GaetanMichelet's picture
Model save
0fad488 verified
{
"best_metric": 0.39944469928741455,
"best_model_checkpoint": "data/Gemma-2-2B_task-3_60-samples_config-1_auto/checkpoint-46",
"epoch": 14.956521739130435,
"eval_steps": 500,
"global_step": 86,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.17391304347826086,
"grad_norm": 10.231386184692383,
"learning_rate": 4.000000000000001e-06,
"loss": 4.3925,
"step": 1
},
{
"epoch": 0.34782608695652173,
"grad_norm": 3.5924782752990723,
"learning_rate": 8.000000000000001e-06,
"loss": 2.3537,
"step": 2
},
{
"epoch": 0.6956521739130435,
"grad_norm": 4.74439001083374,
"learning_rate": 1.6000000000000003e-05,
"loss": 3.0596,
"step": 4
},
{
"epoch": 0.8695652173913043,
"eval_loss": 2.9978291988372803,
"eval_runtime": 4.4095,
"eval_samples_per_second": 2.721,
"eval_steps_per_second": 2.721,
"step": 5
},
{
"epoch": 1.0434782608695652,
"grad_norm": 6.4048943519592285,
"learning_rate": 2.4e-05,
"loss": 3.0606,
"step": 6
},
{
"epoch": 1.391304347826087,
"grad_norm": 7.171980857849121,
"learning_rate": 3.2000000000000005e-05,
"loss": 2.762,
"step": 8
},
{
"epoch": 1.7391304347826086,
"grad_norm": 6.314815998077393,
"learning_rate": 4e-05,
"loss": 2.455,
"step": 10
},
{
"epoch": 1.9130434782608696,
"eval_loss": 1.6208577156066895,
"eval_runtime": 4.1806,
"eval_samples_per_second": 2.87,
"eval_steps_per_second": 2.87,
"step": 11
},
{
"epoch": 2.0869565217391304,
"grad_norm": 2.6356723308563232,
"learning_rate": 4.8e-05,
"loss": 1.8354,
"step": 12
},
{
"epoch": 2.4347826086956523,
"grad_norm": 4.518723487854004,
"learning_rate": 5.6000000000000006e-05,
"loss": 1.4291,
"step": 14
},
{
"epoch": 2.782608695652174,
"grad_norm": 2.733262300491333,
"learning_rate": 6.400000000000001e-05,
"loss": 1.032,
"step": 16
},
{
"epoch": 2.9565217391304346,
"eval_loss": 0.739932119846344,
"eval_runtime": 4.1819,
"eval_samples_per_second": 2.869,
"eval_steps_per_second": 2.869,
"step": 17
},
{
"epoch": 3.130434782608696,
"grad_norm": 1.3501038551330566,
"learning_rate": 7.2e-05,
"loss": 0.9292,
"step": 18
},
{
"epoch": 3.4782608695652173,
"grad_norm": 1.0682424306869507,
"learning_rate": 8e-05,
"loss": 0.6387,
"step": 20
},
{
"epoch": 3.8260869565217392,
"grad_norm": 0.7127682566642761,
"learning_rate": 8.800000000000001e-05,
"loss": 0.4025,
"step": 22
},
{
"epoch": 4.0,
"eval_loss": 0.4877440631389618,
"eval_runtime": 4.1802,
"eval_samples_per_second": 2.871,
"eval_steps_per_second": 2.871,
"step": 23
},
{
"epoch": 4.173913043478261,
"grad_norm": 0.6702598333358765,
"learning_rate": 9.6e-05,
"loss": 0.4641,
"step": 24
},
{
"epoch": 4.521739130434782,
"grad_norm": 0.6460157036781311,
"learning_rate": 9.999512620046522e-05,
"loss": 0.4543,
"step": 26
},
{
"epoch": 4.869565217391305,
"grad_norm": 0.6406768560409546,
"learning_rate": 9.995614150494293e-05,
"loss": 0.2824,
"step": 28
},
{
"epoch": 4.869565217391305,
"eval_loss": 0.4447990357875824,
"eval_runtime": 4.1833,
"eval_samples_per_second": 2.869,
"eval_steps_per_second": 2.869,
"step": 28
},
{
"epoch": 5.217391304347826,
"grad_norm": 0.6102786064147949,
"learning_rate": 9.987820251299122e-05,
"loss": 0.3553,
"step": 30
},
{
"epoch": 5.565217391304348,
"grad_norm": 0.5026488304138184,
"learning_rate": 9.976136999909156e-05,
"loss": 0.3528,
"step": 32
},
{
"epoch": 5.913043478260869,
"grad_norm": 0.8358061909675598,
"learning_rate": 9.96057350657239e-05,
"loss": 0.2665,
"step": 34
},
{
"epoch": 5.913043478260869,
"eval_loss": 0.4215783178806305,
"eval_runtime": 4.1847,
"eval_samples_per_second": 2.868,
"eval_steps_per_second": 2.868,
"step": 34
},
{
"epoch": 6.260869565217392,
"grad_norm": 0.44195014238357544,
"learning_rate": 9.941141907232765e-05,
"loss": 0.2478,
"step": 36
},
{
"epoch": 6.608695652173913,
"grad_norm": 0.5455642342567444,
"learning_rate": 9.917857354066931e-05,
"loss": 0.3024,
"step": 38
},
{
"epoch": 6.956521739130435,
"grad_norm": 0.536037266254425,
"learning_rate": 9.890738003669029e-05,
"loss": 0.209,
"step": 40
},
{
"epoch": 6.956521739130435,
"eval_loss": 0.4002387523651123,
"eval_runtime": 4.1874,
"eval_samples_per_second": 2.866,
"eval_steps_per_second": 2.866,
"step": 40
},
{
"epoch": 7.304347826086957,
"grad_norm": 0.4730531573295593,
"learning_rate": 9.859805002892732e-05,
"loss": 0.1766,
"step": 42
},
{
"epoch": 7.6521739130434785,
"grad_norm": 0.36501502990722656,
"learning_rate": 9.825082472361557e-05,
"loss": 0.1574,
"step": 44
},
{
"epoch": 8.0,
"grad_norm": 0.6335607767105103,
"learning_rate": 9.786597487660337e-05,
"loss": 0.2776,
"step": 46
},
{
"epoch": 8.0,
"eval_loss": 0.39944469928741455,
"eval_runtime": 4.1812,
"eval_samples_per_second": 2.87,
"eval_steps_per_second": 2.87,
"step": 46
},
{
"epoch": 8.347826086956522,
"grad_norm": 0.4695717394351959,
"learning_rate": 9.744380058222483e-05,
"loss": 0.129,
"step": 48
},
{
"epoch": 8.695652173913043,
"grad_norm": 0.8587205410003662,
"learning_rate": 9.698463103929542e-05,
"loss": 0.1608,
"step": 50
},
{
"epoch": 8.869565217391305,
"eval_loss": 0.4351060390472412,
"eval_runtime": 4.1865,
"eval_samples_per_second": 2.866,
"eval_steps_per_second": 2.866,
"step": 51
},
{
"epoch": 9.043478260869565,
"grad_norm": 0.6147332191467285,
"learning_rate": 9.648882429441257e-05,
"loss": 0.152,
"step": 52
},
{
"epoch": 9.391304347826088,
"grad_norm": 0.4804113507270813,
"learning_rate": 9.595676696276172e-05,
"loss": 0.0825,
"step": 54
},
{
"epoch": 9.73913043478261,
"grad_norm": 0.8708305358886719,
"learning_rate": 9.538887392664544e-05,
"loss": 0.1258,
"step": 56
},
{
"epoch": 9.91304347826087,
"eval_loss": 0.5196338295936584,
"eval_runtime": 4.1793,
"eval_samples_per_second": 2.871,
"eval_steps_per_second": 2.871,
"step": 57
},
{
"epoch": 10.08695652173913,
"grad_norm": 0.727849543094635,
"learning_rate": 9.478558801197065e-05,
"loss": 0.1048,
"step": 58
},
{
"epoch": 10.434782608695652,
"grad_norm": 0.7062841653823853,
"learning_rate": 9.414737964294636e-05,
"loss": 0.0515,
"step": 60
},
{
"epoch": 10.782608695652174,
"grad_norm": 0.9390588402748108,
"learning_rate": 9.347474647526095e-05,
"loss": 0.0748,
"step": 62
},
{
"epoch": 10.956521739130435,
"eval_loss": 0.622729480266571,
"eval_runtime": 4.1936,
"eval_samples_per_second": 2.862,
"eval_steps_per_second": 2.862,
"step": 63
},
{
"epoch": 11.130434782608695,
"grad_norm": 0.4920307397842407,
"learning_rate": 9.276821300802534e-05,
"loss": 0.0674,
"step": 64
},
{
"epoch": 11.478260869565217,
"grad_norm": 0.3085463047027588,
"learning_rate": 9.202833017478422e-05,
"loss": 0.0247,
"step": 66
},
{
"epoch": 11.826086956521738,
"grad_norm": 0.7659479379653931,
"learning_rate": 9.125567491391476e-05,
"loss": 0.0581,
"step": 68
},
{
"epoch": 12.0,
"eval_loss": 0.7061209082603455,
"eval_runtime": 4.1826,
"eval_samples_per_second": 2.869,
"eval_steps_per_second": 2.869,
"step": 69
},
{
"epoch": 12.173913043478262,
"grad_norm": 0.6105247139930725,
"learning_rate": 9.045084971874738e-05,
"loss": 0.0311,
"step": 70
},
{
"epoch": 12.521739130434783,
"grad_norm": 0.4759242534637451,
"learning_rate": 8.961448216775954e-05,
"loss": 0.0347,
"step": 72
},
{
"epoch": 12.869565217391305,
"grad_norm": 0.7405514121055603,
"learning_rate": 8.874722443520899e-05,
"loss": 0.0281,
"step": 74
},
{
"epoch": 12.869565217391305,
"eval_loss": 0.7657582759857178,
"eval_runtime": 4.1886,
"eval_samples_per_second": 2.865,
"eval_steps_per_second": 2.865,
"step": 74
},
{
"epoch": 13.217391304347826,
"grad_norm": 0.5165314674377441,
"learning_rate": 8.784975278258783e-05,
"loss": 0.0201,
"step": 76
},
{
"epoch": 13.565217391304348,
"grad_norm": 0.8582074642181396,
"learning_rate": 8.692276703129421e-05,
"loss": 0.0171,
"step": 78
},
{
"epoch": 13.91304347826087,
"grad_norm": 0.28264862298965454,
"learning_rate": 8.596699001693255e-05,
"loss": 0.0178,
"step": 80
},
{
"epoch": 13.91304347826087,
"eval_loss": 0.7745249271392822,
"eval_runtime": 4.1849,
"eval_samples_per_second": 2.867,
"eval_steps_per_second": 2.867,
"step": 80
},
{
"epoch": 14.26086956521739,
"grad_norm": 0.2634141743183136,
"learning_rate": 8.498316702566828e-05,
"loss": 0.0119,
"step": 82
},
{
"epoch": 14.608695652173914,
"grad_norm": 0.6244635581970215,
"learning_rate": 8.397206521307584e-05,
"loss": 0.0131,
"step": 84
},
{
"epoch": 14.956521739130435,
"grad_norm": 0.1751631647348404,
"learning_rate": 8.293447300593402e-05,
"loss": 0.006,
"step": 86
},
{
"epoch": 14.956521739130435,
"eval_loss": 0.817257821559906,
"eval_runtime": 4.1898,
"eval_samples_per_second": 2.864,
"eval_steps_per_second": 2.864,
"step": 86
},
{
"epoch": 14.956521739130435,
"step": 86,
"total_flos": 1.898813586617139e+16,
"train_loss": 0.5986227188692537,
"train_runtime": 815.6084,
"train_samples_per_second": 2.82,
"train_steps_per_second": 0.307
}
],
"logging_steps": 2,
"max_steps": 250,
"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": 1.898813586617139e+16,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}