GaetanMichelet's picture
Model save
aee264d verified
{
"best_metric": 0.7086067795753479,
"best_model_checkpoint": "data/Llama-31-8B_task-2_60-samples_config-2/checkpoint-28",
"epoch": 16.695652173913043,
"eval_steps": 500,
"global_step": 48,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.34782608695652173,
"grad_norm": 0.6984471678733826,
"learning_rate": 1e-05,
"loss": 1.1182,
"step": 1
},
{
"epoch": 0.6956521739130435,
"grad_norm": 0.6600484251976013,
"learning_rate": 2e-05,
"loss": 1.0749,
"step": 2
},
{
"epoch": 0.6956521739130435,
"eval_loss": 1.0892785787582397,
"eval_runtime": 12.9786,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 2
},
{
"epoch": 1.391304347826087,
"grad_norm": 0.6943985223770142,
"learning_rate": 4e-05,
"loss": 1.0585,
"step": 4
},
{
"epoch": 1.7391304347826086,
"eval_loss": 1.0146803855895996,
"eval_runtime": 12.9779,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 5
},
{
"epoch": 2.0869565217391304,
"grad_norm": 0.6072419285774231,
"learning_rate": 6e-05,
"loss": 1.0156,
"step": 6
},
{
"epoch": 2.782608695652174,
"grad_norm": 0.4303366541862488,
"learning_rate": 8e-05,
"loss": 0.935,
"step": 8
},
{
"epoch": 2.782608695652174,
"eval_loss": 0.8861821293830872,
"eval_runtime": 12.9778,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 8
},
{
"epoch": 3.4782608695652173,
"grad_norm": 0.2728838622570038,
"learning_rate": 0.0001,
"loss": 0.8267,
"step": 10
},
{
"epoch": 3.8260869565217392,
"eval_loss": 0.8298996090888977,
"eval_runtime": 12.9806,
"eval_samples_per_second": 0.924,
"eval_steps_per_second": 0.924,
"step": 11
},
{
"epoch": 4.173913043478261,
"grad_norm": 0.35264384746551514,
"learning_rate": 9.987820251299122e-05,
"loss": 0.8308,
"step": 12
},
{
"epoch": 4.869565217391305,
"grad_norm": 0.37614378333091736,
"learning_rate": 9.951340343707852e-05,
"loss": 0.7982,
"step": 14
},
{
"epoch": 4.869565217391305,
"eval_loss": 0.7848379015922546,
"eval_runtime": 12.9787,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 14
},
{
"epoch": 5.565217391304348,
"grad_norm": 0.2864533066749573,
"learning_rate": 9.890738003669029e-05,
"loss": 0.7295,
"step": 16
},
{
"epoch": 5.913043478260869,
"eval_loss": 0.7512074112892151,
"eval_runtime": 12.9823,
"eval_samples_per_second": 0.924,
"eval_steps_per_second": 0.924,
"step": 17
},
{
"epoch": 6.260869565217392,
"grad_norm": 0.22847534716129303,
"learning_rate": 9.806308479691595e-05,
"loss": 0.6918,
"step": 18
},
{
"epoch": 6.956521739130435,
"grad_norm": 0.1993773877620697,
"learning_rate": 9.698463103929542e-05,
"loss": 0.7036,
"step": 20
},
{
"epoch": 6.956521739130435,
"eval_loss": 0.7341786026954651,
"eval_runtime": 12.9779,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 20
},
{
"epoch": 7.6521739130434785,
"grad_norm": 0.20846480131149292,
"learning_rate": 9.567727288213005e-05,
"loss": 0.6598,
"step": 22
},
{
"epoch": 8.0,
"eval_loss": 0.72257000207901,
"eval_runtime": 12.979,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 23
},
{
"epoch": 8.347826086956522,
"grad_norm": 0.21805794537067413,
"learning_rate": 9.414737964294636e-05,
"loss": 0.6493,
"step": 24
},
{
"epoch": 8.695652173913043,
"eval_loss": 0.7157177925109863,
"eval_runtime": 12.9788,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 25
},
{
"epoch": 9.043478260869565,
"grad_norm": 0.24037529528141022,
"learning_rate": 9.24024048078213e-05,
"loss": 0.6126,
"step": 26
},
{
"epoch": 9.73913043478261,
"grad_norm": 0.2167801856994629,
"learning_rate": 9.045084971874738e-05,
"loss": 0.5819,
"step": 28
},
{
"epoch": 9.73913043478261,
"eval_loss": 0.7086067795753479,
"eval_runtime": 12.9757,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 28
},
{
"epoch": 10.434782608695652,
"grad_norm": 0.24030882120132446,
"learning_rate": 8.83022221559489e-05,
"loss": 0.5695,
"step": 30
},
{
"epoch": 10.782608695652174,
"eval_loss": 0.708713710308075,
"eval_runtime": 12.9863,
"eval_samples_per_second": 0.924,
"eval_steps_per_second": 0.924,
"step": 31
},
{
"epoch": 11.130434782608695,
"grad_norm": 0.27788665890693665,
"learning_rate": 8.596699001693255e-05,
"loss": 0.54,
"step": 32
},
{
"epoch": 11.826086956521738,
"grad_norm": 0.2642171084880829,
"learning_rate": 8.345653031794292e-05,
"loss": 0.5157,
"step": 34
},
{
"epoch": 11.826086956521738,
"eval_loss": 0.7256595492362976,
"eval_runtime": 12.9755,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 34
},
{
"epoch": 12.521739130434783,
"grad_norm": 0.31618693470954895,
"learning_rate": 8.07830737662829e-05,
"loss": 0.4755,
"step": 36
},
{
"epoch": 12.869565217391305,
"eval_loss": 0.744239091873169,
"eval_runtime": 12.9737,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 37
},
{
"epoch": 13.217391304347826,
"grad_norm": 0.35100746154785156,
"learning_rate": 7.795964517353735e-05,
"loss": 0.4416,
"step": 38
},
{
"epoch": 13.91304347826087,
"grad_norm": 0.33555322885513306,
"learning_rate": 7.500000000000001e-05,
"loss": 0.4184,
"step": 40
},
{
"epoch": 13.91304347826087,
"eval_loss": 0.7554243206977844,
"eval_runtime": 12.9794,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 40
},
{
"epoch": 14.608695652173914,
"grad_norm": 0.33266159892082214,
"learning_rate": 7.191855733945387e-05,
"loss": 0.3399,
"step": 42
},
{
"epoch": 14.956521739130435,
"eval_loss": 0.8075953125953674,
"eval_runtime": 12.9775,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 43
},
{
"epoch": 15.304347826086957,
"grad_norm": 0.39590662717819214,
"learning_rate": 6.873032967079561e-05,
"loss": 0.3511,
"step": 44
},
{
"epoch": 16.0,
"grad_norm": 0.4365744888782501,
"learning_rate": 6.545084971874738e-05,
"loss": 0.2958,
"step": 46
},
{
"epoch": 16.0,
"eval_loss": 0.8420116901397705,
"eval_runtime": 12.9783,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 46
},
{
"epoch": 16.695652173913043,
"grad_norm": 0.4204292297363281,
"learning_rate": 6.209609477998338e-05,
"loss": 0.2339,
"step": 48
},
{
"epoch": 16.695652173913043,
"eval_loss": 0.9178810119628906,
"eval_runtime": 12.978,
"eval_samples_per_second": 0.925,
"eval_steps_per_second": 0.925,
"step": 48
},
{
"epoch": 16.695652173913043,
"step": 48,
"total_flos": 7.557781991614054e+16,
"train_loss": 0.644991461498042,
"train_runtime": 2716.8392,
"train_samples_per_second": 0.847,
"train_steps_per_second": 0.037
}
],
"logging_steps": 2,
"max_steps": 100,
"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": 7.557781991614054e+16,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}