punctuation-tedtalk2012-t5-base / trainer_state.json
Tiago Barbosa de Lima
End of training
b86502e
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"global_step": 77894,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.967905101805017e-05,
"loss": 0.778,
"step": 500
},
{
"epoch": 0.01,
"learning_rate": 4.9358102036100344e-05,
"loss": 0.1406,
"step": 1000
},
{
"epoch": 0.02,
"learning_rate": 4.9037153054150515e-05,
"loss": 0.1084,
"step": 1500
},
{
"epoch": 0.03,
"learning_rate": 4.8716204072200686e-05,
"loss": 0.0908,
"step": 2000
},
{
"epoch": 0.03,
"learning_rate": 4.8395255090250857e-05,
"loss": 0.0767,
"step": 2500
},
{
"epoch": 0.04,
"learning_rate": 4.807430610830103e-05,
"loss": 0.0759,
"step": 3000
},
{
"epoch": 0.04,
"learning_rate": 4.775335712635119e-05,
"loss": 0.073,
"step": 3500
},
{
"epoch": 0.05,
"learning_rate": 4.743240814440137e-05,
"loss": 0.0655,
"step": 4000
},
{
"epoch": 0.06,
"learning_rate": 4.711145916245154e-05,
"loss": 0.0653,
"step": 4500
},
{
"epoch": 0.06,
"learning_rate": 4.679051018050171e-05,
"loss": 0.0602,
"step": 5000
},
{
"epoch": 0.07,
"learning_rate": 4.646956119855188e-05,
"loss": 0.061,
"step": 5500
},
{
"epoch": 0.08,
"learning_rate": 4.614861221660205e-05,
"loss": 0.0632,
"step": 6000
},
{
"epoch": 0.08,
"learning_rate": 4.582766323465222e-05,
"loss": 0.057,
"step": 6500
},
{
"epoch": 0.09,
"learning_rate": 4.5506714252702394e-05,
"loss": 0.0575,
"step": 7000
},
{
"epoch": 0.1,
"learning_rate": 4.518576527075256e-05,
"loss": 0.0555,
"step": 7500
},
{
"epoch": 0.1,
"learning_rate": 4.4864816288802736e-05,
"loss": 0.0557,
"step": 8000
},
{
"epoch": 0.11,
"learning_rate": 4.4543867306852907e-05,
"loss": 0.0547,
"step": 8500
},
{
"epoch": 0.12,
"learning_rate": 4.422291832490307e-05,
"loss": 0.0534,
"step": 9000
},
{
"epoch": 0.12,
"learning_rate": 4.390196934295325e-05,
"loss": 0.0513,
"step": 9500
},
{
"epoch": 0.13,
"learning_rate": 4.358102036100342e-05,
"loss": 0.0526,
"step": 10000
},
{
"epoch": 0.13,
"learning_rate": 4.326007137905358e-05,
"loss": 0.0502,
"step": 10500
},
{
"epoch": 0.14,
"learning_rate": 4.293912239710376e-05,
"loss": 0.0503,
"step": 11000
},
{
"epoch": 0.15,
"learning_rate": 4.2618173415153925e-05,
"loss": 0.0486,
"step": 11500
},
{
"epoch": 0.15,
"learning_rate": 4.22972244332041e-05,
"loss": 0.0471,
"step": 12000
},
{
"epoch": 0.16,
"learning_rate": 4.197627545125427e-05,
"loss": 0.0484,
"step": 12500
},
{
"epoch": 0.17,
"learning_rate": 4.165532646930444e-05,
"loss": 0.0486,
"step": 13000
},
{
"epoch": 0.17,
"learning_rate": 4.1334377487354615e-05,
"loss": 0.0502,
"step": 13500
},
{
"epoch": 0.18,
"learning_rate": 4.1013428505404786e-05,
"loss": 0.0452,
"step": 14000
},
{
"epoch": 0.19,
"learning_rate": 4.069247952345495e-05,
"loss": 0.0477,
"step": 14500
},
{
"epoch": 0.19,
"learning_rate": 4.037153054150513e-05,
"loss": 0.0446,
"step": 15000
},
{
"epoch": 0.2,
"learning_rate": 4.005058155955529e-05,
"loss": 0.0419,
"step": 15500
},
{
"epoch": 0.21,
"learning_rate": 3.972963257760546e-05,
"loss": 0.0477,
"step": 16000
},
{
"epoch": 0.21,
"learning_rate": 3.940868359565564e-05,
"loss": 0.0466,
"step": 16500
},
{
"epoch": 0.22,
"learning_rate": 3.9087734613705804e-05,
"loss": 0.047,
"step": 17000
},
{
"epoch": 0.22,
"learning_rate": 3.8766785631755975e-05,
"loss": 0.0424,
"step": 17500
},
{
"epoch": 0.23,
"learning_rate": 3.844583664980615e-05,
"loss": 0.0442,
"step": 18000
},
{
"epoch": 0.24,
"learning_rate": 3.8124887667856317e-05,
"loss": 0.0427,
"step": 18500
},
{
"epoch": 0.24,
"learning_rate": 3.7803938685906494e-05,
"loss": 0.0444,
"step": 19000
},
{
"epoch": 0.25,
"learning_rate": 3.748298970395666e-05,
"loss": 0.0413,
"step": 19500
},
{
"epoch": 0.26,
"learning_rate": 3.716204072200683e-05,
"loss": 0.0421,
"step": 20000
},
{
"epoch": 0.26,
"learning_rate": 3.684109174005701e-05,
"loss": 0.0419,
"step": 20500
},
{
"epoch": 0.27,
"learning_rate": 3.652014275810717e-05,
"loss": 0.0419,
"step": 21000
},
{
"epoch": 0.28,
"learning_rate": 3.619919377615734e-05,
"loss": 0.042,
"step": 21500
},
{
"epoch": 0.28,
"learning_rate": 3.587824479420752e-05,
"loss": 0.0375,
"step": 22000
},
{
"epoch": 0.29,
"learning_rate": 3.555729581225768e-05,
"loss": 0.0409,
"step": 22500
},
{
"epoch": 0.3,
"learning_rate": 3.5236346830307854e-05,
"loss": 0.047,
"step": 23000
},
{
"epoch": 0.3,
"learning_rate": 3.4915397848358025e-05,
"loss": 0.0383,
"step": 23500
},
{
"epoch": 0.31,
"learning_rate": 3.4594448866408196e-05,
"loss": 0.0453,
"step": 24000
},
{
"epoch": 0.31,
"learning_rate": 3.4273499884458367e-05,
"loss": 0.0396,
"step": 24500
},
{
"epoch": 0.32,
"learning_rate": 3.395255090250854e-05,
"loss": 0.0441,
"step": 25000
},
{
"epoch": 0.33,
"learning_rate": 3.363160192055871e-05,
"loss": 0.0416,
"step": 25500
},
{
"epoch": 0.33,
"learning_rate": 3.3310652938608886e-05,
"loss": 0.0414,
"step": 26000
},
{
"epoch": 0.34,
"learning_rate": 3.298970395665905e-05,
"loss": 0.0404,
"step": 26500
},
{
"epoch": 0.35,
"learning_rate": 3.266875497470922e-05,
"loss": 0.0412,
"step": 27000
},
{
"epoch": 0.35,
"learning_rate": 3.234780599275939e-05,
"loss": 0.0416,
"step": 27500
},
{
"epoch": 0.36,
"learning_rate": 3.202685701080956e-05,
"loss": 0.0411,
"step": 28000
},
{
"epoch": 0.37,
"learning_rate": 3.170590802885973e-05,
"loss": 0.042,
"step": 28500
},
{
"epoch": 0.37,
"learning_rate": 3.1384959046909904e-05,
"loss": 0.0392,
"step": 29000
},
{
"epoch": 0.38,
"learning_rate": 3.1064010064960075e-05,
"loss": 0.0386,
"step": 29500
},
{
"epoch": 0.39,
"learning_rate": 3.0743061083010246e-05,
"loss": 0.0418,
"step": 30000
},
{
"epoch": 0.39,
"learning_rate": 3.0422112101060417e-05,
"loss": 0.0417,
"step": 30500
},
{
"epoch": 0.4,
"learning_rate": 3.0101163119110587e-05,
"loss": 0.0379,
"step": 31000
},
{
"epoch": 0.4,
"learning_rate": 2.9780214137160755e-05,
"loss": 0.0418,
"step": 31500
},
{
"epoch": 0.41,
"learning_rate": 2.945926515521093e-05,
"loss": 0.0416,
"step": 32000
},
{
"epoch": 0.42,
"learning_rate": 2.91383161732611e-05,
"loss": 0.0396,
"step": 32500
},
{
"epoch": 0.42,
"learning_rate": 2.8817367191311274e-05,
"loss": 0.0436,
"step": 33000
},
{
"epoch": 0.43,
"learning_rate": 2.849641820936144e-05,
"loss": 0.0387,
"step": 33500
},
{
"epoch": 0.44,
"learning_rate": 2.817546922741161e-05,
"loss": 0.0367,
"step": 34000
},
{
"epoch": 0.44,
"learning_rate": 2.7854520245461783e-05,
"loss": 0.039,
"step": 34500
},
{
"epoch": 0.45,
"learning_rate": 2.7533571263511954e-05,
"loss": 0.0383,
"step": 35000
},
{
"epoch": 0.46,
"learning_rate": 2.721262228156212e-05,
"loss": 0.0401,
"step": 35500
},
{
"epoch": 0.46,
"learning_rate": 2.6891673299612296e-05,
"loss": 0.0413,
"step": 36000
},
{
"epoch": 0.47,
"learning_rate": 2.6570724317662467e-05,
"loss": 0.0411,
"step": 36500
},
{
"epoch": 0.48,
"learning_rate": 2.6249775335712634e-05,
"loss": 0.0379,
"step": 37000
},
{
"epoch": 0.48,
"learning_rate": 2.5928826353762808e-05,
"loss": 0.0378,
"step": 37500
},
{
"epoch": 0.49,
"learning_rate": 2.5607877371812976e-05,
"loss": 0.0365,
"step": 38000
},
{
"epoch": 0.49,
"learning_rate": 2.5286928389863147e-05,
"loss": 0.0361,
"step": 38500
},
{
"epoch": 0.5,
"learning_rate": 2.496597940791332e-05,
"loss": 0.0373,
"step": 39000
},
{
"epoch": 0.51,
"learning_rate": 2.4645030425963488e-05,
"loss": 0.0369,
"step": 39500
},
{
"epoch": 0.51,
"learning_rate": 2.432408144401366e-05,
"loss": 0.0395,
"step": 40000
},
{
"epoch": 0.52,
"learning_rate": 2.4003132462063833e-05,
"loss": 0.0401,
"step": 40500
},
{
"epoch": 0.53,
"learning_rate": 2.3682183480114004e-05,
"loss": 0.0375,
"step": 41000
},
{
"epoch": 0.53,
"learning_rate": 2.336123449816417e-05,
"loss": 0.0348,
"step": 41500
},
{
"epoch": 0.54,
"learning_rate": 2.3040285516214342e-05,
"loss": 0.0369,
"step": 42000
},
{
"epoch": 0.55,
"learning_rate": 2.2719336534264517e-05,
"loss": 0.0407,
"step": 42500
},
{
"epoch": 0.55,
"learning_rate": 2.2398387552314684e-05,
"loss": 0.0365,
"step": 43000
},
{
"epoch": 0.56,
"learning_rate": 2.2077438570364855e-05,
"loss": 0.0378,
"step": 43500
},
{
"epoch": 0.56,
"learning_rate": 2.1756489588415026e-05,
"loss": 0.0373,
"step": 44000
},
{
"epoch": 0.57,
"learning_rate": 2.14355406064652e-05,
"loss": 0.0363,
"step": 44500
},
{
"epoch": 0.58,
"learning_rate": 2.1114591624515367e-05,
"loss": 0.0355,
"step": 45000
},
{
"epoch": 0.58,
"learning_rate": 2.0793642642565538e-05,
"loss": 0.0371,
"step": 45500
},
{
"epoch": 0.59,
"learning_rate": 2.047269366061571e-05,
"loss": 0.0363,
"step": 46000
},
{
"epoch": 0.6,
"learning_rate": 2.015174467866588e-05,
"loss": 0.0361,
"step": 46500
},
{
"epoch": 0.6,
"learning_rate": 1.983079569671605e-05,
"loss": 0.0378,
"step": 47000
},
{
"epoch": 0.61,
"learning_rate": 1.950984671476622e-05,
"loss": 0.038,
"step": 47500
},
{
"epoch": 0.62,
"learning_rate": 1.9188897732816392e-05,
"loss": 0.0371,
"step": 48000
},
{
"epoch": 0.62,
"learning_rate": 1.8867948750866563e-05,
"loss": 0.0358,
"step": 48500
},
{
"epoch": 0.63,
"learning_rate": 1.8546999768916734e-05,
"loss": 0.0365,
"step": 49000
},
{
"epoch": 0.64,
"learning_rate": 1.8226050786966905e-05,
"loss": 0.0343,
"step": 49500
},
{
"epoch": 0.64,
"learning_rate": 1.7905101805017076e-05,
"loss": 0.0364,
"step": 50000
},
{
"epoch": 0.65,
"learning_rate": 1.7584152823067247e-05,
"loss": 0.0363,
"step": 50500
},
{
"epoch": 0.65,
"learning_rate": 1.7263203841117417e-05,
"loss": 0.0338,
"step": 51000
},
{
"epoch": 0.66,
"learning_rate": 1.6942254859167585e-05,
"loss": 0.0361,
"step": 51500
},
{
"epoch": 0.67,
"learning_rate": 1.662130587721776e-05,
"loss": 0.0361,
"step": 52000
},
{
"epoch": 0.67,
"learning_rate": 1.630035689526793e-05,
"loss": 0.0329,
"step": 52500
},
{
"epoch": 0.68,
"learning_rate": 1.59794079133181e-05,
"loss": 0.0339,
"step": 53000
},
{
"epoch": 0.69,
"learning_rate": 1.5658458931368268e-05,
"loss": 0.0346,
"step": 53500
},
{
"epoch": 0.69,
"learning_rate": 1.5337509949418442e-05,
"loss": 0.0369,
"step": 54000
},
{
"epoch": 0.7,
"learning_rate": 1.5016560967468613e-05,
"loss": 0.0348,
"step": 54500
},
{
"epoch": 0.71,
"learning_rate": 1.469561198551878e-05,
"loss": 0.0362,
"step": 55000
},
{
"epoch": 0.71,
"learning_rate": 1.4374663003568953e-05,
"loss": 0.0353,
"step": 55500
},
{
"epoch": 0.72,
"learning_rate": 1.4053714021619124e-05,
"loss": 0.0345,
"step": 56000
},
{
"epoch": 0.73,
"learning_rate": 1.3732765039669297e-05,
"loss": 0.0338,
"step": 56500
},
{
"epoch": 0.73,
"learning_rate": 1.3411816057719464e-05,
"loss": 0.0349,
"step": 57000
},
{
"epoch": 0.74,
"learning_rate": 1.3090867075769637e-05,
"loss": 0.0333,
"step": 57500
},
{
"epoch": 0.74,
"learning_rate": 1.2769918093819807e-05,
"loss": 0.0334,
"step": 58000
},
{
"epoch": 0.75,
"learning_rate": 1.2448969111869978e-05,
"loss": 0.0331,
"step": 58500
},
{
"epoch": 0.76,
"learning_rate": 1.2128020129920147e-05,
"loss": 0.0363,
"step": 59000
},
{
"epoch": 0.76,
"learning_rate": 1.180707114797032e-05,
"loss": 0.0361,
"step": 59500
},
{
"epoch": 0.77,
"learning_rate": 1.1486122166020489e-05,
"loss": 0.0358,
"step": 60000
},
{
"epoch": 0.78,
"learning_rate": 1.1165173184070662e-05,
"loss": 0.0346,
"step": 60500
},
{
"epoch": 0.78,
"learning_rate": 1.084422420212083e-05,
"loss": 0.0336,
"step": 61000
},
{
"epoch": 0.79,
"learning_rate": 1.0523275220171003e-05,
"loss": 0.0343,
"step": 61500
},
{
"epoch": 0.8,
"learning_rate": 1.0202326238221172e-05,
"loss": 0.0323,
"step": 62000
},
{
"epoch": 0.8,
"learning_rate": 9.881377256271343e-06,
"loss": 0.0361,
"step": 62500
},
{
"epoch": 0.81,
"learning_rate": 9.560428274321514e-06,
"loss": 0.0322,
"step": 63000
},
{
"epoch": 0.82,
"learning_rate": 9.239479292371685e-06,
"loss": 0.0344,
"step": 63500
},
{
"epoch": 0.82,
"learning_rate": 8.918530310421856e-06,
"loss": 0.0325,
"step": 64000
},
{
"epoch": 0.83,
"learning_rate": 8.597581328472027e-06,
"loss": 0.0324,
"step": 64500
},
{
"epoch": 0.83,
"learning_rate": 8.276632346522197e-06,
"loss": 0.036,
"step": 65000
},
{
"epoch": 0.84,
"learning_rate": 7.955683364572368e-06,
"loss": 0.0368,
"step": 65500
},
{
"epoch": 0.85,
"learning_rate": 7.634734382622537e-06,
"loss": 0.0341,
"step": 66000
},
{
"epoch": 0.85,
"learning_rate": 7.31378540067271e-06,
"loss": 0.0357,
"step": 66500
},
{
"epoch": 0.86,
"learning_rate": 6.99283641872288e-06,
"loss": 0.0353,
"step": 67000
},
{
"epoch": 0.87,
"learning_rate": 6.671887436773052e-06,
"loss": 0.0341,
"step": 67500
},
{
"epoch": 0.87,
"learning_rate": 6.3509384548232216e-06,
"loss": 0.0338,
"step": 68000
},
{
"epoch": 0.88,
"learning_rate": 6.029989472873392e-06,
"loss": 0.0387,
"step": 68500
},
{
"epoch": 0.89,
"learning_rate": 5.709040490923563e-06,
"loss": 0.0335,
"step": 69000
},
{
"epoch": 0.89,
"learning_rate": 5.388091508973734e-06,
"loss": 0.0367,
"step": 69500
},
{
"epoch": 0.9,
"learning_rate": 5.067142527023905e-06,
"loss": 0.0344,
"step": 70000
},
{
"epoch": 0.91,
"learning_rate": 4.746193545074075e-06,
"loss": 0.0341,
"step": 70500
},
{
"epoch": 0.91,
"learning_rate": 4.425244563124246e-06,
"loss": 0.0357,
"step": 71000
},
{
"epoch": 0.92,
"learning_rate": 4.1042955811744166e-06,
"loss": 0.036,
"step": 71500
},
{
"epoch": 0.92,
"learning_rate": 3.7833465992245874e-06,
"loss": 0.0345,
"step": 72000
},
{
"epoch": 0.93,
"learning_rate": 3.4623976172747582e-06,
"loss": 0.0353,
"step": 72500
},
{
"epoch": 0.94,
"learning_rate": 3.141448635324929e-06,
"loss": 0.0365,
"step": 73000
},
{
"epoch": 0.94,
"learning_rate": 2.8204996533750995e-06,
"loss": 0.0354,
"step": 73500
},
{
"epoch": 0.95,
"learning_rate": 2.4995506714252703e-06,
"loss": 0.0328,
"step": 74000
},
{
"epoch": 0.96,
"learning_rate": 2.178601689475441e-06,
"loss": 0.0318,
"step": 74500
},
{
"epoch": 0.96,
"learning_rate": 1.8576527075256118e-06,
"loss": 0.0349,
"step": 75000
},
{
"epoch": 0.97,
"learning_rate": 1.5367037255757826e-06,
"loss": 0.0328,
"step": 75500
},
{
"epoch": 0.98,
"learning_rate": 1.2157547436259532e-06,
"loss": 0.0328,
"step": 76000
},
{
"epoch": 0.98,
"learning_rate": 8.948057616761241e-07,
"loss": 0.0364,
"step": 76500
},
{
"epoch": 0.99,
"learning_rate": 5.738567797262948e-07,
"loss": 0.0334,
"step": 77000
},
{
"epoch": 0.99,
"learning_rate": 2.529077977764655e-07,
"loss": 0.0348,
"step": 77500
},
{
"epoch": 1.0,
"eval_loss": 0.03994523733854294,
"eval_runtime": 15.5614,
"eval_samples_per_second": 67.346,
"eval_steps_per_second": 33.673,
"step": 77894
},
{
"epoch": 1.0,
"step": 77894,
"total_flos": 4718099882050560.0,
"train_loss": 0.04686258947643893,
"train_runtime": 14067.3463,
"train_samples_per_second": 11.074,
"train_steps_per_second": 5.537
}
],
"max_steps": 77894,
"num_train_epochs": 1,
"total_flos": 4718099882050560.0,
"trial_name": null,
"trial_params": null
}