{
  "best_metric": 0.9252543449401855,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.46323103647944414,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0023161551823972205,
      "grad_norm": 1.2503989934921265,
      "learning_rate": 7e-06,
      "loss": 2.2545,
      "step": 1
    },
    {
      "epoch": 0.0023161551823972205,
      "eval_loss": 3.0283942222595215,
      "eval_runtime": 55.3246,
      "eval_samples_per_second": 13.141,
      "eval_steps_per_second": 3.29,
      "step": 1
    },
    {
      "epoch": 0.004632310364794441,
      "grad_norm": 1.3826779127120972,
      "learning_rate": 1.4e-05,
      "loss": 2.3452,
      "step": 2
    },
    {
      "epoch": 0.006948465547191662,
      "grad_norm": 1.4480987787246704,
      "learning_rate": 2.1e-05,
      "loss": 2.5906,
      "step": 3
    },
    {
      "epoch": 0.009264620729588882,
      "grad_norm": 1.3076990842819214,
      "learning_rate": 2.8e-05,
      "loss": 2.624,
      "step": 4
    },
    {
      "epoch": 0.011580775911986103,
      "grad_norm": 1.1654059886932373,
      "learning_rate": 3.5e-05,
      "loss": 2.4437,
      "step": 5
    },
    {
      "epoch": 0.013896931094383324,
      "grad_norm": 1.1110953092575073,
      "learning_rate": 4.2e-05,
      "loss": 2.6337,
      "step": 6
    },
    {
      "epoch": 0.016213086276780544,
      "grad_norm": 1.1773900985717773,
      "learning_rate": 4.899999999999999e-05,
      "loss": 2.7596,
      "step": 7
    },
    {
      "epoch": 0.018529241459177764,
      "grad_norm": 1.8836665153503418,
      "learning_rate": 5.6e-05,
      "loss": 2.6959,
      "step": 8
    },
    {
      "epoch": 0.020845396641574986,
      "grad_norm": 1.7037103176116943,
      "learning_rate": 6.3e-05,
      "loss": 2.3329,
      "step": 9
    },
    {
      "epoch": 0.023161551823972205,
      "grad_norm": 1.4448015689849854,
      "learning_rate": 7e-05,
      "loss": 2.3956,
      "step": 10
    },
    {
      "epoch": 0.025477707006369428,
      "grad_norm": 1.1605085134506226,
      "learning_rate": 6.999521567473641e-05,
      "loss": 2.3339,
      "step": 11
    },
    {
      "epoch": 0.027793862188766647,
      "grad_norm": 1.33013117313385,
      "learning_rate": 6.998086400693241e-05,
      "loss": 2.2781,
      "step": 12
    },
    {
      "epoch": 0.030110017371163866,
      "grad_norm": 1.4077861309051514,
      "learning_rate": 6.995694892019065e-05,
      "loss": 2.3166,
      "step": 13
    },
    {
      "epoch": 0.03242617255356109,
      "grad_norm": 1.2181549072265625,
      "learning_rate": 6.99234769526571e-05,
      "loss": 2.2675,
      "step": 14
    },
    {
      "epoch": 0.03474232773595831,
      "grad_norm": 1.1249672174453735,
      "learning_rate": 6.988045725523343e-05,
      "loss": 2.1643,
      "step": 15
    },
    {
      "epoch": 0.03705848291835553,
      "grad_norm": 1.5914721488952637,
      "learning_rate": 6.982790158907539e-05,
      "loss": 2.2916,
      "step": 16
    },
    {
      "epoch": 0.03937463810075275,
      "grad_norm": 1.4551798105239868,
      "learning_rate": 6.976582432237733e-05,
      "loss": 2.1199,
      "step": 17
    },
    {
      "epoch": 0.04169079328314997,
      "grad_norm": 1.2567614316940308,
      "learning_rate": 6.969424242644413e-05,
      "loss": 2.0909,
      "step": 18
    },
    {
      "epoch": 0.04400694846554719,
      "grad_norm": 1.2276580333709717,
      "learning_rate": 6.961317547105138e-05,
      "loss": 1.8407,
      "step": 19
    },
    {
      "epoch": 0.04632310364794441,
      "grad_norm": 1.128091812133789,
      "learning_rate": 6.952264561909527e-05,
      "loss": 2.0004,
      "step": 20
    },
    {
      "epoch": 0.04863925883034163,
      "grad_norm": 1.1621633768081665,
      "learning_rate": 6.942267762053337e-05,
      "loss": 2.0205,
      "step": 21
    },
    {
      "epoch": 0.050955414012738856,
      "grad_norm": 1.2083147764205933,
      "learning_rate": 6.931329880561832e-05,
      "loss": 1.8645,
      "step": 22
    },
    {
      "epoch": 0.05327156919513607,
      "grad_norm": 1.1491119861602783,
      "learning_rate": 6.919453907742597e-05,
      "loss": 2.0392,
      "step": 23
    },
    {
      "epoch": 0.055587724377533294,
      "grad_norm": 1.2552692890167236,
      "learning_rate": 6.90664309036802e-05,
      "loss": 1.9381,
      "step": 24
    },
    {
      "epoch": 0.05790387955993052,
      "grad_norm": 1.2837193012237549,
      "learning_rate": 6.892900930787656e-05,
      "loss": 1.8941,
      "step": 25
    },
    {
      "epoch": 0.06022003474232773,
      "grad_norm": 1.2813526391983032,
      "learning_rate": 6.87823118597072e-05,
      "loss": 1.9548,
      "step": 26
    },
    {
      "epoch": 0.06253618992472496,
      "grad_norm": 1.272463083267212,
      "learning_rate": 6.862637866478969e-05,
      "loss": 1.9528,
      "step": 27
    },
    {
      "epoch": 0.06485234510712218,
      "grad_norm": 1.2786381244659424,
      "learning_rate": 6.846125235370252e-05,
      "loss": 1.7617,
      "step": 28
    },
    {
      "epoch": 0.0671685002895194,
      "grad_norm": 1.199897289276123,
      "learning_rate": 6.828697807033038e-05,
      "loss": 1.7425,
      "step": 29
    },
    {
      "epoch": 0.06948465547191662,
      "grad_norm": 1.1612555980682373,
      "learning_rate": 6.81036034595222e-05,
      "loss": 1.6898,
      "step": 30
    },
    {
      "epoch": 0.07180081065431385,
      "grad_norm": 1.4991174936294556,
      "learning_rate": 6.791117865406564e-05,
      "loss": 1.4727,
      "step": 31
    },
    {
      "epoch": 0.07411696583671105,
      "grad_norm": 1.406917929649353,
      "learning_rate": 6.770975626098112e-05,
      "loss": 1.5182,
      "step": 32
    },
    {
      "epoch": 0.07643312101910828,
      "grad_norm": 1.5695204734802246,
      "learning_rate": 6.749939134713974e-05,
      "loss": 1.6764,
      "step": 33
    },
    {
      "epoch": 0.0787492762015055,
      "grad_norm": 1.4783178567886353,
      "learning_rate": 6.728014142420846e-05,
      "loss": 1.4336,
      "step": 34
    },
    {
      "epoch": 0.08106543138390272,
      "grad_norm": 1.5821142196655273,
      "learning_rate": 6.7052066432927e-05,
      "loss": 1.7416,
      "step": 35
    },
    {
      "epoch": 0.08338158656629995,
      "grad_norm": 1.3772616386413574,
      "learning_rate": 6.681522872672069e-05,
      "loss": 1.5644,
      "step": 36
    },
    {
      "epoch": 0.08569774174869717,
      "grad_norm": 1.5498988628387451,
      "learning_rate": 6.656969305465356e-05,
      "loss": 1.5495,
      "step": 37
    },
    {
      "epoch": 0.08801389693109438,
      "grad_norm": 1.627152681350708,
      "learning_rate": 6.631552654372672e-05,
      "loss": 1.6281,
      "step": 38
    },
    {
      "epoch": 0.0903300521134916,
      "grad_norm": 1.6475415229797363,
      "learning_rate": 6.60527986805264e-05,
      "loss": 1.3677,
      "step": 39
    },
    {
      "epoch": 0.09264620729588882,
      "grad_norm": 1.7694090604782104,
      "learning_rate": 6.578158129222711e-05,
      "loss": 1.4586,
      "step": 40
    },
    {
      "epoch": 0.09496236247828604,
      "grad_norm": 1.7969664335250854,
      "learning_rate": 6.550194852695469e-05,
      "loss": 1.3842,
      "step": 41
    },
    {
      "epoch": 0.09727851766068327,
      "grad_norm": 1.6679091453552246,
      "learning_rate": 6.521397683351509e-05,
      "loss": 1.3623,
      "step": 42
    },
    {
      "epoch": 0.09959467284308049,
      "grad_norm": 1.8218728303909302,
      "learning_rate": 6.491774494049386e-05,
      "loss": 1.4426,
      "step": 43
    },
    {
      "epoch": 0.10191082802547771,
      "grad_norm": 1.8028229475021362,
      "learning_rate": 6.461333383473272e-05,
      "loss": 1.5289,
      "step": 44
    },
    {
      "epoch": 0.10422698320787492,
      "grad_norm": 1.7978172302246094,
      "learning_rate": 6.430082673918849e-05,
      "loss": 1.365,
      "step": 45
    },
    {
      "epoch": 0.10654313839027214,
      "grad_norm": 2.145550489425659,
      "learning_rate": 6.398030909018069e-05,
      "loss": 1.3449,
      "step": 46
    },
    {
      "epoch": 0.10885929357266937,
      "grad_norm": 2.116974115371704,
      "learning_rate": 6.365186851403423e-05,
      "loss": 1.3279,
      "step": 47
    },
    {
      "epoch": 0.11117544875506659,
      "grad_norm": 2.0331828594207764,
      "learning_rate": 6.331559480312315e-05,
      "loss": 1.2943,
      "step": 48
    },
    {
      "epoch": 0.11349160393746381,
      "grad_norm": 2.129798650741577,
      "learning_rate": 6.297157989132236e-05,
      "loss": 1.0358,
      "step": 49
    },
    {
      "epoch": 0.11580775911986103,
      "grad_norm": 2.835911750793457,
      "learning_rate": 6.261991782887377e-05,
      "loss": 1.2203,
      "step": 50
    },
    {
      "epoch": 0.11580775911986103,
      "eval_loss": 1.6897388696670532,
      "eval_runtime": 56.0898,
      "eval_samples_per_second": 12.961,
      "eval_steps_per_second": 3.245,
      "step": 50
    },
    {
      "epoch": 0.11812391430225826,
      "grad_norm": 2.849514961242676,
      "learning_rate": 6.226070475667393e-05,
      "loss": 2.0836,
      "step": 51
    },
    {
      "epoch": 0.12044006948465547,
      "grad_norm": 2.162397623062134,
      "learning_rate": 6.189403887999006e-05,
      "loss": 2.0693,
      "step": 52
    },
    {
      "epoch": 0.12275622466705269,
      "grad_norm": 1.8029311895370483,
      "learning_rate": 6.152002044161171e-05,
      "loss": 2.0471,
      "step": 53
    },
    {
      "epoch": 0.1250723798494499,
      "grad_norm": 1.3188810348510742,
      "learning_rate": 6.113875169444539e-05,
      "loss": 2.0245,
      "step": 54
    },
    {
      "epoch": 0.12738853503184713,
      "grad_norm": 1.1737282276153564,
      "learning_rate": 6.0750336873559605e-05,
      "loss": 1.8553,
      "step": 55
    },
    {
      "epoch": 0.12970469021424436,
      "grad_norm": 1.0292160511016846,
      "learning_rate": 6.035488216768811e-05,
      "loss": 1.868,
      "step": 56
    },
    {
      "epoch": 0.13202084539664158,
      "grad_norm": 0.96671462059021,
      "learning_rate": 5.9952495690198894e-05,
      "loss": 1.913,
      "step": 57
    },
    {
      "epoch": 0.1343370005790388,
      "grad_norm": 1.1832921504974365,
      "learning_rate": 5.954328744953709e-05,
      "loss": 1.7736,
      "step": 58
    },
    {
      "epoch": 0.13665315576143602,
      "grad_norm": 1.0528168678283691,
      "learning_rate": 5.91273693191498e-05,
      "loss": 1.9491,
      "step": 59
    },
    {
      "epoch": 0.13896931094383325,
      "grad_norm": 1.0293359756469727,
      "learning_rate": 5.870485500690094e-05,
      "loss": 1.7518,
      "step": 60
    },
    {
      "epoch": 0.14128546612623047,
      "grad_norm": 1.0944894552230835,
      "learning_rate": 5.827586002398468e-05,
      "loss": 1.7148,
      "step": 61
    },
    {
      "epoch": 0.1436016213086277,
      "grad_norm": 1.0049891471862793,
      "learning_rate": 5.784050165334589e-05,
      "loss": 1.6303,
      "step": 62
    },
    {
      "epoch": 0.1459177764910249,
      "grad_norm": 1.0505611896514893,
      "learning_rate": 5.739889891761608e-05,
      "loss": 1.6231,
      "step": 63
    },
    {
      "epoch": 0.1482339316734221,
      "grad_norm": 1.2406843900680542,
      "learning_rate": 5.6951172546573794e-05,
      "loss": 1.8787,
      "step": 64
    },
    {
      "epoch": 0.15055008685581933,
      "grad_norm": 1.1927564144134521,
      "learning_rate": 5.6497444944138376e-05,
      "loss": 1.7534,
      "step": 65
    },
    {
      "epoch": 0.15286624203821655,
      "grad_norm": 1.2264256477355957,
      "learning_rate": 5.603784015490587e-05,
      "loss": 1.6633,
      "step": 66
    },
    {
      "epoch": 0.15518239722061378,
      "grad_norm": 1.3675661087036133,
      "learning_rate": 5.557248383023655e-05,
      "loss": 1.5166,
      "step": 67
    },
    {
      "epoch": 0.157498552403011,
      "grad_norm": 1.3764818906784058,
      "learning_rate": 5.510150319390302e-05,
      "loss": 1.4764,
      "step": 68
    },
    {
      "epoch": 0.15981470758540822,
      "grad_norm": 1.3945223093032837,
      "learning_rate": 5.4625027007308546e-05,
      "loss": 1.5581,
      "step": 69
    },
    {
      "epoch": 0.16213086276780544,
      "grad_norm": 1.2444453239440918,
      "learning_rate": 5.414318553428494e-05,
      "loss": 1.4439,
      "step": 70
    },
    {
      "epoch": 0.16444701795020267,
      "grad_norm": 1.3003394603729248,
      "learning_rate": 5.3656110505479776e-05,
      "loss": 1.6489,
      "step": 71
    },
    {
      "epoch": 0.1667631731325999,
      "grad_norm": 1.247867226600647,
      "learning_rate": 5.316393508234253e-05,
      "loss": 1.4745,
      "step": 72
    },
    {
      "epoch": 0.1690793283149971,
      "grad_norm": 1.249312162399292,
      "learning_rate": 5.266679382071953e-05,
      "loss": 1.5594,
      "step": 73
    },
    {
      "epoch": 0.17139548349739434,
      "grad_norm": 1.2661000490188599,
      "learning_rate": 5.216482263406778e-05,
      "loss": 1.4676,
      "step": 74
    },
    {
      "epoch": 0.17371163867979156,
      "grad_norm": 1.3261014223098755,
      "learning_rate": 5.1658158756297576e-05,
      "loss": 1.4847,
      "step": 75
    },
    {
      "epoch": 0.17602779386218875,
      "grad_norm": 1.2597907781600952,
      "learning_rate": 5.114694070425407e-05,
      "loss": 1.3025,
      "step": 76
    },
    {
      "epoch": 0.17834394904458598,
      "grad_norm": 1.215097427368164,
      "learning_rate": 5.063130823984823e-05,
      "loss": 1.2496,
      "step": 77
    },
    {
      "epoch": 0.1806601042269832,
      "grad_norm": 1.2596161365509033,
      "learning_rate": 5.011140233184724e-05,
      "loss": 1.2445,
      "step": 78
    },
    {
      "epoch": 0.18297625940938042,
      "grad_norm": 1.4308902025222778,
      "learning_rate": 4.958736511733516e-05,
      "loss": 1.2937,
      "step": 79
    },
    {
      "epoch": 0.18529241459177764,
      "grad_norm": 1.5676943063735962,
      "learning_rate": 4.905933986285393e-05,
      "loss": 1.1867,
      "step": 80
    },
    {
      "epoch": 0.18760856977417487,
      "grad_norm": 1.5051430463790894,
      "learning_rate": 4.8527470925235824e-05,
      "loss": 1.1516,
      "step": 81
    },
    {
      "epoch": 0.1899247249565721,
      "grad_norm": 1.4236093759536743,
      "learning_rate": 4.799190371213772e-05,
      "loss": 0.9429,
      "step": 82
    },
    {
      "epoch": 0.1922408801389693,
      "grad_norm": 1.6276626586914062,
      "learning_rate": 4.745278464228808e-05,
      "loss": 1.1588,
      "step": 83
    },
    {
      "epoch": 0.19455703532136653,
      "grad_norm": 1.5213537216186523,
      "learning_rate": 4.69102611054575e-05,
      "loss": 1.1604,
      "step": 84
    },
    {
      "epoch": 0.19687319050376376,
      "grad_norm": 1.7984035015106201,
      "learning_rate": 4.6364481422163926e-05,
      "loss": 1.1797,
      "step": 85
    },
    {
      "epoch": 0.19918934568616098,
      "grad_norm": 1.4924752712249756,
      "learning_rate": 4.581559480312316e-05,
      "loss": 1.0718,
      "step": 86
    },
    {
      "epoch": 0.2015055008685582,
      "grad_norm": 1.7816964387893677,
      "learning_rate": 4.526375130845627e-05,
      "loss": 1.1378,
      "step": 87
    },
    {
      "epoch": 0.20382165605095542,
      "grad_norm": 1.7848283052444458,
      "learning_rate": 4.4709101806664554e-05,
      "loss": 0.9752,
      "step": 88
    },
    {
      "epoch": 0.20613781123335265,
      "grad_norm": 1.836548089981079,
      "learning_rate": 4.4151797933383685e-05,
      "loss": 0.9158,
      "step": 89
    },
    {
      "epoch": 0.20845396641574984,
      "grad_norm": 1.678355097770691,
      "learning_rate": 4.359199204992797e-05,
      "loss": 0.9598,
      "step": 90
    },
    {
      "epoch": 0.21077012159814706,
      "grad_norm": 1.9656362533569336,
      "learning_rate": 4.30298372016363e-05,
      "loss": 1.1572,
      "step": 91
    },
    {
      "epoch": 0.2130862767805443,
      "grad_norm": 2.5709915161132812,
      "learning_rate": 4.246548707603114e-05,
      "loss": 0.8351,
      "step": 92
    },
    {
      "epoch": 0.2154024319629415,
      "grad_norm": 1.8190523386001587,
      "learning_rate": 4.1899095960801805e-05,
      "loss": 0.9647,
      "step": 93
    },
    {
      "epoch": 0.21771858714533873,
      "grad_norm": 2.0618791580200195,
      "learning_rate": 4.133081870162385e-05,
      "loss": 1.0824,
      "step": 94
    },
    {
      "epoch": 0.22003474232773595,
      "grad_norm": 2.0086538791656494,
      "learning_rate": 4.076081065982569e-05,
      "loss": 1.093,
      "step": 95
    },
    {
      "epoch": 0.22235089751013318,
      "grad_norm": 2.2027931213378906,
      "learning_rate": 4.018922766991447e-05,
      "loss": 1.1038,
      "step": 96
    },
    {
      "epoch": 0.2246670526925304,
      "grad_norm": 1.8367047309875488,
      "learning_rate": 3.961622599697241e-05,
      "loss": 0.9804,
      "step": 97
    },
    {
      "epoch": 0.22698320787492762,
      "grad_norm": 2.6135077476501465,
      "learning_rate": 3.9041962293935516e-05,
      "loss": 0.9048,
      "step": 98
    },
    {
      "epoch": 0.22929936305732485,
      "grad_norm": 2.061969518661499,
      "learning_rate": 3.84665935587662e-05,
      "loss": 0.8975,
      "step": 99
    },
    {
      "epoch": 0.23161551823972207,
      "grad_norm": 2.850947141647339,
      "learning_rate": 3.7890277091531636e-05,
      "loss": 0.8124,
      "step": 100
    },
    {
      "epoch": 0.23161551823972207,
      "eval_loss": 1.2259724140167236,
      "eval_runtime": 56.0663,
      "eval_samples_per_second": 12.967,
      "eval_steps_per_second": 3.246,
      "step": 100
    },
    {
      "epoch": 0.2339316734221193,
      "grad_norm": 1.7626456022262573,
      "learning_rate": 3.7313170451399475e-05,
      "loss": 1.9503,
      "step": 101
    },
    {
      "epoch": 0.2362478286045165,
      "grad_norm": 1.505061388015747,
      "learning_rate": 3.673543141356278e-05,
      "loss": 1.8245,
      "step": 102
    },
    {
      "epoch": 0.23856398378691374,
      "grad_norm": 1.3618402481079102,
      "learning_rate": 3.6157217926105783e-05,
      "loss": 1.7775,
      "step": 103
    },
    {
      "epoch": 0.24088013896931093,
      "grad_norm": 1.3016786575317383,
      "learning_rate": 3.557868806682255e-05,
      "loss": 1.6679,
      "step": 104
    },
    {
      "epoch": 0.24319629415170815,
      "grad_norm": 1.2957885265350342,
      "learning_rate": 3.5e-05,
      "loss": 1.6692,
      "step": 105
    },
    {
      "epoch": 0.24551244933410538,
      "grad_norm": 1.004072666168213,
      "learning_rate": 3.442131193317745e-05,
      "loss": 1.521,
      "step": 106
    },
    {
      "epoch": 0.2478286045165026,
      "grad_norm": 1.0314702987670898,
      "learning_rate": 3.384278207389421e-05,
      "loss": 1.6253,
      "step": 107
    },
    {
      "epoch": 0.2501447596988998,
      "grad_norm": 1.1805181503295898,
      "learning_rate": 3.3264568586437216e-05,
      "loss": 1.4228,
      "step": 108
    },
    {
      "epoch": 0.25246091488129707,
      "grad_norm": 1.028948187828064,
      "learning_rate": 3.268682954860052e-05,
      "loss": 1.4977,
      "step": 109
    },
    {
      "epoch": 0.25477707006369427,
      "grad_norm": 1.0726194381713867,
      "learning_rate": 3.210972290846837e-05,
      "loss": 1.3065,
      "step": 110
    },
    {
      "epoch": 0.25709322524609146,
      "grad_norm": 1.1146329641342163,
      "learning_rate": 3.15334064412338e-05,
      "loss": 1.0397,
      "step": 111
    },
    {
      "epoch": 0.2594093804284887,
      "grad_norm": 1.1305572986602783,
      "learning_rate": 3.0958037706064485e-05,
      "loss": 1.2692,
      "step": 112
    },
    {
      "epoch": 0.2617255356108859,
      "grad_norm": 1.2434415817260742,
      "learning_rate": 3.038377400302758e-05,
      "loss": 1.3937,
      "step": 113
    },
    {
      "epoch": 0.26404169079328316,
      "grad_norm": 1.2750946283340454,
      "learning_rate": 2.9810772330085524e-05,
      "loss": 1.3832,
      "step": 114
    },
    {
      "epoch": 0.26635784597568035,
      "grad_norm": 1.4428255558013916,
      "learning_rate": 2.9239189340174306e-05,
      "loss": 1.134,
      "step": 115
    },
    {
      "epoch": 0.2686740011580776,
      "grad_norm": 1.4373363256454468,
      "learning_rate": 2.8669181298376163e-05,
      "loss": 1.2468,
      "step": 116
    },
    {
      "epoch": 0.2709901563404748,
      "grad_norm": 1.548967719078064,
      "learning_rate": 2.8100904039198193e-05,
      "loss": 1.0152,
      "step": 117
    },
    {
      "epoch": 0.27330631152287205,
      "grad_norm": 1.930369257926941,
      "learning_rate": 2.7534512923968863e-05,
      "loss": 1.2061,
      "step": 118
    },
    {
      "epoch": 0.27562246670526924,
      "grad_norm": 1.7386815547943115,
      "learning_rate": 2.6970162798363695e-05,
      "loss": 1.1742,
      "step": 119
    },
    {
      "epoch": 0.2779386218876665,
      "grad_norm": 1.7060924768447876,
      "learning_rate": 2.640800795007203e-05,
      "loss": 1.2716,
      "step": 120
    },
    {
      "epoch": 0.2802547770700637,
      "grad_norm": 1.5493820905685425,
      "learning_rate": 2.5848202066616305e-05,
      "loss": 1.2847,
      "step": 121
    },
    {
      "epoch": 0.28257093225246094,
      "grad_norm": 1.513087272644043,
      "learning_rate": 2.5290898193335446e-05,
      "loss": 1.2291,
      "step": 122
    },
    {
      "epoch": 0.28488708743485813,
      "grad_norm": 1.405207633972168,
      "learning_rate": 2.4736248691543736e-05,
      "loss": 1.0475,
      "step": 123
    },
    {
      "epoch": 0.2872032426172554,
      "grad_norm": 1.4663727283477783,
      "learning_rate": 2.4184405196876842e-05,
      "loss": 0.9763,
      "step": 124
    },
    {
      "epoch": 0.2895193977996526,
      "grad_norm": 1.33937406539917,
      "learning_rate": 2.363551857783608e-05,
      "loss": 0.981,
      "step": 125
    },
    {
      "epoch": 0.2918355529820498,
      "grad_norm": 1.4397436380386353,
      "learning_rate": 2.308973889454249e-05,
      "loss": 1.1096,
      "step": 126
    },
    {
      "epoch": 0.294151708164447,
      "grad_norm": 1.3392612934112549,
      "learning_rate": 2.2547215357711918e-05,
      "loss": 0.9199,
      "step": 127
    },
    {
      "epoch": 0.2964678633468442,
      "grad_norm": 1.3941303491592407,
      "learning_rate": 2.2008096287862266e-05,
      "loss": 0.885,
      "step": 128
    },
    {
      "epoch": 0.29878401852924147,
      "grad_norm": 1.5315736532211304,
      "learning_rate": 2.1472529074764177e-05,
      "loss": 0.9312,
      "step": 129
    },
    {
      "epoch": 0.30110017371163866,
      "grad_norm": 1.461169719696045,
      "learning_rate": 2.0940660137146074e-05,
      "loss": 0.8541,
      "step": 130
    },
    {
      "epoch": 0.3034163288940359,
      "grad_norm": 1.5014770030975342,
      "learning_rate": 2.041263488266484e-05,
      "loss": 0.8172,
      "step": 131
    },
    {
      "epoch": 0.3057324840764331,
      "grad_norm": 1.8396813869476318,
      "learning_rate": 1.988859766815275e-05,
      "loss": 0.8973,
      "step": 132
    },
    {
      "epoch": 0.30804863925883036,
      "grad_norm": 1.7794278860092163,
      "learning_rate": 1.9368691760151773e-05,
      "loss": 0.9055,
      "step": 133
    },
    {
      "epoch": 0.31036479444122755,
      "grad_norm": 1.544495701789856,
      "learning_rate": 1.885305929574593e-05,
      "loss": 0.8416,
      "step": 134
    },
    {
      "epoch": 0.3126809496236248,
      "grad_norm": 1.71176278591156,
      "learning_rate": 1.8341841243702424e-05,
      "loss": 0.8015,
      "step": 135
    },
    {
      "epoch": 0.314997104806022,
      "grad_norm": 1.837409257888794,
      "learning_rate": 1.7835177365932225e-05,
      "loss": 0.7487,
      "step": 136
    },
    {
      "epoch": 0.31731325998841925,
      "grad_norm": 1.8642661571502686,
      "learning_rate": 1.7333206179280478e-05,
      "loss": 0.8194,
      "step": 137
    },
    {
      "epoch": 0.31962941517081644,
      "grad_norm": 1.9652878046035767,
      "learning_rate": 1.6836064917657478e-05,
      "loss": 0.7471,
      "step": 138
    },
    {
      "epoch": 0.32194557035321364,
      "grad_norm": 1.8793476819992065,
      "learning_rate": 1.6343889494520224e-05,
      "loss": 0.5699,
      "step": 139
    },
    {
      "epoch": 0.3242617255356109,
      "grad_norm": 1.8087778091430664,
      "learning_rate": 1.5856814465715064e-05,
      "loss": 0.702,
      "step": 140
    },
    {
      "epoch": 0.3265778807180081,
      "grad_norm": 1.814977765083313,
      "learning_rate": 1.5374972992691458e-05,
      "loss": 0.6915,
      "step": 141
    },
    {
      "epoch": 0.32889403590040533,
      "grad_norm": 1.9736248254776,
      "learning_rate": 1.4898496806096974e-05,
      "loss": 0.6077,
      "step": 142
    },
    {
      "epoch": 0.33121019108280253,
      "grad_norm": 2.0833942890167236,
      "learning_rate": 1.4427516169763444e-05,
      "loss": 0.6608,
      "step": 143
    },
    {
      "epoch": 0.3335263462651998,
      "grad_norm": 2.1390953063964844,
      "learning_rate": 1.396215984509412e-05,
      "loss": 0.614,
      "step": 144
    },
    {
      "epoch": 0.335842501447597,
      "grad_norm": 2.002488851547241,
      "learning_rate": 1.3502555055861625e-05,
      "loss": 0.6135,
      "step": 145
    },
    {
      "epoch": 0.3381586566299942,
      "grad_norm": 2.446239471435547,
      "learning_rate": 1.3048827453426203e-05,
      "loss": 0.7834,
      "step": 146
    },
    {
      "epoch": 0.3404748118123914,
      "grad_norm": 2.466515302658081,
      "learning_rate": 1.2601101082383917e-05,
      "loss": 0.7394,
      "step": 147
    },
    {
      "epoch": 0.34279096699478867,
      "grad_norm": 2.9753072261810303,
      "learning_rate": 1.2159498346654094e-05,
      "loss": 0.6323,
      "step": 148
    },
    {
      "epoch": 0.34510712217718587,
      "grad_norm": 2.352726459503174,
      "learning_rate": 1.1724139976015306e-05,
      "loss": 0.5433,
      "step": 149
    },
    {
      "epoch": 0.3474232773595831,
      "grad_norm": 2.822594404220581,
      "learning_rate": 1.1295144993099068e-05,
      "loss": 0.6062,
      "step": 150
    },
    {
      "epoch": 0.3474232773595831,
      "eval_loss": 0.9819753766059875,
      "eval_runtime": 56.0681,
      "eval_samples_per_second": 12.966,
      "eval_steps_per_second": 3.246,
      "step": 150
    },
    {
      "epoch": 0.3497394325419803,
      "grad_norm": 1.6045421361923218,
      "learning_rate": 1.0872630680850196e-05,
      "loss": 1.6882,
      "step": 151
    },
    {
      "epoch": 0.3520555877243775,
      "grad_norm": 1.5667160749435425,
      "learning_rate": 1.0456712550462898e-05,
      "loss": 1.4824,
      "step": 152
    },
    {
      "epoch": 0.35437174290677476,
      "grad_norm": 1.938839316368103,
      "learning_rate": 1.0047504309801104e-05,
      "loss": 1.4432,
      "step": 153
    },
    {
      "epoch": 0.35668789808917195,
      "grad_norm": 1.8820574283599854,
      "learning_rate": 9.645117832311886e-06,
      "loss": 1.3595,
      "step": 154
    },
    {
      "epoch": 0.3590040532715692,
      "grad_norm": 1.5141949653625488,
      "learning_rate": 9.249663126440394e-06,
      "loss": 1.442,
      "step": 155
    },
    {
      "epoch": 0.3613202084539664,
      "grad_norm": 1.4562164545059204,
      "learning_rate": 8.861248305554624e-06,
      "loss": 1.4735,
      "step": 156
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 1.6026912927627563,
      "learning_rate": 8.47997955838829e-06,
      "loss": 1.2689,
      "step": 157
    },
    {
      "epoch": 0.36595251881876084,
      "grad_norm": 1.5428178310394287,
      "learning_rate": 8.10596112000994e-06,
      "loss": 1.3727,
      "step": 158
    },
    {
      "epoch": 0.3682686740011581,
      "grad_norm": 1.2998900413513184,
      "learning_rate": 7.739295243326067e-06,
      "loss": 1.1046,
      "step": 159
    },
    {
      "epoch": 0.3705848291835553,
      "grad_norm": 1.3930641412734985,
      "learning_rate": 7.380082171126228e-06,
      "loss": 1.2705,
      "step": 160
    },
    {
      "epoch": 0.37290098436595254,
      "grad_norm": 1.318785309791565,
      "learning_rate": 7.028420108677635e-06,
      "loss": 1.2158,
      "step": 161
    },
    {
      "epoch": 0.37521713954834973,
      "grad_norm": 1.42453932762146,
      "learning_rate": 6.684405196876842e-06,
      "loss": 1.3409,
      "step": 162
    },
    {
      "epoch": 0.377533294730747,
      "grad_norm": 1.6251355409622192,
      "learning_rate": 6.3481314859657675e-06,
      "loss": 1.1197,
      "step": 163
    },
    {
      "epoch": 0.3798494499131442,
      "grad_norm": 1.3056257963180542,
      "learning_rate": 6.019690909819298e-06,
      "loss": 1.1391,
      "step": 164
    },
    {
      "epoch": 0.3821656050955414,
      "grad_norm": 1.4243314266204834,
      "learning_rate": 5.6991732608115e-06,
      "loss": 1.3049,
      "step": 165
    },
    {
      "epoch": 0.3844817602779386,
      "grad_norm": 1.621582269668579,
      "learning_rate": 5.386666165267256e-06,
      "loss": 0.9413,
      "step": 166
    },
    {
      "epoch": 0.3867979154603358,
      "grad_norm": 1.5309045314788818,
      "learning_rate": 5.08225505950613e-06,
      "loss": 1.024,
      "step": 167
    },
    {
      "epoch": 0.38911407064273307,
      "grad_norm": 1.4823757410049438,
      "learning_rate": 4.786023166484913e-06,
      "loss": 0.9542,
      "step": 168
    },
    {
      "epoch": 0.39143022582513026,
      "grad_norm": 1.5925859212875366,
      "learning_rate": 4.498051473045291e-06,
      "loss": 1.0656,
      "step": 169
    },
    {
      "epoch": 0.3937463810075275,
      "grad_norm": 1.4511696100234985,
      "learning_rate": 4.218418707772886e-06,
      "loss": 1.1376,
      "step": 170
    },
    {
      "epoch": 0.3960625361899247,
      "grad_norm": 1.4989674091339111,
      "learning_rate": 3.947201319473587e-06,
      "loss": 1.1977,
      "step": 171
    },
    {
      "epoch": 0.39837869137232196,
      "grad_norm": 1.7597720623016357,
      "learning_rate": 3.684473456273278e-06,
      "loss": 1.1539,
      "step": 172
    },
    {
      "epoch": 0.40069484655471915,
      "grad_norm": 1.5664502382278442,
      "learning_rate": 3.4303069453464383e-06,
      "loss": 1.035,
      "step": 173
    },
    {
      "epoch": 0.4030110017371164,
      "grad_norm": 1.5649906396865845,
      "learning_rate": 3.184771273279312e-06,
      "loss": 0.8856,
      "step": 174
    },
    {
      "epoch": 0.4053271569195136,
      "grad_norm": 1.6556047201156616,
      "learning_rate": 2.947933567072987e-06,
      "loss": 0.972,
      "step": 175
    },
    {
      "epoch": 0.40764331210191085,
      "grad_norm": 1.6981557607650757,
      "learning_rate": 2.719858575791534e-06,
      "loss": 0.7624,
      "step": 176
    },
    {
      "epoch": 0.40995946728430804,
      "grad_norm": 1.6652178764343262,
      "learning_rate": 2.500608652860256e-06,
      "loss": 0.9003,
      "step": 177
    },
    {
      "epoch": 0.4122756224667053,
      "grad_norm": 1.6681879758834839,
      "learning_rate": 2.2902437390188737e-06,
      "loss": 0.6842,
      "step": 178
    },
    {
      "epoch": 0.4145917776491025,
      "grad_norm": 1.7632416486740112,
      "learning_rate": 2.0888213459343587e-06,
      "loss": 0.7585,
      "step": 179
    },
    {
      "epoch": 0.4169079328314997,
      "grad_norm": 1.8788822889328003,
      "learning_rate": 1.8963965404777875e-06,
      "loss": 0.7509,
      "step": 180
    },
    {
      "epoch": 0.41922408801389693,
      "grad_norm": 1.7253082990646362,
      "learning_rate": 1.7130219296696263e-06,
      "loss": 0.8523,
      "step": 181
    },
    {
      "epoch": 0.42154024319629413,
      "grad_norm": 1.6087950468063354,
      "learning_rate": 1.5387476462974824e-06,
      "loss": 0.7647,
      "step": 182
    },
    {
      "epoch": 0.4238563983786914,
      "grad_norm": 1.601576328277588,
      "learning_rate": 1.3736213352103147e-06,
      "loss": 0.6134,
      "step": 183
    },
    {
      "epoch": 0.4261725535610886,
      "grad_norm": 1.6278523206710815,
      "learning_rate": 1.2176881402928002e-06,
      "loss": 0.5783,
      "step": 184
    },
    {
      "epoch": 0.4284887087434858,
      "grad_norm": 1.9031941890716553,
      "learning_rate": 1.0709906921234367e-06,
      "loss": 0.6664,
      "step": 185
    },
    {
      "epoch": 0.430804863925883,
      "grad_norm": 1.792665719985962,
      "learning_rate": 9.33569096319799e-07,
      "loss": 0.5366,
      "step": 186
    },
    {
      "epoch": 0.43312101910828027,
      "grad_norm": 2.003882884979248,
      "learning_rate": 8.054609225740255e-07,
      "loss": 0.7031,
      "step": 187
    },
    {
      "epoch": 0.43543717429067746,
      "grad_norm": 1.9855848550796509,
      "learning_rate": 6.867011943816724e-07,
      "loss": 0.5054,
      "step": 188
    },
    {
      "epoch": 0.4377533294730747,
      "grad_norm": 1.944218635559082,
      "learning_rate": 5.77322379466617e-07,
      "loss": 0.5625,
      "step": 189
    },
    {
      "epoch": 0.4400694846554719,
      "grad_norm": 1.8458359241485596,
      "learning_rate": 4.773543809047186e-07,
      "loss": 0.7162,
      "step": 190
    },
    {
      "epoch": 0.44238563983786916,
      "grad_norm": 1.8968647718429565,
      "learning_rate": 3.868245289486027e-07,
      "loss": 0.5324,
      "step": 191
    },
    {
      "epoch": 0.44470179502026635,
      "grad_norm": 1.8599846363067627,
      "learning_rate": 3.0575757355586817e-07,
      "loss": 0.6424,
      "step": 192
    },
    {
      "epoch": 0.4470179502026636,
      "grad_norm": 2.0066614151000977,
      "learning_rate": 2.3417567762266497e-07,
      "loss": 0.6599,
      "step": 193
    },
    {
      "epoch": 0.4493341053850608,
      "grad_norm": 2.488804340362549,
      "learning_rate": 1.7209841092460043e-07,
      "loss": 0.748,
      "step": 194
    },
    {
      "epoch": 0.451650260567458,
      "grad_norm": 2.2025880813598633,
      "learning_rate": 1.1954274476655534e-07,
      "loss": 0.6648,
      "step": 195
    },
    {
      "epoch": 0.45396641574985525,
      "grad_norm": 2.2725353240966797,
      "learning_rate": 7.652304734289127e-08,
      "loss": 0.6651,
      "step": 196
    },
    {
      "epoch": 0.45628257093225244,
      "grad_norm": 2.63472318649292,
      "learning_rate": 4.30510798093342e-08,
      "loss": 0.734,
      "step": 197
    },
    {
      "epoch": 0.4585987261146497,
      "grad_norm": 2.484386444091797,
      "learning_rate": 1.9135993067588284e-08,
      "loss": 0.6181,
      "step": 198
    },
    {
      "epoch": 0.4609148812970469,
      "grad_norm": 2.705622911453247,
      "learning_rate": 4.784325263584854e-09,
      "loss": 0.4998,
      "step": 199
    },
    {
      "epoch": 0.46323103647944414,
      "grad_norm": 3.950650215148926,
      "learning_rate": 0.0,
      "loss": 0.5755,
      "step": 200
    },
    {
      "epoch": 0.46323103647944414,
      "eval_loss": 0.9252543449401855,
      "eval_runtime": 56.0402,
      "eval_samples_per_second": 12.973,
      "eval_steps_per_second": 3.248,
      "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": 3.017027157491712e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}