epoch": 0.4014336917562724,
      "grad_norm": 2.3916211128234863,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.7454,
      "step": 28
    },
    {
      "epoch": 0.4157706093189964,
      "grad_norm": 5.038878917694092,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.8929,
      "step": 29
    },
    {
      "epoch": 0.43010752688172044,
      "grad_norm": 2.351196765899658,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.9044,
      "step": 30
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 3.2885828018188477,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.1265,
      "step": 31
    },
    {
      "epoch": 0.45878136200716846,
      "grad_norm": 5.5460429191589355,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.7935,
      "step": 32
    },
    {
      "epoch": 0.4731182795698925,
      "grad_norm": 4.941125392913818,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.2779,
      "step": 33
    },
    {
      "epoch": 0.4874551971326165,
      "grad_norm": 4.175712585449219,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.0494,
      "step": 34
    },
    {
      "epoch": 0.5017921146953405,
      "grad_norm": 1.914487361907959,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.046,
      "step": 35
    },
    {
      "epoch": 0.5161290322580645,
      "grad_norm": 2.078220844268799,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.7801,
      "step": 36
    },
    {
      "epoch": 0.5304659498207885,
      "grad_norm": 2.3253490924835205,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.0495,
      "step": 37
    },
    {
      "epoch": 0.5448028673835126,
      "grad_norm": 1.8537297248840332,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.9072,
      "step": 38
    },
    {
      "epoch": 0.5591397849462365,
      "grad_norm": 1.554869532585144,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.5753,
      "step": 39
    },
    {
      "epoch": 0.5734767025089605,
      "grad_norm": 1.505541205406189,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.7123,
      "step": 40
    },
    {
      "epoch": 0.5878136200716846,
      "grad_norm": 2.7118098735809326,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.5844,
      "step": 41
    },
    {
      "epoch": 0.6021505376344086,
      "grad_norm": 1.9012823104858398,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.6834,
      "step": 42
    },
    {
      "epoch": 0.6164874551971327,
      "grad_norm": 2.6380529403686523,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.9228,
      "step": 43
    },
    {
      "epoch": 0.6308243727598566,
      "grad_norm": 2.996264934539795,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.75,
      "step": 44
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 2.810377597808838,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.7923,
      "step": 45
    },
    {
      "epoch": 0.6594982078853047,
      "grad_norm": 2.8084309101104736,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.4627,
      "step": 46
    },
    {
      "epoch": 0.6738351254480287,
      "grad_norm": 3.203434944152832,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.7908,
      "step": 47
    },
    {
      "epoch": 0.6881720430107527,
      "grad_norm": 2.6524767875671387,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.7677,
      "step": 48
    },
    {
      "epoch": 0.7025089605734767,
      "grad_norm": 3.983142137527466,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.7745,
      "step": 49
    },
    {
      "epoch": 0.7168458781362007,
      "grad_norm": 5.699389934539795,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.1839,
      "step": 50
    },
    {
      "epoch": 0.7168458781362007,
      "eval_loss": 0.6850312352180481,
      "eval_runtime": 9.0431,
      "eval_samples_per_second": 13.049,
      "eval_steps_per_second": 3.317,
      "step": 50
    },
    {
      "epoch": 0.7311827956989247,
      "grad_norm": 6.081845760345459,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.9081,
      "step": 51
    },
    {
      "epoch": 0.7455197132616488,
      "grad_norm": 2.100945234298706,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.9846,
      "step": 52
    },
    {
      "epoch": 0.7598566308243727,
      "grad_norm": 1.7365548610687256,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.7357,
      "step": 53
    },
    {
      "epoch": 0.7741935483870968,
      "grad_norm": 2.0256400108337402,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.8173,
      "step": 54
    },
    {
      "epoch": 0.7885304659498208,
      "grad_norm": 2.565948247909546,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.7689,
      "step": 55
    },
    {
      "epoch": 0.8028673835125448,
      "grad_norm": 1.4809150695800781,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.4844,
      "step": 56
    },
    {
      "epoch": 0.8172043010752689,
      "grad_norm": 1.7772880792617798,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.7141,
      "step": 57
    },
    {
      "epoch": 0.8315412186379928,
      "grad_norm": 1.4801161289215088,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.3704,
      "step": 58
    },
    {
      "epoch": 0.8458781362007168,
      "grad_norm": 1.3622130155563354,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.3023,
      "step": 59
    },
    {
      "epoch": 0.8602150537634409,
      "grad_norm": 1.783310055732727,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.5713,
      "step": 60
    },
    {
      "epoch": 0.8745519713261649,
      "grad_norm": 1.728874683380127,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.3739,
      "step": 61
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 1.7142260074615479,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.3124,
      "step": 62
    },
    {
      "epoch": 0.9032258064516129,
      "grad_norm": 2.797607183456421,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.5819,
      "step": 63
    },
    {
      "epoch": 0.9175627240143369,
      "grad_norm": 3.8919594287872314,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.0355,
      "step": 64
    },
    {
      "epoch": 0.931899641577061,
      "grad_norm": 2.8246655464172363,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.5985,
      "step": 65
    },
    {
      "epoch": 0.946236559139785,
      "grad_norm": 3.3262031078338623,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.6244,
      "step": 66
    },
    {
      "epoch": 0.9605734767025089,
      "grad_norm": 6.562505722045898,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.5774,
      "step": 67
    },
    {
      "epoch": 0.974910394265233,
      "grad_norm": 8.12563705444336,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.0758,
      "step": 68
    },
    {
      "epoch": 0.989247311827957,
      "grad_norm": 1.5006723403930664,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.4908,
      "step": 69
    },
    {
      "epoch": 1.003584229390681,
      "grad_norm": 2.735072612762451,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.6535,
      "step": 70
    },
    {
      "epoch": 1.017921146953405,
      "grad_norm": 1.5043795108795166,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.7292,
      "step": 71
    },
    {
      "epoch": 1.032258064516129,
      "grad_norm": 1.4688665866851807,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.3972,
      "step": 72
    },
    {
      "epoch": 1.0465949820788532,
      "grad_norm": 1.337483286857605,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.3203,
      "step": 73
    },
    {
      "epoch": 1.060931899641577,
      "grad_norm": 2.1305058002471924,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.2926,
      "step": 74
    },
    {
      "epoch": 1.075268817204301,
      "grad_norm": 1.156721591949463,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.1418,
      "step": 75
    },
    {
      "epoch": 1.0896057347670252,
      "grad_norm": 1.1064209938049316,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.1707,
      "step": 76
    },
    {
      "epoch": 1.103942652329749,
      "grad_norm": 0.7985288500785828,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.0836,
      "step": 77
    },
    {
      "epoch": 1.118279569892473,
      "grad_norm": 4.694607257843018,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.0965,
      "step": 78
    },
    {
      "epoch": 1.1326164874551972,
      "grad_norm": 1.8385326862335205,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.2258,
      "step": 79
    },
    {
      "epoch": 1.146953405017921,
      "grad_norm": 1.635382890701294,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.1219,
      "step": 80
    },
    {
      "epoch": 1.1612903225806452,
      "grad_norm": 1.9358904361724854,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.15,
      "step": 81
    },
    {
      "epoch": 1.1756272401433692,
      "grad_norm": 2.070408821105957,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.1857,
      "step": 82
    },
    {
      "epoch": 1.189964157706093,
      "grad_norm": 2.193985939025879,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.2241,
      "step": 83
    },
    {
      "epoch": 1.2043010752688172,
      "grad_norm": 0.8806750178337097,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.061,
      "step": 84
    },
    {
      "epoch": 1.2186379928315412,
      "grad_norm": 2.840721845626831,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.3011,
      "step": 85
    },
    {
      "epoch": 1.2329749103942653,
      "grad_norm": 2.4514873027801514,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.2026,
      "step": 86
    },
    {
      "epoch": 1.2473118279569892,
      "grad_norm": 3.0623319149017334,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.3004,
      "step": 87
    },
    {
      "epoch": 1.2616487455197132,
      "grad_norm": 2.500314950942993,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.4167,
      "step": 88
    },
    {
      "epoch": 1.2759856630824373,
      "grad_norm": 1.8271678686141968,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.2206,
      "step": 89
    },
    {
      "epoch": 1.2903225806451613,
      "grad_norm": 3.1398000717163086,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.4415,
      "step": 90
    },
    {
      "epoch": 1.3046594982078852,
      "grad_norm": 2.0039515495300293,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.2216,
      "step": 91
    },
    {
      "epoch": 1.3189964157706093,
      "grad_norm": 1.2938251495361328,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.1492,
      "step": 92
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 1.488156795501709,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.0973,
      "step": 93
    },
    {
      "epoch": 1.3476702508960574,
      "grad_norm": 1.7337243556976318,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.0838,
      "step": 94
    },
    {
      "epoch": 1.3620071684587813,
      "grad_norm": 2.756385326385498,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.1313,
      "step": 95
    },
    {
      "epoch": 1.3763440860215055,
      "grad_norm": 3.0886261463165283,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.0979,
      "step": 96
    },
    {
      "epoch": 1.3906810035842294,
      "grad_norm": 1.405226469039917,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.1325,
      "step": 97
    },
    {
      "epoch": 1.4050179211469533,
      "grad_norm": 1.9139821529388428,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.1189,
      "step": 98
    },
    {
      "epoch": 1.4193548387096775,
      "grad_norm": 2.0083534717559814,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.1132,
      "step": 99
    },
    {
      "epoch": 1.4336917562724014,
      "grad_norm": 1.4845714569091797,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.0941,
      "step": 100
    },
    {
      "epoch": 1.4336917562724014,
      "eval_loss": 0.25433480739593506,
      "eval_runtime": 9.0253,
      "eval_samples_per_second": 13.074,
      "eval_steps_per_second": 3.324,
      "step": 100
    },
    {
      "epoch": 1.4480286738351253,
      "grad_norm": 2.1107029914855957,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.0907,
      "step": 101
    },
    {
      "epoch": 1.4623655913978495,
      "grad_norm": 1.7067404985427856,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.0527,
      "step": 102
    },
    {
      "epoch": 1.4767025089605734,
      "grad_norm": 3.6597208976745605,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.3302,
      "step": 103
    },
    {
      "epoch": 1.4910394265232976,
      "grad_norm": 4.162262916564941,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.2532,
      "step": 104
    },
    {
      "epoch": 1.5053763440860215,
      "grad_norm": 1.1525654792785645,
      "learning_rate": 5e-05,
      "loss": 0.3028,
      "step": 105
    },
    {
      "epoch": 1.5197132616487457,
      "grad_norm": 1.064614176750183,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.1768,
      "step": 106
    },
    {
      "epoch": 1.5340501792114696,
      "grad_norm": 1.2395060062408447,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.2393,
      "step": 107
    },
    {
      "epoch": 1.5483870967741935,
      "grad_norm": 0.8635635375976562,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.1047,
      "step": 108
    },
    {
      "epoch": 1.5627240143369177,
      "grad_norm": 0.6571157574653625,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.0573,
      "step": 109
    },
    {
      "epoch": 1.5770609318996416,
      "grad_norm": 0.8516560196876526,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.0907,
      "step": 110
    },
    {
      "epoch": 1.5913978494623655,
      "grad_norm": 1.9296753406524658,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.1394,
      "step": 111
    },
    {
      "epoch": 1.6057347670250897,
      "grad_norm": 0.7486221790313721,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.0604,
      "step": 112
    },
    {
      "epoch": 1.6200716845878136,
      "grad_norm": 0.767206609249115,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.0552,
      "step": 113
    },
    {
      "epoch": 1.6344086021505375,
      "grad_norm": 1.1909141540527344,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.0516,
      "step": 114
    },
    {
      "epoch": 1.6487455197132617,
      "grad_norm": 2.412421703338623,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.0754,
      "step": 115
    },
    {
      "epoch": 1.6630824372759858,
      "grad_norm": 2.385477066040039,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.1189,
      "step": 116
    },
    {
      "epoch": 1.6774193548387095,
      "grad_norm": 1.372868299484253,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.0898,
      "step": 117
    },
    {
      "epoch": 1.6917562724014337,
      "grad_norm": 2.122359037399292,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.109,
      "step": 118
    },
    {
      "epoch": 1.7060931899641578,
      "grad_norm": 2.169341802597046,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.0898,
      "step": 119
    },
    {
      "epoch": 1.7204301075268817,
      "grad_norm": 3.5736114978790283,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.1211,
      "step": 120
    },
    {
      "epoch": 1.7347670250896057,
      "grad_norm": 0.7986944317817688,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.0362,
      "step": 121
    },
    {
      "epoch": 1.7491039426523298,
      "grad_norm": 0.9726013541221619,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.1572,
      "step": 122
    },
    {
      "epoch": 1.7634408602150538,
      "grad_norm": 0.9983823895454407,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.0835,
      "step": 123
    },
    {
      "epoch": 1.7777777777777777,
      "grad_norm": 1.2275230884552002,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.0786,
      "step": 124
    },
    {
      "epoch": 1.7921146953405018,
      "grad_norm": 1.4080216884613037,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.0653,
      "step": 125
    },
    {
      "epoch": 1.8064516129032258,
      "grad_norm": 0.8735175132751465,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.0408,
      "step": 126
    },
    {
      "epoch": 1.8207885304659497,
      "grad_norm": 0.7024149298667908,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.0401,
      "step": 127
    },
    {
      "epoch": 1.8351254480286738,
      "grad_norm": 2.560001850128174,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.0633,
      "step": 128
    },
    {
      "epoch": 1.849462365591398,
      "grad_norm": 1.4625259637832642,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.0346,
      "step": 129
    },
    {
      "epoch": 1.863799283154122,
      "grad_norm": 0.9953064918518066,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.0283,
      "step": 130
    },
    {
      "epoch": 1.8781362007168458,
      "grad_norm": 1.8523070812225342,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.0486,
      "step": 131
    },
    {
      "epoch": 1.89247311827957,
      "grad_norm": 1.2022563219070435,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.0617,
      "step": 132
    },
    {
      "epoch": 1.906810035842294,
      "grad_norm": 1.7056341171264648,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.0885,
      "step": 133
    },
    {
      "epoch": 1.9211469534050178,
      "grad_norm": 2.650871515274048,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.1461,
      "step": 134
    },
    {
      "epoch": 1.935483870967742,
      "grad_norm": 1.7407726049423218,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.0542,
      "step": 135
    },
    {
      "epoch": 1.949820788530466,
      "grad_norm": 2.8300583362579346,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.1114,
      "step": 136
    },
    {
      "epoch": 1.9641577060931898,
      "grad_norm": 2.184911012649536,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.0572,
      "step": 137
    },
    {
      "epoch": 1.978494623655914,
      "grad_norm": 0.7571486234664917,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.0362,
      "step": 138
    },
    {
      "epoch": 1.9928315412186381,
      "grad_norm": 0.6461971998214722,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.0241,
      "step": 139
    },
    {
      "epoch": 2.007168458781362,
      "grad_norm": 3.7571802139282227,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.146,
      "step": 140
    },
    {
      "epoch": 2.021505376344086,
      "grad_norm": 1.5061140060424805,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.0877,
      "step": 141
    },
    {
      "epoch": 2.03584229390681,
      "grad_norm": 0.7904454469680786,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.0269,
      "step": 142
    },
    {
      "epoch": 2.050179211469534,
      "grad_norm": 0.527105450630188,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.0284,
      "step": 143
    },
    {
      "epoch": 2.064516129032258,
      "grad_norm": 0.7070644497871399,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.0447,
      "step": 144
    },
    {
      "epoch": 2.078853046594982,
      "grad_norm": 0.31195706129074097,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.0059,
      "step": 145
    },
    {
      "epoch": 2.0931899641577063,
      "grad_norm": 1.6050896644592285,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.0635,
      "step": 146
    },
    {
      "epoch": 2.10752688172043,
      "grad_norm": 0.2570977509021759,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.0103,
      "step": 147
    },
    {
      "epoch": 2.121863799283154,
      "grad_norm": 0.5432910919189453,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.0163,
      "step": 148
    },
    {
      "epoch": 2.1362007168458783,
      "grad_norm": 0.6007760167121887,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.0149,
      "step": 149
    },
    {
      "epoch": 2.150537634408602,
      "grad_norm": 0.5216220021247864,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.0104,
      "step": 150
    },
    {
      "epoch": 2.150537634408602,
      "eval_loss": 0.10088521242141724,
      "eval_runtime": 9.0439,
      "eval_samples_per_second": 13.047,
      "eval_steps_per_second": 3.317,
      "step": 150
    },
    {
      "epoch": 2.164874551971326,
      "grad_norm": 0.48872295022010803,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.0109,
      "step": 151
    },
    {
      "epoch": 2.1792114695340503,
      "grad_norm": 0.30729997158050537,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.0122,
      "step": 152
    },
    {
      "epoch": 2.193548387096774,
      "grad_norm": 0.3366848826408386,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.0085,
      "step": 153
    },
    {
      "epoch": 2.207885304659498,
      "grad_norm": 1.2246568202972412,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.0233,
      "step": 154
    },
    {
      "epoch": 2.2222222222222223,
      "grad_norm": 0.7484713196754456,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.0245,
      "step": 155
    },
    {
      "epoch": 2.236559139784946,
      "grad_norm": 1.060377836227417,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.0302,
      "step": 156
    },
    {
      "epoch": 2.25089605734767,
      "grad_norm": 0.5308035016059875,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.0236,
      "step": 157
    },
    {
      "epoch": 2.2652329749103943,
      "grad_norm": 0.7446233630180359,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.0715,
      "step": 158
    },
    {
      "epoch": 2.279569892473118,
      "grad_norm": 0.6268203854560852,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.0364,
      "step": 159
    },
    {
      "epoch": 2.293906810035842,
      "grad_norm": 0.2845878303050995,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.0116,
      "step": 160
    },
    {
      "epoch": 2.3082437275985663,
      "grad_norm": 0.6574869751930237,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.0211,
      "step": 161
    },
    {
      "epoch": 2.3225806451612905,
      "grad_norm": 0.10483105480670929,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.0034,
      "step": 162
    },
    {
      "epoch": 2.336917562724014,
      "grad_norm": 0.5200598835945129,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.0195,
      "step": 163
    },
    {
      "epoch": 2.3512544802867383,
      "grad_norm": 0.46628037095069885,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.0118,
      "step": 164
    },
    {
      "epoch": 2.3655913978494625,
      "grad_norm": 0.10381802916526794,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.0039,
      "step": 165
    },
    {
      "epoch": 2.379928315412186,
      "grad_norm": 0.09438429027795792,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.0028,
      "step": 166
    },
    {
      "epoch": 2.3942652329749103,
      "grad_norm": 0.16798050701618195,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.0035,
      "step": 167
    },
    {
      "epoch": 2.4086021505376345,
      "grad_norm": 0.3789800703525543,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.0061,
      "step": 168
    },
    {
      "epoch": 2.4229390681003586,
      "grad_norm": 0.15485787391662598,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.0041,
      "step": 169
    },
    {
      "epoch": 2.4372759856630823,
      "grad_norm": 0.06331079453229904,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.002,
      "step": 170
    },
    {
      "epoch": 2.4516129032258065,
      "grad_norm": 0.11097713559865952,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.0031,
      "step": 171
    },
    {
      "epoch": 2.4659498207885306,
      "grad_norm": 1.3297219276428223,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.0234,
      "step": 172
    },
    {
      "epoch": 2.4802867383512543,
      "grad_norm": 0.22801584005355835,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.0051,
      "step": 173
    },
    {
      "epoch": 2.4946236559139785,
      "grad_norm": 1.6642920970916748,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.062,
      "step": 174
    },
    {
      "epoch": 2.5089605734767026,
      "grad_norm": 0.7889490127563477,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.0277,
      "step": 175
    },
    {
      "epoch": 2.5232974910394264,
      "grad_norm": 1.2105519771575928,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.0479,
      "step": 176
    },
    {
      "epoch": 2.5376344086021505,
      "grad_norm": 0.24619156122207642,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.0122,
      "step": 177
    },
    {
      "epoch": 2.5519713261648747,
      "grad_norm": 0.14964176714420319,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.0059,
      "step": 178
    },
    {
      "epoch": 2.5663082437275984,
      "grad_norm": 0.28255024552345276,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.0056,
      "step": 179
    },
    {
      "epoch": 2.5806451612903225,
      "grad_norm": 1.3876649141311646,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.0224,
      "step": 180
    },
    {
      "epoch": 2.5949820788530467,
      "grad_norm": 0.17212137579917908,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.0043,
      "step": 181
    },
    {
      "epoch": 2.6093189964157704,
      "grad_norm": 1.8873724937438965,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.0248,
      "step": 182
    },
    {
      "epoch": 2.6236559139784945,
      "grad_norm": 0.043813642114400864,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.0014,
      "step": 183
    },
    {
      "epoch": 2.6379928315412187,
      "grad_norm": 0.5559656023979187,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.0139,
      "step": 184
    },
    {
      "epoch": 2.652329749103943,
      "grad_norm": 0.5161683559417725,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.0111,
      "step": 185
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 0.5870691537857056,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.009,
      "step": 186
    },
    {
      "epoch": 2.6810035842293907,
      "grad_norm": 0.4506298303604126,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.0035,
      "step": 187
    },
    {
      "epoch": 2.695340501792115,
      "grad_norm": 0.2792058289051056,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.0062,
      "step": 188
    },
    {
      "epoch": 2.709677419354839,
      "grad_norm": 0.2612236440181732,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.0047,
      "step": 189
    },
    {
      "epoch": 2.7240143369175627,
      "grad_norm": 0.12661564350128174,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.0043,
      "step": 190
    },
    {
      "epoch": 2.738351254480287,
      "grad_norm": 0.5674365162849426,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.039,
      "step": 191
    },
    {
      "epoch": 2.752688172043011,
      "grad_norm": 0.38780298829078674,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.0317,
      "step": 192
    },
    {
      "epoch": 2.7670250896057347,
      "grad_norm": 0.6319880485534668,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.0412,
      "step": 193
    },
    {
      "epoch": 2.781362007168459,
      "grad_norm": 1.7055460214614868,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.0595,
      "step": 194
    },
    {
      "epoch": 2.795698924731183,
      "grad_norm": 0.1415621042251587,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.0061,
      "step": 195
    },
    {
      "epoch": 2.8100358422939067,
      "grad_norm": 0.24341678619384766,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.0056,
      "step": 196
    },
    {
      "epoch": 2.824372759856631,
      "grad_norm": 0.4508267939090729,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.0058,
      "step": 197
    },
    {
      "epoch": 2.838709677419355,
      "grad_norm": 0.24219681322574615,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.0199,
      "step": 198
    },
    {
      "epoch": 2.8530465949820787,
      "grad_norm": 0.19702303409576416,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.0089,
      "step": 199
    },
    {
      "epoch": 2.867383512544803,
      "grad_norm": 0.1203891783952713,
      "learning_rate": 0.0,
      "loss": 0.004,
      "step": 200
    },
    {
      "epoch": 2.867383512544803,
      "eval_loss": 0.06954215466976166,
      "eval_runtime": 9.0456,
      "eval_samples_per_second": 13.045,
      "eval_steps_per_second": 3.317,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.017027157491712e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}