tiny-bert-sst2 / run-3 /checkpoint-2635 /trainer_state.json
jason-zhoou's picture
Training in progress, step 500
6925bae verified
raw
history blame
12.2 kB
{
"best_metric": 0.8188073394495413,
"best_model_checkpoint": "tiny-bert-sst2/run-3/checkpoint-1000",
"epoch": 5.0,
"eval_steps": 100,
"global_step": 2635,
"is_hyper_param_search": true,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.18975332068311196,
"grad_norm": 7.5281982421875,
"learning_rate": 0.00022048666969254715,
"loss": 1.6888,
"step": 100
},
{
"epoch": 0.18975332068311196,
"eval_accuracy": 0.7740825688073395,
"eval_loss": 1.2137372493743896,
"eval_runtime": 2.512,
"eval_samples_per_second": 347.137,
"eval_steps_per_second": 2.787,
"step": 100
},
{
"epoch": 0.3795066413662239,
"grad_norm": 8.994091033935547,
"learning_rate": 0.00022849311940488983,
"loss": 1.0387,
"step": 200
},
{
"epoch": 0.3795066413662239,
"eval_accuracy": 0.7958715596330275,
"eval_loss": 1.0297660827636719,
"eval_runtime": 2.4479,
"eval_samples_per_second": 356.218,
"eval_steps_per_second": 2.86,
"step": 200
},
{
"epoch": 0.5692599620493358,
"grad_norm": 4.1480183601379395,
"learning_rate": 0.00022592984315022126,
"loss": 0.7891,
"step": 300
},
{
"epoch": 0.5692599620493358,
"eval_accuracy": 0.7935779816513762,
"eval_loss": 0.9698419570922852,
"eval_runtime": 2.4189,
"eval_samples_per_second": 360.492,
"eval_steps_per_second": 2.894,
"step": 300
},
{
"epoch": 0.7590132827324478,
"grad_norm": 6.172699451446533,
"learning_rate": 0.00022165433603986336,
"loss": 0.6767,
"step": 400
},
{
"epoch": 0.7590132827324478,
"eval_accuracy": 0.8222477064220184,
"eval_loss": 0.9135227799415588,
"eval_runtime": 2.4368,
"eval_samples_per_second": 357.851,
"eval_steps_per_second": 2.873,
"step": 400
},
{
"epoch": 0.9487666034155597,
"grad_norm": 6.589296340942383,
"learning_rate": 0.00021573238588267377,
"loss": 0.617,
"step": 500
},
{
"epoch": 0.9487666034155597,
"eval_accuracy": 0.8073394495412844,
"eval_loss": 0.9335416555404663,
"eval_runtime": 2.4972,
"eval_samples_per_second": 349.185,
"eval_steps_per_second": 2.803,
"step": 500
},
{
"epoch": 1.1385199240986716,
"grad_norm": 4.637527942657471,
"learning_rate": 0.00020825511453038828,
"loss": 0.5142,
"step": 600
},
{
"epoch": 1.1385199240986716,
"eval_accuracy": 0.8073394495412844,
"eval_loss": 1.0174880027770996,
"eval_runtime": 2.431,
"eval_samples_per_second": 358.694,
"eval_steps_per_second": 2.879,
"step": 600
},
{
"epoch": 1.3282732447817835,
"grad_norm": 3.807858467102051,
"learning_rate": 0.00019933757577330664,
"loss": 0.4511,
"step": 700
},
{
"epoch": 1.3282732447817835,
"eval_accuracy": 0.8142201834862385,
"eval_loss": 0.9702788591384888,
"eval_runtime": 2.4299,
"eval_samples_per_second": 358.858,
"eval_steps_per_second": 2.881,
"step": 700
},
{
"epoch": 1.5180265654648957,
"grad_norm": 5.257256984710693,
"learning_rate": 0.000189116984991988,
"loss": 0.4292,
"step": 800
},
{
"epoch": 1.5180265654648957,
"eval_accuracy": 0.8027522935779816,
"eval_loss": 1.0670255422592163,
"eval_runtime": 2.4833,
"eval_samples_per_second": 351.149,
"eval_steps_per_second": 2.819,
"step": 800
},
{
"epoch": 1.7077798861480076,
"grad_norm": 4.167674541473389,
"learning_rate": 0.0001777506078055455,
"loss": 0.4201,
"step": 900
},
{
"epoch": 1.7077798861480076,
"eval_accuracy": 0.8165137614678899,
"eval_loss": 0.966061532497406,
"eval_runtime": 2.4276,
"eval_samples_per_second": 359.207,
"eval_steps_per_second": 2.884,
"step": 900
},
{
"epoch": 1.8975332068311195,
"grad_norm": 7.251418590545654,
"learning_rate": 0.00016541334020419587,
"loss": 0.4173,
"step": 1000
},
{
"epoch": 1.8975332068311195,
"eval_accuracy": 0.8188073394495413,
"eval_loss": 0.9646121263504028,
"eval_runtime": 2.4466,
"eval_samples_per_second": 356.409,
"eval_steps_per_second": 2.861,
"step": 1000
},
{
"epoch": 2.0872865275142316,
"grad_norm": 3.3101117610931396,
"learning_rate": 0.0001522950174008934,
"loss": 0.3537,
"step": 1100
},
{
"epoch": 2.0872865275142316,
"eval_accuracy": 0.8165137614678899,
"eval_loss": 1.00448477268219,
"eval_runtime": 2.4388,
"eval_samples_per_second": 357.559,
"eval_steps_per_second": 2.87,
"step": 1100
},
{
"epoch": 2.2770398481973433,
"grad_norm": 4.73748254776001,
"learning_rate": 0.00013859749281111374,
"loss": 0.3095,
"step": 1200
},
{
"epoch": 2.2770398481973433,
"eval_accuracy": 0.8153669724770642,
"eval_loss": 1.0533220767974854,
"eval_runtime": 2.4451,
"eval_samples_per_second": 356.635,
"eval_steps_per_second": 2.863,
"step": 1200
},
{
"epoch": 2.4667931688804554,
"grad_norm": 2.5525803565979004,
"learning_rate": 0.00012453153210692205,
"loss": 0.3254,
"step": 1300
},
{
"epoch": 2.4667931688804554,
"eval_accuracy": 0.8130733944954128,
"eval_loss": 1.1290571689605713,
"eval_runtime": 2.463,
"eval_samples_per_second": 354.04,
"eval_steps_per_second": 2.842,
"step": 1300
},
{
"epoch": 2.656546489563567,
"grad_norm": 5.130295276641846,
"learning_rate": 0.00011031357013693877,
"loss": 0.3135,
"step": 1400
},
{
"epoch": 2.656546489563567,
"eval_accuracy": 0.8073394495412844,
"eval_loss": 1.0808497667312622,
"eval_runtime": 2.434,
"eval_samples_per_second": 358.263,
"eval_steps_per_second": 2.876,
"step": 1400
},
{
"epoch": 2.846299810246679,
"grad_norm": 2.434037685394287,
"learning_rate": 9.616238061391467e-05,
"loss": 0.3162,
"step": 1500
},
{
"epoch": 2.846299810246679,
"eval_accuracy": 0.8096330275229358,
"eval_loss": 1.1154645681381226,
"eval_runtime": 2.4873,
"eval_samples_per_second": 350.575,
"eval_steps_per_second": 2.814,
"step": 1500
},
{
"epoch": 3.0360531309297913,
"grad_norm": 2.9498291015625,
"learning_rate": 8.229570981388642e-05,
"loss": 0.2907,
"step": 1600
},
{
"epoch": 3.0360531309297913,
"eval_accuracy": 0.823394495412844,
"eval_loss": 1.059111475944519,
"eval_runtime": 2.4279,
"eval_samples_per_second": 359.161,
"eval_steps_per_second": 2.883,
"step": 1600
},
{
"epoch": 3.225806451612903,
"grad_norm": 2.830878496170044,
"learning_rate": 6.892692608463966e-05,
"loss": 0.2325,
"step": 1700
},
{
"epoch": 3.225806451612903,
"eval_accuracy": 0.8153669724770642,
"eval_loss": 1.0795413255691528,
"eval_runtime": 2.4332,
"eval_samples_per_second": 358.383,
"eval_steps_per_second": 2.877,
"step": 1700
},
{
"epoch": 3.415559772296015,
"grad_norm": 2.6231229305267334,
"learning_rate": 5.62617367179498e-05,
"loss": 0.2423,
"step": 1800
},
{
"epoch": 3.415559772296015,
"eval_accuracy": 0.8211009174311926,
"eval_loss": 1.0745121240615845,
"eval_runtime": 2.5016,
"eval_samples_per_second": 348.574,
"eval_steps_per_second": 2.798,
"step": 1800
},
{
"epoch": 3.6053130929791273,
"grad_norm": 3.7570250034332275,
"learning_rate": 4.4495022703532034e-05,
"loss": 0.2608,
"step": 1900
},
{
"epoch": 3.6053130929791273,
"eval_accuracy": 0.8142201834862385,
"eval_loss": 1.0897266864776611,
"eval_runtime": 2.4393,
"eval_samples_per_second": 357.481,
"eval_steps_per_second": 2.87,
"step": 1900
},
{
"epoch": 3.795066413662239,
"grad_norm": 3.726491689682007,
"learning_rate": 3.3807840068772395e-05,
"loss": 0.2518,
"step": 2000
},
{
"epoch": 3.795066413662239,
"eval_accuracy": 0.8153669724770642,
"eval_loss": 1.0948952436447144,
"eval_runtime": 2.4397,
"eval_samples_per_second": 357.425,
"eval_steps_per_second": 2.869,
"step": 2000
},
{
"epoch": 3.984819734345351,
"grad_norm": 3.4763576984405518,
"learning_rate": 2.436463394503288e-05,
"loss": 0.2529,
"step": 2100
},
{
"epoch": 3.984819734345351,
"eval_accuracy": 0.8176605504587156,
"eval_loss": 1.1064103841781616,
"eval_runtime": 2.5075,
"eval_samples_per_second": 347.752,
"eval_steps_per_second": 2.792,
"step": 2100
},
{
"epoch": 4.174573055028463,
"grad_norm": 6.195148468017578,
"learning_rate": 1.631070822807034e-05,
"loss": 0.2251,
"step": 2200
},
{
"epoch": 4.174573055028463,
"eval_accuracy": 0.8142201834862385,
"eval_loss": 1.1260789632797241,
"eval_runtime": 2.4849,
"eval_samples_per_second": 350.92,
"eval_steps_per_second": 2.817,
"step": 2200
},
{
"epoch": 4.364326375711575,
"grad_norm": 2.506267547607422,
"learning_rate": 9.769989767248965e-06,
"loss": 0.221,
"step": 2300
},
{
"epoch": 4.364326375711575,
"eval_accuracy": 0.8130733944954128,
"eval_loss": 1.1184971332550049,
"eval_runtime": 2.448,
"eval_samples_per_second": 356.205,
"eval_steps_per_second": 2.859,
"step": 2300
},
{
"epoch": 4.554079696394687,
"grad_norm": 3.0686843395233154,
"learning_rate": 4.843121486269946e-06,
"loss": 0.2266,
"step": 2400
},
{
"epoch": 4.554079696394687,
"eval_accuracy": 0.8188073394495413,
"eval_loss": 1.1176307201385498,
"eval_runtime": 2.4651,
"eval_samples_per_second": 353.741,
"eval_steps_per_second": 2.84,
"step": 2400
},
{
"epoch": 4.743833017077799,
"grad_norm": 3.0805752277374268,
"learning_rate": 1.6059137768293674e-06,
"loss": 0.218,
"step": 2500
},
{
"epoch": 4.743833017077799,
"eval_accuracy": 0.8153669724770642,
"eval_loss": 1.1191787719726562,
"eval_runtime": 2.4416,
"eval_samples_per_second": 357.14,
"eval_steps_per_second": 2.867,
"step": 2500
},
{
"epoch": 4.933586337760911,
"grad_norm": 2.7237091064453125,
"learning_rate": 1.0817799382007904e-07,
"loss": 0.2257,
"step": 2600
},
{
"epoch": 4.933586337760911,
"eval_accuracy": 0.8119266055045872,
"eval_loss": 1.118699550628662,
"eval_runtime": 2.4777,
"eval_samples_per_second": 351.943,
"eval_steps_per_second": 2.825,
"step": 2600
}
],
"logging_steps": 100,
"max_steps": 2635,
"num_input_tokens_seen": 0,
"num_train_epochs": 5,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 39911918594520.0,
"train_batch_size": 128,
"trial_name": null,
"trial_params": {
"alpha": 0.7363133901666971,
"learning_rate": 0.00022930613648024903,
"num_train_epochs": 5,
"temperature": 13,
"warmup_ratio": 0.03930811279168249
}
}