GaetanMichelet's picture
Model save
9f76e0c verified
{
"best_metric": 0.7875569462776184,
"best_model_checkpoint": "data/Mistral-7B_task-2_180-samples_config-1_full/checkpoint-85",
"epoch": 12.0,
"eval_steps": 500,
"global_step": 204,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.058823529411764705,
"grad_norm": 1.0816072225570679,
"learning_rate": 1.1764705882352942e-06,
"loss": 1.1758,
"step": 1
},
{
"epoch": 0.11764705882352941,
"grad_norm": 1.0684632062911987,
"learning_rate": 2.3529411764705885e-06,
"loss": 1.2284,
"step": 2
},
{
"epoch": 0.23529411764705882,
"grad_norm": 1.3180915117263794,
"learning_rate": 4.705882352941177e-06,
"loss": 1.2252,
"step": 4
},
{
"epoch": 0.35294117647058826,
"grad_norm": 1.1849663257598877,
"learning_rate": 7.058823529411765e-06,
"loss": 1.2332,
"step": 6
},
{
"epoch": 0.47058823529411764,
"grad_norm": 1.0098148584365845,
"learning_rate": 9.411764705882354e-06,
"loss": 1.1832,
"step": 8
},
{
"epoch": 0.5882352941176471,
"grad_norm": 0.8811877965927124,
"learning_rate": 1.1764705882352942e-05,
"loss": 1.1311,
"step": 10
},
{
"epoch": 0.7058823529411765,
"grad_norm": 0.8064162731170654,
"learning_rate": 1.411764705882353e-05,
"loss": 1.0949,
"step": 12
},
{
"epoch": 0.8235294117647058,
"grad_norm": 0.7510309815406799,
"learning_rate": 1.647058823529412e-05,
"loss": 1.1021,
"step": 14
},
{
"epoch": 0.9411764705882353,
"grad_norm": 0.5527868866920471,
"learning_rate": 1.8823529411764708e-05,
"loss": 1.0543,
"step": 16
},
{
"epoch": 1.0,
"eval_loss": 1.0411328077316284,
"eval_runtime": 45.5779,
"eval_samples_per_second": 0.79,
"eval_steps_per_second": 0.79,
"step": 17
},
{
"epoch": 1.0588235294117647,
"grad_norm": 0.4396335780620575,
"learning_rate": 2.1176470588235296e-05,
"loss": 1.0491,
"step": 18
},
{
"epoch": 1.1764705882352942,
"grad_norm": 0.342247873544693,
"learning_rate": 2.3529411764705884e-05,
"loss": 1.0331,
"step": 20
},
{
"epoch": 1.2941176470588236,
"grad_norm": 0.328950434923172,
"learning_rate": 2.5882352941176475e-05,
"loss": 1.001,
"step": 22
},
{
"epoch": 1.4117647058823528,
"grad_norm": 0.3215286135673523,
"learning_rate": 2.823529411764706e-05,
"loss": 0.9663,
"step": 24
},
{
"epoch": 1.5294117647058822,
"grad_norm": 0.2940995395183563,
"learning_rate": 3.058823529411765e-05,
"loss": 0.9488,
"step": 26
},
{
"epoch": 1.6470588235294117,
"grad_norm": 0.3360099196434021,
"learning_rate": 3.294117647058824e-05,
"loss": 0.9415,
"step": 28
},
{
"epoch": 1.7647058823529411,
"grad_norm": 0.45030349493026733,
"learning_rate": 3.529411764705883e-05,
"loss": 0.9408,
"step": 30
},
{
"epoch": 1.8823529411764706,
"grad_norm": 0.5008979439735413,
"learning_rate": 3.7647058823529415e-05,
"loss": 0.8956,
"step": 32
},
{
"epoch": 2.0,
"grad_norm": 0.4257796108722687,
"learning_rate": 4e-05,
"loss": 0.9001,
"step": 34
},
{
"epoch": 2.0,
"eval_loss": 0.8847403526306152,
"eval_runtime": 45.585,
"eval_samples_per_second": 0.79,
"eval_steps_per_second": 0.79,
"step": 34
},
{
"epoch": 2.1176470588235294,
"grad_norm": 0.42106279730796814,
"learning_rate": 4.235294117647059e-05,
"loss": 0.8453,
"step": 36
},
{
"epoch": 2.235294117647059,
"grad_norm": 0.30742546916007996,
"learning_rate": 4.470588235294118e-05,
"loss": 0.8186,
"step": 38
},
{
"epoch": 2.3529411764705883,
"grad_norm": 0.29315757751464844,
"learning_rate": 4.705882352941177e-05,
"loss": 0.8045,
"step": 40
},
{
"epoch": 2.4705882352941178,
"grad_norm": 0.2516879141330719,
"learning_rate": 4.9411764705882355e-05,
"loss": 0.8069,
"step": 42
},
{
"epoch": 2.588235294117647,
"grad_norm": 0.27411141991615295,
"learning_rate": 5.176470588235295e-05,
"loss": 0.7798,
"step": 44
},
{
"epoch": 2.7058823529411766,
"grad_norm": 0.3091156482696533,
"learning_rate": 5.411764705882353e-05,
"loss": 0.8091,
"step": 46
},
{
"epoch": 2.8235294117647056,
"grad_norm": 0.24321137368679047,
"learning_rate": 5.647058823529412e-05,
"loss": 0.8014,
"step": 48
},
{
"epoch": 2.9411764705882355,
"grad_norm": 0.2646884024143219,
"learning_rate": 5.882352941176471e-05,
"loss": 0.791,
"step": 50
},
{
"epoch": 3.0,
"eval_loss": 0.8165853023529053,
"eval_runtime": 45.5911,
"eval_samples_per_second": 0.79,
"eval_steps_per_second": 0.79,
"step": 51
},
{
"epoch": 3.0588235294117645,
"grad_norm": 0.2296939641237259,
"learning_rate": 6.11764705882353e-05,
"loss": 0.7695,
"step": 52
},
{
"epoch": 3.176470588235294,
"grad_norm": 0.22971342504024506,
"learning_rate": 6.352941176470588e-05,
"loss": 0.752,
"step": 54
},
{
"epoch": 3.2941176470588234,
"grad_norm": 0.22274336218833923,
"learning_rate": 6.588235294117648e-05,
"loss": 0.7908,
"step": 56
},
{
"epoch": 3.411764705882353,
"grad_norm": 0.24630722403526306,
"learning_rate": 6.823529411764707e-05,
"loss": 0.771,
"step": 58
},
{
"epoch": 3.5294117647058822,
"grad_norm": 0.23917770385742188,
"learning_rate": 7.058823529411765e-05,
"loss": 0.7183,
"step": 60
},
{
"epoch": 3.6470588235294117,
"grad_norm": 0.22868651151657104,
"learning_rate": 7.294117647058823e-05,
"loss": 0.7985,
"step": 62
},
{
"epoch": 3.764705882352941,
"grad_norm": 0.3208144009113312,
"learning_rate": 7.529411764705883e-05,
"loss": 0.742,
"step": 64
},
{
"epoch": 3.8823529411764706,
"grad_norm": 0.25168377161026,
"learning_rate": 7.764705882352942e-05,
"loss": 0.72,
"step": 66
},
{
"epoch": 4.0,
"grad_norm": 0.2622774541378021,
"learning_rate": 8e-05,
"loss": 0.7142,
"step": 68
},
{
"epoch": 4.0,
"eval_loss": 0.796552300453186,
"eval_runtime": 45.5817,
"eval_samples_per_second": 0.79,
"eval_steps_per_second": 0.79,
"step": 68
},
{
"epoch": 4.117647058823529,
"grad_norm": 0.24447962641716003,
"learning_rate": 8.23529411764706e-05,
"loss": 0.7762,
"step": 70
},
{
"epoch": 4.235294117647059,
"grad_norm": 0.2675190567970276,
"learning_rate": 8.470588235294118e-05,
"loss": 0.6752,
"step": 72
},
{
"epoch": 4.352941176470588,
"grad_norm": 0.2746583819389343,
"learning_rate": 8.705882352941177e-05,
"loss": 0.7374,
"step": 74
},
{
"epoch": 4.470588235294118,
"grad_norm": 0.30379965901374817,
"learning_rate": 8.941176470588236e-05,
"loss": 0.6883,
"step": 76
},
{
"epoch": 4.588235294117647,
"grad_norm": 0.3467769920825958,
"learning_rate": 9.176470588235295e-05,
"loss": 0.7017,
"step": 78
},
{
"epoch": 4.705882352941177,
"grad_norm": 0.2998846471309662,
"learning_rate": 9.411764705882353e-05,
"loss": 0.7178,
"step": 80
},
{
"epoch": 4.823529411764706,
"grad_norm": 0.3738335967063904,
"learning_rate": 9.647058823529412e-05,
"loss": 0.6912,
"step": 82
},
{
"epoch": 4.9411764705882355,
"grad_norm": 0.34114784002304077,
"learning_rate": 9.882352941176471e-05,
"loss": 0.692,
"step": 84
},
{
"epoch": 5.0,
"eval_loss": 0.7875569462776184,
"eval_runtime": 45.5825,
"eval_samples_per_second": 0.79,
"eval_steps_per_second": 0.79,
"step": 85
},
{
"epoch": 5.0588235294117645,
"grad_norm": 0.448775053024292,
"learning_rate": 9.99995783847866e-05,
"loss": 0.6368,
"step": 86
},
{
"epoch": 5.176470588235294,
"grad_norm": 0.39181679487228394,
"learning_rate": 9.999620550574153e-05,
"loss": 0.6427,
"step": 88
},
{
"epoch": 5.294117647058823,
"grad_norm": 0.4281361997127533,
"learning_rate": 9.998945997517956e-05,
"loss": 0.6759,
"step": 90
},
{
"epoch": 5.411764705882353,
"grad_norm": 0.46260866522789,
"learning_rate": 9.997934224814173e-05,
"loss": 0.6789,
"step": 92
},
{
"epoch": 5.529411764705882,
"grad_norm": 0.42020484805107117,
"learning_rate": 9.996585300715116e-05,
"loss": 0.6379,
"step": 94
},
{
"epoch": 5.647058823529412,
"grad_norm": 0.442160040140152,
"learning_rate": 9.994899316216708e-05,
"loss": 0.6411,
"step": 96
},
{
"epoch": 5.764705882352941,
"grad_norm": 0.46501389145851135,
"learning_rate": 9.992876385052345e-05,
"loss": 0.6596,
"step": 98
},
{
"epoch": 5.882352941176471,
"grad_norm": 0.41799578070640564,
"learning_rate": 9.990516643685222e-05,
"loss": 0.6616,
"step": 100
},
{
"epoch": 6.0,
"grad_norm": 0.4698163866996765,
"learning_rate": 9.987820251299122e-05,
"loss": 0.6117,
"step": 102
},
{
"epoch": 6.0,
"eval_loss": 0.8012238144874573,
"eval_runtime": 45.5829,
"eval_samples_per_second": 0.79,
"eval_steps_per_second": 0.79,
"step": 102
},
{
"epoch": 6.117647058823529,
"grad_norm": 0.45167994499206543,
"learning_rate": 9.984787389787688e-05,
"loss": 0.5643,
"step": 104
},
{
"epoch": 6.235294117647059,
"grad_norm": 0.6128388047218323,
"learning_rate": 9.981418263742148e-05,
"loss": 0.568,
"step": 106
},
{
"epoch": 6.352941176470588,
"grad_norm": 0.5976178050041199,
"learning_rate": 9.977713100437509e-05,
"loss": 0.5279,
"step": 108
},
{
"epoch": 6.470588235294118,
"grad_norm": 0.6520164012908936,
"learning_rate": 9.973672149817232e-05,
"loss": 0.5771,
"step": 110
},
{
"epoch": 6.588235294117647,
"grad_norm": 0.6118868589401245,
"learning_rate": 9.96929568447637e-05,
"loss": 0.5479,
"step": 112
},
{
"epoch": 6.705882352941177,
"grad_norm": 0.5594154596328735,
"learning_rate": 9.964583999643174e-05,
"loss": 0.6118,
"step": 114
},
{
"epoch": 6.823529411764706,
"grad_norm": 0.5402033925056458,
"learning_rate": 9.95953741315919e-05,
"loss": 0.616,
"step": 116
},
{
"epoch": 6.9411764705882355,
"grad_norm": 0.60446697473526,
"learning_rate": 9.954156265457801e-05,
"loss": 0.5747,
"step": 118
},
{
"epoch": 7.0,
"eval_loss": 0.8235461115837097,
"eval_runtime": 45.5785,
"eval_samples_per_second": 0.79,
"eval_steps_per_second": 0.79,
"step": 119
},
{
"epoch": 7.0588235294117645,
"grad_norm": 0.676578164100647,
"learning_rate": 9.948440919541278e-05,
"loss": 0.5514,
"step": 120
},
{
"epoch": 7.176470588235294,
"grad_norm": 1.1455576419830322,
"learning_rate": 9.942391760956277e-05,
"loss": 0.4932,
"step": 122
},
{
"epoch": 7.294117647058823,
"grad_norm": 0.608930766582489,
"learning_rate": 9.936009197767845e-05,
"loss": 0.4935,
"step": 124
},
{
"epoch": 7.411764705882353,
"grad_norm": 0.6835198402404785,
"learning_rate": 9.929293660531888e-05,
"loss": 0.4686,
"step": 126
},
{
"epoch": 7.529411764705882,
"grad_norm": 0.8285886645317078,
"learning_rate": 9.922245602266118e-05,
"loss": 0.4765,
"step": 128
},
{
"epoch": 7.647058823529412,
"grad_norm": 0.7334555983543396,
"learning_rate": 9.91486549841951e-05,
"loss": 0.5045,
"step": 130
},
{
"epoch": 7.764705882352941,
"grad_norm": 0.7504845261573792,
"learning_rate": 9.90715384684021e-05,
"loss": 0.448,
"step": 132
},
{
"epoch": 7.882352941176471,
"grad_norm": 0.7206101417541504,
"learning_rate": 9.899111167741966e-05,
"loss": 0.5379,
"step": 134
},
{
"epoch": 8.0,
"grad_norm": 0.7412144541740417,
"learning_rate": 9.890738003669029e-05,
"loss": 0.4667,
"step": 136
},
{
"epoch": 8.0,
"eval_loss": 0.8642133474349976,
"eval_runtime": 45.5763,
"eval_samples_per_second": 0.79,
"eval_steps_per_second": 0.79,
"step": 136
},
{
"epoch": 8.117647058823529,
"grad_norm": 0.7694720029830933,
"learning_rate": 9.882034919459555e-05,
"loss": 0.4025,
"step": 138
},
{
"epoch": 8.235294117647058,
"grad_norm": 1.309736728668213,
"learning_rate": 9.873002502207503e-05,
"loss": 0.3983,
"step": 140
},
{
"epoch": 8.352941176470589,
"grad_norm": 0.7858896255493164,
"learning_rate": 9.863641361223024e-05,
"loss": 0.4195,
"step": 142
},
{
"epoch": 8.470588235294118,
"grad_norm": 0.7856791019439697,
"learning_rate": 9.853952127991372e-05,
"loss": 0.4257,
"step": 144
},
{
"epoch": 8.588235294117647,
"grad_norm": 0.8528513312339783,
"learning_rate": 9.843935456130295e-05,
"loss": 0.4466,
"step": 146
},
{
"epoch": 8.705882352941176,
"grad_norm": 0.9507191777229309,
"learning_rate": 9.833592021345937e-05,
"loss": 0.4243,
"step": 148
},
{
"epoch": 8.823529411764707,
"grad_norm": 0.8049846291542053,
"learning_rate": 9.822922521387276e-05,
"loss": 0.3831,
"step": 150
},
{
"epoch": 8.941176470588236,
"grad_norm": 0.7895063757896423,
"learning_rate": 9.811927675999036e-05,
"loss": 0.3799,
"step": 152
},
{
"epoch": 9.0,
"eval_loss": 0.9077615737915039,
"eval_runtime": 45.5893,
"eval_samples_per_second": 0.79,
"eval_steps_per_second": 0.79,
"step": 153
},
{
"epoch": 9.058823529411764,
"grad_norm": 0.8904362320899963,
"learning_rate": 9.800608226873142e-05,
"loss": 0.3983,
"step": 154
},
{
"epoch": 9.176470588235293,
"grad_norm": 1.0312925577163696,
"learning_rate": 9.788964937598689e-05,
"loss": 0.3326,
"step": 156
},
{
"epoch": 9.294117647058824,
"grad_norm": 0.899383544921875,
"learning_rate": 9.776998593610428e-05,
"loss": 0.3302,
"step": 158
},
{
"epoch": 9.411764705882353,
"grad_norm": 0.8629933595657349,
"learning_rate": 9.764710002135784e-05,
"loss": 0.3441,
"step": 160
},
{
"epoch": 9.529411764705882,
"grad_norm": 0.8594860434532166,
"learning_rate": 9.752099992140399e-05,
"loss": 0.3291,
"step": 162
},
{
"epoch": 9.647058823529411,
"grad_norm": 0.8668737411499023,
"learning_rate": 9.739169414272217e-05,
"loss": 0.3489,
"step": 164
},
{
"epoch": 9.764705882352942,
"grad_norm": 0.9916549324989319,
"learning_rate": 9.725919140804099e-05,
"loss": 0.3535,
"step": 166
},
{
"epoch": 9.882352941176471,
"grad_norm": 0.8480520844459534,
"learning_rate": 9.71235006557497e-05,
"loss": 0.2896,
"step": 168
},
{
"epoch": 10.0,
"grad_norm": 0.8887491822242737,
"learning_rate": 9.698463103929542e-05,
"loss": 0.3324,
"step": 170
},
{
"epoch": 10.0,
"eval_loss": 0.9619680047035217,
"eval_runtime": 45.5842,
"eval_samples_per_second": 0.79,
"eval_steps_per_second": 0.79,
"step": 170
},
{
"epoch": 10.117647058823529,
"grad_norm": 0.8702213168144226,
"learning_rate": 9.684259192656553e-05,
"loss": 0.244,
"step": 172
},
{
"epoch": 10.235294117647058,
"grad_norm": 1.2155134677886963,
"learning_rate": 9.669739289925577e-05,
"loss": 0.2679,
"step": 174
},
{
"epoch": 10.352941176470589,
"grad_norm": 0.8592737913131714,
"learning_rate": 9.654904375222385e-05,
"loss": 0.2821,
"step": 176
},
{
"epoch": 10.470588235294118,
"grad_norm": 0.8212958574295044,
"learning_rate": 9.639755449282875e-05,
"loss": 0.2441,
"step": 178
},
{
"epoch": 10.588235294117647,
"grad_norm": 0.8830901980400085,
"learning_rate": 9.62429353402556e-05,
"loss": 0.2793,
"step": 180
},
{
"epoch": 10.705882352941176,
"grad_norm": 0.8993498682975769,
"learning_rate": 9.608519672482636e-05,
"loss": 0.261,
"step": 182
},
{
"epoch": 10.823529411764707,
"grad_norm": 0.9086623191833496,
"learning_rate": 9.592434928729616e-05,
"loss": 0.2835,
"step": 184
},
{
"epoch": 10.941176470588236,
"grad_norm": 0.8232172727584839,
"learning_rate": 9.576040387813552e-05,
"loss": 0.2891,
"step": 186
},
{
"epoch": 11.0,
"eval_loss": 1.0193877220153809,
"eval_runtime": 45.5838,
"eval_samples_per_second": 0.79,
"eval_steps_per_second": 0.79,
"step": 187
},
{
"epoch": 11.058823529411764,
"grad_norm": 0.7882196307182312,
"learning_rate": 9.559337155679842e-05,
"loss": 0.2641,
"step": 188
},
{
"epoch": 11.176470588235293,
"grad_norm": 1.0639575719833374,
"learning_rate": 9.542326359097619e-05,
"loss": 0.217,
"step": 190
},
{
"epoch": 11.294117647058824,
"grad_norm": 0.9592022895812988,
"learning_rate": 9.525009145583745e-05,
"loss": 0.2453,
"step": 192
},
{
"epoch": 11.411764705882353,
"grad_norm": 0.8479567170143127,
"learning_rate": 9.507386683325404e-05,
"loss": 0.1575,
"step": 194
},
{
"epoch": 11.529411764705882,
"grad_norm": 0.9146374464035034,
"learning_rate": 9.489460161101291e-05,
"loss": 0.2364,
"step": 196
},
{
"epoch": 11.647058823529411,
"grad_norm": 1.0637890100479126,
"learning_rate": 9.471230788201429e-05,
"loss": 0.2216,
"step": 198
},
{
"epoch": 11.764705882352942,
"grad_norm": 1.0326683521270752,
"learning_rate": 9.452699794345581e-05,
"loss": 0.2194,
"step": 200
},
{
"epoch": 11.882352941176471,
"grad_norm": 0.9039857983589172,
"learning_rate": 9.43386842960031e-05,
"loss": 0.2148,
"step": 202
},
{
"epoch": 12.0,
"grad_norm": 0.9226338267326355,
"learning_rate": 9.414737964294636e-05,
"loss": 0.19,
"step": 204
},
{
"epoch": 12.0,
"eval_loss": 1.074280858039856,
"eval_runtime": 45.5989,
"eval_samples_per_second": 0.789,
"eval_steps_per_second": 0.789,
"step": 204
},
{
"epoch": 12.0,
"step": 204,
"total_flos": 1.9338296791190733e+17,
"train_loss": 0.6092646595309762,
"train_runtime": 7014.9082,
"train_samples_per_second": 0.969,
"train_steps_per_second": 0.121
}
],
"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": 1.9338296791190733e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}