{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 30.0,
  "eval_steps": 500,
  "global_step": 9390,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_loss": 3.8149499893188477,
      "eval_mse_0": 3.755868434906006,
      "eval_mse_1": 3.9979312419891357,
      "eval_mse_2": 3.7539875507354736,
      "eval_mse_3": 3.752009868621826,
      "eval_pcc_0": 0.19238500149154686,
      "eval_pcc_1": 0.1776398118553242,
      "eval_pcc_2": 0.007480538067184007,
      "eval_pcc_3": -0.3489487228560554,
      "eval_runtime": 102.287,
      "eval_samples_per_second": 24.441,
      "eval_steps_per_second": 3.06,
      "step": 313
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.467518636847711e-06,
      "loss": 10.7002,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_loss": 1.9381023645401,
      "eval_mse_0": 2.1219639778137207,
      "eval_mse_1": 1.756824016571045,
      "eval_mse_2": 1.7819912433624268,
      "eval_mse_3": 2.091628074645996,
      "eval_pcc_0": 0.3882147701970152,
      "eval_pcc_1": 0.44674624068925833,
      "eval_pcc_2": 0.4466174095825964,
      "eval_pcc_3": 0.3952736055599715,
      "eval_runtime": 101.3791,
      "eval_samples_per_second": 24.66,
      "eval_steps_per_second": 3.087,
      "step": 626
    },
    {
      "epoch": 3.0,
      "eval_loss": 1.68776273727417,
      "eval_mse_0": 1.873080849647522,
      "eval_mse_1": 1.4604036808013916,
      "eval_mse_2": 1.5174270868301392,
      "eval_mse_3": 1.9001433849334717,
      "eval_pcc_0": 0.4870196345841211,
      "eval_pcc_1": 0.5445910750056667,
      "eval_pcc_2": 0.5375179097868217,
      "eval_pcc_3": 0.4978717809473544,
      "eval_runtime": 101.2516,
      "eval_samples_per_second": 24.691,
      "eval_steps_per_second": 3.091,
      "step": 939
    },
    {
      "epoch": 3.19,
      "learning_rate": 8.935037273695422e-06,
      "loss": 2.2251,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "eval_loss": 1.4689956903457642,
      "eval_mse_0": 1.6715960502624512,
      "eval_mse_1": 1.2319759130477905,
      "eval_mse_2": 1.34701406955719,
      "eval_mse_3": 1.6253972053527832,
      "eval_pcc_0": 0.6013148331951307,
      "eval_pcc_1": 0.6605847059188266,
      "eval_pcc_2": 0.6530310261210702,
      "eval_pcc_3": 0.6166350338813703,
      "eval_runtime": 101.2565,
      "eval_samples_per_second": 24.69,
      "eval_steps_per_second": 3.091,
      "step": 1252
    },
    {
      "epoch": 4.79,
      "learning_rate": 8.402555910543132e-06,
      "loss": 1.1372,
      "step": 1500
    },
    {
      "epoch": 5.0,
      "eval_loss": 1.2834810018539429,
      "eval_mse_0": 1.4975959062576294,
      "eval_mse_1": 1.1127667427062988,
      "eval_mse_2": 1.0967236757278442,
      "eval_mse_3": 1.4268358945846558,
      "eval_pcc_0": 0.667617438547489,
      "eval_pcc_1": 0.7272906468479914,
      "eval_pcc_2": 0.7394717314114011,
      "eval_pcc_3": 0.6982645021045678,
      "eval_runtime": 100.9481,
      "eval_samples_per_second": 24.765,
      "eval_steps_per_second": 3.101,
      "step": 1565
    },
    {
      "epoch": 6.0,
      "eval_loss": 1.5220290422439575,
      "eval_mse_0": 1.797974944114685,
      "eval_mse_1": 1.138425588607788,
      "eval_mse_2": 1.3066158294677734,
      "eval_mse_3": 1.845100998878479,
      "eval_pcc_0": 0.6437292562481384,
      "eval_pcc_1": 0.7277782333291264,
      "eval_pcc_2": 0.721712301994763,
      "eval_pcc_3": 0.6702157450411846,
      "eval_runtime": 100.7127,
      "eval_samples_per_second": 24.823,
      "eval_steps_per_second": 3.108,
      "step": 1878
    },
    {
      "epoch": 6.39,
      "learning_rate": 7.870074547390843e-06,
      "loss": 0.8374,
      "step": 2000
    },
    {
      "epoch": 7.0,
      "eval_loss": 1.1292527914047241,
      "eval_mse_0": 1.3234530687332153,
      "eval_mse_1": 0.974384069442749,
      "eval_mse_2": 0.9388296008110046,
      "eval_mse_3": 1.2803441286087036,
      "eval_pcc_0": 0.6891364312392516,
      "eval_pcc_1": 0.7499496458933641,
      "eval_pcc_2": 0.7549399472585467,
      "eval_pcc_3": 0.7128765406376988,
      "eval_runtime": 100.5658,
      "eval_samples_per_second": 24.859,
      "eval_steps_per_second": 3.112,
      "step": 2191
    },
    {
      "epoch": 7.99,
      "learning_rate": 7.337593184238552e-06,
      "loss": 0.6712,
      "step": 2500
    },
    {
      "epoch": 8.0,
      "eval_loss": 1.5684882402420044,
      "eval_mse_0": 1.8322019577026367,
      "eval_mse_1": 1.2989739179611206,
      "eval_mse_2": 1.323211669921875,
      "eval_mse_3": 1.8195669651031494,
      "eval_pcc_0": 0.6737439236904719,
      "eval_pcc_1": 0.7488637790730267,
      "eval_pcc_2": 0.7514121200560983,
      "eval_pcc_3": 0.6994098335223952,
      "eval_runtime": 100.8109,
      "eval_samples_per_second": 24.799,
      "eval_steps_per_second": 3.105,
      "step": 2504
    },
    {
      "epoch": 9.0,
      "eval_loss": 1.4102668762207031,
      "eval_mse_0": 1.6206148862838745,
      "eval_mse_1": 1.1973247528076172,
      "eval_mse_2": 1.2341444492340088,
      "eval_mse_3": 1.5889830589294434,
      "eval_pcc_0": 0.6851788133643094,
      "eval_pcc_1": 0.7581205361413508,
      "eval_pcc_2": 0.7615756986314213,
      "eval_pcc_3": 0.708741002662432,
      "eval_runtime": 100.6389,
      "eval_samples_per_second": 24.841,
      "eval_steps_per_second": 3.11,
      "step": 2817
    },
    {
      "epoch": 9.58,
      "learning_rate": 6.805111821086262e-06,
      "loss": 0.57,
      "step": 3000
    },
    {
      "epoch": 10.0,
      "eval_loss": 1.348330020904541,
      "eval_mse_0": 1.5797193050384521,
      "eval_mse_1": 1.1507281064987183,
      "eval_mse_2": 1.1433807611465454,
      "eval_mse_3": 1.5194944143295288,
      "eval_pcc_0": 0.6852188493270641,
      "eval_pcc_1": 0.7562537239574154,
      "eval_pcc_2": 0.7557024973237004,
      "eval_pcc_3": 0.7054120588235016,
      "eval_runtime": 100.5701,
      "eval_samples_per_second": 24.858,
      "eval_steps_per_second": 3.112,
      "step": 3130
    },
    {
      "epoch": 11.0,
      "eval_loss": 1.5709850788116455,
      "eval_mse_0": 1.7214244604110718,
      "eval_mse_1": 1.412024736404419,
      "eval_mse_2": 1.4254834651947021,
      "eval_mse_3": 1.7250086069107056,
      "eval_pcc_0": 0.677518409167023,
      "eval_pcc_1": 0.7571105568376092,
      "eval_pcc_2": 0.7600582646570031,
      "eval_pcc_3": 0.7045871868047965,
      "eval_runtime": 100.5813,
      "eval_samples_per_second": 24.856,
      "eval_steps_per_second": 3.112,
      "step": 3443
    },
    {
      "epoch": 11.18,
      "learning_rate": 6.272630457933972e-06,
      "loss": 0.5098,
      "step": 3500
    },
    {
      "epoch": 12.0,
      "eval_loss": 1.4235678911209106,
      "eval_mse_0": 1.7117582559585571,
      "eval_mse_1": 1.1990758180618286,
      "eval_mse_2": 1.233838438987732,
      "eval_mse_3": 1.549597978591919,
      "eval_pcc_0": 0.6972914883828407,
      "eval_pcc_1": 0.7674880831251762,
      "eval_pcc_2": 0.7730368089204195,
      "eval_pcc_3": 0.7219812663434483,
      "eval_runtime": 100.9265,
      "eval_samples_per_second": 24.771,
      "eval_steps_per_second": 3.101,
      "step": 3756
    },
    {
      "epoch": 12.78,
      "learning_rate": 5.7401490947816826e-06,
      "loss": 0.4561,
      "step": 4000
    },
    {
      "epoch": 13.0,
      "eval_loss": 1.3512660264968872,
      "eval_mse_0": 1.6560535430908203,
      "eval_mse_1": 1.0737866163253784,
      "eval_mse_2": 1.167884111404419,
      "eval_mse_3": 1.507339596748352,
      "eval_pcc_0": 0.6978020685983576,
      "eval_pcc_1": 0.7669595572984247,
      "eval_pcc_2": 0.7704479293605421,
      "eval_pcc_3": 0.7218605199092147,
      "eval_runtime": 100.7483,
      "eval_samples_per_second": 24.814,
      "eval_steps_per_second": 3.107,
      "step": 4069
    },
    {
      "epoch": 14.0,
      "eval_loss": 1.4450573921203613,
      "eval_mse_0": 1.7005558013916016,
      "eval_mse_1": 1.2587356567382812,
      "eval_mse_2": 1.264267086982727,
      "eval_mse_3": 1.5566717386245728,
      "eval_pcc_0": 0.69044979553412,
      "eval_pcc_1": 0.769899037280411,
      "eval_pcc_2": 0.7715228679160522,
      "eval_pcc_3": 0.7180620133606046,
      "eval_runtime": 100.74,
      "eval_samples_per_second": 24.816,
      "eval_steps_per_second": 3.107,
      "step": 4382
    },
    {
      "epoch": 14.38,
      "learning_rate": 5.207667731629393e-06,
      "loss": 0.4262,
      "step": 4500
    },
    {
      "epoch": 15.0,
      "eval_loss": 1.2482588291168213,
      "eval_mse_0": 1.5385853052139282,
      "eval_mse_1": 0.9972391724586487,
      "eval_mse_2": 1.0317867994308472,
      "eval_mse_3": 1.4254233837127686,
      "eval_pcc_0": 0.7066790588440384,
      "eval_pcc_1": 0.7721327175086767,
      "eval_pcc_2": 0.7759734776016606,
      "eval_pcc_3": 0.7353232362019365,
      "eval_runtime": 100.6177,
      "eval_samples_per_second": 24.847,
      "eval_steps_per_second": 3.111,
      "step": 4695
    },
    {
      "epoch": 15.97,
      "learning_rate": 4.675186368477103e-06,
      "loss": 0.3855,
      "step": 5000
    },
    {
      "epoch": 16.0,
      "eval_loss": 1.2972311973571777,
      "eval_mse_0": 1.5596505403518677,
      "eval_mse_1": 1.1698797941207886,
      "eval_mse_2": 1.006371021270752,
      "eval_mse_3": 1.4530200958251953,
      "eval_pcc_0": 0.6952806951896114,
      "eval_pcc_1": 0.7707645603019087,
      "eval_pcc_2": 0.77274772478712,
      "eval_pcc_3": 0.7223344399086041,
      "eval_runtime": 101.0829,
      "eval_samples_per_second": 24.732,
      "eval_steps_per_second": 3.096,
      "step": 5008
    },
    {
      "epoch": 17.0,
      "eval_loss": 1.1730587482452393,
      "eval_mse_0": 1.4187856912612915,
      "eval_mse_1": 0.958073616027832,
      "eval_mse_2": 0.9783998727798462,
      "eval_mse_3": 1.3369756937026978,
      "eval_pcc_0": 0.7180267227388664,
      "eval_pcc_1": 0.7730168663887997,
      "eval_pcc_2": 0.7747266394371812,
      "eval_pcc_3": 0.7386234213820474,
      "eval_runtime": 100.8439,
      "eval_samples_per_second": 24.791,
      "eval_steps_per_second": 3.104,
      "step": 5321
    },
    {
      "epoch": 17.57,
      "learning_rate": 4.142705005324814e-06,
      "loss": 0.3572,
      "step": 5500
    },
    {
      "epoch": 18.0,
      "eval_loss": 1.3291434049606323,
      "eval_mse_0": 1.5203914642333984,
      "eval_mse_1": 1.1383605003356934,
      "eval_mse_2": 1.206380009651184,
      "eval_mse_3": 1.4514374732971191,
      "eval_pcc_0": 0.6933471923810368,
      "eval_pcc_1": 0.769459601309813,
      "eval_pcc_2": 0.7712010361671546,
      "eval_pcc_3": 0.7209381847866932,
      "eval_runtime": 100.7791,
      "eval_samples_per_second": 24.807,
      "eval_steps_per_second": 3.106,
      "step": 5634
    },
    {
      "epoch": 19.0,
      "eval_loss": 1.312319278717041,
      "eval_mse_0": 1.602527379989624,
      "eval_mse_1": 1.0087671279907227,
      "eval_mse_2": 1.1071666479110718,
      "eval_mse_3": 1.5308144092559814,
      "eval_pcc_0": 0.6996240750220193,
      "eval_pcc_1": 0.7725455333159118,
      "eval_pcc_2": 0.7726030288434409,
      "eval_pcc_3": 0.7283744247063056,
      "eval_runtime": 100.9276,
      "eval_samples_per_second": 24.77,
      "eval_steps_per_second": 3.101,
      "step": 5947
    },
    {
      "epoch": 19.17,
      "learning_rate": 3.6102236421725247e-06,
      "loss": 0.3685,
      "step": 6000
    },
    {
      "epoch": 20.0,
      "eval_loss": 1.194506049156189,
      "eval_mse_0": 1.382067084312439,
      "eval_mse_1": 1.0061663389205933,
      "eval_mse_2": 1.0188947916030884,
      "eval_mse_3": 1.370891809463501,
      "eval_pcc_0": 0.7058221298639312,
      "eval_pcc_1": 0.7695843992718115,
      "eval_pcc_2": 0.7696140899835595,
      "eval_pcc_3": 0.728296502432432,
      "eval_runtime": 100.7002,
      "eval_samples_per_second": 24.826,
      "eval_steps_per_second": 3.108,
      "step": 6260
    },
    {
      "epoch": 20.77,
      "learning_rate": 3.0777422790202343e-06,
      "loss": 0.3088,
      "step": 6500
    },
    {
      "epoch": 21.0,
      "eval_loss": 1.3180090188980103,
      "eval_mse_0": 1.5541187524795532,
      "eval_mse_1": 1.1070194244384766,
      "eval_mse_2": 1.1667213439941406,
      "eval_mse_3": 1.4441734552383423,
      "eval_pcc_0": 0.7058127676527958,
      "eval_pcc_1": 0.7705582261262895,
      "eval_pcc_2": 0.774042720209831,
      "eval_pcc_3": 0.7320268286311518,
      "eval_runtime": 100.5901,
      "eval_samples_per_second": 24.853,
      "eval_steps_per_second": 3.112,
      "step": 6573
    },
    {
      "epoch": 22.0,
      "eval_loss": 1.3018730878829956,
      "eval_mse_0": 1.6211662292480469,
      "eval_mse_1": 1.0196444988250732,
      "eval_mse_2": 1.0701396465301514,
      "eval_mse_3": 1.4965418577194214,
      "eval_pcc_0": 0.7032254446171953,
      "eval_pcc_1": 0.7763021450998202,
      "eval_pcc_2": 0.7768905768460305,
      "eval_pcc_3": 0.7307131933544537,
      "eval_runtime": 100.4268,
      "eval_samples_per_second": 24.894,
      "eval_steps_per_second": 3.117,
      "step": 6886
    },
    {
      "epoch": 22.36,
      "learning_rate": 2.5452609158679447e-06,
      "loss": 0.3233,
      "step": 7000
    },
    {
      "epoch": 23.0,
      "eval_loss": 1.24806547164917,
      "eval_mse_0": 1.4808491468429565,
      "eval_mse_1": 1.0740902423858643,
      "eval_mse_2": 1.075546383857727,
      "eval_mse_3": 1.3617769479751587,
      "eval_pcc_0": 0.7083667911361838,
      "eval_pcc_1": 0.7728779871674696,
      "eval_pcc_2": 0.772914855053998,
      "eval_pcc_3": 0.7355157998059723,
      "eval_runtime": 100.6985,
      "eval_samples_per_second": 24.827,
      "eval_steps_per_second": 3.108,
      "step": 7199
    },
    {
      "epoch": 23.96,
      "learning_rate": 2.012779552715655e-06,
      "loss": 0.2833,
      "step": 7500
    },
    {
      "epoch": 24.0,
      "eval_loss": 1.2991966009140015,
      "eval_mse_0": 1.5615781545639038,
      "eval_mse_1": 1.0941851139068604,
      "eval_mse_2": 1.1104274988174438,
      "eval_mse_3": 1.4305994510650635,
      "eval_pcc_0": 0.7051879609373185,
      "eval_pcc_1": 0.7743239129791413,
      "eval_pcc_2": 0.7744441249826322,
      "eval_pcc_3": 0.7341415407223592,
      "eval_runtime": 100.4516,
      "eval_samples_per_second": 24.888,
      "eval_steps_per_second": 3.116,
      "step": 7512
    },
    {
      "epoch": 25.0,
      "eval_loss": 1.2408746480941772,
      "eval_mse_0": 1.5619524717330933,
      "eval_mse_1": 0.9850296974182129,
      "eval_mse_2": 1.0179213285446167,
      "eval_mse_3": 1.3985931873321533,
      "eval_pcc_0": 0.7060029463516057,
      "eval_pcc_1": 0.7733307038103625,
      "eval_pcc_2": 0.772266549126521,
      "eval_pcc_3": 0.7339611885634963,
      "eval_runtime": 100.8095,
      "eval_samples_per_second": 24.799,
      "eval_steps_per_second": 3.105,
      "step": 7825
    },
    {
      "epoch": 25.56,
      "learning_rate": 1.4802981895633654e-06,
      "loss": 0.2638,
      "step": 8000
    },
    {
      "epoch": 26.0,
      "eval_loss": 1.2704391479492188,
      "eval_mse_0": 1.5212668180465698,
      "eval_mse_1": 1.081721544265747,
      "eval_mse_2": 1.0785852670669556,
      "eval_mse_3": 1.4001832008361816,
      "eval_pcc_0": 0.7106093307015088,
      "eval_pcc_1": 0.7789024607932206,
      "eval_pcc_2": 0.778950859240916,
      "eval_pcc_3": 0.737807152463041,
      "eval_runtime": 100.6436,
      "eval_samples_per_second": 24.84,
      "eval_steps_per_second": 3.11,
      "step": 8138
    },
    {
      "epoch": 27.0,
      "eval_loss": 1.3225528001785278,
      "eval_mse_0": 1.5737501382827759,
      "eval_mse_1": 1.1116234064102173,
      "eval_mse_2": 1.1209893226623535,
      "eval_mse_3": 1.4838504791259766,
      "eval_pcc_0": 0.7037694468899893,
      "eval_pcc_1": 0.7743690194599888,
      "eval_pcc_2": 0.774563190337004,
      "eval_pcc_3": 0.732017624810135,
      "eval_runtime": 100.5945,
      "eval_samples_per_second": 24.852,
      "eval_steps_per_second": 3.112,
      "step": 8451
    },
    {
      "epoch": 27.16,
      "learning_rate": 9.478168264110757e-07,
      "loss": 0.2495,
      "step": 8500
    },
    {
      "epoch": 28.0,
      "eval_loss": 1.244736909866333,
      "eval_mse_0": 1.4972796440124512,
      "eval_mse_1": 1.041703701019287,
      "eval_mse_2": 1.0512263774871826,
      "eval_mse_3": 1.3887364864349365,
      "eval_pcc_0": 0.7094004718076047,
      "eval_pcc_1": 0.7743070082705473,
      "eval_pcc_2": 0.7746505211725243,
      "eval_pcc_3": 0.7355049412770104,
      "eval_runtime": 100.7018,
      "eval_samples_per_second": 24.826,
      "eval_steps_per_second": 3.108,
      "step": 8764
    },
    {
      "epoch": 28.75,
      "learning_rate": 4.1533546325878595e-07,
      "loss": 0.2396,
      "step": 9000
    },
    {
      "epoch": 29.0,
      "eval_loss": 1.2441821098327637,
      "eval_mse_0": 1.4906530380249023,
      "eval_mse_1": 1.0265562534332275,
      "eval_mse_2": 1.0737992525100708,
      "eval_mse_3": 1.3857206106185913,
      "eval_pcc_0": 0.7046759605484965,
      "eval_pcc_1": 0.7732490309579964,
      "eval_pcc_2": 0.7728792272728902,
      "eval_pcc_3": 0.7323298208640173,
      "eval_runtime": 100.5549,
      "eval_samples_per_second": 24.862,
      "eval_steps_per_second": 3.113,
      "step": 9077
    },
    {
      "epoch": 30.0,
      "eval_loss": 1.257910966873169,
      "eval_mse_0": 1.490696668624878,
      "eval_mse_1": 1.0560848712921143,
      "eval_mse_2": 1.087384819984436,
      "eval_mse_3": 1.3974806070327759,
      "eval_pcc_0": 0.7062959885893274,
      "eval_pcc_1": 0.7733662364545745,
      "eval_pcc_2": 0.7731921896926559,
      "eval_pcc_3": 0.7334781892016459,
      "eval_runtime": 100.5894,
      "eval_samples_per_second": 24.854,
      "eval_steps_per_second": 3.112,
      "step": 9390
    },
    {
      "epoch": 30.0,
      "step": 9390,
      "total_flos": 0.0,
      "train_loss": 1.0932516819118057,
      "train_runtime": 9115.9528,
      "train_samples_per_second": 8.227,
      "train_steps_per_second": 1.03
    }
  ],
  "logging_steps": 500,
  "max_steps": 9390,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 30,
  "save_steps": 500,
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}