mrm8488's picture
Initial commit from mrm8488
8e52830
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 11.465683130211675,
"global_step": 71500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.1603592046183451,
"learning_rate": 4.942728855493449e-05,
"loss": 3.79368212890625,
"step": 1000
},
{
"epoch": 0.3207184092366902,
"learning_rate": 4.8854577109868964e-05,
"loss": 2.90124365234375,
"step": 2000
},
{
"epoch": 0.4810776138550353,
"learning_rate": 4.828186566480345e-05,
"loss": 2.601152587890625,
"step": 3000
},
{
"epoch": 0.6414368184733804,
"learning_rate": 4.7709154219737926e-05,
"loss": 2.300880615234375,
"step": 4000
},
{
"epoch": 0.8017960230917255,
"learning_rate": 4.713644277467241e-05,
"loss": 2.096479736328125,
"step": 5000
},
{
"epoch": 0.9621552277100706,
"learning_rate": 4.6563731329606894e-05,
"loss": 1.9469154052734374,
"step": 6000
},
{
"epoch": 1.1225144323284157,
"learning_rate": 4.5991019884541375e-05,
"loss": 1.593583740234375,
"step": 7000
},
{
"epoch": 1.2828736369467608,
"learning_rate": 4.5418308439475855e-05,
"loss": 1.4818087158203126,
"step": 8000
},
{
"epoch": 1.443232841565106,
"learning_rate": 4.4845596994410336e-05,
"loss": 1.446810791015625,
"step": 9000
},
{
"epoch": 1.603592046183451,
"learning_rate": 4.4272885549344824e-05,
"loss": 1.425852783203125,
"step": 10000
},
{
"epoch": 1.763951250801796,
"learning_rate": 4.37001741042793e-05,
"loss": 1.4071231689453125,
"step": 11000
},
{
"epoch": 1.9243104554201411,
"learning_rate": 4.3127462659213785e-05,
"loss": 1.3911290283203126,
"step": 12000
},
{
"epoch": 2.0846696600384864,
"learning_rate": 4.2554751214148266e-05,
"loss": 1.1586458740234375,
"step": 13000
},
{
"epoch": 2.2450288646568315,
"learning_rate": 4.198203976908275e-05,
"loss": 1.0172818603515625,
"step": 14000
},
{
"epoch": 2.4053880692751766,
"learning_rate": 4.140932832401723e-05,
"loss": 1.037733154296875,
"step": 15000
},
{
"epoch": 2.5657472738935216,
"learning_rate": 4.083661687895171e-05,
"loss": 1.0479439697265625,
"step": 16000
},
{
"epoch": 2.7261064785118667,
"learning_rate": 4.0263905433886196e-05,
"loss": 1.0455704345703125,
"step": 17000
},
{
"epoch": 2.886465683130212,
"learning_rate": 3.969119398882068e-05,
"loss": 1.047005859375,
"step": 18000
},
{
"epoch": 3.046824887748557,
"learning_rate": 3.911848254375516e-05,
"loss": 0.9432368774414063,
"step": 19000
},
{
"epoch": 3.207184092366902,
"learning_rate": 3.854577109868964e-05,
"loss": 0.7485797119140625,
"step": 20000
},
{
"epoch": 3.367543296985247,
"learning_rate": 3.797305965362412e-05,
"loss": 0.7684258422851562,
"step": 21000
},
{
"epoch": 3.527902501603592,
"learning_rate": 3.74003482085586e-05,
"loss": 0.7839947509765625,
"step": 22000
},
{
"epoch": 3.688261706221937,
"learning_rate": 3.682763676349308e-05,
"loss": 0.7889647216796875,
"step": 23000
},
{
"epoch": 3.8486209108402822,
"learning_rate": 3.625492531842756e-05,
"loss": 0.808726806640625,
"step": 24000
},
{
"epoch": 4.008980115458628,
"learning_rate": 3.568221387336205e-05,
"loss": 0.8007049560546875,
"step": 25000
},
{
"epoch": 4.169339320076973,
"learning_rate": 3.510950242829653e-05,
"loss": 0.5585819091796875,
"step": 26000
},
{
"epoch": 4.329698524695318,
"learning_rate": 3.453679098323101e-05,
"loss": 0.5828939819335938,
"step": 27000
},
{
"epoch": 4.490057729313663,
"learning_rate": 3.396407953816549e-05,
"loss": 0.5914722900390625,
"step": 28000
},
{
"epoch": 4.650416933932008,
"learning_rate": 3.339136809309997e-05,
"loss": 0.6112024536132813,
"step": 29000
},
{
"epoch": 4.810776138550353,
"learning_rate": 3.281865664803446e-05,
"loss": 0.6254320678710937,
"step": 30000
},
{
"epoch": 4.971135343168698,
"learning_rate": 3.2245945202968934e-05,
"loss": 0.6254859619140625,
"step": 31000
},
{
"epoch": 5.131494547787043,
"learning_rate": 3.167323375790342e-05,
"loss": 0.4557585144042969,
"step": 32000
},
{
"epoch": 5.291853752405388,
"learning_rate": 3.1100522312837895e-05,
"loss": 0.43634527587890626,
"step": 33000
},
{
"epoch": 5.452212957023733,
"learning_rate": 3.052781086777238e-05,
"loss": 0.46431121826171873,
"step": 34000
},
{
"epoch": 5.6125721616420785,
"learning_rate": 2.9955099422706867e-05,
"loss": 0.46590155029296876,
"step": 35000
},
{
"epoch": 5.772931366260424,
"learning_rate": 2.9382387977641345e-05,
"loss": 0.47823056030273436,
"step": 36000
},
{
"epoch": 5.933290570878769,
"learning_rate": 2.880967653257583e-05,
"loss": 0.4822598571777344,
"step": 37000
},
{
"epoch": 6.093649775497114,
"learning_rate": 2.823696508751031e-05,
"loss": 0.38333685302734377,
"step": 38000
},
{
"epoch": 6.254008980115459,
"learning_rate": 2.7664253642444794e-05,
"loss": 0.3381705017089844,
"step": 39000
},
{
"epoch": 6.414368184733804,
"learning_rate": 2.709154219737927e-05,
"loss": 0.3528754272460938,
"step": 40000
},
{
"epoch": 6.574727389352149,
"learning_rate": 2.6518830752313755e-05,
"loss": 0.358309326171875,
"step": 41000
},
{
"epoch": 6.735086593970494,
"learning_rate": 2.594611930724824e-05,
"loss": 0.36756768798828127,
"step": 42000
},
{
"epoch": 6.895445798588839,
"learning_rate": 2.537340786218272e-05,
"loss": 0.3713406677246094,
"step": 43000
},
{
"epoch": 7.055805003207184,
"learning_rate": 2.48006964171172e-05,
"loss": 0.330199462890625,
"step": 44000
},
{
"epoch": 7.216164207825529,
"learning_rate": 2.4227984972051682e-05,
"loss": 0.25607583618164065,
"step": 45000
},
{
"epoch": 7.376523412443874,
"learning_rate": 2.3655273526986163e-05,
"loss": 0.2684461975097656,
"step": 46000
},
{
"epoch": 7.536882617062219,
"learning_rate": 2.3082562081920647e-05,
"loss": 0.27459906005859375,
"step": 47000
},
{
"epoch": 7.6972418216805645,
"learning_rate": 2.2509850636855128e-05,
"loss": 0.2851370239257813,
"step": 48000
},
{
"epoch": 7.8576010262989096,
"learning_rate": 2.1937139191789612e-05,
"loss": 0.2865502624511719,
"step": 49000
},
{
"epoch": 8.017960230917256,
"learning_rate": 2.1364427746724092e-05,
"loss": 0.2776366882324219,
"step": 50000
},
{
"epoch": 8.1783194355356,
"learning_rate": 2.0791716301658573e-05,
"loss": 0.1949043884277344,
"step": 51000
},
{
"epoch": 8.338678640153946,
"learning_rate": 2.0219004856593054e-05,
"loss": 0.20391067504882812,
"step": 52000
},
{
"epoch": 8.49903784477229,
"learning_rate": 1.9646293411527535e-05,
"loss": 0.21169953918457032,
"step": 53000
},
{
"epoch": 8.659397049390636,
"learning_rate": 1.907358196646202e-05,
"loss": 0.21576847839355467,
"step": 54000
},
{
"epoch": 8.819756254008981,
"learning_rate": 1.8500870521396503e-05,
"loss": 0.21953797912597656,
"step": 55000
},
{
"epoch": 8.980115458627326,
"learning_rate": 1.7928159076330984e-05,
"loss": 0.22510685729980467,
"step": 56000
},
{
"epoch": 9.140474663245671,
"learning_rate": 1.7355447631265465e-05,
"loss": 0.15727980041503906,
"step": 57000
},
{
"epoch": 9.300833867864016,
"learning_rate": 1.6782736186199946e-05,
"loss": 0.15759164428710937,
"step": 58000
},
{
"epoch": 9.461193072482361,
"learning_rate": 1.6210024741134426e-05,
"loss": 0.16151931762695312,
"step": 59000
},
{
"epoch": 9.621552277100706,
"learning_rate": 1.563731329606891e-05,
"loss": 0.16533021545410156,
"step": 60000
},
{
"epoch": 9.781911481719051,
"learning_rate": 1.506460185100339e-05,
"loss": 0.16902993774414063,
"step": 61000
},
{
"epoch": 9.942270686337396,
"learning_rate": 1.4491890405937872e-05,
"loss": 0.16983213806152345,
"step": 62000
},
{
"epoch": 10.102629890955741,
"learning_rate": 1.3919178960872356e-05,
"loss": 0.1350119171142578,
"step": 63000
},
{
"epoch": 10.262989095574087,
"learning_rate": 1.3346467515806837e-05,
"loss": 0.1185531997680664,
"step": 64000
},
{
"epoch": 10.423348300192432,
"learning_rate": 1.277375607074132e-05,
"loss": 0.12531481170654296,
"step": 65000
},
{
"epoch": 10.583707504810777,
"learning_rate": 1.22010446256758e-05,
"loss": 0.1266140365600586,
"step": 66000
},
{
"epoch": 10.744066709429122,
"learning_rate": 1.1628333180610281e-05,
"loss": 0.12920794677734376,
"step": 67000
},
{
"epoch": 10.904425914047467,
"learning_rate": 1.1055621735544764e-05,
"loss": 0.1271785125732422,
"step": 68000
},
{
"epoch": 11.064785118665812,
"learning_rate": 1.0482910290479244e-05,
"loss": 0.11574956512451172,
"step": 69000
},
{
"epoch": 11.225144323284157,
"learning_rate": 9.910198845413727e-06,
"loss": 0.09364027404785157,
"step": 70000
},
{
"epoch": 11.385503527902502,
"learning_rate": 9.33748740034821e-06,
"loss": 0.09231264495849609,
"step": 71000
}
],
"max_steps": 87304,
"num_train_epochs": 14,
"total_flos": 409014659202678144,
"trial_name": null,
"trial_params": null
}