GaetanMichelet's picture
Model save
c7c2035 verified
{
"best_metric": 0.8520960807800293,
"best_model_checkpoint": "data/Mistral-7B_task-3_120-samples_config-2_full_auto/checkpoint-44",
"epoch": 14.909090909090908,
"eval_steps": 500,
"global_step": 82,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.18181818181818182,
"grad_norm": 1.0692335367202759,
"learning_rate": 4.000000000000001e-06,
"loss": 1.2465,
"step": 1
},
{
"epoch": 0.36363636363636365,
"grad_norm": 0.9576011300086975,
"learning_rate": 8.000000000000001e-06,
"loss": 1.2172,
"step": 2
},
{
"epoch": 0.7272727272727273,
"grad_norm": 0.8975992202758789,
"learning_rate": 1.6000000000000003e-05,
"loss": 1.246,
"step": 4
},
{
"epoch": 0.9090909090909091,
"eval_loss": 1.1940295696258545,
"eval_runtime": 25.8928,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 5
},
{
"epoch": 1.0909090909090908,
"grad_norm": 0.860111653804779,
"learning_rate": 2.4e-05,
"loss": 1.2192,
"step": 6
},
{
"epoch": 1.4545454545454546,
"grad_norm": 0.570076584815979,
"learning_rate": 3.2000000000000005e-05,
"loss": 1.1504,
"step": 8
},
{
"epoch": 1.8181818181818183,
"grad_norm": 0.4158064126968384,
"learning_rate": 4e-05,
"loss": 1.1422,
"step": 10
},
{
"epoch": 2.0,
"eval_loss": 1.0793336629867554,
"eval_runtime": 25.8928,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 11
},
{
"epoch": 2.1818181818181817,
"grad_norm": 0.3297309875488281,
"learning_rate": 4.8e-05,
"loss": 1.0569,
"step": 12
},
{
"epoch": 2.5454545454545454,
"grad_norm": 0.3553498089313507,
"learning_rate": 5.6000000000000006e-05,
"loss": 1.0317,
"step": 14
},
{
"epoch": 2.909090909090909,
"grad_norm": 0.40063291788101196,
"learning_rate": 6.400000000000001e-05,
"loss": 1.0294,
"step": 16
},
{
"epoch": 2.909090909090909,
"eval_loss": 1.0065960884094238,
"eval_runtime": 25.8899,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 16
},
{
"epoch": 3.2727272727272725,
"grad_norm": 0.4419811964035034,
"learning_rate": 7.2e-05,
"loss": 0.9902,
"step": 18
},
{
"epoch": 3.6363636363636362,
"grad_norm": 0.43059974908828735,
"learning_rate": 8e-05,
"loss": 0.9199,
"step": 20
},
{
"epoch": 4.0,
"grad_norm": 0.30248573422431946,
"learning_rate": 8.800000000000001e-05,
"loss": 0.8899,
"step": 22
},
{
"epoch": 4.0,
"eval_loss": 0.9020195007324219,
"eval_runtime": 25.9265,
"eval_samples_per_second": 0.926,
"eval_steps_per_second": 0.926,
"step": 22
},
{
"epoch": 4.363636363636363,
"grad_norm": 0.26780542731285095,
"learning_rate": 9.6e-05,
"loss": 0.8597,
"step": 24
},
{
"epoch": 4.7272727272727275,
"grad_norm": 0.25114116072654724,
"learning_rate": 9.999512620046522e-05,
"loss": 0.8298,
"step": 26
},
{
"epoch": 4.909090909090909,
"eval_loss": 0.8736847043037415,
"eval_runtime": 25.8924,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 27
},
{
"epoch": 5.090909090909091,
"grad_norm": 0.21722856163978577,
"learning_rate": 9.995614150494293e-05,
"loss": 0.8389,
"step": 28
},
{
"epoch": 5.454545454545454,
"grad_norm": 0.2524879276752472,
"learning_rate": 9.987820251299122e-05,
"loss": 0.8138,
"step": 30
},
{
"epoch": 5.818181818181818,
"grad_norm": 0.20729583501815796,
"learning_rate": 9.976136999909156e-05,
"loss": 0.7997,
"step": 32
},
{
"epoch": 6.0,
"eval_loss": 0.8607468605041504,
"eval_runtime": 25.887,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 33
},
{
"epoch": 6.181818181818182,
"grad_norm": 0.21508027613162994,
"learning_rate": 9.96057350657239e-05,
"loss": 0.7913,
"step": 34
},
{
"epoch": 6.545454545454545,
"grad_norm": 0.21418441832065582,
"learning_rate": 9.941141907232765e-05,
"loss": 0.774,
"step": 36
},
{
"epoch": 6.909090909090909,
"grad_norm": 0.18334874510765076,
"learning_rate": 9.917857354066931e-05,
"loss": 0.7828,
"step": 38
},
{
"epoch": 6.909090909090909,
"eval_loss": 0.852733314037323,
"eval_runtime": 25.8882,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 38
},
{
"epoch": 7.2727272727272725,
"grad_norm": 0.2507544457912445,
"learning_rate": 9.890738003669029e-05,
"loss": 0.7758,
"step": 40
},
{
"epoch": 7.636363636363637,
"grad_norm": 0.22936642169952393,
"learning_rate": 9.859805002892732e-05,
"loss": 0.7366,
"step": 42
},
{
"epoch": 8.0,
"grad_norm": 0.22656971216201782,
"learning_rate": 9.825082472361557e-05,
"loss": 0.7537,
"step": 44
},
{
"epoch": 8.0,
"eval_loss": 0.8520960807800293,
"eval_runtime": 25.8859,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 44
},
{
"epoch": 8.363636363636363,
"grad_norm": 0.2110353410243988,
"learning_rate": 9.786597487660337e-05,
"loss": 0.7203,
"step": 46
},
{
"epoch": 8.727272727272727,
"grad_norm": 0.2937312722206116,
"learning_rate": 9.744380058222483e-05,
"loss": 0.701,
"step": 48
},
{
"epoch": 8.909090909090908,
"eval_loss": 0.8577341437339783,
"eval_runtime": 25.8858,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 49
},
{
"epoch": 9.090909090909092,
"grad_norm": 0.31091007590293884,
"learning_rate": 9.698463103929542e-05,
"loss": 0.7131,
"step": 50
},
{
"epoch": 9.454545454545455,
"grad_norm": 0.29342061281204224,
"learning_rate": 9.648882429441257e-05,
"loss": 0.6491,
"step": 52
},
{
"epoch": 9.818181818181818,
"grad_norm": 0.2867165803909302,
"learning_rate": 9.595676696276172e-05,
"loss": 0.6948,
"step": 54
},
{
"epoch": 10.0,
"eval_loss": 0.8666787147521973,
"eval_runtime": 25.8908,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 55
},
{
"epoch": 10.181818181818182,
"grad_norm": 0.3385396897792816,
"learning_rate": 9.538887392664544e-05,
"loss": 0.6249,
"step": 56
},
{
"epoch": 10.545454545454545,
"grad_norm": 0.5323635935783386,
"learning_rate": 9.478558801197065e-05,
"loss": 0.6032,
"step": 58
},
{
"epoch": 10.909090909090908,
"grad_norm": 0.4243149757385254,
"learning_rate": 9.414737964294636e-05,
"loss": 0.622,
"step": 60
},
{
"epoch": 10.909090909090908,
"eval_loss": 0.8851115107536316,
"eval_runtime": 25.8839,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 60
},
{
"epoch": 11.272727272727273,
"grad_norm": 0.523710310459137,
"learning_rate": 9.347474647526095e-05,
"loss": 0.5815,
"step": 62
},
{
"epoch": 11.636363636363637,
"grad_norm": 0.5354682207107544,
"learning_rate": 9.276821300802534e-05,
"loss": 0.574,
"step": 64
},
{
"epoch": 12.0,
"grad_norm": 0.6191315054893494,
"learning_rate": 9.202833017478422e-05,
"loss": 0.542,
"step": 66
},
{
"epoch": 12.0,
"eval_loss": 0.9303509593009949,
"eval_runtime": 25.8884,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 66
},
{
"epoch": 12.363636363636363,
"grad_norm": 0.7306078672409058,
"learning_rate": 9.125567491391476e-05,
"loss": 0.506,
"step": 68
},
{
"epoch": 12.727272727272727,
"grad_norm": 0.6805274486541748,
"learning_rate": 9.045084971874738e-05,
"loss": 0.4829,
"step": 70
},
{
"epoch": 12.909090909090908,
"eval_loss": 0.9684502482414246,
"eval_runtime": 25.8932,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 71
},
{
"epoch": 13.090909090909092,
"grad_norm": 0.6913831233978271,
"learning_rate": 8.961448216775954e-05,
"loss": 0.4696,
"step": 72
},
{
"epoch": 13.454545454545455,
"grad_norm": 0.6931716799736023,
"learning_rate": 8.874722443520899e-05,
"loss": 0.4739,
"step": 74
},
{
"epoch": 13.818181818181818,
"grad_norm": 0.847641110420227,
"learning_rate": 8.784975278258783e-05,
"loss": 0.4214,
"step": 76
},
{
"epoch": 14.0,
"eval_loss": 1.003703236579895,
"eval_runtime": 25.8886,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 77
},
{
"epoch": 14.181818181818182,
"grad_norm": 0.848358154296875,
"learning_rate": 8.692276703129421e-05,
"loss": 0.3705,
"step": 78
},
{
"epoch": 14.545454545454545,
"grad_norm": 0.9739524722099304,
"learning_rate": 8.596699001693255e-05,
"loss": 0.4026,
"step": 80
},
{
"epoch": 14.909090909090908,
"grad_norm": 1.1165748834609985,
"learning_rate": 8.498316702566828e-05,
"loss": 0.314,
"step": 82
},
{
"epoch": 14.909090909090908,
"eval_loss": 1.033321499824524,
"eval_runtime": 25.9019,
"eval_samples_per_second": 0.927,
"eval_steps_per_second": 0.927,
"step": 82
},
{
"epoch": 14.909090909090908,
"step": 82,
"total_flos": 1.2926137520042803e+17,
"train_loss": 0.761626343719843,
"train_runtime": 4805.8482,
"train_samples_per_second": 0.916,
"train_steps_per_second": 0.052
}
],
"logging_steps": 2,
"max_steps": 250,
"num_input_tokens_seen": 0,
"num_train_epochs": 50,
"save_steps": 25,
"stateful_callbacks": {
"EarlyStoppingCallback": {
"args": {
"early_stopping_patience": 7,
"early_stopping_threshold": 0.0
},
"attributes": {
"early_stopping_patience_counter": 0
}
},
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 1.2926137520042803e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}