{
  "best_metric": 10.34589672088623,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.3386960203217612,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001693480101608806,
      "grad_norm": 0.05767215043306351,
      "learning_rate": 7e-06,
      "loss": 10.3761,
      "step": 1
    },
    {
      "epoch": 0.001693480101608806,
      "eval_loss": 10.376470565795898,
      "eval_runtime": 2.2633,
      "eval_samples_per_second": 439.62,
      "eval_steps_per_second": 110.016,
      "step": 1
    },
    {
      "epoch": 0.003386960203217612,
      "grad_norm": 0.05682261288166046,
      "learning_rate": 1.4e-05,
      "loss": 10.3756,
      "step": 2
    },
    {
      "epoch": 0.005080440304826418,
      "grad_norm": 0.05533347651362419,
      "learning_rate": 2.1e-05,
      "loss": 10.3759,
      "step": 3
    },
    {
      "epoch": 0.006773920406435224,
      "grad_norm": 0.06400913745164871,
      "learning_rate": 2.8e-05,
      "loss": 10.3753,
      "step": 4
    },
    {
      "epoch": 0.00846740050804403,
      "grad_norm": 0.06217538192868233,
      "learning_rate": 3.5e-05,
      "loss": 10.3766,
      "step": 5
    },
    {
      "epoch": 0.010160880609652836,
      "grad_norm": 0.059953831136226654,
      "learning_rate": 4.2e-05,
      "loss": 10.3753,
      "step": 6
    },
    {
      "epoch": 0.011854360711261643,
      "grad_norm": 0.06322812288999557,
      "learning_rate": 4.899999999999999e-05,
      "loss": 10.3751,
      "step": 7
    },
    {
      "epoch": 0.013547840812870448,
      "grad_norm": 0.06448973715305328,
      "learning_rate": 5.6e-05,
      "loss": 10.3755,
      "step": 8
    },
    {
      "epoch": 0.015241320914479255,
      "grad_norm": 0.06553128361701965,
      "learning_rate": 6.3e-05,
      "loss": 10.3754,
      "step": 9
    },
    {
      "epoch": 0.01693480101608806,
      "grad_norm": 0.06717841327190399,
      "learning_rate": 7e-05,
      "loss": 10.3749,
      "step": 10
    },
    {
      "epoch": 0.018628281117696866,
      "grad_norm": 0.06435838341712952,
      "learning_rate": 6.999521567473641e-05,
      "loss": 10.3753,
      "step": 11
    },
    {
      "epoch": 0.02032176121930567,
      "grad_norm": 0.06390493363142014,
      "learning_rate": 6.998086400693241e-05,
      "loss": 10.3744,
      "step": 12
    },
    {
      "epoch": 0.02201524132091448,
      "grad_norm": 0.06731149554252625,
      "learning_rate": 6.995694892019065e-05,
      "loss": 10.3745,
      "step": 13
    },
    {
      "epoch": 0.023708721422523286,
      "grad_norm": 0.08515335619449615,
      "learning_rate": 6.99234769526571e-05,
      "loss": 10.3737,
      "step": 14
    },
    {
      "epoch": 0.02540220152413209,
      "grad_norm": 0.08068357408046722,
      "learning_rate": 6.988045725523343e-05,
      "loss": 10.3727,
      "step": 15
    },
    {
      "epoch": 0.027095681625740897,
      "grad_norm": 0.08304785192012787,
      "learning_rate": 6.982790158907539e-05,
      "loss": 10.3722,
      "step": 16
    },
    {
      "epoch": 0.028789161727349702,
      "grad_norm": 0.08605454862117767,
      "learning_rate": 6.976582432237733e-05,
      "loss": 10.3725,
      "step": 17
    },
    {
      "epoch": 0.03048264182895851,
      "grad_norm": 0.0890476256608963,
      "learning_rate": 6.969424242644413e-05,
      "loss": 10.3728,
      "step": 18
    },
    {
      "epoch": 0.03217612193056731,
      "grad_norm": 0.09603014588356018,
      "learning_rate": 6.961317547105138e-05,
      "loss": 10.3707,
      "step": 19
    },
    {
      "epoch": 0.03386960203217612,
      "grad_norm": 0.09333094954490662,
      "learning_rate": 6.952264561909527e-05,
      "loss": 10.3714,
      "step": 20
    },
    {
      "epoch": 0.03556308213378493,
      "grad_norm": 0.11809182167053223,
      "learning_rate": 6.942267762053337e-05,
      "loss": 10.3699,
      "step": 21
    },
    {
      "epoch": 0.03725656223539373,
      "grad_norm": 0.12310867756605148,
      "learning_rate": 6.931329880561832e-05,
      "loss": 10.3696,
      "step": 22
    },
    {
      "epoch": 0.03895004233700254,
      "grad_norm": 0.12797869741916656,
      "learning_rate": 6.919453907742597e-05,
      "loss": 10.3694,
      "step": 23
    },
    {
      "epoch": 0.04064352243861134,
      "grad_norm": 0.14211532473564148,
      "learning_rate": 6.90664309036802e-05,
      "loss": 10.3684,
      "step": 24
    },
    {
      "epoch": 0.04233700254022015,
      "grad_norm": 0.13544224202632904,
      "learning_rate": 6.892900930787656e-05,
      "loss": 10.3687,
      "step": 25
    },
    {
      "epoch": 0.04403048264182896,
      "grad_norm": 0.15587803721427917,
      "learning_rate": 6.87823118597072e-05,
      "loss": 10.3667,
      "step": 26
    },
    {
      "epoch": 0.04572396274343776,
      "grad_norm": 0.15392404794692993,
      "learning_rate": 6.862637866478969e-05,
      "loss": 10.3679,
      "step": 27
    },
    {
      "epoch": 0.04741744284504657,
      "grad_norm": 0.15195734798908234,
      "learning_rate": 6.846125235370252e-05,
      "loss": 10.3674,
      "step": 28
    },
    {
      "epoch": 0.04911092294665537,
      "grad_norm": 0.1591203212738037,
      "learning_rate": 6.828697807033038e-05,
      "loss": 10.3646,
      "step": 29
    },
    {
      "epoch": 0.05080440304826418,
      "grad_norm": 0.17264410853385925,
      "learning_rate": 6.81036034595222e-05,
      "loss": 10.3642,
      "step": 30
    },
    {
      "epoch": 0.05249788314987299,
      "grad_norm": 0.1709284484386444,
      "learning_rate": 6.791117865406564e-05,
      "loss": 10.3635,
      "step": 31
    },
    {
      "epoch": 0.05419136325148179,
      "grad_norm": 0.17630036175251007,
      "learning_rate": 6.770975626098112e-05,
      "loss": 10.365,
      "step": 32
    },
    {
      "epoch": 0.0558848433530906,
      "grad_norm": 0.1866825520992279,
      "learning_rate": 6.749939134713974e-05,
      "loss": 10.3628,
      "step": 33
    },
    {
      "epoch": 0.057578323454699404,
      "grad_norm": 0.17809787392616272,
      "learning_rate": 6.728014142420846e-05,
      "loss": 10.3629,
      "step": 34
    },
    {
      "epoch": 0.05927180355630821,
      "grad_norm": 0.20281663537025452,
      "learning_rate": 6.7052066432927e-05,
      "loss": 10.3578,
      "step": 35
    },
    {
      "epoch": 0.06096528365791702,
      "grad_norm": 0.19204723834991455,
      "learning_rate": 6.681522872672069e-05,
      "loss": 10.3585,
      "step": 36
    },
    {
      "epoch": 0.06265876375952582,
      "grad_norm": 0.1810905933380127,
      "learning_rate": 6.656969305465356e-05,
      "loss": 10.3592,
      "step": 37
    },
    {
      "epoch": 0.06435224386113463,
      "grad_norm": 0.19030340015888214,
      "learning_rate": 6.631552654372672e-05,
      "loss": 10.3601,
      "step": 38
    },
    {
      "epoch": 0.06604572396274344,
      "grad_norm": 0.16998638212680817,
      "learning_rate": 6.60527986805264e-05,
      "loss": 10.3573,
      "step": 39
    },
    {
      "epoch": 0.06773920406435224,
      "grad_norm": 0.1517295241355896,
      "learning_rate": 6.578158129222711e-05,
      "loss": 10.3549,
      "step": 40
    },
    {
      "epoch": 0.06943268416596104,
      "grad_norm": 0.15249769389629364,
      "learning_rate": 6.550194852695469e-05,
      "loss": 10.3559,
      "step": 41
    },
    {
      "epoch": 0.07112616426756986,
      "grad_norm": 0.15435636043548584,
      "learning_rate": 6.521397683351509e-05,
      "loss": 10.354,
      "step": 42
    },
    {
      "epoch": 0.07281964436917866,
      "grad_norm": 0.14684386551380157,
      "learning_rate": 6.491774494049386e-05,
      "loss": 10.355,
      "step": 43
    },
    {
      "epoch": 0.07451312447078746,
      "grad_norm": 0.15682685375213623,
      "learning_rate": 6.461333383473272e-05,
      "loss": 10.3522,
      "step": 44
    },
    {
      "epoch": 0.07620660457239628,
      "grad_norm": 0.14739371836185455,
      "learning_rate": 6.430082673918849e-05,
      "loss": 10.3518,
      "step": 45
    },
    {
      "epoch": 0.07790008467400508,
      "grad_norm": 0.14083759486675262,
      "learning_rate": 6.398030909018069e-05,
      "loss": 10.3527,
      "step": 46
    },
    {
      "epoch": 0.07959356477561388,
      "grad_norm": 0.11909397691488266,
      "learning_rate": 6.365186851403423e-05,
      "loss": 10.3539,
      "step": 47
    },
    {
      "epoch": 0.08128704487722269,
      "grad_norm": 0.11432235687971115,
      "learning_rate": 6.331559480312315e-05,
      "loss": 10.3531,
      "step": 48
    },
    {
      "epoch": 0.0829805249788315,
      "grad_norm": 0.13134874403476715,
      "learning_rate": 6.297157989132236e-05,
      "loss": 10.3544,
      "step": 49
    },
    {
      "epoch": 0.0846740050804403,
      "grad_norm": 0.19639384746551514,
      "learning_rate": 6.261991782887377e-05,
      "loss": 10.3542,
      "step": 50
    },
    {
      "epoch": 0.0846740050804403,
      "eval_loss": 10.351482391357422,
      "eval_runtime": 2.2614,
      "eval_samples_per_second": 439.995,
      "eval_steps_per_second": 110.109,
      "step": 50
    },
    {
      "epoch": 0.0863674851820491,
      "grad_norm": 0.10659899562597275,
      "learning_rate": 6.226070475667393e-05,
      "loss": 10.352,
      "step": 51
    },
    {
      "epoch": 0.08806096528365792,
      "grad_norm": 0.10386061668395996,
      "learning_rate": 6.189403887999006e-05,
      "loss": 10.353,
      "step": 52
    },
    {
      "epoch": 0.08975444538526672,
      "grad_norm": 0.08967398852109909,
      "learning_rate": 6.152002044161171e-05,
      "loss": 10.3509,
      "step": 53
    },
    {
      "epoch": 0.09144792548687553,
      "grad_norm": 0.09589025378227234,
      "learning_rate": 6.113875169444539e-05,
      "loss": 10.3516,
      "step": 54
    },
    {
      "epoch": 0.09314140558848434,
      "grad_norm": 0.08911201357841492,
      "learning_rate": 6.0750336873559605e-05,
      "loss": 10.352,
      "step": 55
    },
    {
      "epoch": 0.09483488569009314,
      "grad_norm": 0.08531937748193741,
      "learning_rate": 6.035488216768811e-05,
      "loss": 10.3506,
      "step": 56
    },
    {
      "epoch": 0.09652836579170194,
      "grad_norm": 0.07509951293468475,
      "learning_rate": 5.9952495690198894e-05,
      "loss": 10.3501,
      "step": 57
    },
    {
      "epoch": 0.09822184589331075,
      "grad_norm": 0.07540036737918854,
      "learning_rate": 5.954328744953709e-05,
      "loss": 10.3506,
      "step": 58
    },
    {
      "epoch": 0.09991532599491956,
      "grad_norm": 0.07696950435638428,
      "learning_rate": 5.91273693191498e-05,
      "loss": 10.3511,
      "step": 59
    },
    {
      "epoch": 0.10160880609652836,
      "grad_norm": 0.0718412697315216,
      "learning_rate": 5.870485500690094e-05,
      "loss": 10.3504,
      "step": 60
    },
    {
      "epoch": 0.10330228619813717,
      "grad_norm": 0.08057332038879395,
      "learning_rate": 5.827586002398468e-05,
      "loss": 10.3502,
      "step": 61
    },
    {
      "epoch": 0.10499576629974598,
      "grad_norm": 0.06910684704780579,
      "learning_rate": 5.784050165334589e-05,
      "loss": 10.3511,
      "step": 62
    },
    {
      "epoch": 0.10668924640135478,
      "grad_norm": 0.06749627739191055,
      "learning_rate": 5.739889891761608e-05,
      "loss": 10.3491,
      "step": 63
    },
    {
      "epoch": 0.10838272650296359,
      "grad_norm": 0.06321804970502853,
      "learning_rate": 5.6951172546573794e-05,
      "loss": 10.349,
      "step": 64
    },
    {
      "epoch": 0.1100762066045724,
      "grad_norm": 0.07615376263856888,
      "learning_rate": 5.6497444944138376e-05,
      "loss": 10.3487,
      "step": 65
    },
    {
      "epoch": 0.1117696867061812,
      "grad_norm": 0.06743479520082474,
      "learning_rate": 5.603784015490587e-05,
      "loss": 10.3504,
      "step": 66
    },
    {
      "epoch": 0.11346316680779,
      "grad_norm": 0.060844793915748596,
      "learning_rate": 5.557248383023655e-05,
      "loss": 10.3501,
      "step": 67
    },
    {
      "epoch": 0.11515664690939881,
      "grad_norm": 0.048706915229558945,
      "learning_rate": 5.510150319390302e-05,
      "loss": 10.349,
      "step": 68
    },
    {
      "epoch": 0.11685012701100762,
      "grad_norm": 0.04675814509391785,
      "learning_rate": 5.4625027007308546e-05,
      "loss": 10.3491,
      "step": 69
    },
    {
      "epoch": 0.11854360711261643,
      "grad_norm": 0.05300172045826912,
      "learning_rate": 5.414318553428494e-05,
      "loss": 10.3494,
      "step": 70
    },
    {
      "epoch": 0.12023708721422523,
      "grad_norm": 0.042747609317302704,
      "learning_rate": 5.3656110505479776e-05,
      "loss": 10.3503,
      "step": 71
    },
    {
      "epoch": 0.12193056731583404,
      "grad_norm": 0.04265451431274414,
      "learning_rate": 5.316393508234253e-05,
      "loss": 10.3482,
      "step": 72
    },
    {
      "epoch": 0.12362404741744284,
      "grad_norm": 0.06077653169631958,
      "learning_rate": 5.266679382071953e-05,
      "loss": 10.3482,
      "step": 73
    },
    {
      "epoch": 0.12531752751905165,
      "grad_norm": 0.05237172171473503,
      "learning_rate": 5.216482263406778e-05,
      "loss": 10.3489,
      "step": 74
    },
    {
      "epoch": 0.12701100762066045,
      "grad_norm": 0.061119917780160904,
      "learning_rate": 5.1658158756297576e-05,
      "loss": 10.3472,
      "step": 75
    },
    {
      "epoch": 0.12870448772226925,
      "grad_norm": 0.03865138441324234,
      "learning_rate": 5.114694070425407e-05,
      "loss": 10.3497,
      "step": 76
    },
    {
      "epoch": 0.13039796782387808,
      "grad_norm": 0.03798210248351097,
      "learning_rate": 5.063130823984823e-05,
      "loss": 10.3472,
      "step": 77
    },
    {
      "epoch": 0.13209144792548688,
      "grad_norm": 0.03693859279155731,
      "learning_rate": 5.011140233184724e-05,
      "loss": 10.3487,
      "step": 78
    },
    {
      "epoch": 0.13378492802709568,
      "grad_norm": 0.0393260158598423,
      "learning_rate": 4.958736511733516e-05,
      "loss": 10.3483,
      "step": 79
    },
    {
      "epoch": 0.1354784081287045,
      "grad_norm": 0.04144861549139023,
      "learning_rate": 4.905933986285393e-05,
      "loss": 10.3489,
      "step": 80
    },
    {
      "epoch": 0.1371718882303133,
      "grad_norm": 0.057905759662389755,
      "learning_rate": 4.8527470925235824e-05,
      "loss": 10.3488,
      "step": 81
    },
    {
      "epoch": 0.1388653683319221,
      "grad_norm": 0.0418972410261631,
      "learning_rate": 4.799190371213772e-05,
      "loss": 10.3495,
      "step": 82
    },
    {
      "epoch": 0.14055884843353092,
      "grad_norm": 0.03675097972154617,
      "learning_rate": 4.745278464228808e-05,
      "loss": 10.3468,
      "step": 83
    },
    {
      "epoch": 0.14225232853513972,
      "grad_norm": 0.05150232091546059,
      "learning_rate": 4.69102611054575e-05,
      "loss": 10.347,
      "step": 84
    },
    {
      "epoch": 0.14394580863674852,
      "grad_norm": 0.05643484368920326,
      "learning_rate": 4.6364481422163926e-05,
      "loss": 10.3483,
      "step": 85
    },
    {
      "epoch": 0.14563928873835733,
      "grad_norm": 0.053120311349630356,
      "learning_rate": 4.581559480312316e-05,
      "loss": 10.3476,
      "step": 86
    },
    {
      "epoch": 0.14733276883996613,
      "grad_norm": 0.03632555529475212,
      "learning_rate": 4.526375130845627e-05,
      "loss": 10.3475,
      "step": 87
    },
    {
      "epoch": 0.14902624894157493,
      "grad_norm": 0.04733320325613022,
      "learning_rate": 4.4709101806664554e-05,
      "loss": 10.3487,
      "step": 88
    },
    {
      "epoch": 0.15071972904318373,
      "grad_norm": 0.03701463341712952,
      "learning_rate": 4.4151797933383685e-05,
      "loss": 10.3458,
      "step": 89
    },
    {
      "epoch": 0.15241320914479256,
      "grad_norm": 0.040851932018995285,
      "learning_rate": 4.359199204992797e-05,
      "loss": 10.3481,
      "step": 90
    },
    {
      "epoch": 0.15410668924640136,
      "grad_norm": 0.04483847692608833,
      "learning_rate": 4.30298372016363e-05,
      "loss": 10.3478,
      "step": 91
    },
    {
      "epoch": 0.15580016934801016,
      "grad_norm": 0.05173524469137192,
      "learning_rate": 4.246548707603114e-05,
      "loss": 10.3461,
      "step": 92
    },
    {
      "epoch": 0.15749364944961897,
      "grad_norm": 0.0741141214966774,
      "learning_rate": 4.1899095960801805e-05,
      "loss": 10.345,
      "step": 93
    },
    {
      "epoch": 0.15918712955122777,
      "grad_norm": 0.052725061774253845,
      "learning_rate": 4.133081870162385e-05,
      "loss": 10.3481,
      "step": 94
    },
    {
      "epoch": 0.16088060965283657,
      "grad_norm": 0.057170942425727844,
      "learning_rate": 4.076081065982569e-05,
      "loss": 10.346,
      "step": 95
    },
    {
      "epoch": 0.16257408975444537,
      "grad_norm": 0.07554444670677185,
      "learning_rate": 4.018922766991447e-05,
      "loss": 10.3477,
      "step": 96
    },
    {
      "epoch": 0.1642675698560542,
      "grad_norm": 0.07930599898099899,
      "learning_rate": 3.961622599697241e-05,
      "loss": 10.3516,
      "step": 97
    },
    {
      "epoch": 0.165961049957663,
      "grad_norm": 0.07189879566431046,
      "learning_rate": 3.9041962293935516e-05,
      "loss": 10.3461,
      "step": 98
    },
    {
      "epoch": 0.1676545300592718,
      "grad_norm": 0.0871279165148735,
      "learning_rate": 3.84665935587662e-05,
      "loss": 10.3498,
      "step": 99
    },
    {
      "epoch": 0.1693480101608806,
      "grad_norm": 0.17175425589084625,
      "learning_rate": 3.7890277091531636e-05,
      "loss": 10.3469,
      "step": 100
    },
    {
      "epoch": 0.1693480101608806,
      "eval_loss": 10.34717845916748,
      "eval_runtime": 2.3148,
      "eval_samples_per_second": 429.838,
      "eval_steps_per_second": 107.568,
      "step": 100
    },
    {
      "epoch": 0.1710414902624894,
      "grad_norm": 0.04662218317389488,
      "learning_rate": 3.7313170451399475e-05,
      "loss": 10.3474,
      "step": 101
    },
    {
      "epoch": 0.1727349703640982,
      "grad_norm": 0.04416072368621826,
      "learning_rate": 3.673543141356278e-05,
      "loss": 10.3473,
      "step": 102
    },
    {
      "epoch": 0.17442845046570704,
      "grad_norm": 0.039319951087236404,
      "learning_rate": 3.6157217926105783e-05,
      "loss": 10.3474,
      "step": 103
    },
    {
      "epoch": 0.17612193056731584,
      "grad_norm": 0.047466833144426346,
      "learning_rate": 3.557868806682255e-05,
      "loss": 10.3483,
      "step": 104
    },
    {
      "epoch": 0.17781541066892464,
      "grad_norm": 0.04195760190486908,
      "learning_rate": 3.5e-05,
      "loss": 10.3475,
      "step": 105
    },
    {
      "epoch": 0.17950889077053345,
      "grad_norm": 0.052754271775484085,
      "learning_rate": 3.442131193317745e-05,
      "loss": 10.348,
      "step": 106
    },
    {
      "epoch": 0.18120237087214225,
      "grad_norm": 0.03500129655003548,
      "learning_rate": 3.384278207389421e-05,
      "loss": 10.3492,
      "step": 107
    },
    {
      "epoch": 0.18289585097375105,
      "grad_norm": 0.0344226248562336,
      "learning_rate": 3.3264568586437216e-05,
      "loss": 10.3483,
      "step": 108
    },
    {
      "epoch": 0.18458933107535985,
      "grad_norm": 0.0443701297044754,
      "learning_rate": 3.268682954860052e-05,
      "loss": 10.3489,
      "step": 109
    },
    {
      "epoch": 0.18628281117696868,
      "grad_norm": 0.04030259698629379,
      "learning_rate": 3.210972290846837e-05,
      "loss": 10.3469,
      "step": 110
    },
    {
      "epoch": 0.18797629127857748,
      "grad_norm": 0.03760414570569992,
      "learning_rate": 3.15334064412338e-05,
      "loss": 10.3481,
      "step": 111
    },
    {
      "epoch": 0.1896697713801863,
      "grad_norm": 0.0433247834444046,
      "learning_rate": 3.0958037706064485e-05,
      "loss": 10.3479,
      "step": 112
    },
    {
      "epoch": 0.1913632514817951,
      "grad_norm": 0.0393412746489048,
      "learning_rate": 3.038377400302758e-05,
      "loss": 10.3459,
      "step": 113
    },
    {
      "epoch": 0.1930567315834039,
      "grad_norm": 0.0358240120112896,
      "learning_rate": 2.9810772330085524e-05,
      "loss": 10.3471,
      "step": 114
    },
    {
      "epoch": 0.1947502116850127,
      "grad_norm": 0.04923977330327034,
      "learning_rate": 2.9239189340174306e-05,
      "loss": 10.349,
      "step": 115
    },
    {
      "epoch": 0.1964436917866215,
      "grad_norm": 0.03750025853514671,
      "learning_rate": 2.8669181298376163e-05,
      "loss": 10.3473,
      "step": 116
    },
    {
      "epoch": 0.19813717188823032,
      "grad_norm": 0.029165159910917282,
      "learning_rate": 2.8100904039198193e-05,
      "loss": 10.3473,
      "step": 117
    },
    {
      "epoch": 0.19983065198983913,
      "grad_norm": 0.038380276411771774,
      "learning_rate": 2.7534512923968863e-05,
      "loss": 10.3476,
      "step": 118
    },
    {
      "epoch": 0.20152413209144793,
      "grad_norm": 0.03772229328751564,
      "learning_rate": 2.6970162798363695e-05,
      "loss": 10.347,
      "step": 119
    },
    {
      "epoch": 0.20321761219305673,
      "grad_norm": 0.042656853795051575,
      "learning_rate": 2.640800795007203e-05,
      "loss": 10.3464,
      "step": 120
    },
    {
      "epoch": 0.20491109229466553,
      "grad_norm": 0.038278091698884964,
      "learning_rate": 2.5848202066616305e-05,
      "loss": 10.3465,
      "step": 121
    },
    {
      "epoch": 0.20660457239627433,
      "grad_norm": 0.03428620845079422,
      "learning_rate": 2.5290898193335446e-05,
      "loss": 10.3461,
      "step": 122
    },
    {
      "epoch": 0.20829805249788316,
      "grad_norm": 0.038639917969703674,
      "learning_rate": 2.4736248691543736e-05,
      "loss": 10.3467,
      "step": 123
    },
    {
      "epoch": 0.20999153259949196,
      "grad_norm": 0.03810051828622818,
      "learning_rate": 2.4184405196876842e-05,
      "loss": 10.347,
      "step": 124
    },
    {
      "epoch": 0.21168501270110077,
      "grad_norm": 0.03695661202073097,
      "learning_rate": 2.363551857783608e-05,
      "loss": 10.3476,
      "step": 125
    },
    {
      "epoch": 0.21337849280270957,
      "grad_norm": 0.046606481075286865,
      "learning_rate": 2.308973889454249e-05,
      "loss": 10.347,
      "step": 126
    },
    {
      "epoch": 0.21507197290431837,
      "grad_norm": 0.04259473457932472,
      "learning_rate": 2.2547215357711918e-05,
      "loss": 10.3466,
      "step": 127
    },
    {
      "epoch": 0.21676545300592717,
      "grad_norm": 0.034811653196811676,
      "learning_rate": 2.2008096287862266e-05,
      "loss": 10.347,
      "step": 128
    },
    {
      "epoch": 0.21845893310753597,
      "grad_norm": 0.03522142395377159,
      "learning_rate": 2.1472529074764177e-05,
      "loss": 10.3465,
      "step": 129
    },
    {
      "epoch": 0.2201524132091448,
      "grad_norm": 0.030583711341023445,
      "learning_rate": 2.0940660137146074e-05,
      "loss": 10.3461,
      "step": 130
    },
    {
      "epoch": 0.2218458933107536,
      "grad_norm": 0.036702465265989304,
      "learning_rate": 2.041263488266484e-05,
      "loss": 10.3487,
      "step": 131
    },
    {
      "epoch": 0.2235393734123624,
      "grad_norm": 0.03403886407613754,
      "learning_rate": 1.988859766815275e-05,
      "loss": 10.3443,
      "step": 132
    },
    {
      "epoch": 0.2252328535139712,
      "grad_norm": 0.033561062067747116,
      "learning_rate": 1.9368691760151773e-05,
      "loss": 10.3463,
      "step": 133
    },
    {
      "epoch": 0.22692633361558,
      "grad_norm": 0.04279007017612457,
      "learning_rate": 1.885305929574593e-05,
      "loss": 10.3474,
      "step": 134
    },
    {
      "epoch": 0.2286198137171888,
      "grad_norm": 0.03716912120580673,
      "learning_rate": 1.8341841243702424e-05,
      "loss": 10.3432,
      "step": 135
    },
    {
      "epoch": 0.23031329381879762,
      "grad_norm": 0.03685978800058365,
      "learning_rate": 1.7835177365932225e-05,
      "loss": 10.3482,
      "step": 136
    },
    {
      "epoch": 0.23200677392040644,
      "grad_norm": 0.03692598268389702,
      "learning_rate": 1.7333206179280478e-05,
      "loss": 10.3452,
      "step": 137
    },
    {
      "epoch": 0.23370025402201525,
      "grad_norm": 0.05442724749445915,
      "learning_rate": 1.6836064917657478e-05,
      "loss": 10.3441,
      "step": 138
    },
    {
      "epoch": 0.23539373412362405,
      "grad_norm": 0.03732023015618324,
      "learning_rate": 1.6343889494520224e-05,
      "loss": 10.346,
      "step": 139
    },
    {
      "epoch": 0.23708721422523285,
      "grad_norm": 0.041189733892679214,
      "learning_rate": 1.5856814465715064e-05,
      "loss": 10.3449,
      "step": 140
    },
    {
      "epoch": 0.23878069432684165,
      "grad_norm": 0.039623793214559555,
      "learning_rate": 1.5374972992691458e-05,
      "loss": 10.3455,
      "step": 141
    },
    {
      "epoch": 0.24047417442845045,
      "grad_norm": 0.05379751697182655,
      "learning_rate": 1.4898496806096974e-05,
      "loss": 10.3445,
      "step": 142
    },
    {
      "epoch": 0.24216765453005928,
      "grad_norm": 0.04194023460149765,
      "learning_rate": 1.4427516169763444e-05,
      "loss": 10.3449,
      "step": 143
    },
    {
      "epoch": 0.2438611346316681,
      "grad_norm": 0.056030042469501495,
      "learning_rate": 1.396215984509412e-05,
      "loss": 10.3429,
      "step": 144
    },
    {
      "epoch": 0.2455546147332769,
      "grad_norm": 0.0768880546092987,
      "learning_rate": 1.3502555055861625e-05,
      "loss": 10.3461,
      "step": 145
    },
    {
      "epoch": 0.2472480948348857,
      "grad_norm": 0.04330800473690033,
      "learning_rate": 1.3048827453426203e-05,
      "loss": 10.3455,
      "step": 146
    },
    {
      "epoch": 0.2489415749364945,
      "grad_norm": 0.07401636242866516,
      "learning_rate": 1.2601101082383917e-05,
      "loss": 10.3477,
      "step": 147
    },
    {
      "epoch": 0.2506350550381033,
      "grad_norm": 0.07112563401460648,
      "learning_rate": 1.2159498346654094e-05,
      "loss": 10.3463,
      "step": 148
    },
    {
      "epoch": 0.2523285351397121,
      "grad_norm": 0.09609571099281311,
      "learning_rate": 1.1724139976015306e-05,
      "loss": 10.3451,
      "step": 149
    },
    {
      "epoch": 0.2540220152413209,
      "grad_norm": 0.1710243821144104,
      "learning_rate": 1.1295144993099068e-05,
      "loss": 10.348,
      "step": 150
    },
    {
      "epoch": 0.2540220152413209,
      "eval_loss": 10.346138954162598,
      "eval_runtime": 2.2706,
      "eval_samples_per_second": 438.212,
      "eval_steps_per_second": 109.663,
      "step": 150
    },
    {
      "epoch": 0.2557154953429297,
      "grad_norm": 0.03954402729868889,
      "learning_rate": 1.0872630680850196e-05,
      "loss": 10.3483,
      "step": 151
    },
    {
      "epoch": 0.2574089754445385,
      "grad_norm": 0.0462455190718174,
      "learning_rate": 1.0456712550462898e-05,
      "loss": 10.3468,
      "step": 152
    },
    {
      "epoch": 0.25910245554614736,
      "grad_norm": 0.03721482306718826,
      "learning_rate": 1.0047504309801104e-05,
      "loss": 10.3471,
      "step": 153
    },
    {
      "epoch": 0.26079593564775616,
      "grad_norm": 0.0438302680850029,
      "learning_rate": 9.645117832311886e-06,
      "loss": 10.3467,
      "step": 154
    },
    {
      "epoch": 0.26248941574936496,
      "grad_norm": 0.04052369296550751,
      "learning_rate": 9.249663126440394e-06,
      "loss": 10.3475,
      "step": 155
    },
    {
      "epoch": 0.26418289585097376,
      "grad_norm": 0.0368545800447464,
      "learning_rate": 8.861248305554624e-06,
      "loss": 10.3469,
      "step": 156
    },
    {
      "epoch": 0.26587637595258257,
      "grad_norm": 0.033342041075229645,
      "learning_rate": 8.47997955838829e-06,
      "loss": 10.3472,
      "step": 157
    },
    {
      "epoch": 0.26756985605419137,
      "grad_norm": 0.040352921932935715,
      "learning_rate": 8.10596112000994e-06,
      "loss": 10.3476,
      "step": 158
    },
    {
      "epoch": 0.26926333615580017,
      "grad_norm": 0.045244328677654266,
      "learning_rate": 7.739295243326067e-06,
      "loss": 10.3481,
      "step": 159
    },
    {
      "epoch": 0.270956816257409,
      "grad_norm": 0.03528599813580513,
      "learning_rate": 7.380082171126228e-06,
      "loss": 10.3464,
      "step": 160
    },
    {
      "epoch": 0.2726502963590178,
      "grad_norm": 0.051067374646663666,
      "learning_rate": 7.028420108677635e-06,
      "loss": 10.3466,
      "step": 161
    },
    {
      "epoch": 0.2743437764606266,
      "grad_norm": 0.03743496537208557,
      "learning_rate": 6.684405196876842e-06,
      "loss": 10.3461,
      "step": 162
    },
    {
      "epoch": 0.2760372565622354,
      "grad_norm": 0.03153526782989502,
      "learning_rate": 6.3481314859657675e-06,
      "loss": 10.3471,
      "step": 163
    },
    {
      "epoch": 0.2777307366638442,
      "grad_norm": 0.0363004244863987,
      "learning_rate": 6.019690909819298e-06,
      "loss": 10.3473,
      "step": 164
    },
    {
      "epoch": 0.279424216765453,
      "grad_norm": 0.03496118262410164,
      "learning_rate": 5.6991732608115e-06,
      "loss": 10.3472,
      "step": 165
    },
    {
      "epoch": 0.28111769686706184,
      "grad_norm": 0.04263157397508621,
      "learning_rate": 5.386666165267256e-06,
      "loss": 10.3462,
      "step": 166
    },
    {
      "epoch": 0.28281117696867064,
      "grad_norm": 0.03566034883260727,
      "learning_rate": 5.08225505950613e-06,
      "loss": 10.3481,
      "step": 167
    },
    {
      "epoch": 0.28450465707027944,
      "grad_norm": 0.03986247256398201,
      "learning_rate": 4.786023166484913e-06,
      "loss": 10.3463,
      "step": 168
    },
    {
      "epoch": 0.28619813717188824,
      "grad_norm": 0.039538342505693436,
      "learning_rate": 4.498051473045291e-06,
      "loss": 10.3464,
      "step": 169
    },
    {
      "epoch": 0.28789161727349705,
      "grad_norm": 0.03614843636751175,
      "learning_rate": 4.218418707772886e-06,
      "loss": 10.3472,
      "step": 170
    },
    {
      "epoch": 0.28958509737510585,
      "grad_norm": 0.027718421071767807,
      "learning_rate": 3.947201319473587e-06,
      "loss": 10.3445,
      "step": 171
    },
    {
      "epoch": 0.29127857747671465,
      "grad_norm": 0.04426714777946472,
      "learning_rate": 3.684473456273278e-06,
      "loss": 10.3459,
      "step": 172
    },
    {
      "epoch": 0.29297205757832345,
      "grad_norm": 0.027950307354331017,
      "learning_rate": 3.4303069453464383e-06,
      "loss": 10.3464,
      "step": 173
    },
    {
      "epoch": 0.29466553767993225,
      "grad_norm": 0.03315555676817894,
      "learning_rate": 3.184771273279312e-06,
      "loss": 10.3455,
      "step": 174
    },
    {
      "epoch": 0.29635901778154106,
      "grad_norm": 0.026465769857168198,
      "learning_rate": 2.947933567072987e-06,
      "loss": 10.347,
      "step": 175
    },
    {
      "epoch": 0.29805249788314986,
      "grad_norm": 0.033119168132543564,
      "learning_rate": 2.719858575791534e-06,
      "loss": 10.3467,
      "step": 176
    },
    {
      "epoch": 0.29974597798475866,
      "grad_norm": 0.034391850233078,
      "learning_rate": 2.500608652860256e-06,
      "loss": 10.3449,
      "step": 177
    },
    {
      "epoch": 0.30143945808636746,
      "grad_norm": 0.03679900988936424,
      "learning_rate": 2.2902437390188737e-06,
      "loss": 10.3463,
      "step": 178
    },
    {
      "epoch": 0.30313293818797626,
      "grad_norm": 0.041116807609796524,
      "learning_rate": 2.0888213459343587e-06,
      "loss": 10.3455,
      "step": 179
    },
    {
      "epoch": 0.3048264182895851,
      "grad_norm": 0.03838905692100525,
      "learning_rate": 1.8963965404777875e-06,
      "loss": 10.3452,
      "step": 180
    },
    {
      "epoch": 0.3065198983911939,
      "grad_norm": 0.039337340742349625,
      "learning_rate": 1.7130219296696263e-06,
      "loss": 10.3459,
      "step": 181
    },
    {
      "epoch": 0.3082133784928027,
      "grad_norm": 0.032874945551157,
      "learning_rate": 1.5387476462974824e-06,
      "loss": 10.346,
      "step": 182
    },
    {
      "epoch": 0.3099068585944115,
      "grad_norm": 0.04064587503671646,
      "learning_rate": 1.3736213352103147e-06,
      "loss": 10.3462,
      "step": 183
    },
    {
      "epoch": 0.31160033869602033,
      "grad_norm": 0.03022320568561554,
      "learning_rate": 1.2176881402928002e-06,
      "loss": 10.3447,
      "step": 184
    },
    {
      "epoch": 0.31329381879762913,
      "grad_norm": 0.04129958897829056,
      "learning_rate": 1.0709906921234367e-06,
      "loss": 10.344,
      "step": 185
    },
    {
      "epoch": 0.31498729889923793,
      "grad_norm": 0.04146081581711769,
      "learning_rate": 9.33569096319799e-07,
      "loss": 10.3444,
      "step": 186
    },
    {
      "epoch": 0.31668077900084673,
      "grad_norm": 0.045574747025966644,
      "learning_rate": 8.054609225740255e-07,
      "loss": 10.3443,
      "step": 187
    },
    {
      "epoch": 0.31837425910245554,
      "grad_norm": 0.04817044734954834,
      "learning_rate": 6.867011943816724e-07,
      "loss": 10.3444,
      "step": 188
    },
    {
      "epoch": 0.32006773920406434,
      "grad_norm": 0.04357161372900009,
      "learning_rate": 5.77322379466617e-07,
      "loss": 10.3441,
      "step": 189
    },
    {
      "epoch": 0.32176121930567314,
      "grad_norm": 0.04117486625909805,
      "learning_rate": 4.773543809047186e-07,
      "loss": 10.3451,
      "step": 190
    },
    {
      "epoch": 0.32345469940728194,
      "grad_norm": 0.04571401700377464,
      "learning_rate": 3.868245289486027e-07,
      "loss": 10.3444,
      "step": 191
    },
    {
      "epoch": 0.32514817950889074,
      "grad_norm": 0.0635538250207901,
      "learning_rate": 3.0575757355586817e-07,
      "loss": 10.3452,
      "step": 192
    },
    {
      "epoch": 0.3268416596104996,
      "grad_norm": 0.046295084059238434,
      "learning_rate": 2.3417567762266497e-07,
      "loss": 10.3457,
      "step": 193
    },
    {
      "epoch": 0.3285351397121084,
      "grad_norm": 0.05600731447339058,
      "learning_rate": 1.7209841092460043e-07,
      "loss": 10.3441,
      "step": 194
    },
    {
      "epoch": 0.3302286198137172,
      "grad_norm": 0.05414403975009918,
      "learning_rate": 1.1954274476655534e-07,
      "loss": 10.3467,
      "step": 195
    },
    {
      "epoch": 0.331922099915326,
      "grad_norm": 0.07247443497180939,
      "learning_rate": 7.652304734289127e-08,
      "loss": 10.3455,
      "step": 196
    },
    {
      "epoch": 0.3336155800169348,
      "grad_norm": 0.08474735170602798,
      "learning_rate": 4.30510798093342e-08,
      "loss": 10.344,
      "step": 197
    },
    {
      "epoch": 0.3353090601185436,
      "grad_norm": 0.07254213094711304,
      "learning_rate": 1.9135993067588284e-08,
      "loss": 10.3455,
      "step": 198
    },
    {
      "epoch": 0.3370025402201524,
      "grad_norm": 0.11007220298051834,
      "learning_rate": 4.784325263584854e-09,
      "loss": 10.3485,
      "step": 199
    },
    {
      "epoch": 0.3386960203217612,
      "grad_norm": 0.1537920981645584,
      "learning_rate": 0.0,
      "loss": 10.3449,
      "step": 200
    },
    {
      "epoch": 0.3386960203217612,
      "eval_loss": 10.34589672088623,
      "eval_runtime": 2.2612,
      "eval_samples_per_second": 440.04,
      "eval_steps_per_second": 110.121,
      "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": 4,
        "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": 22923660754944.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}