GaetanMichelet's picture
Model save
59c6752 verified
{
"best_metric": 1.2263596057891846,
"best_model_checkpoint": "data/Llama-31-8B_task-1_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": 1.806323766708374,
"learning_rate": 1.1764705882352942e-06,
"loss": 2.1281,
"step": 1
},
{
"epoch": 0.11764705882352941,
"grad_norm": 1.5079092979431152,
"learning_rate": 2.3529411764705885e-06,
"loss": 1.9223,
"step": 2
},
{
"epoch": 0.23529411764705882,
"grad_norm": 1.8070296049118042,
"learning_rate": 4.705882352941177e-06,
"loss": 2.0084,
"step": 4
},
{
"epoch": 0.35294117647058826,
"grad_norm": 1.6644213199615479,
"learning_rate": 7.058823529411765e-06,
"loss": 2.1225,
"step": 6
},
{
"epoch": 0.47058823529411764,
"grad_norm": 1.774314284324646,
"learning_rate": 9.411764705882354e-06,
"loss": 1.9249,
"step": 8
},
{
"epoch": 0.5882352941176471,
"grad_norm": 1.6763962507247925,
"learning_rate": 1.1764705882352942e-05,
"loss": 2.06,
"step": 10
},
{
"epoch": 0.7058823529411765,
"grad_norm": 1.8542298078536987,
"learning_rate": 1.411764705882353e-05,
"loss": 1.981,
"step": 12
},
{
"epoch": 0.8235294117647058,
"grad_norm": 1.5285791158676147,
"learning_rate": 1.647058823529412e-05,
"loss": 1.9257,
"step": 14
},
{
"epoch": 0.9411764705882353,
"grad_norm": 1.7371580600738525,
"learning_rate": 1.8823529411764708e-05,
"loss": 1.8393,
"step": 16
},
{
"epoch": 1.0,
"eval_loss": 1.7238094806671143,
"eval_runtime": 14.4902,
"eval_samples_per_second": 2.484,
"eval_steps_per_second": 2.484,
"step": 17
},
{
"epoch": 1.0588235294117647,
"grad_norm": 1.277214765548706,
"learning_rate": 2.1176470588235296e-05,
"loss": 1.713,
"step": 18
},
{
"epoch": 1.1764705882352942,
"grad_norm": 0.8376792073249817,
"learning_rate": 2.3529411764705884e-05,
"loss": 1.7011,
"step": 20
},
{
"epoch": 1.2941176470588236,
"grad_norm": 0.6922251582145691,
"learning_rate": 2.5882352941176475e-05,
"loss": 1.5788,
"step": 22
},
{
"epoch": 1.4117647058823528,
"grad_norm": 0.7128008008003235,
"learning_rate": 2.823529411764706e-05,
"loss": 1.6274,
"step": 24
},
{
"epoch": 1.5294117647058822,
"grad_norm": 0.8218591809272766,
"learning_rate": 3.058823529411765e-05,
"loss": 1.5799,
"step": 26
},
{
"epoch": 1.6470588235294117,
"grad_norm": 0.7004019021987915,
"learning_rate": 3.294117647058824e-05,
"loss": 1.4887,
"step": 28
},
{
"epoch": 1.7647058823529411,
"grad_norm": 0.6600505113601685,
"learning_rate": 3.529411764705883e-05,
"loss": 1.4354,
"step": 30
},
{
"epoch": 1.8823529411764706,
"grad_norm": 0.6615216135978699,
"learning_rate": 3.7647058823529415e-05,
"loss": 1.4252,
"step": 32
},
{
"epoch": 2.0,
"grad_norm": 0.7450706958770752,
"learning_rate": 4e-05,
"loss": 1.3906,
"step": 34
},
{
"epoch": 2.0,
"eval_loss": 1.4334121942520142,
"eval_runtime": 14.4955,
"eval_samples_per_second": 2.484,
"eval_steps_per_second": 2.484,
"step": 34
},
{
"epoch": 2.1176470588235294,
"grad_norm": 0.6278265714645386,
"learning_rate": 4.235294117647059e-05,
"loss": 1.305,
"step": 36
},
{
"epoch": 2.235294117647059,
"grad_norm": 0.7144865989685059,
"learning_rate": 4.470588235294118e-05,
"loss": 1.4263,
"step": 38
},
{
"epoch": 2.3529411764705883,
"grad_norm": 0.7090567350387573,
"learning_rate": 4.705882352941177e-05,
"loss": 1.2102,
"step": 40
},
{
"epoch": 2.4705882352941178,
"grad_norm": 0.7017014622688293,
"learning_rate": 4.9411764705882355e-05,
"loss": 1.3959,
"step": 42
},
{
"epoch": 2.588235294117647,
"grad_norm": 0.7466076016426086,
"learning_rate": 5.176470588235295e-05,
"loss": 1.2635,
"step": 44
},
{
"epoch": 2.7058823529411766,
"grad_norm": 0.750141978263855,
"learning_rate": 5.411764705882353e-05,
"loss": 1.2825,
"step": 46
},
{
"epoch": 2.8235294117647056,
"grad_norm": 0.7849295139312744,
"learning_rate": 5.647058823529412e-05,
"loss": 1.2437,
"step": 48
},
{
"epoch": 2.9411764705882355,
"grad_norm": 0.814879298210144,
"learning_rate": 5.882352941176471e-05,
"loss": 1.1829,
"step": 50
},
{
"epoch": 3.0,
"eval_loss": 1.272010326385498,
"eval_runtime": 14.4968,
"eval_samples_per_second": 2.483,
"eval_steps_per_second": 2.483,
"step": 51
},
{
"epoch": 3.0588235294117645,
"grad_norm": 0.9217017889022827,
"learning_rate": 6.11764705882353e-05,
"loss": 1.1552,
"step": 52
},
{
"epoch": 3.176470588235294,
"grad_norm": 0.8735676407814026,
"learning_rate": 6.352941176470588e-05,
"loss": 1.1407,
"step": 54
},
{
"epoch": 3.2941176470588234,
"grad_norm": 0.8801168203353882,
"learning_rate": 6.588235294117648e-05,
"loss": 1.1258,
"step": 56
},
{
"epoch": 3.411764705882353,
"grad_norm": 0.9490126967430115,
"learning_rate": 6.823529411764707e-05,
"loss": 1.0762,
"step": 58
},
{
"epoch": 3.5294117647058822,
"grad_norm": 1.0785832405090332,
"learning_rate": 7.058823529411765e-05,
"loss": 1.1216,
"step": 60
},
{
"epoch": 3.6470588235294117,
"grad_norm": 1.0801024436950684,
"learning_rate": 7.294117647058823e-05,
"loss": 1.0717,
"step": 62
},
{
"epoch": 3.764705882352941,
"grad_norm": 0.9483073353767395,
"learning_rate": 7.529411764705883e-05,
"loss": 0.996,
"step": 64
},
{
"epoch": 3.8823529411764706,
"grad_norm": 1.1348040103912354,
"learning_rate": 7.764705882352942e-05,
"loss": 1.0813,
"step": 66
},
{
"epoch": 4.0,
"grad_norm": 1.2702502012252808,
"learning_rate": 8e-05,
"loss": 1.0037,
"step": 68
},
{
"epoch": 4.0,
"eval_loss": 1.2263596057891846,
"eval_runtime": 14.4887,
"eval_samples_per_second": 2.485,
"eval_steps_per_second": 2.485,
"step": 68
},
{
"epoch": 4.117647058823529,
"grad_norm": 1.0951895713806152,
"learning_rate": 8.23529411764706e-05,
"loss": 0.9193,
"step": 70
},
{
"epoch": 4.235294117647059,
"grad_norm": 1.1319009065628052,
"learning_rate": 8.470588235294118e-05,
"loss": 0.9073,
"step": 72
},
{
"epoch": 4.352941176470588,
"grad_norm": 1.3507709503173828,
"learning_rate": 8.705882352941177e-05,
"loss": 0.9094,
"step": 74
},
{
"epoch": 4.470588235294118,
"grad_norm": 1.5976216793060303,
"learning_rate": 8.941176470588236e-05,
"loss": 0.8,
"step": 76
},
{
"epoch": 4.588235294117647,
"grad_norm": 1.5247750282287598,
"learning_rate": 9.176470588235295e-05,
"loss": 0.8411,
"step": 78
},
{
"epoch": 4.705882352941177,
"grad_norm": 1.7212128639221191,
"learning_rate": 9.411764705882353e-05,
"loss": 0.7832,
"step": 80
},
{
"epoch": 4.823529411764706,
"grad_norm": 1.636215090751648,
"learning_rate": 9.647058823529412e-05,
"loss": 0.7942,
"step": 82
},
{
"epoch": 4.9411764705882355,
"grad_norm": 1.6557462215423584,
"learning_rate": 9.882352941176471e-05,
"loss": 0.875,
"step": 84
},
{
"epoch": 5.0,
"eval_loss": 1.2729443311691284,
"eval_runtime": 14.4917,
"eval_samples_per_second": 2.484,
"eval_steps_per_second": 2.484,
"step": 85
},
{
"epoch": 5.0588235294117645,
"grad_norm": 1.4015008211135864,
"learning_rate": 9.99995783847866e-05,
"loss": 0.7261,
"step": 86
},
{
"epoch": 5.176470588235294,
"grad_norm": 1.3426791429519653,
"learning_rate": 9.999620550574153e-05,
"loss": 0.595,
"step": 88
},
{
"epoch": 5.294117647058823,
"grad_norm": 1.7509989738464355,
"learning_rate": 9.998945997517956e-05,
"loss": 0.5419,
"step": 90
},
{
"epoch": 5.411764705882353,
"grad_norm": 3.1275417804718018,
"learning_rate": 9.997934224814173e-05,
"loss": 0.6423,
"step": 92
},
{
"epoch": 5.529411764705882,
"grad_norm": 2.2710916996002197,
"learning_rate": 9.996585300715116e-05,
"loss": 0.5543,
"step": 94
},
{
"epoch": 5.647058823529412,
"grad_norm": 1.9362425804138184,
"learning_rate": 9.994899316216708e-05,
"loss": 0.4271,
"step": 96
},
{
"epoch": 5.764705882352941,
"grad_norm": 1.7837473154067993,
"learning_rate": 9.992876385052345e-05,
"loss": 0.5637,
"step": 98
},
{
"epoch": 5.882352941176471,
"grad_norm": 1.80534827709198,
"learning_rate": 9.990516643685222e-05,
"loss": 0.6042,
"step": 100
},
{
"epoch": 6.0,
"grad_norm": 1.8158038854599,
"learning_rate": 9.987820251299122e-05,
"loss": 0.5286,
"step": 102
},
{
"epoch": 6.0,
"eval_loss": 1.4640731811523438,
"eval_runtime": 14.4917,
"eval_samples_per_second": 2.484,
"eval_steps_per_second": 2.484,
"step": 102
},
{
"epoch": 6.117647058823529,
"grad_norm": 1.8034647703170776,
"learning_rate": 9.984787389787688e-05,
"loss": 0.3217,
"step": 104
},
{
"epoch": 6.235294117647059,
"grad_norm": 1.959482192993164,
"learning_rate": 9.981418263742148e-05,
"loss": 0.3218,
"step": 106
},
{
"epoch": 6.352941176470588,
"grad_norm": 2.72263240814209,
"learning_rate": 9.977713100437509e-05,
"loss": 0.3157,
"step": 108
},
{
"epoch": 6.470588235294118,
"grad_norm": 3.339064359664917,
"learning_rate": 9.973672149817232e-05,
"loss": 0.3228,
"step": 110
},
{
"epoch": 6.588235294117647,
"grad_norm": 3.0396878719329834,
"learning_rate": 9.96929568447637e-05,
"loss": 0.2937,
"step": 112
},
{
"epoch": 6.705882352941177,
"grad_norm": 3.0496692657470703,
"learning_rate": 9.964583999643174e-05,
"loss": 0.3314,
"step": 114
},
{
"epoch": 6.823529411764706,
"grad_norm": 1.9532092809677124,
"learning_rate": 9.95953741315919e-05,
"loss": 0.2716,
"step": 116
},
{
"epoch": 6.9411764705882355,
"grad_norm": 1.7998892068862915,
"learning_rate": 9.954156265457801e-05,
"loss": 0.2721,
"step": 118
},
{
"epoch": 7.0,
"eval_loss": 1.7060846090316772,
"eval_runtime": 14.4931,
"eval_samples_per_second": 2.484,
"eval_steps_per_second": 2.484,
"step": 119
},
{
"epoch": 7.0588235294117645,
"grad_norm": 1.2864255905151367,
"learning_rate": 9.948440919541278e-05,
"loss": 0.2517,
"step": 120
},
{
"epoch": 7.176470588235294,
"grad_norm": 1.3337100744247437,
"learning_rate": 9.942391760956277e-05,
"loss": 0.1339,
"step": 122
},
{
"epoch": 7.294117647058823,
"grad_norm": 2.677187442779541,
"learning_rate": 9.936009197767845e-05,
"loss": 0.1415,
"step": 124
},
{
"epoch": 7.411764705882353,
"grad_norm": 2.17299485206604,
"learning_rate": 9.929293660531888e-05,
"loss": 0.1373,
"step": 126
},
{
"epoch": 7.529411764705882,
"grad_norm": 3.5403952598571777,
"learning_rate": 9.922245602266118e-05,
"loss": 0.1608,
"step": 128
},
{
"epoch": 7.647058823529412,
"grad_norm": 2.2326877117156982,
"learning_rate": 9.91486549841951e-05,
"loss": 0.1371,
"step": 130
},
{
"epoch": 7.764705882352941,
"grad_norm": 2.6239540576934814,
"learning_rate": 9.90715384684021e-05,
"loss": 0.1604,
"step": 132
},
{
"epoch": 7.882352941176471,
"grad_norm": 2.034834623336792,
"learning_rate": 9.899111167741966e-05,
"loss": 0.1671,
"step": 134
},
{
"epoch": 8.0,
"grad_norm": 2.1497554779052734,
"learning_rate": 9.890738003669029e-05,
"loss": 0.16,
"step": 136
},
{
"epoch": 8.0,
"eval_loss": 2.0944266319274902,
"eval_runtime": 14.498,
"eval_samples_per_second": 2.483,
"eval_steps_per_second": 2.483,
"step": 136
},
{
"epoch": 8.117647058823529,
"grad_norm": 1.502654790878296,
"learning_rate": 9.882034919459555e-05,
"loss": 0.081,
"step": 138
},
{
"epoch": 8.235294117647058,
"grad_norm": 1.151986837387085,
"learning_rate": 9.873002502207503e-05,
"loss": 0.0554,
"step": 140
},
{
"epoch": 8.352941176470589,
"grad_norm": 3.554004669189453,
"learning_rate": 9.863641361223024e-05,
"loss": 0.0771,
"step": 142
},
{
"epoch": 8.470588235294118,
"grad_norm": 2.4202420711517334,
"learning_rate": 9.853952127991372e-05,
"loss": 0.0956,
"step": 144
},
{
"epoch": 8.588235294117647,
"grad_norm": 3.394904613494873,
"learning_rate": 9.843935456130295e-05,
"loss": 0.0781,
"step": 146
},
{
"epoch": 8.705882352941176,
"grad_norm": 1.7927554845809937,
"learning_rate": 9.833592021345937e-05,
"loss": 0.0944,
"step": 148
},
{
"epoch": 8.823529411764707,
"grad_norm": 1.5801514387130737,
"learning_rate": 9.822922521387276e-05,
"loss": 0.0829,
"step": 150
},
{
"epoch": 8.941176470588236,
"grad_norm": 2.423281669616699,
"learning_rate": 9.811927675999036e-05,
"loss": 0.0814,
"step": 152
},
{
"epoch": 9.0,
"eval_loss": 2.3201403617858887,
"eval_runtime": 14.4986,
"eval_samples_per_second": 2.483,
"eval_steps_per_second": 2.483,
"step": 153
},
{
"epoch": 9.058823529411764,
"grad_norm": 0.9261131882667542,
"learning_rate": 9.800608226873142e-05,
"loss": 0.0459,
"step": 154
},
{
"epoch": 9.176470588235293,
"grad_norm": 0.9879596829414368,
"learning_rate": 9.788964937598689e-05,
"loss": 0.0279,
"step": 156
},
{
"epoch": 9.294117647058824,
"grad_norm": 1.363547444343567,
"learning_rate": 9.776998593610428e-05,
"loss": 0.0334,
"step": 158
},
{
"epoch": 9.411764705882353,
"grad_norm": 1.988921046257019,
"learning_rate": 9.764710002135784e-05,
"loss": 0.0548,
"step": 160
},
{
"epoch": 9.529411764705882,
"grad_norm": 1.7726337909698486,
"learning_rate": 9.752099992140399e-05,
"loss": 0.047,
"step": 162
},
{
"epoch": 9.647058823529411,
"grad_norm": 1.5547761917114258,
"learning_rate": 9.739169414272217e-05,
"loss": 0.0633,
"step": 164
},
{
"epoch": 9.764705882352942,
"grad_norm": 2.4369709491729736,
"learning_rate": 9.725919140804099e-05,
"loss": 0.0684,
"step": 166
},
{
"epoch": 9.882352941176471,
"grad_norm": 2.0575246810913086,
"learning_rate": 9.71235006557497e-05,
"loss": 0.0708,
"step": 168
},
{
"epoch": 10.0,
"grad_norm": 1.5280945301055908,
"learning_rate": 9.698463103929542e-05,
"loss": 0.0608,
"step": 170
},
{
"epoch": 10.0,
"eval_loss": 2.290346145629883,
"eval_runtime": 14.4996,
"eval_samples_per_second": 2.483,
"eval_steps_per_second": 2.483,
"step": 170
},
{
"epoch": 10.117647058823529,
"grad_norm": 1.086666464805603,
"learning_rate": 9.684259192656553e-05,
"loss": 0.0315,
"step": 172
},
{
"epoch": 10.235294117647058,
"grad_norm": 1.520897626876831,
"learning_rate": 9.669739289925577e-05,
"loss": 0.0324,
"step": 174
},
{
"epoch": 10.352941176470589,
"grad_norm": 2.3758738040924072,
"learning_rate": 9.654904375222385e-05,
"loss": 0.0401,
"step": 176
},
{
"epoch": 10.470588235294118,
"grad_norm": 2.0643303394317627,
"learning_rate": 9.639755449282875e-05,
"loss": 0.0572,
"step": 178
},
{
"epoch": 10.588235294117647,
"grad_norm": 1.8688856363296509,
"learning_rate": 9.62429353402556e-05,
"loss": 0.0354,
"step": 180
},
{
"epoch": 10.705882352941176,
"grad_norm": 1.3134160041809082,
"learning_rate": 9.608519672482636e-05,
"loss": 0.0585,
"step": 182
},
{
"epoch": 10.823529411764707,
"grad_norm": 1.692589521408081,
"learning_rate": 9.592434928729616e-05,
"loss": 0.0269,
"step": 184
},
{
"epoch": 10.941176470588236,
"grad_norm": 1.541475534439087,
"learning_rate": 9.576040387813552e-05,
"loss": 0.033,
"step": 186
},
{
"epoch": 11.0,
"eval_loss": 2.681903839111328,
"eval_runtime": 14.4882,
"eval_samples_per_second": 2.485,
"eval_steps_per_second": 2.485,
"step": 187
},
{
"epoch": 11.0,
"step": 187,
"total_flos": 2.381284302336819e+16,
"train_loss": 0.7195047194506714,
"train_runtime": 1765.5808,
"train_samples_per_second": 3.851,
"train_steps_per_second": 0.481
}
],
"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": 2.381284302336819e+16,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}