GaetanMichelet's picture
Model save
cf36f0b verified
{
"best_metric": 0.9374688267707825,
"best_model_checkpoint": "data/Gemma-2-2B_task-1_180-samples_config-2_full_auto/checkpoint-59",
"epoch": 14.0,
"eval_steps": 500,
"global_step": 119,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.11764705882352941,
"grad_norm": 1.1734390258789062,
"learning_rate": 2.5e-06,
"loss": 2.3064,
"step": 1
},
{
"epoch": 0.23529411764705882,
"grad_norm": 1.2410606145858765,
"learning_rate": 5e-06,
"loss": 2.2132,
"step": 2
},
{
"epoch": 0.47058823529411764,
"grad_norm": 1.2488532066345215,
"learning_rate": 1e-05,
"loss": 2.2463,
"step": 4
},
{
"epoch": 0.7058823529411765,
"grad_norm": 1.297817349433899,
"learning_rate": 1.5e-05,
"loss": 2.2842,
"step": 6
},
{
"epoch": 0.9411764705882353,
"grad_norm": 1.255168080329895,
"learning_rate": 2e-05,
"loss": 2.2783,
"step": 8
},
{
"epoch": 0.9411764705882353,
"eval_loss": 2.2319889068603516,
"eval_runtime": 3.2225,
"eval_samples_per_second": 11.171,
"eval_steps_per_second": 11.171,
"step": 8
},
{
"epoch": 1.1764705882352942,
"grad_norm": 1.1765460968017578,
"learning_rate": 2.5e-05,
"loss": 2.1902,
"step": 10
},
{
"epoch": 1.4117647058823528,
"grad_norm": 1.1392910480499268,
"learning_rate": 3e-05,
"loss": 2.1265,
"step": 12
},
{
"epoch": 1.6470588235294117,
"grad_norm": 1.1089133024215698,
"learning_rate": 3.5e-05,
"loss": 2.0401,
"step": 14
},
{
"epoch": 1.8823529411764706,
"grad_norm": 1.0177971124649048,
"learning_rate": 4e-05,
"loss": 1.8925,
"step": 16
},
{
"epoch": 2.0,
"eval_loss": 1.7373692989349365,
"eval_runtime": 3.2031,
"eval_samples_per_second": 11.239,
"eval_steps_per_second": 11.239,
"step": 17
},
{
"epoch": 2.1176470588235294,
"grad_norm": 0.847246527671814,
"learning_rate": 4.5e-05,
"loss": 1.777,
"step": 18
},
{
"epoch": 2.3529411764705883,
"grad_norm": 0.7961128354072571,
"learning_rate": 5e-05,
"loss": 1.6087,
"step": 20
},
{
"epoch": 2.588235294117647,
"grad_norm": 0.7852380275726318,
"learning_rate": 5.500000000000001e-05,
"loss": 1.5294,
"step": 22
},
{
"epoch": 2.8235294117647056,
"grad_norm": 0.8271363973617554,
"learning_rate": 6e-05,
"loss": 1.479,
"step": 24
},
{
"epoch": 2.9411764705882355,
"eval_loss": 1.3322114944458008,
"eval_runtime": 3.2039,
"eval_samples_per_second": 11.236,
"eval_steps_per_second": 11.236,
"step": 25
},
{
"epoch": 3.0588235294117645,
"grad_norm": 0.8656835556030273,
"learning_rate": 6.500000000000001e-05,
"loss": 1.3077,
"step": 26
},
{
"epoch": 3.2941176470588234,
"grad_norm": 0.7790645360946655,
"learning_rate": 7e-05,
"loss": 1.2478,
"step": 28
},
{
"epoch": 3.5294117647058822,
"grad_norm": 0.8071153163909912,
"learning_rate": 7.500000000000001e-05,
"loss": 1.1602,
"step": 30
},
{
"epoch": 3.764705882352941,
"grad_norm": 0.5582975745201111,
"learning_rate": 8e-05,
"loss": 1.107,
"step": 32
},
{
"epoch": 4.0,
"grad_norm": 0.5295036435127258,
"learning_rate": 8.5e-05,
"loss": 1.0469,
"step": 34
},
{
"epoch": 4.0,
"eval_loss": 1.0631213188171387,
"eval_runtime": 3.2075,
"eval_samples_per_second": 11.224,
"eval_steps_per_second": 11.224,
"step": 34
},
{
"epoch": 4.235294117647059,
"grad_norm": 0.46653056144714355,
"learning_rate": 9e-05,
"loss": 1.0102,
"step": 36
},
{
"epoch": 4.470588235294118,
"grad_norm": 0.4525793790817261,
"learning_rate": 9.5e-05,
"loss": 0.9945,
"step": 38
},
{
"epoch": 4.705882352941177,
"grad_norm": 0.42871928215026855,
"learning_rate": 0.0001,
"loss": 0.9855,
"step": 40
},
{
"epoch": 4.9411764705882355,
"grad_norm": 0.43947210907936096,
"learning_rate": 9.999238475781957e-05,
"loss": 0.962,
"step": 42
},
{
"epoch": 4.9411764705882355,
"eval_loss": 0.993911623954773,
"eval_runtime": 3.1422,
"eval_samples_per_second": 11.457,
"eval_steps_per_second": 11.457,
"step": 42
},
{
"epoch": 5.176470588235294,
"grad_norm": 0.41288837790489197,
"learning_rate": 9.99695413509548e-05,
"loss": 0.9483,
"step": 44
},
{
"epoch": 5.411764705882353,
"grad_norm": 0.40874138474464417,
"learning_rate": 9.99314767377287e-05,
"loss": 0.8921,
"step": 46
},
{
"epoch": 5.647058823529412,
"grad_norm": 0.4313524663448334,
"learning_rate": 9.987820251299122e-05,
"loss": 0.8607,
"step": 48
},
{
"epoch": 5.882352941176471,
"grad_norm": 0.40737274289131165,
"learning_rate": 9.980973490458728e-05,
"loss": 0.9343,
"step": 50
},
{
"epoch": 6.0,
"eval_loss": 0.9544721245765686,
"eval_runtime": 3.1309,
"eval_samples_per_second": 11.498,
"eval_steps_per_second": 11.498,
"step": 51
},
{
"epoch": 6.117647058823529,
"grad_norm": 0.39538058638572693,
"learning_rate": 9.972609476841367e-05,
"loss": 0.847,
"step": 52
},
{
"epoch": 6.352941176470588,
"grad_norm": 0.41286802291870117,
"learning_rate": 9.962730758206611e-05,
"loss": 0.8491,
"step": 54
},
{
"epoch": 6.588235294117647,
"grad_norm": 0.40585649013519287,
"learning_rate": 9.951340343707852e-05,
"loss": 0.8341,
"step": 56
},
{
"epoch": 6.823529411764706,
"grad_norm": 0.4548417627811432,
"learning_rate": 9.938441702975689e-05,
"loss": 0.8325,
"step": 58
},
{
"epoch": 6.9411764705882355,
"eval_loss": 0.9374688267707825,
"eval_runtime": 3.2246,
"eval_samples_per_second": 11.164,
"eval_steps_per_second": 11.164,
"step": 59
},
{
"epoch": 7.0588235294117645,
"grad_norm": 0.4099650979042053,
"learning_rate": 9.924038765061042e-05,
"loss": 0.8346,
"step": 60
},
{
"epoch": 7.294117647058823,
"grad_norm": 0.4285833537578583,
"learning_rate": 9.908135917238321e-05,
"loss": 0.7878,
"step": 62
},
{
"epoch": 7.529411764705882,
"grad_norm": 0.4640856385231018,
"learning_rate": 9.890738003669029e-05,
"loss": 0.7322,
"step": 64
},
{
"epoch": 7.764705882352941,
"grad_norm": 0.4892260730266571,
"learning_rate": 9.871850323926177e-05,
"loss": 0.7475,
"step": 66
},
{
"epoch": 8.0,
"grad_norm": 0.49551746249198914,
"learning_rate": 9.851478631379982e-05,
"loss": 0.7698,
"step": 68
},
{
"epoch": 8.0,
"eval_loss": 0.9379422664642334,
"eval_runtime": 3.1521,
"eval_samples_per_second": 11.421,
"eval_steps_per_second": 11.421,
"step": 68
},
{
"epoch": 8.235294117647058,
"grad_norm": 0.4985809922218323,
"learning_rate": 9.829629131445342e-05,
"loss": 0.6919,
"step": 70
},
{
"epoch": 8.470588235294118,
"grad_norm": 0.519849956035614,
"learning_rate": 9.806308479691595e-05,
"loss": 0.7148,
"step": 72
},
{
"epoch": 8.705882352941176,
"grad_norm": 0.6658029556274414,
"learning_rate": 9.781523779815179e-05,
"loss": 0.6521,
"step": 74
},
{
"epoch": 8.941176470588236,
"grad_norm": 0.6914530992507935,
"learning_rate": 9.755282581475769e-05,
"loss": 0.6897,
"step": 76
},
{
"epoch": 8.941176470588236,
"eval_loss": 0.9577365517616272,
"eval_runtime": 3.1388,
"eval_samples_per_second": 11.469,
"eval_steps_per_second": 11.469,
"step": 76
},
{
"epoch": 9.176470588235293,
"grad_norm": 0.6358802914619446,
"learning_rate": 9.727592877996585e-05,
"loss": 0.6228,
"step": 78
},
{
"epoch": 9.411764705882353,
"grad_norm": 0.6340999603271484,
"learning_rate": 9.698463103929542e-05,
"loss": 0.5725,
"step": 80
},
{
"epoch": 9.647058823529411,
"grad_norm": 0.7093944549560547,
"learning_rate": 9.667902132486009e-05,
"loss": 0.5951,
"step": 82
},
{
"epoch": 9.882352941176471,
"grad_norm": 0.7260375022888184,
"learning_rate": 9.635919272833938e-05,
"loss": 0.5925,
"step": 84
},
{
"epoch": 10.0,
"eval_loss": 1.0181899070739746,
"eval_runtime": 3.1482,
"eval_samples_per_second": 11.435,
"eval_steps_per_second": 11.435,
"step": 85
},
{
"epoch": 10.117647058823529,
"grad_norm": 0.6957091093063354,
"learning_rate": 9.602524267262203e-05,
"loss": 0.5215,
"step": 86
},
{
"epoch": 10.352941176470589,
"grad_norm": 0.7796018719673157,
"learning_rate": 9.567727288213005e-05,
"loss": 0.4734,
"step": 88
},
{
"epoch": 10.588235294117647,
"grad_norm": 0.8650311827659607,
"learning_rate": 9.53153893518325e-05,
"loss": 0.4979,
"step": 90
},
{
"epoch": 10.823529411764707,
"grad_norm": 0.9703614115715027,
"learning_rate": 9.493970231495835e-05,
"loss": 0.506,
"step": 92
},
{
"epoch": 10.941176470588236,
"eval_loss": 1.1055245399475098,
"eval_runtime": 3.212,
"eval_samples_per_second": 11.208,
"eval_steps_per_second": 11.208,
"step": 93
},
{
"epoch": 11.058823529411764,
"grad_norm": 0.8423256278038025,
"learning_rate": 9.45503262094184e-05,
"loss": 0.4772,
"step": 94
},
{
"epoch": 11.294117647058824,
"grad_norm": 0.9342357516288757,
"learning_rate": 9.414737964294636e-05,
"loss": 0.4045,
"step": 96
},
{
"epoch": 11.529411764705882,
"grad_norm": 1.0554617643356323,
"learning_rate": 9.373098535696979e-05,
"loss": 0.3993,
"step": 98
},
{
"epoch": 11.764705882352942,
"grad_norm": 1.1891016960144043,
"learning_rate": 9.330127018922194e-05,
"loss": 0.3595,
"step": 100
},
{
"epoch": 12.0,
"grad_norm": 1.214532494544983,
"learning_rate": 9.285836503510562e-05,
"loss": 0.3319,
"step": 102
},
{
"epoch": 12.0,
"eval_loss": 1.2682422399520874,
"eval_runtime": 3.1308,
"eval_samples_per_second": 11.499,
"eval_steps_per_second": 11.499,
"step": 102
},
{
"epoch": 12.235294117647058,
"grad_norm": 1.1215403079986572,
"learning_rate": 9.24024048078213e-05,
"loss": 0.3175,
"step": 104
},
{
"epoch": 12.470588235294118,
"grad_norm": 1.2607516050338745,
"learning_rate": 9.193352839727121e-05,
"loss": 0.2979,
"step": 106
},
{
"epoch": 12.705882352941176,
"grad_norm": 1.4650073051452637,
"learning_rate": 9.145187862775209e-05,
"loss": 0.2532,
"step": 108
},
{
"epoch": 12.941176470588236,
"grad_norm": 1.2881687879562378,
"learning_rate": 9.09576022144496e-05,
"loss": 0.2649,
"step": 110
},
{
"epoch": 12.941176470588236,
"eval_loss": 1.4429007768630981,
"eval_runtime": 3.1464,
"eval_samples_per_second": 11.442,
"eval_steps_per_second": 11.442,
"step": 110
},
{
"epoch": 13.176470588235293,
"grad_norm": 1.1737487316131592,
"learning_rate": 9.045084971874738e-05,
"loss": 0.246,
"step": 112
},
{
"epoch": 13.411764705882353,
"grad_norm": 1.406424641609192,
"learning_rate": 8.993177550236464e-05,
"loss": 0.192,
"step": 114
},
{
"epoch": 13.647058823529411,
"grad_norm": 1.2458537817001343,
"learning_rate": 8.940053768033609e-05,
"loss": 0.196,
"step": 116
},
{
"epoch": 13.882352941176471,
"grad_norm": 1.2824193239212036,
"learning_rate": 8.885729807284856e-05,
"loss": 0.2121,
"step": 118
},
{
"epoch": 14.0,
"eval_loss": 1.648590326309204,
"eval_runtime": 3.1273,
"eval_samples_per_second": 11.511,
"eval_steps_per_second": 11.511,
"step": 119
},
{
"epoch": 14.0,
"step": 119,
"total_flos": 8440897798668288.0,
"train_loss": 0.9342566123780083,
"train_runtime": 815.0277,
"train_samples_per_second": 8.343,
"train_steps_per_second": 0.491
}
],
"logging_steps": 2,
"max_steps": 400,
"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": 8440897798668288.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}