GaetanMichelet's picture
Model save
d9684a7 verified
{
"best_metric": 0.7019143104553223,
"best_model_checkpoint": "data/Llama-31-8B_task-2_120-samples_config-1/checkpoint-55",
"epoch": 12.0,
"eval_steps": 500,
"global_step": 132,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.09090909090909091,
"grad_norm": 0.7728537321090698,
"learning_rate": 1.818181818181818e-06,
"loss": 1.102,
"step": 1
},
{
"epoch": 0.18181818181818182,
"grad_norm": 0.8361679315567017,
"learning_rate": 3.636363636363636e-06,
"loss": 1.2412,
"step": 2
},
{
"epoch": 0.36363636363636365,
"grad_norm": 0.7789838910102844,
"learning_rate": 7.272727272727272e-06,
"loss": 1.0306,
"step": 4
},
{
"epoch": 0.5454545454545454,
"grad_norm": 0.7142726182937622,
"learning_rate": 1.0909090909090909e-05,
"loss": 1.0953,
"step": 6
},
{
"epoch": 0.7272727272727273,
"grad_norm": 0.7831631898880005,
"learning_rate": 1.4545454545454545e-05,
"loss": 1.1798,
"step": 8
},
{
"epoch": 0.9090909090909091,
"grad_norm": 0.6790210008621216,
"learning_rate": 1.8181818181818182e-05,
"loss": 1.0038,
"step": 10
},
{
"epoch": 1.0,
"eval_loss": 1.0449191331863403,
"eval_runtime": 24.3618,
"eval_samples_per_second": 0.985,
"eval_steps_per_second": 0.985,
"step": 11
},
{
"epoch": 1.0909090909090908,
"grad_norm": 0.704658031463623,
"learning_rate": 2.1818181818181818e-05,
"loss": 1.0256,
"step": 12
},
{
"epoch": 1.2727272727272727,
"grad_norm": 0.5898718237876892,
"learning_rate": 2.5454545454545454e-05,
"loss": 1.0022,
"step": 14
},
{
"epoch": 1.4545454545454546,
"grad_norm": 0.5323920249938965,
"learning_rate": 2.909090909090909e-05,
"loss": 1.0023,
"step": 16
},
{
"epoch": 1.6363636363636362,
"grad_norm": 0.4282853305339813,
"learning_rate": 3.272727272727273e-05,
"loss": 0.9719,
"step": 18
},
{
"epoch": 1.8181818181818183,
"grad_norm": 0.3479868769645691,
"learning_rate": 3.6363636363636364e-05,
"loss": 0.8616,
"step": 20
},
{
"epoch": 2.0,
"grad_norm": 0.3353016674518585,
"learning_rate": 4e-05,
"loss": 0.7984,
"step": 22
},
{
"epoch": 2.0,
"eval_loss": 0.8688964247703552,
"eval_runtime": 24.3695,
"eval_samples_per_second": 0.985,
"eval_steps_per_second": 0.985,
"step": 22
},
{
"epoch": 2.1818181818181817,
"grad_norm": 0.3350922763347626,
"learning_rate": 4.3636363636363636e-05,
"loss": 0.8245,
"step": 24
},
{
"epoch": 2.3636363636363638,
"grad_norm": 0.3857496976852417,
"learning_rate": 4.7272727272727275e-05,
"loss": 0.9021,
"step": 26
},
{
"epoch": 2.5454545454545454,
"grad_norm": 0.339462012052536,
"learning_rate": 5.090909090909091e-05,
"loss": 0.7947,
"step": 28
},
{
"epoch": 2.7272727272727275,
"grad_norm": 0.31833556294441223,
"learning_rate": 5.4545454545454546e-05,
"loss": 0.7581,
"step": 30
},
{
"epoch": 2.909090909090909,
"grad_norm": 0.33274346590042114,
"learning_rate": 5.818181818181818e-05,
"loss": 0.7296,
"step": 32
},
{
"epoch": 3.0,
"eval_loss": 0.7799892425537109,
"eval_runtime": 24.3558,
"eval_samples_per_second": 0.985,
"eval_steps_per_second": 0.985,
"step": 33
},
{
"epoch": 3.090909090909091,
"grad_norm": 0.3070923388004303,
"learning_rate": 6.181818181818182e-05,
"loss": 0.8024,
"step": 34
},
{
"epoch": 3.2727272727272725,
"grad_norm": 0.2655418813228607,
"learning_rate": 6.545454545454546e-05,
"loss": 0.743,
"step": 36
},
{
"epoch": 3.4545454545454546,
"grad_norm": 0.33472537994384766,
"learning_rate": 6.90909090909091e-05,
"loss": 0.726,
"step": 38
},
{
"epoch": 3.6363636363636362,
"grad_norm": 0.2951119542121887,
"learning_rate": 7.272727272727273e-05,
"loss": 0.6722,
"step": 40
},
{
"epoch": 3.8181818181818183,
"grad_norm": 0.3927033543586731,
"learning_rate": 7.636363636363637e-05,
"loss": 0.7468,
"step": 42
},
{
"epoch": 4.0,
"grad_norm": 0.29957640171051025,
"learning_rate": 8e-05,
"loss": 0.6657,
"step": 44
},
{
"epoch": 4.0,
"eval_loss": 0.7303240895271301,
"eval_runtime": 24.3595,
"eval_samples_per_second": 0.985,
"eval_steps_per_second": 0.985,
"step": 44
},
{
"epoch": 4.181818181818182,
"grad_norm": 0.36573946475982666,
"learning_rate": 8.363636363636364e-05,
"loss": 0.6997,
"step": 46
},
{
"epoch": 4.363636363636363,
"grad_norm": 0.3403160870075226,
"learning_rate": 8.727272727272727e-05,
"loss": 0.6135,
"step": 48
},
{
"epoch": 4.545454545454545,
"grad_norm": 0.3396548628807068,
"learning_rate": 9.090909090909092e-05,
"loss": 0.6566,
"step": 50
},
{
"epoch": 4.7272727272727275,
"grad_norm": 0.40658122301101685,
"learning_rate": 9.454545454545455e-05,
"loss": 0.6637,
"step": 52
},
{
"epoch": 4.909090909090909,
"grad_norm": 0.3593677878379822,
"learning_rate": 9.818181818181818e-05,
"loss": 0.6182,
"step": 54
},
{
"epoch": 5.0,
"eval_loss": 0.7019143104553223,
"eval_runtime": 24.3643,
"eval_samples_per_second": 0.985,
"eval_steps_per_second": 0.985,
"step": 55
},
{
"epoch": 5.090909090909091,
"grad_norm": 0.3561100661754608,
"learning_rate": 9.999899300364532e-05,
"loss": 0.5637,
"step": 56
},
{
"epoch": 5.2727272727272725,
"grad_norm": 0.4029925465583801,
"learning_rate": 9.99909372761763e-05,
"loss": 0.599,
"step": 58
},
{
"epoch": 5.454545454545454,
"grad_norm": 0.4341828227043152,
"learning_rate": 9.997482711915927e-05,
"loss": 0.549,
"step": 60
},
{
"epoch": 5.636363636363637,
"grad_norm": 0.49505615234375,
"learning_rate": 9.99506651282272e-05,
"loss": 0.5447,
"step": 62
},
{
"epoch": 5.818181818181818,
"grad_norm": 0.5305821299552917,
"learning_rate": 9.991845519630678e-05,
"loss": 0.5396,
"step": 64
},
{
"epoch": 6.0,
"grad_norm": 0.5642025470733643,
"learning_rate": 9.987820251299122e-05,
"loss": 0.5273,
"step": 66
},
{
"epoch": 6.0,
"eval_loss": 0.7108838558197021,
"eval_runtime": 24.3517,
"eval_samples_per_second": 0.986,
"eval_steps_per_second": 0.986,
"step": 66
},
{
"epoch": 6.181818181818182,
"grad_norm": 0.5778025388717651,
"learning_rate": 9.982991356370404e-05,
"loss": 0.4703,
"step": 68
},
{
"epoch": 6.363636363636363,
"grad_norm": 0.5730472803115845,
"learning_rate": 9.977359612865423e-05,
"loss": 0.4494,
"step": 70
},
{
"epoch": 6.545454545454545,
"grad_norm": 0.6630522608757019,
"learning_rate": 9.970925928158274e-05,
"loss": 0.4058,
"step": 72
},
{
"epoch": 6.7272727272727275,
"grad_norm": 0.6100122928619385,
"learning_rate": 9.963691338830044e-05,
"loss": 0.4222,
"step": 74
},
{
"epoch": 6.909090909090909,
"grad_norm": 0.6278058290481567,
"learning_rate": 9.955657010501806e-05,
"loss": 0.4129,
"step": 76
},
{
"epoch": 7.0,
"eval_loss": 0.7554300427436829,
"eval_runtime": 24.3483,
"eval_samples_per_second": 0.986,
"eval_steps_per_second": 0.986,
"step": 77
},
{
"epoch": 7.090909090909091,
"grad_norm": 0.6618012189865112,
"learning_rate": 9.946824237646824e-05,
"loss": 0.3733,
"step": 78
},
{
"epoch": 7.2727272727272725,
"grad_norm": 0.8544771075248718,
"learning_rate": 9.937194443381972e-05,
"loss": 0.2606,
"step": 80
},
{
"epoch": 7.454545454545454,
"grad_norm": 1.1671147346496582,
"learning_rate": 9.926769179238466e-05,
"loss": 0.2835,
"step": 82
},
{
"epoch": 7.636363636363637,
"grad_norm": 0.7617112994194031,
"learning_rate": 9.915550124911866e-05,
"loss": 0.281,
"step": 84
},
{
"epoch": 7.818181818181818,
"grad_norm": 0.6874366402626038,
"learning_rate": 9.903539087991462e-05,
"loss": 0.2826,
"step": 86
},
{
"epoch": 8.0,
"grad_norm": 1.045663833618164,
"learning_rate": 9.890738003669029e-05,
"loss": 0.3213,
"step": 88
},
{
"epoch": 8.0,
"eval_loss": 0.8616926074028015,
"eval_runtime": 24.3589,
"eval_samples_per_second": 0.985,
"eval_steps_per_second": 0.985,
"step": 88
},
{
"epoch": 8.181818181818182,
"grad_norm": 0.6918639540672302,
"learning_rate": 9.877148934427037e-05,
"loss": 0.1949,
"step": 90
},
{
"epoch": 8.363636363636363,
"grad_norm": 1.0379951000213623,
"learning_rate": 9.862774069706346e-05,
"loss": 0.1642,
"step": 92
},
{
"epoch": 8.545454545454545,
"grad_norm": 1.3061919212341309,
"learning_rate": 9.847615725553456e-05,
"loss": 0.1498,
"step": 94
},
{
"epoch": 8.727272727272727,
"grad_norm": 1.0143979787826538,
"learning_rate": 9.831676344247342e-05,
"loss": 0.1278,
"step": 96
},
{
"epoch": 8.909090909090908,
"grad_norm": 0.9734245538711548,
"learning_rate": 9.814958493905963e-05,
"loss": 0.1596,
"step": 98
},
{
"epoch": 9.0,
"eval_loss": 0.9911222457885742,
"eval_runtime": 24.3602,
"eval_samples_per_second": 0.985,
"eval_steps_per_second": 0.985,
"step": 99
},
{
"epoch": 9.090909090909092,
"grad_norm": 0.7598123550415039,
"learning_rate": 9.797464868072488e-05,
"loss": 0.136,
"step": 100
},
{
"epoch": 9.272727272727273,
"grad_norm": 0.9402742385864258,
"learning_rate": 9.779198285281325e-05,
"loss": 0.0738,
"step": 102
},
{
"epoch": 9.454545454545455,
"grad_norm": 0.9576045870780945,
"learning_rate": 9.760161688604008e-05,
"loss": 0.0585,
"step": 104
},
{
"epoch": 9.636363636363637,
"grad_norm": 0.9261518120765686,
"learning_rate": 9.740358145174998e-05,
"loss": 0.0808,
"step": 106
},
{
"epoch": 9.818181818181818,
"grad_norm": 1.1570649147033691,
"learning_rate": 9.719790845697533e-05,
"loss": 0.0656,
"step": 108
},
{
"epoch": 10.0,
"grad_norm": 0.9755429029464722,
"learning_rate": 9.698463103929542e-05,
"loss": 0.0669,
"step": 110
},
{
"epoch": 10.0,
"eval_loss": 1.2426633834838867,
"eval_runtime": 24.3514,
"eval_samples_per_second": 0.986,
"eval_steps_per_second": 0.986,
"step": 110
},
{
"epoch": 10.181818181818182,
"grad_norm": 0.5552050471305847,
"learning_rate": 9.676378356149734e-05,
"loss": 0.0282,
"step": 112
},
{
"epoch": 10.363636363636363,
"grad_norm": 0.8133194446563721,
"learning_rate": 9.653540160603956e-05,
"loss": 0.0337,
"step": 114
},
{
"epoch": 10.545454545454545,
"grad_norm": 1.0760459899902344,
"learning_rate": 9.629952196931901e-05,
"loss": 0.0398,
"step": 116
},
{
"epoch": 10.727272727272727,
"grad_norm": 0.9968352913856506,
"learning_rate": 9.60561826557425e-05,
"loss": 0.0362,
"step": 118
},
{
"epoch": 10.909090909090908,
"grad_norm": 0.6872853636741638,
"learning_rate": 9.580542287160348e-05,
"loss": 0.0274,
"step": 120
},
{
"epoch": 11.0,
"eval_loss": 1.383212924003601,
"eval_runtime": 24.3601,
"eval_samples_per_second": 0.985,
"eval_steps_per_second": 0.985,
"step": 121
},
{
"epoch": 11.090909090909092,
"grad_norm": 0.5362421274185181,
"learning_rate": 9.554728301876526e-05,
"loss": 0.0214,
"step": 122
},
{
"epoch": 11.272727272727273,
"grad_norm": 0.8885002136230469,
"learning_rate": 9.528180468815155e-05,
"loss": 0.0218,
"step": 124
},
{
"epoch": 11.454545454545455,
"grad_norm": 1.4024193286895752,
"learning_rate": 9.50090306530454e-05,
"loss": 0.017,
"step": 126
},
{
"epoch": 11.636363636363637,
"grad_norm": 0.7060501575469971,
"learning_rate": 9.472900486219769e-05,
"loss": 0.0226,
"step": 128
},
{
"epoch": 11.818181818181818,
"grad_norm": 0.6948946714401245,
"learning_rate": 9.444177243274618e-05,
"loss": 0.0226,
"step": 130
},
{
"epoch": 12.0,
"grad_norm": 0.9274241328239441,
"learning_rate": 9.414737964294636e-05,
"loss": 0.0273,
"step": 132
},
{
"epoch": 12.0,
"eval_loss": 1.3989367485046387,
"eval_runtime": 24.3521,
"eval_samples_per_second": 0.986,
"eval_steps_per_second": 0.986,
"step": 132
},
{
"epoch": 12.0,
"step": 132,
"total_flos": 1.0037430664246067e+17,
"train_loss": 0.47911660449410026,
"train_runtime": 3603.2128,
"train_samples_per_second": 1.221,
"train_steps_per_second": 0.153
}
],
"logging_steps": 2,
"max_steps": 550,
"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": 1.0037430664246067e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}