GaetanMichelet's picture
Model save
671269b verified
{
"best_metric": 0.915133535861969,
"best_model_checkpoint": "data/Mistral-7B_task-1_120-samples_config-1_full/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": 2.566887855529785,
"learning_rate": 1.818181818181818e-06,
"loss": 2.0791,
"step": 1
},
{
"epoch": 0.18181818181818182,
"grad_norm": 2.830773115158081,
"learning_rate": 3.636363636363636e-06,
"loss": 2.1354,
"step": 2
},
{
"epoch": 0.36363636363636365,
"grad_norm": 2.6159067153930664,
"learning_rate": 7.272727272727272e-06,
"loss": 2.1214,
"step": 4
},
{
"epoch": 0.5454545454545454,
"grad_norm": 2.952951669692993,
"learning_rate": 1.0909090909090909e-05,
"loss": 2.0604,
"step": 6
},
{
"epoch": 0.7272727272727273,
"grad_norm": 2.1884078979492188,
"learning_rate": 1.4545454545454545e-05,
"loss": 1.9963,
"step": 8
},
{
"epoch": 0.9090909090909091,
"grad_norm": 2.105299949645996,
"learning_rate": 1.8181818181818182e-05,
"loss": 1.9008,
"step": 10
},
{
"epoch": 1.0,
"eval_loss": 1.7749409675598145,
"eval_runtime": 9.6515,
"eval_samples_per_second": 2.487,
"eval_steps_per_second": 2.487,
"step": 11
},
{
"epoch": 1.0909090909090908,
"grad_norm": 1.8031773567199707,
"learning_rate": 2.1818181818181818e-05,
"loss": 1.846,
"step": 12
},
{
"epoch": 1.2727272727272727,
"grad_norm": 1.4205222129821777,
"learning_rate": 2.5454545454545454e-05,
"loss": 1.7576,
"step": 14
},
{
"epoch": 1.4545454545454546,
"grad_norm": 1.291684865951538,
"learning_rate": 2.909090909090909e-05,
"loss": 1.5778,
"step": 16
},
{
"epoch": 1.6363636363636362,
"grad_norm": 1.338912844657898,
"learning_rate": 3.272727272727273e-05,
"loss": 1.5468,
"step": 18
},
{
"epoch": 1.8181818181818183,
"grad_norm": 1.7353410720825195,
"learning_rate": 3.6363636363636364e-05,
"loss": 1.3755,
"step": 20
},
{
"epoch": 2.0,
"grad_norm": 1.1533783674240112,
"learning_rate": 4e-05,
"loss": 1.2368,
"step": 22
},
{
"epoch": 2.0,
"eval_loss": 1.204591155052185,
"eval_runtime": 9.6569,
"eval_samples_per_second": 2.485,
"eval_steps_per_second": 2.485,
"step": 22
},
{
"epoch": 2.1818181818181817,
"grad_norm": 1.1981287002563477,
"learning_rate": 4.3636363636363636e-05,
"loss": 1.1336,
"step": 24
},
{
"epoch": 2.3636363636363638,
"grad_norm": 1.0418447256088257,
"learning_rate": 4.7272727272727275e-05,
"loss": 1.1206,
"step": 26
},
{
"epoch": 2.5454545454545454,
"grad_norm": 0.9300098419189453,
"learning_rate": 5.090909090909091e-05,
"loss": 0.9848,
"step": 28
},
{
"epoch": 2.7272727272727275,
"grad_norm": 0.876102864742279,
"learning_rate": 5.4545454545454546e-05,
"loss": 0.9623,
"step": 30
},
{
"epoch": 2.909090909090909,
"grad_norm": 0.8831844329833984,
"learning_rate": 5.818181818181818e-05,
"loss": 0.9597,
"step": 32
},
{
"epoch": 3.0,
"eval_loss": 0.976677417755127,
"eval_runtime": 9.648,
"eval_samples_per_second": 2.488,
"eval_steps_per_second": 2.488,
"step": 33
},
{
"epoch": 3.090909090909091,
"grad_norm": 0.8121486306190491,
"learning_rate": 6.181818181818182e-05,
"loss": 0.9711,
"step": 34
},
{
"epoch": 3.2727272727272725,
"grad_norm": 0.7559528946876526,
"learning_rate": 6.545454545454546e-05,
"loss": 0.8857,
"step": 36
},
{
"epoch": 3.4545454545454546,
"grad_norm": 0.7981972098350525,
"learning_rate": 6.90909090909091e-05,
"loss": 0.8503,
"step": 38
},
{
"epoch": 3.6363636363636362,
"grad_norm": 0.8330462574958801,
"learning_rate": 7.272727272727273e-05,
"loss": 0.8669,
"step": 40
},
{
"epoch": 3.8181818181818183,
"grad_norm": 1.0780049562454224,
"learning_rate": 7.636363636363637e-05,
"loss": 0.8486,
"step": 42
},
{
"epoch": 4.0,
"grad_norm": 1.0033646821975708,
"learning_rate": 8e-05,
"loss": 0.8364,
"step": 44
},
{
"epoch": 4.0,
"eval_loss": 0.9267880916595459,
"eval_runtime": 9.6473,
"eval_samples_per_second": 2.488,
"eval_steps_per_second": 2.488,
"step": 44
},
{
"epoch": 4.181818181818182,
"grad_norm": 0.9593367576599121,
"learning_rate": 8.363636363636364e-05,
"loss": 0.7425,
"step": 46
},
{
"epoch": 4.363636363636363,
"grad_norm": 1.122158408164978,
"learning_rate": 8.727272727272727e-05,
"loss": 0.7308,
"step": 48
},
{
"epoch": 4.545454545454545,
"grad_norm": 1.0962390899658203,
"learning_rate": 9.090909090909092e-05,
"loss": 0.7428,
"step": 50
},
{
"epoch": 4.7272727272727275,
"grad_norm": 1.0357098579406738,
"learning_rate": 9.454545454545455e-05,
"loss": 0.7131,
"step": 52
},
{
"epoch": 4.909090909090909,
"grad_norm": 1.1160370111465454,
"learning_rate": 9.818181818181818e-05,
"loss": 0.6751,
"step": 54
},
{
"epoch": 5.0,
"eval_loss": 0.915133535861969,
"eval_runtime": 9.6465,
"eval_samples_per_second": 2.488,
"eval_steps_per_second": 2.488,
"step": 55
},
{
"epoch": 5.090909090909091,
"grad_norm": 1.1304196119308472,
"learning_rate": 9.999899300364532e-05,
"loss": 0.6196,
"step": 56
},
{
"epoch": 5.2727272727272725,
"grad_norm": 1.3260724544525146,
"learning_rate": 9.99909372761763e-05,
"loss": 0.5389,
"step": 58
},
{
"epoch": 5.454545454545454,
"grad_norm": 2.0141329765319824,
"learning_rate": 9.997482711915927e-05,
"loss": 0.5146,
"step": 60
},
{
"epoch": 5.636363636363637,
"grad_norm": 1.6033296585083008,
"learning_rate": 9.99506651282272e-05,
"loss": 0.5495,
"step": 62
},
{
"epoch": 5.818181818181818,
"grad_norm": 1.500605821609497,
"learning_rate": 9.991845519630678e-05,
"loss": 0.4957,
"step": 64
},
{
"epoch": 6.0,
"grad_norm": 1.3610997200012207,
"learning_rate": 9.987820251299122e-05,
"loss": 0.492,
"step": 66
},
{
"epoch": 6.0,
"eval_loss": 0.96142578125,
"eval_runtime": 9.6466,
"eval_samples_per_second": 2.488,
"eval_steps_per_second": 2.488,
"step": 66
},
{
"epoch": 6.181818181818182,
"grad_norm": 1.1826406717300415,
"learning_rate": 9.982991356370404e-05,
"loss": 0.395,
"step": 68
},
{
"epoch": 6.363636363636363,
"grad_norm": 1.3635964393615723,
"learning_rate": 9.977359612865423e-05,
"loss": 0.3639,
"step": 70
},
{
"epoch": 6.545454545454545,
"grad_norm": 1.9806699752807617,
"learning_rate": 9.970925928158274e-05,
"loss": 0.3793,
"step": 72
},
{
"epoch": 6.7272727272727275,
"grad_norm": 2.0373482704162598,
"learning_rate": 9.963691338830044e-05,
"loss": 0.386,
"step": 74
},
{
"epoch": 6.909090909090909,
"grad_norm": 1.5265707969665527,
"learning_rate": 9.955657010501806e-05,
"loss": 0.3136,
"step": 76
},
{
"epoch": 7.0,
"eval_loss": 1.0793741941452026,
"eval_runtime": 9.6816,
"eval_samples_per_second": 2.479,
"eval_steps_per_second": 2.479,
"step": 77
},
{
"epoch": 7.090909090909091,
"grad_norm": 1.2268736362457275,
"learning_rate": 9.946824237646824e-05,
"loss": 0.288,
"step": 78
},
{
"epoch": 7.2727272727272725,
"grad_norm": 1.1878007650375366,
"learning_rate": 9.937194443381972e-05,
"loss": 0.2682,
"step": 80
},
{
"epoch": 7.454545454545454,
"grad_norm": 1.615853190422058,
"learning_rate": 9.926769179238466e-05,
"loss": 0.2332,
"step": 82
},
{
"epoch": 7.636363636363637,
"grad_norm": 1.8352854251861572,
"learning_rate": 9.915550124911866e-05,
"loss": 0.2292,
"step": 84
},
{
"epoch": 7.818181818181818,
"grad_norm": 1.9822676181793213,
"learning_rate": 9.903539087991462e-05,
"loss": 0.2604,
"step": 86
},
{
"epoch": 8.0,
"grad_norm": 2.0019595623016357,
"learning_rate": 9.890738003669029e-05,
"loss": 0.2439,
"step": 88
},
{
"epoch": 8.0,
"eval_loss": 1.2905783653259277,
"eval_runtime": 9.6438,
"eval_samples_per_second": 2.489,
"eval_steps_per_second": 2.489,
"step": 88
},
{
"epoch": 8.181818181818182,
"grad_norm": 1.3164584636688232,
"learning_rate": 9.877148934427037e-05,
"loss": 0.1722,
"step": 90
},
{
"epoch": 8.363636363636363,
"grad_norm": 1.6368311643600464,
"learning_rate": 9.862774069706346e-05,
"loss": 0.177,
"step": 92
},
{
"epoch": 8.545454545454545,
"grad_norm": 1.649357795715332,
"learning_rate": 9.847615725553456e-05,
"loss": 0.1658,
"step": 94
},
{
"epoch": 8.727272727272727,
"grad_norm": 1.3624839782714844,
"learning_rate": 9.831676344247342e-05,
"loss": 0.1729,
"step": 96
},
{
"epoch": 8.909090909090908,
"grad_norm": 1.5218144655227661,
"learning_rate": 9.814958493905963e-05,
"loss": 0.1635,
"step": 98
},
{
"epoch": 9.0,
"eval_loss": 1.2869492769241333,
"eval_runtime": 9.6504,
"eval_samples_per_second": 2.487,
"eval_steps_per_second": 2.487,
"step": 99
},
{
"epoch": 9.090909090909092,
"grad_norm": 1.094982624053955,
"learning_rate": 9.797464868072488e-05,
"loss": 0.1559,
"step": 100
},
{
"epoch": 9.272727272727273,
"grad_norm": 1.2318857908248901,
"learning_rate": 9.779198285281325e-05,
"loss": 0.1275,
"step": 102
},
{
"epoch": 9.454545454545455,
"grad_norm": 2.4848134517669678,
"learning_rate": 9.760161688604008e-05,
"loss": 0.1259,
"step": 104
},
{
"epoch": 9.636363636363637,
"grad_norm": 1.4462814331054688,
"learning_rate": 9.740358145174998e-05,
"loss": 0.1264,
"step": 106
},
{
"epoch": 9.818181818181818,
"grad_norm": 1.9087587594985962,
"learning_rate": 9.719790845697533e-05,
"loss": 0.1368,
"step": 108
},
{
"epoch": 10.0,
"grad_norm": 1.2537627220153809,
"learning_rate": 9.698463103929542e-05,
"loss": 0.1396,
"step": 110
},
{
"epoch": 10.0,
"eval_loss": 1.3566713333129883,
"eval_runtime": 9.651,
"eval_samples_per_second": 2.487,
"eval_steps_per_second": 2.487,
"step": 110
},
{
"epoch": 10.181818181818182,
"grad_norm": 1.6255460977554321,
"learning_rate": 9.676378356149734e-05,
"loss": 0.1033,
"step": 112
},
{
"epoch": 10.363636363636363,
"grad_norm": 1.0529149770736694,
"learning_rate": 9.653540160603956e-05,
"loss": 0.1009,
"step": 114
},
{
"epoch": 10.545454545454545,
"grad_norm": 1.5140531063079834,
"learning_rate": 9.629952196931901e-05,
"loss": 0.1062,
"step": 116
},
{
"epoch": 10.727272727272727,
"grad_norm": 1.8233051300048828,
"learning_rate": 9.60561826557425e-05,
"loss": 0.1153,
"step": 118
},
{
"epoch": 10.909090909090908,
"grad_norm": 1.4265450239181519,
"learning_rate": 9.580542287160348e-05,
"loss": 0.116,
"step": 120
},
{
"epoch": 11.0,
"eval_loss": 1.34817636013031,
"eval_runtime": 9.6478,
"eval_samples_per_second": 2.488,
"eval_steps_per_second": 2.488,
"step": 121
},
{
"epoch": 11.090909090909092,
"grad_norm": 0.8790718913078308,
"learning_rate": 9.554728301876526e-05,
"loss": 0.0999,
"step": 122
},
{
"epoch": 11.272727272727273,
"grad_norm": 0.858034074306488,
"learning_rate": 9.528180468815155e-05,
"loss": 0.0895,
"step": 124
},
{
"epoch": 11.454545454545455,
"grad_norm": 0.832482099533081,
"learning_rate": 9.50090306530454e-05,
"loss": 0.089,
"step": 126
},
{
"epoch": 11.636363636363637,
"grad_norm": 1.222090244293213,
"learning_rate": 9.472900486219769e-05,
"loss": 0.0922,
"step": 128
},
{
"epoch": 11.818181818181818,
"grad_norm": 1.581915020942688,
"learning_rate": 9.444177243274618e-05,
"loss": 0.1027,
"step": 130
},
{
"epoch": 12.0,
"grad_norm": 1.0929337739944458,
"learning_rate": 9.414737964294636e-05,
"loss": 0.104,
"step": 132
},
{
"epoch": 12.0,
"eval_loss": 1.4276927709579468,
"eval_runtime": 9.6483,
"eval_samples_per_second": 2.487,
"eval_steps_per_second": 2.487,
"step": 132
},
{
"epoch": 12.0,
"step": 132,
"total_flos": 1.5907229681057792e+16,
"train_loss": 0.6500745423589692,
"train_runtime": 1244.1938,
"train_samples_per_second": 3.536,
"train_steps_per_second": 0.442
}
],
"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": 1.5907229681057792e+16,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}