327 lines
7.7 KiB
JSON
327 lines
7.7 KiB
JSON
|
{
|
||
|
|
"best_global_step": 2030,
|
||
|
|
"best_metric": 3.28729248046875,
|
||
|
|
"best_model_checkpoint": "/home/p318482/syntactic-bootstrapping/model_trained/cds_sh1_fr_30/checkpoint-2030",
|
||
|
|
"epoch": 19.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 2755,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.006896551724137931,
|
||
|
|
"grad_norm": 34.13589859008789,
|
||
|
|
"learning_rate": 0.0,
|
||
|
|
"loss": 9.8164,
|
||
|
|
"step": 1
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"grad_norm": 1.4214282035827637,
|
||
|
|
"learning_rate": 2.88e-05,
|
||
|
|
"loss": 6.0017,
|
||
|
|
"step": 145
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"eval_loss": 4.516501426696777,
|
||
|
|
"eval_runtime": 4.3416,
|
||
|
|
"eval_samples_per_second": 810.983,
|
||
|
|
"eval_steps_per_second": 3.225,
|
||
|
|
"step": 145
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0,
|
||
|
|
"grad_norm": 1.7844622135162354,
|
||
|
|
"learning_rate": 5.7799999999999995e-05,
|
||
|
|
"loss": 4.1747,
|
||
|
|
"step": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0,
|
||
|
|
"eval_loss": 4.028092384338379,
|
||
|
|
"eval_runtime": 4.4671,
|
||
|
|
"eval_samples_per_second": 788.2,
|
||
|
|
"eval_steps_per_second": 3.134,
|
||
|
|
"step": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0,
|
||
|
|
"grad_norm": 1.2072691917419434,
|
||
|
|
"learning_rate": 8.680000000000001e-05,
|
||
|
|
"loss": 3.8777,
|
||
|
|
"step": 435
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0,
|
||
|
|
"eval_loss": 3.7704145908355713,
|
||
|
|
"eval_runtime": 4.3729,
|
||
|
|
"eval_samples_per_second": 805.181,
|
||
|
|
"eval_steps_per_second": 3.202,
|
||
|
|
"step": 435
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0,
|
||
|
|
"grad_norm": 1.2117047309875488,
|
||
|
|
"learning_rate": 9.668067226890756e-05,
|
||
|
|
"loss": 3.6723,
|
||
|
|
"step": 580
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0,
|
||
|
|
"eval_loss": 3.6042661666870117,
|
||
|
|
"eval_runtime": 4.4715,
|
||
|
|
"eval_samples_per_second": 787.434,
|
||
|
|
"eval_steps_per_second": 3.131,
|
||
|
|
"step": 580
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"grad_norm": 1.2041577100753784,
|
||
|
|
"learning_rate": 9.058823529411765e-05,
|
||
|
|
"loss": 3.5366,
|
||
|
|
"step": 725
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"eval_loss": 3.5113701820373535,
|
||
|
|
"eval_runtime": 4.511,
|
||
|
|
"eval_samples_per_second": 780.542,
|
||
|
|
"eval_steps_per_second": 3.104,
|
||
|
|
"step": 725
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.0,
|
||
|
|
"grad_norm": 1.1716428995132446,
|
||
|
|
"learning_rate": 8.449579831932774e-05,
|
||
|
|
"loss": 3.4452,
|
||
|
|
"step": 870
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.0,
|
||
|
|
"eval_loss": 3.4479944705963135,
|
||
|
|
"eval_runtime": 4.4139,
|
||
|
|
"eval_samples_per_second": 797.713,
|
||
|
|
"eval_steps_per_second": 3.172,
|
||
|
|
"step": 870
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.0,
|
||
|
|
"grad_norm": 1.1970024108886719,
|
||
|
|
"learning_rate": 7.840336134453782e-05,
|
||
|
|
"loss": 3.3743,
|
||
|
|
"step": 1015
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.0,
|
||
|
|
"eval_loss": 3.4008681774139404,
|
||
|
|
"eval_runtime": 4.385,
|
||
|
|
"eval_samples_per_second": 802.963,
|
||
|
|
"eval_steps_per_second": 3.193,
|
||
|
|
"step": 1015
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 8.0,
|
||
|
|
"grad_norm": 1.2204298973083496,
|
||
|
|
"learning_rate": 7.23109243697479e-05,
|
||
|
|
"loss": 3.3153,
|
||
|
|
"step": 1160
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 8.0,
|
||
|
|
"eval_loss": 3.366744041442871,
|
||
|
|
"eval_runtime": 4.437,
|
||
|
|
"eval_samples_per_second": 793.546,
|
||
|
|
"eval_steps_per_second": 3.155,
|
||
|
|
"step": 1160
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 9.0,
|
||
|
|
"grad_norm": 1.3044636249542236,
|
||
|
|
"learning_rate": 6.621848739495798e-05,
|
||
|
|
"loss": 3.2657,
|
||
|
|
"step": 1305
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 9.0,
|
||
|
|
"eval_loss": 3.3413963317871094,
|
||
|
|
"eval_runtime": 4.694,
|
||
|
|
"eval_samples_per_second": 750.11,
|
||
|
|
"eval_steps_per_second": 2.983,
|
||
|
|
"step": 1305
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 10.0,
|
||
|
|
"grad_norm": 1.4050198793411255,
|
||
|
|
"learning_rate": 6.012605042016807e-05,
|
||
|
|
"loss": 3.2193,
|
||
|
|
"step": 1450
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 10.0,
|
||
|
|
"eval_loss": 3.3189010620117188,
|
||
|
|
"eval_runtime": 4.3858,
|
||
|
|
"eval_samples_per_second": 802.816,
|
||
|
|
"eval_steps_per_second": 3.192,
|
||
|
|
"step": 1450
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 11.0,
|
||
|
|
"grad_norm": 1.4973615407943726,
|
||
|
|
"learning_rate": 5.403361344537815e-05,
|
||
|
|
"loss": 3.1773,
|
||
|
|
"step": 1595
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 11.0,
|
||
|
|
"eval_loss": 3.30843448638916,
|
||
|
|
"eval_runtime": 4.4298,
|
||
|
|
"eval_samples_per_second": 794.852,
|
||
|
|
"eval_steps_per_second": 3.16,
|
||
|
|
"step": 1595
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 12.0,
|
||
|
|
"grad_norm": 1.5111000537872314,
|
||
|
|
"learning_rate": 4.794117647058824e-05,
|
||
|
|
"loss": 3.1366,
|
||
|
|
"step": 1740
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 12.0,
|
||
|
|
"eval_loss": 3.2951903343200684,
|
||
|
|
"eval_runtime": 4.3718,
|
||
|
|
"eval_samples_per_second": 805.398,
|
||
|
|
"eval_steps_per_second": 3.202,
|
||
|
|
"step": 1740
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 13.0,
|
||
|
|
"grad_norm": 1.6258167028427124,
|
||
|
|
"learning_rate": 4.184873949579832e-05,
|
||
|
|
"loss": 3.0985,
|
||
|
|
"step": 1885
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 13.0,
|
||
|
|
"eval_loss": 3.2935917377471924,
|
||
|
|
"eval_runtime": 4.4103,
|
||
|
|
"eval_samples_per_second": 798.356,
|
||
|
|
"eval_steps_per_second": 3.174,
|
||
|
|
"step": 1885
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 14.0,
|
||
|
|
"grad_norm": 1.7234424352645874,
|
||
|
|
"learning_rate": 3.575630252100841e-05,
|
||
|
|
"loss": 3.061,
|
||
|
|
"step": 2030
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 14.0,
|
||
|
|
"eval_loss": 3.28729248046875,
|
||
|
|
"eval_runtime": 4.3993,
|
||
|
|
"eval_samples_per_second": 800.353,
|
||
|
|
"eval_steps_per_second": 3.182,
|
||
|
|
"step": 2030
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 15.0,
|
||
|
|
"grad_norm": 1.8610221147537231,
|
||
|
|
"learning_rate": 2.966386554621849e-05,
|
||
|
|
"loss": 3.0252,
|
||
|
|
"step": 2175
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 15.0,
|
||
|
|
"eval_loss": 3.2955870628356934,
|
||
|
|
"eval_runtime": 4.5452,
|
||
|
|
"eval_samples_per_second": 774.661,
|
||
|
|
"eval_steps_per_second": 3.08,
|
||
|
|
"step": 2175
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 16.0,
|
||
|
|
"grad_norm": 1.9218178987503052,
|
||
|
|
"learning_rate": 2.357142857142857e-05,
|
||
|
|
"loss": 2.9897,
|
||
|
|
"step": 2320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 16.0,
|
||
|
|
"eval_loss": 3.2937674522399902,
|
||
|
|
"eval_runtime": 4.4093,
|
||
|
|
"eval_samples_per_second": 798.538,
|
||
|
|
"eval_steps_per_second": 3.175,
|
||
|
|
"step": 2320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 17.0,
|
||
|
|
"grad_norm": 1.9699417352676392,
|
||
|
|
"learning_rate": 1.7478991596638656e-05,
|
||
|
|
"loss": 2.9574,
|
||
|
|
"step": 2465
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 17.0,
|
||
|
|
"eval_loss": 3.297903537750244,
|
||
|
|
"eval_runtime": 4.6469,
|
||
|
|
"eval_samples_per_second": 757.711,
|
||
|
|
"eval_steps_per_second": 3.013,
|
||
|
|
"step": 2465
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 18.0,
|
||
|
|
"grad_norm": 2.197739362716675,
|
||
|
|
"learning_rate": 1.138655462184874e-05,
|
||
|
|
"loss": 2.9302,
|
||
|
|
"step": 2610
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 18.0,
|
||
|
|
"eval_loss": 3.3014707565307617,
|
||
|
|
"eval_runtime": 4.4191,
|
||
|
|
"eval_samples_per_second": 796.766,
|
||
|
|
"eval_steps_per_second": 3.168,
|
||
|
|
"step": 2610
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 19.0,
|
||
|
|
"grad_norm": 2.1990742683410645,
|
||
|
|
"learning_rate": 5.294117647058824e-06,
|
||
|
|
"loss": 2.9063,
|
||
|
|
"step": 2755
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 19.0,
|
||
|
|
"eval_loss": 3.307129383087158,
|
||
|
|
"eval_runtime": 4.6156,
|
||
|
|
"eval_samples_per_second": 762.845,
|
||
|
|
"eval_steps_per_second": 3.033,
|
||
|
|
"step": 2755
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 500,
|
||
|
|
"max_steps": 2900,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 20,
|
||
|
|
"save_steps": 500,
|
||
|
|
"stateful_callbacks": {
|
||
|
|
"TrainerControl": {
|
||
|
|
"args": {
|
||
|
|
"should_epoch_stop": false,
|
||
|
|
"should_evaluate": false,
|
||
|
|
"should_log": false,
|
||
|
|
"should_save": true,
|
||
|
|
"should_training_stop": false
|
||
|
|
},
|
||
|
|
"attributes": {}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"total_flos": 4.5836436160512e+16,
|
||
|
|
"train_batch_size": 256,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|