BERT-SBIC-targetcategory / trainer_state.json
Cameron Clarke
add fixed model files
bd0ca45
{
"best_metric": 0.7773735903965078,
"best_model_checkpoint": "/scratch/ccc779/sbic_targetcategory_model/checkpoint-32985",
"epoch": 2.0,
"global_step": 65970,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02,
"learning_rate": 1.2089215154878955e-05,
"loss": 1.5864,
"step": 500
},
{
"epoch": 0.03,
"learning_rate": 1.199688878283925e-05,
"loss": 1.2104,
"step": 1000
},
{
"epoch": 0.05,
"learning_rate": 1.1904562410799545e-05,
"loss": 1.2276,
"step": 1500
},
{
"epoch": 0.06,
"learning_rate": 1.181223603875984e-05,
"loss": 1.2612,
"step": 2000
},
{
"epoch": 0.08,
"learning_rate": 1.1719909666720135e-05,
"loss": 1.1934,
"step": 2500
},
{
"epoch": 0.09,
"learning_rate": 1.1627583294680432e-05,
"loss": 1.1754,
"step": 3000
},
{
"epoch": 0.11,
"learning_rate": 1.1535256922640727e-05,
"loss": 1.2006,
"step": 3500
},
{
"epoch": 0.12,
"learning_rate": 1.144293055060102e-05,
"loss": 1.1315,
"step": 4000
},
{
"epoch": 0.14,
"learning_rate": 1.1350604178561317e-05,
"loss": 1.1612,
"step": 4500
},
{
"epoch": 0.15,
"learning_rate": 1.1258277806521612e-05,
"loss": 1.1651,
"step": 5000
},
{
"epoch": 0.17,
"learning_rate": 1.1165951434481907e-05,
"loss": 1.2153,
"step": 5500
},
{
"epoch": 0.18,
"learning_rate": 1.1073625062442202e-05,
"loss": 1.1769,
"step": 6000
},
{
"epoch": 0.2,
"learning_rate": 1.0981298690402497e-05,
"loss": 0.9871,
"step": 6500
},
{
"epoch": 0.21,
"learning_rate": 1.0888972318362792e-05,
"loss": 1.0295,
"step": 7000
},
{
"epoch": 0.23,
"learning_rate": 1.0796645946323087e-05,
"loss": 1.0582,
"step": 7500
},
{
"epoch": 0.24,
"learning_rate": 1.0704319574283382e-05,
"loss": 1.1231,
"step": 8000
},
{
"epoch": 0.26,
"learning_rate": 1.0611993202243677e-05,
"loss": 1.036,
"step": 8500
},
{
"epoch": 0.27,
"learning_rate": 1.0519666830203972e-05,
"loss": 1.0158,
"step": 9000
},
{
"epoch": 0.29,
"learning_rate": 1.0427340458164267e-05,
"loss": 0.9902,
"step": 9500
},
{
"epoch": 0.3,
"learning_rate": 1.0335014086124564e-05,
"loss": 1.0596,
"step": 10000
},
{
"epoch": 0.32,
"learning_rate": 1.0242687714084859e-05,
"loss": 1.0969,
"step": 10500
},
{
"epoch": 0.33,
"learning_rate": 1.0150361342045152e-05,
"loss": 1.0643,
"step": 11000
},
{
"epoch": 0.35,
"learning_rate": 1.0058034970005447e-05,
"loss": 0.9806,
"step": 11500
},
{
"epoch": 0.36,
"learning_rate": 9.965708597965744e-06,
"loss": 1.057,
"step": 12000
},
{
"epoch": 0.38,
"learning_rate": 9.873382225926039e-06,
"loss": 1.0682,
"step": 12500
},
{
"epoch": 0.39,
"learning_rate": 9.781055853886334e-06,
"loss": 1.1143,
"step": 13000
},
{
"epoch": 0.41,
"learning_rate": 9.688729481846629e-06,
"loss": 0.9714,
"step": 13500
},
{
"epoch": 0.42,
"learning_rate": 9.596403109806924e-06,
"loss": 1.0199,
"step": 14000
},
{
"epoch": 0.44,
"learning_rate": 9.504076737767219e-06,
"loss": 0.9187,
"step": 14500
},
{
"epoch": 0.45,
"learning_rate": 9.411750365727514e-06,
"loss": 0.9385,
"step": 15000
},
{
"epoch": 0.47,
"learning_rate": 9.31942399368781e-06,
"loss": 0.9898,
"step": 15500
},
{
"epoch": 0.49,
"learning_rate": 9.227097621648104e-06,
"loss": 1.036,
"step": 16000
},
{
"epoch": 0.5,
"learning_rate": 9.134771249608399e-06,
"loss": 0.928,
"step": 16500
},
{
"epoch": 0.52,
"learning_rate": 9.042444877568694e-06,
"loss": 0.9821,
"step": 17000
},
{
"epoch": 0.53,
"learning_rate": 8.95011850552899e-06,
"loss": 0.9844,
"step": 17500
},
{
"epoch": 0.55,
"learning_rate": 8.857792133489285e-06,
"loss": 0.9938,
"step": 18000
},
{
"epoch": 0.56,
"learning_rate": 8.765465761449579e-06,
"loss": 0.8945,
"step": 18500
},
{
"epoch": 0.58,
"learning_rate": 8.673139389409876e-06,
"loss": 1.0655,
"step": 19000
},
{
"epoch": 0.59,
"learning_rate": 8.58081301737017e-06,
"loss": 0.9115,
"step": 19500
},
{
"epoch": 0.61,
"learning_rate": 8.488486645330466e-06,
"loss": 0.9287,
"step": 20000
},
{
"epoch": 0.62,
"learning_rate": 8.39616027329076e-06,
"loss": 0.9015,
"step": 20500
},
{
"epoch": 0.64,
"learning_rate": 8.303833901251056e-06,
"loss": 0.9908,
"step": 21000
},
{
"epoch": 0.65,
"learning_rate": 8.21150752921135e-06,
"loss": 1.0955,
"step": 21500
},
{
"epoch": 0.67,
"learning_rate": 8.119181157171646e-06,
"loss": 0.9731,
"step": 22000
},
{
"epoch": 0.68,
"learning_rate": 8.026854785131942e-06,
"loss": 0.8865,
"step": 22500
},
{
"epoch": 0.7,
"learning_rate": 7.934528413092236e-06,
"loss": 0.9823,
"step": 23000
},
{
"epoch": 0.71,
"learning_rate": 7.84220204105253e-06,
"loss": 0.9258,
"step": 23500
},
{
"epoch": 0.73,
"learning_rate": 7.749875669012826e-06,
"loss": 0.8959,
"step": 24000
},
{
"epoch": 0.74,
"learning_rate": 7.657549296973122e-06,
"loss": 1.0291,
"step": 24500
},
{
"epoch": 0.76,
"learning_rate": 7.5652229249334165e-06,
"loss": 0.8294,
"step": 25000
},
{
"epoch": 0.77,
"learning_rate": 7.4728965528937115e-06,
"loss": 0.9083,
"step": 25500
},
{
"epoch": 0.79,
"learning_rate": 7.3805701808540065e-06,
"loss": 1.0509,
"step": 26000
},
{
"epoch": 0.8,
"learning_rate": 7.288243808814302e-06,
"loss": 0.9337,
"step": 26500
},
{
"epoch": 0.82,
"learning_rate": 7.195917436774597e-06,
"loss": 0.9458,
"step": 27000
},
{
"epoch": 0.83,
"learning_rate": 7.1035910647348915e-06,
"loss": 0.9063,
"step": 27500
},
{
"epoch": 0.85,
"learning_rate": 7.011264692695188e-06,
"loss": 0.8929,
"step": 28000
},
{
"epoch": 0.86,
"learning_rate": 6.918938320655482e-06,
"loss": 0.8642,
"step": 28500
},
{
"epoch": 0.88,
"learning_rate": 6.826611948615777e-06,
"loss": 0.845,
"step": 29000
},
{
"epoch": 0.89,
"learning_rate": 6.7342855765760724e-06,
"loss": 0.8494,
"step": 29500
},
{
"epoch": 0.91,
"learning_rate": 6.641959204536368e-06,
"loss": 0.8934,
"step": 30000
},
{
"epoch": 0.92,
"learning_rate": 6.549632832496663e-06,
"loss": 0.8529,
"step": 30500
},
{
"epoch": 0.94,
"learning_rate": 6.457306460456958e-06,
"loss": 0.9276,
"step": 31000
},
{
"epoch": 0.95,
"learning_rate": 6.364980088417254e-06,
"loss": 0.9077,
"step": 31500
},
{
"epoch": 0.97,
"learning_rate": 6.272653716377549e-06,
"loss": 0.8587,
"step": 32000
},
{
"epoch": 0.99,
"learning_rate": 6.180327344337843e-06,
"loss": 0.9533,
"step": 32500
},
{
"epoch": 1.0,
"eval_accuracy": 0.7773735903965078,
"eval_f1": 0.7773735903965078,
"eval_loss": 1.080080270767212,
"eval_precision": 0.7773735903965078,
"eval_recall": 0.7773735903965078,
"eval_runtime": 45.2439,
"eval_samples_per_second": 243.038,
"step": 32985
},
{
"epoch": 1.0,
"learning_rate": 6.088000972298139e-06,
"loss": 0.7506,
"step": 33000
},
{
"epoch": 1.02,
"learning_rate": 5.995674600258433e-06,
"loss": 0.6419,
"step": 33500
},
{
"epoch": 1.03,
"learning_rate": 5.903348228218729e-06,
"loss": 0.5962,
"step": 34000
},
{
"epoch": 1.05,
"learning_rate": 5.811021856179024e-06,
"loss": 0.6539,
"step": 34500
},
{
"epoch": 1.06,
"learning_rate": 5.718695484139319e-06,
"loss": 0.6921,
"step": 35000
},
{
"epoch": 1.08,
"learning_rate": 5.626369112099615e-06,
"loss": 0.5754,
"step": 35500
},
{
"epoch": 1.09,
"learning_rate": 5.534042740059909e-06,
"loss": 0.6421,
"step": 36000
},
{
"epoch": 1.11,
"learning_rate": 5.441716368020205e-06,
"loss": 0.6803,
"step": 36500
},
{
"epoch": 1.12,
"learning_rate": 5.3493899959805e-06,
"loss": 0.7151,
"step": 37000
},
{
"epoch": 1.14,
"learning_rate": 5.257063623940795e-06,
"loss": 0.5593,
"step": 37500
},
{
"epoch": 1.15,
"learning_rate": 5.16473725190109e-06,
"loss": 0.5911,
"step": 38000
},
{
"epoch": 1.17,
"learning_rate": 5.072410879861385e-06,
"loss": 0.628,
"step": 38500
},
{
"epoch": 1.18,
"learning_rate": 4.98008450782168e-06,
"loss": 0.6334,
"step": 39000
},
{
"epoch": 1.2,
"learning_rate": 4.887758135781975e-06,
"loss": 0.6261,
"step": 39500
},
{
"epoch": 1.21,
"learning_rate": 4.795431763742271e-06,
"loss": 0.6338,
"step": 40000
},
{
"epoch": 1.23,
"learning_rate": 4.703105391702566e-06,
"loss": 0.6761,
"step": 40500
},
{
"epoch": 1.24,
"learning_rate": 4.610779019662861e-06,
"loss": 0.6878,
"step": 41000
},
{
"epoch": 1.26,
"learning_rate": 4.518452647623156e-06,
"loss": 0.6907,
"step": 41500
},
{
"epoch": 1.27,
"learning_rate": 4.426126275583451e-06,
"loss": 0.6187,
"step": 42000
},
{
"epoch": 1.29,
"learning_rate": 4.333799903543746e-06,
"loss": 0.6238,
"step": 42500
},
{
"epoch": 1.3,
"learning_rate": 4.241473531504042e-06,
"loss": 0.6785,
"step": 43000
},
{
"epoch": 1.32,
"learning_rate": 4.149147159464336e-06,
"loss": 0.6033,
"step": 43500
},
{
"epoch": 1.33,
"learning_rate": 4.056820787424632e-06,
"loss": 0.6451,
"step": 44000
},
{
"epoch": 1.35,
"learning_rate": 3.964494415384927e-06,
"loss": 0.6872,
"step": 44500
},
{
"epoch": 1.36,
"learning_rate": 3.872168043345222e-06,
"loss": 0.6623,
"step": 45000
},
{
"epoch": 1.38,
"learning_rate": 3.7798416713055174e-06,
"loss": 0.6111,
"step": 45500
},
{
"epoch": 1.39,
"learning_rate": 3.687515299265812e-06,
"loss": 0.7073,
"step": 46000
},
{
"epoch": 1.41,
"learning_rate": 3.5951889272261075e-06,
"loss": 0.6318,
"step": 46500
},
{
"epoch": 1.42,
"learning_rate": 3.5028625551864025e-06,
"loss": 0.6551,
"step": 47000
},
{
"epoch": 1.44,
"learning_rate": 3.410536183146698e-06,
"loss": 0.5685,
"step": 47500
},
{
"epoch": 1.46,
"learning_rate": 3.3182098111069925e-06,
"loss": 0.6836,
"step": 48000
},
{
"epoch": 1.47,
"learning_rate": 3.225883439067288e-06,
"loss": 0.6461,
"step": 48500
},
{
"epoch": 1.49,
"learning_rate": 3.1335570670275834e-06,
"loss": 0.5755,
"step": 49000
},
{
"epoch": 1.5,
"learning_rate": 3.041230694987878e-06,
"loss": 0.5359,
"step": 49500
},
{
"epoch": 1.52,
"learning_rate": 2.9489043229481734e-06,
"loss": 0.6455,
"step": 50000
},
{
"epoch": 1.53,
"learning_rate": 2.856577950908469e-06,
"loss": 0.737,
"step": 50500
},
{
"epoch": 1.55,
"learning_rate": 2.764251578868764e-06,
"loss": 0.6399,
"step": 51000
},
{
"epoch": 1.56,
"learning_rate": 2.671925206829059e-06,
"loss": 0.587,
"step": 51500
},
{
"epoch": 1.58,
"learning_rate": 2.579598834789354e-06,
"loss": 0.5731,
"step": 52000
},
{
"epoch": 1.59,
"learning_rate": 2.487272462749649e-06,
"loss": 0.6493,
"step": 52500
},
{
"epoch": 1.61,
"learning_rate": 2.3949460907099443e-06,
"loss": 0.6159,
"step": 53000
},
{
"epoch": 1.62,
"learning_rate": 2.3026197186702393e-06,
"loss": 0.5018,
"step": 53500
},
{
"epoch": 1.64,
"learning_rate": 2.2102933466305343e-06,
"loss": 0.5409,
"step": 54000
},
{
"epoch": 1.65,
"learning_rate": 2.1179669745908293e-06,
"loss": 0.5077,
"step": 54500
},
{
"epoch": 1.67,
"learning_rate": 2.0256406025511248e-06,
"loss": 0.6574,
"step": 55000
},
{
"epoch": 1.68,
"learning_rate": 1.93331423051142e-06,
"loss": 0.5516,
"step": 55500
},
{
"epoch": 1.7,
"learning_rate": 1.840987858471715e-06,
"loss": 0.6192,
"step": 56000
},
{
"epoch": 1.71,
"learning_rate": 1.7486614864320102e-06,
"loss": 0.5709,
"step": 56500
},
{
"epoch": 1.73,
"learning_rate": 1.6563351143923052e-06,
"loss": 0.6052,
"step": 57000
},
{
"epoch": 1.74,
"learning_rate": 1.5640087423526005e-06,
"loss": 0.5838,
"step": 57500
},
{
"epoch": 1.76,
"learning_rate": 1.4716823703128957e-06,
"loss": 0.6324,
"step": 58000
},
{
"epoch": 1.77,
"learning_rate": 1.3793559982731907e-06,
"loss": 0.5654,
"step": 58500
},
{
"epoch": 1.79,
"learning_rate": 1.287029626233486e-06,
"loss": 0.6369,
"step": 59000
},
{
"epoch": 1.8,
"learning_rate": 1.194703254193781e-06,
"loss": 0.6101,
"step": 59500
},
{
"epoch": 1.82,
"learning_rate": 1.102376882154076e-06,
"loss": 0.609,
"step": 60000
},
{
"epoch": 1.83,
"learning_rate": 1.0100505101143714e-06,
"loss": 0.5838,
"step": 60500
},
{
"epoch": 1.85,
"learning_rate": 9.177241380746664e-07,
"loss": 0.5898,
"step": 61000
},
{
"epoch": 1.86,
"learning_rate": 8.253977660349615e-07,
"loss": 0.5207,
"step": 61500
},
{
"epoch": 1.88,
"learning_rate": 7.330713939952566e-07,
"loss": 0.5671,
"step": 62000
},
{
"epoch": 1.89,
"learning_rate": 6.407450219555517e-07,
"loss": 0.5419,
"step": 62500
},
{
"epoch": 1.91,
"learning_rate": 5.48418649915847e-07,
"loss": 0.5781,
"step": 63000
},
{
"epoch": 1.93,
"learning_rate": 4.5609227787614203e-07,
"loss": 0.5255,
"step": 63500
},
{
"epoch": 1.94,
"learning_rate": 3.6376590583643715e-07,
"loss": 0.5419,
"step": 64000
},
{
"epoch": 1.96,
"learning_rate": 2.714395337967323e-07,
"loss": 0.63,
"step": 64500
},
{
"epoch": 1.97,
"learning_rate": 1.7911316175702744e-07,
"loss": 0.5792,
"step": 65000
},
{
"epoch": 1.99,
"learning_rate": 8.678678971732258e-08,
"loss": 0.6106,
"step": 65500
},
{
"epoch": 2.0,
"eval_accuracy": 0.7744634412513641,
"eval_f1": 0.7744634412513641,
"eval_loss": 1.3352124691009521,
"eval_precision": 0.7744634412513641,
"eval_recall": 0.7744634412513641,
"eval_runtime": 55.8298,
"eval_samples_per_second": 196.956,
"step": 65970
}
],
"max_steps": 65970,
"num_train_epochs": 2,
"total_flos": 3284044576091760.0,
"trial_name": null,
"trial_params": null
}