GaetanMichelet's picture
Model save
623ada8 verified
{
"best_metric": 0.3860926926136017,
"best_model_checkpoint": "data/Mistral-7B_task-3_180-samples_config-1/checkpoint-68",
"epoch": 11.0,
"eval_steps": 500,
"global_step": 187,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.058823529411764705,
"grad_norm": 20.036640167236328,
"learning_rate": 1.1764705882352942e-06,
"loss": 3.994,
"step": 1
},
{
"epoch": 0.11764705882352941,
"grad_norm": 10.005636215209961,
"learning_rate": 2.3529411764705885e-06,
"loss": 2.6712,
"step": 2
},
{
"epoch": 0.23529411764705882,
"grad_norm": 18.995513916015625,
"learning_rate": 4.705882352941177e-06,
"loss": 4.2931,
"step": 4
},
{
"epoch": 0.35294117647058826,
"grad_norm": 9.773229598999023,
"learning_rate": 7.058823529411765e-06,
"loss": 3.7351,
"step": 6
},
{
"epoch": 0.47058823529411764,
"grad_norm": 18.577383041381836,
"learning_rate": 9.411764705882354e-06,
"loss": 3.1174,
"step": 8
},
{
"epoch": 0.5882352941176471,
"grad_norm": 16.158309936523438,
"learning_rate": 1.1764705882352942e-05,
"loss": 2.9667,
"step": 10
},
{
"epoch": 0.7058823529411765,
"grad_norm": 8.769892692565918,
"learning_rate": 1.411764705882353e-05,
"loss": 2.0929,
"step": 12
},
{
"epoch": 0.8235294117647058,
"grad_norm": 9.987327575683594,
"learning_rate": 1.647058823529412e-05,
"loss": 2.0934,
"step": 14
},
{
"epoch": 0.9411764705882353,
"grad_norm": 4.576476097106934,
"learning_rate": 1.8823529411764708e-05,
"loss": 1.6703,
"step": 16
},
{
"epoch": 1.0,
"eval_loss": 1.2306901216506958,
"eval_runtime": 39.0674,
"eval_samples_per_second": 0.921,
"eval_steps_per_second": 0.921,
"step": 17
},
{
"epoch": 1.0588235294117647,
"grad_norm": 3.577021360397339,
"learning_rate": 2.1176470588235296e-05,
"loss": 1.3695,
"step": 18
},
{
"epoch": 1.1764705882352942,
"grad_norm": 3.6731462478637695,
"learning_rate": 2.3529411764705884e-05,
"loss": 0.9679,
"step": 20
},
{
"epoch": 1.2941176470588236,
"grad_norm": 2.0665371417999268,
"learning_rate": 2.5882352941176475e-05,
"loss": 0.7898,
"step": 22
},
{
"epoch": 1.4117647058823528,
"grad_norm": 1.932107925415039,
"learning_rate": 2.823529411764706e-05,
"loss": 0.4938,
"step": 24
},
{
"epoch": 1.5294117647058822,
"grad_norm": 1.527052402496338,
"learning_rate": 3.058823529411765e-05,
"loss": 0.4646,
"step": 26
},
{
"epoch": 1.6470588235294117,
"grad_norm": 1.2368085384368896,
"learning_rate": 3.294117647058824e-05,
"loss": 0.4411,
"step": 28
},
{
"epoch": 1.7647058823529411,
"grad_norm": 1.3800956010818481,
"learning_rate": 3.529411764705883e-05,
"loss": 0.7017,
"step": 30
},
{
"epoch": 1.8823529411764706,
"grad_norm": 0.9972752928733826,
"learning_rate": 3.7647058823529415e-05,
"loss": 0.3532,
"step": 32
},
{
"epoch": 2.0,
"grad_norm": 0.667917788028717,
"learning_rate": 4e-05,
"loss": 0.4798,
"step": 34
},
{
"epoch": 2.0,
"eval_loss": 0.48258718848228455,
"eval_runtime": 39.0717,
"eval_samples_per_second": 0.921,
"eval_steps_per_second": 0.921,
"step": 34
},
{
"epoch": 2.1176470588235294,
"grad_norm": 1.0108541250228882,
"learning_rate": 4.235294117647059e-05,
"loss": 0.433,
"step": 36
},
{
"epoch": 2.235294117647059,
"grad_norm": 1.0303305387496948,
"learning_rate": 4.470588235294118e-05,
"loss": 0.483,
"step": 38
},
{
"epoch": 2.3529411764705883,
"grad_norm": 0.8273497223854065,
"learning_rate": 4.705882352941177e-05,
"loss": 0.3875,
"step": 40
},
{
"epoch": 2.4705882352941178,
"grad_norm": 0.7326416969299316,
"learning_rate": 4.9411764705882355e-05,
"loss": 0.2975,
"step": 42
},
{
"epoch": 2.588235294117647,
"grad_norm": 8.435672760009766,
"learning_rate": 5.176470588235295e-05,
"loss": 0.3408,
"step": 44
},
{
"epoch": 2.7058823529411766,
"grad_norm": 0.6872634291648865,
"learning_rate": 5.411764705882353e-05,
"loss": 0.3486,
"step": 46
},
{
"epoch": 2.8235294117647056,
"grad_norm": 1.593389630317688,
"learning_rate": 5.647058823529412e-05,
"loss": 0.3557,
"step": 48
},
{
"epoch": 2.9411764705882355,
"grad_norm": 0.5561437606811523,
"learning_rate": 5.882352941176471e-05,
"loss": 0.2974,
"step": 50
},
{
"epoch": 3.0,
"eval_loss": 0.4028509855270386,
"eval_runtime": 39.0794,
"eval_samples_per_second": 0.921,
"eval_steps_per_second": 0.921,
"step": 51
},
{
"epoch": 3.0588235294117645,
"grad_norm": 0.6127999424934387,
"learning_rate": 6.11764705882353e-05,
"loss": 0.2887,
"step": 52
},
{
"epoch": 3.176470588235294,
"grad_norm": 0.7683257460594177,
"learning_rate": 6.352941176470588e-05,
"loss": 0.3904,
"step": 54
},
{
"epoch": 3.2941176470588234,
"grad_norm": 1.034536361694336,
"learning_rate": 6.588235294117648e-05,
"loss": 0.3144,
"step": 56
},
{
"epoch": 3.411764705882353,
"grad_norm": 0.604548990726471,
"learning_rate": 6.823529411764707e-05,
"loss": 0.3521,
"step": 58
},
{
"epoch": 3.5294117647058822,
"grad_norm": 0.8289085626602173,
"learning_rate": 7.058823529411765e-05,
"loss": 0.1268,
"step": 60
},
{
"epoch": 3.6470588235294117,
"grad_norm": 0.9199197888374329,
"learning_rate": 7.294117647058823e-05,
"loss": 0.2537,
"step": 62
},
{
"epoch": 3.764705882352941,
"grad_norm": 1.2552107572555542,
"learning_rate": 7.529411764705883e-05,
"loss": 0.3479,
"step": 64
},
{
"epoch": 3.8823529411764706,
"grad_norm": 0.8317514657974243,
"learning_rate": 7.764705882352942e-05,
"loss": 0.2015,
"step": 66
},
{
"epoch": 4.0,
"grad_norm": 0.6193119883537292,
"learning_rate": 8e-05,
"loss": 0.1789,
"step": 68
},
{
"epoch": 4.0,
"eval_loss": 0.3860926926136017,
"eval_runtime": 39.0805,
"eval_samples_per_second": 0.921,
"eval_steps_per_second": 0.921,
"step": 68
},
{
"epoch": 4.117647058823529,
"grad_norm": 0.9112085700035095,
"learning_rate": 8.23529411764706e-05,
"loss": 0.1817,
"step": 70
},
{
"epoch": 4.235294117647059,
"grad_norm": 0.9801933169364929,
"learning_rate": 8.470588235294118e-05,
"loss": 0.2136,
"step": 72
},
{
"epoch": 4.352941176470588,
"grad_norm": 1.4230718612670898,
"learning_rate": 8.705882352941177e-05,
"loss": 0.2568,
"step": 74
},
{
"epoch": 4.470588235294118,
"grad_norm": 1.0974682569503784,
"learning_rate": 8.941176470588236e-05,
"loss": 0.137,
"step": 76
},
{
"epoch": 4.588235294117647,
"grad_norm": 1.4483948945999146,
"learning_rate": 9.176470588235295e-05,
"loss": 0.161,
"step": 78
},
{
"epoch": 4.705882352941177,
"grad_norm": 1.0441091060638428,
"learning_rate": 9.411764705882353e-05,
"loss": 0.1432,
"step": 80
},
{
"epoch": 4.823529411764706,
"grad_norm": 1.0316040515899658,
"learning_rate": 9.647058823529412e-05,
"loss": 0.1896,
"step": 82
},
{
"epoch": 4.9411764705882355,
"grad_norm": 0.9002617001533508,
"learning_rate": 9.882352941176471e-05,
"loss": 0.1247,
"step": 84
},
{
"epoch": 5.0,
"eval_loss": 0.4261004328727722,
"eval_runtime": 39.0803,
"eval_samples_per_second": 0.921,
"eval_steps_per_second": 0.921,
"step": 85
},
{
"epoch": 5.0588235294117645,
"grad_norm": 0.8818238377571106,
"learning_rate": 9.99995783847866e-05,
"loss": 0.1043,
"step": 86
},
{
"epoch": 5.176470588235294,
"grad_norm": 0.7031641006469727,
"learning_rate": 9.999620550574153e-05,
"loss": 0.0949,
"step": 88
},
{
"epoch": 5.294117647058823,
"grad_norm": 1.770699381828308,
"learning_rate": 9.998945997517956e-05,
"loss": 0.0677,
"step": 90
},
{
"epoch": 5.411764705882353,
"grad_norm": 0.0026916631031781435,
"learning_rate": 9.997934224814173e-05,
"loss": 0.0357,
"step": 92
},
{
"epoch": 5.529411764705882,
"grad_norm": 2.6577627658843994,
"learning_rate": 9.996585300715116e-05,
"loss": 0.0835,
"step": 94
},
{
"epoch": 5.647058823529412,
"grad_norm": 1.47685706615448,
"learning_rate": 9.994899316216708e-05,
"loss": 0.0339,
"step": 96
},
{
"epoch": 5.764705882352941,
"grad_norm": 1.6694728136062622,
"learning_rate": 9.992876385052345e-05,
"loss": 0.0843,
"step": 98
},
{
"epoch": 5.882352941176471,
"grad_norm": 1.0204684734344482,
"learning_rate": 9.990516643685222e-05,
"loss": 0.1223,
"step": 100
},
{
"epoch": 6.0,
"grad_norm": 1.1496893167495728,
"learning_rate": 9.987820251299122e-05,
"loss": 0.106,
"step": 102
},
{
"epoch": 6.0,
"eval_loss": 0.5085602402687073,
"eval_runtime": 39.0826,
"eval_samples_per_second": 0.921,
"eval_steps_per_second": 0.921,
"step": 102
},
{
"epoch": 6.117647058823529,
"grad_norm": 0.513791024684906,
"learning_rate": 9.984787389787688e-05,
"loss": 0.0246,
"step": 104
},
{
"epoch": 6.235294117647059,
"grad_norm": 0.7581632733345032,
"learning_rate": 9.981418263742148e-05,
"loss": 0.0487,
"step": 106
},
{
"epoch": 6.352941176470588,
"grad_norm": 0.5927326083183289,
"learning_rate": 9.977713100437509e-05,
"loss": 0.036,
"step": 108
},
{
"epoch": 6.470588235294118,
"grad_norm": 0.8820768594741821,
"learning_rate": 9.973672149817232e-05,
"loss": 0.0356,
"step": 110
},
{
"epoch": 6.588235294117647,
"grad_norm": 0.8725058436393738,
"learning_rate": 9.96929568447637e-05,
"loss": 0.0291,
"step": 112
},
{
"epoch": 6.705882352941177,
"grad_norm": 0.8575373291969299,
"learning_rate": 9.964583999643174e-05,
"loss": 0.017,
"step": 114
},
{
"epoch": 6.823529411764706,
"grad_norm": 0.8696101903915405,
"learning_rate": 9.95953741315919e-05,
"loss": 0.058,
"step": 116
},
{
"epoch": 6.9411764705882355,
"grad_norm": 1.933862328529358,
"learning_rate": 9.954156265457801e-05,
"loss": 0.0714,
"step": 118
},
{
"epoch": 7.0,
"eval_loss": 0.5979548096656799,
"eval_runtime": 39.0789,
"eval_samples_per_second": 0.921,
"eval_steps_per_second": 0.921,
"step": 119
},
{
"epoch": 7.0588235294117645,
"grad_norm": 0.5488072633743286,
"learning_rate": 9.948440919541278e-05,
"loss": 0.0337,
"step": 120
},
{
"epoch": 7.176470588235294,
"grad_norm": 1.0012104511260986,
"learning_rate": 9.942391760956277e-05,
"loss": 0.0199,
"step": 122
},
{
"epoch": 7.294117647058823,
"grad_norm": 0.7232321500778198,
"learning_rate": 9.936009197767845e-05,
"loss": 0.0174,
"step": 124
},
{
"epoch": 7.411764705882353,
"grad_norm": 0.43735918402671814,
"learning_rate": 9.929293660531888e-05,
"loss": 0.0152,
"step": 126
},
{
"epoch": 7.529411764705882,
"grad_norm": 0.7728654742240906,
"learning_rate": 9.922245602266118e-05,
"loss": 0.0807,
"step": 128
},
{
"epoch": 7.647058823529412,
"grad_norm": 0.1402147114276886,
"learning_rate": 9.91486549841951e-05,
"loss": 0.0056,
"step": 130
},
{
"epoch": 7.764705882352941,
"grad_norm": 0.46881377696990967,
"learning_rate": 9.90715384684021e-05,
"loss": 0.0094,
"step": 132
},
{
"epoch": 7.882352941176471,
"grad_norm": 0.6516445279121399,
"learning_rate": 9.899111167741966e-05,
"loss": 0.0139,
"step": 134
},
{
"epoch": 8.0,
"grad_norm": 0.7307709455490112,
"learning_rate": 9.890738003669029e-05,
"loss": 0.0255,
"step": 136
},
{
"epoch": 8.0,
"eval_loss": 0.6407665610313416,
"eval_runtime": 39.0838,
"eval_samples_per_second": 0.921,
"eval_steps_per_second": 0.921,
"step": 136
},
{
"epoch": 8.117647058823529,
"grad_norm": 0.4623682498931885,
"learning_rate": 9.882034919459555e-05,
"loss": 0.0077,
"step": 138
},
{
"epoch": 8.235294117647058,
"grad_norm": 0.1462814062833786,
"learning_rate": 9.873002502207503e-05,
"loss": 0.0029,
"step": 140
},
{
"epoch": 8.352941176470589,
"grad_norm": 1.0598957538604736,
"learning_rate": 9.863641361223024e-05,
"loss": 0.0128,
"step": 142
},
{
"epoch": 8.470588235294118,
"grad_norm": 0.0068275160156190395,
"learning_rate": 9.853952127991372e-05,
"loss": 0.0099,
"step": 144
},
{
"epoch": 8.588235294117647,
"grad_norm": 0.26017963886260986,
"learning_rate": 9.843935456130295e-05,
"loss": 0.0046,
"step": 146
},
{
"epoch": 8.705882352941176,
"grad_norm": 0.373006135225296,
"learning_rate": 9.833592021345937e-05,
"loss": 0.0103,
"step": 148
},
{
"epoch": 8.823529411764707,
"grad_norm": 1.1968917846679688,
"learning_rate": 9.822922521387276e-05,
"loss": 0.014,
"step": 150
},
{
"epoch": 8.941176470588236,
"grad_norm": 0.7472420334815979,
"learning_rate": 9.811927675999036e-05,
"loss": 0.0099,
"step": 152
},
{
"epoch": 9.0,
"eval_loss": 0.7027953863143921,
"eval_runtime": 39.0758,
"eval_samples_per_second": 0.921,
"eval_steps_per_second": 0.921,
"step": 153
},
{
"epoch": 9.058823529411764,
"grad_norm": 0.9742393493652344,
"learning_rate": 9.800608226873142e-05,
"loss": 0.0125,
"step": 154
},
{
"epoch": 9.176470588235293,
"grad_norm": 0.899624764919281,
"learning_rate": 9.788964937598689e-05,
"loss": 0.0191,
"step": 156
},
{
"epoch": 9.294117647058824,
"grad_norm": 0.4523799419403076,
"learning_rate": 9.776998593610428e-05,
"loss": 0.0096,
"step": 158
},
{
"epoch": 9.411764705882353,
"grad_norm": 0.35478875041007996,
"learning_rate": 9.764710002135784e-05,
"loss": 0.0055,
"step": 160
},
{
"epoch": 9.529411764705882,
"grad_norm": 0.9173728823661804,
"learning_rate": 9.752099992140399e-05,
"loss": 0.0109,
"step": 162
},
{
"epoch": 9.647058823529411,
"grad_norm": 0.4686696529388428,
"learning_rate": 9.739169414272217e-05,
"loss": 0.0082,
"step": 164
},
{
"epoch": 9.764705882352942,
"grad_norm": 0.599856972694397,
"learning_rate": 9.725919140804099e-05,
"loss": 0.0068,
"step": 166
},
{
"epoch": 9.882352941176471,
"grad_norm": 11.85073471069336,
"learning_rate": 9.71235006557497e-05,
"loss": 0.0118,
"step": 168
},
{
"epoch": 10.0,
"grad_norm": 0.3015131652355194,
"learning_rate": 9.698463103929542e-05,
"loss": 0.0098,
"step": 170
},
{
"epoch": 10.0,
"eval_loss": 0.6469479203224182,
"eval_runtime": 39.0805,
"eval_samples_per_second": 0.921,
"eval_steps_per_second": 0.921,
"step": 170
},
{
"epoch": 10.117647058823529,
"grad_norm": 0.5187628269195557,
"learning_rate": 9.684259192656553e-05,
"loss": 0.0027,
"step": 172
},
{
"epoch": 10.235294117647058,
"grad_norm": 0.1632942110300064,
"learning_rate": 9.669739289925577e-05,
"loss": 0.0039,
"step": 174
},
{
"epoch": 10.352941176470589,
"grad_norm": 0.17772699892520905,
"learning_rate": 9.654904375222385e-05,
"loss": 0.0061,
"step": 176
},
{
"epoch": 10.470588235294118,
"grad_norm": 4.154146671295166,
"learning_rate": 9.639755449282875e-05,
"loss": 0.0073,
"step": 178
},
{
"epoch": 10.588235294117647,
"grad_norm": 0.5235037207603455,
"learning_rate": 9.62429353402556e-05,
"loss": 0.0048,
"step": 180
},
{
"epoch": 10.705882352941176,
"grad_norm": 0.4133511185646057,
"learning_rate": 9.608519672482636e-05,
"loss": 0.0077,
"step": 182
},
{
"epoch": 10.823529411764707,
"grad_norm": 0.5625613331794739,
"learning_rate": 9.592434928729616e-05,
"loss": 0.0059,
"step": 184
},
{
"epoch": 10.941176470588236,
"grad_norm": 0.10571733862161636,
"learning_rate": 9.576040387813552e-05,
"loss": 0.0015,
"step": 186
},
{
"epoch": 11.0,
"eval_loss": 0.8128726482391357,
"eval_runtime": 39.0719,
"eval_samples_per_second": 0.921,
"eval_steps_per_second": 0.921,
"step": 187
},
{
"epoch": 11.0,
"step": 187,
"total_flos": 1.4581013303158374e+17,
"train_loss": 0.4026650073100753,
"train_runtime": 5409.9219,
"train_samples_per_second": 1.257,
"train_steps_per_second": 0.157
}
],
"logging_steps": 2,
"max_steps": 850,
"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": 1.4581013303158374e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}