GaetanMichelet's picture
Model save
defacaa verified
{
"best_metric": 0.8270444869995117,
"best_model_checkpoint": "data/Llama-31-8B_task-1_60-samples_config-2_full_auto/checkpoint-43",
"epoch": 21.91304347826087,
"eval_steps": 500,
"global_step": 63,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.34782608695652173,
"grad_norm": 1.8296887874603271,
"learning_rate": 1e-05,
"loss": 2.2067,
"step": 1
},
{
"epoch": 0.6956521739130435,
"grad_norm": 1.8122031688690186,
"learning_rate": 2e-05,
"loss": 2.2096,
"step": 2
},
{
"epoch": 0.6956521739130435,
"eval_loss": 2.1129257678985596,
"eval_runtime": 5.0286,
"eval_samples_per_second": 2.386,
"eval_steps_per_second": 2.386,
"step": 2
},
{
"epoch": 1.391304347826087,
"grad_norm": 1.456130862236023,
"learning_rate": 4e-05,
"loss": 2.167,
"step": 4
},
{
"epoch": 1.7391304347826086,
"eval_loss": 1.9557948112487793,
"eval_runtime": 5.028,
"eval_samples_per_second": 2.387,
"eval_steps_per_second": 2.387,
"step": 5
},
{
"epoch": 2.0869565217391304,
"grad_norm": 1.2667176723480225,
"learning_rate": 6e-05,
"loss": 2.0722,
"step": 6
},
{
"epoch": 2.782608695652174,
"grad_norm": 0.9311268925666809,
"learning_rate": 8e-05,
"loss": 1.8726,
"step": 8
},
{
"epoch": 2.782608695652174,
"eval_loss": 1.7428107261657715,
"eval_runtime": 5.0288,
"eval_samples_per_second": 2.386,
"eval_steps_per_second": 2.386,
"step": 8
},
{
"epoch": 3.4782608695652173,
"grad_norm": 1.2152765989303589,
"learning_rate": 0.0001,
"loss": 1.7678,
"step": 10
},
{
"epoch": 3.8260869565217392,
"eval_loss": 1.5016740560531616,
"eval_runtime": 5.0523,
"eval_samples_per_second": 2.375,
"eval_steps_per_second": 2.375,
"step": 11
},
{
"epoch": 4.173913043478261,
"grad_norm": 0.8071852326393127,
"learning_rate": 9.987820251299122e-05,
"loss": 1.5772,
"step": 12
},
{
"epoch": 4.869565217391305,
"grad_norm": 0.8298630714416504,
"learning_rate": 9.951340343707852e-05,
"loss": 1.3895,
"step": 14
},
{
"epoch": 4.869565217391305,
"eval_loss": 1.252468466758728,
"eval_runtime": 5.0279,
"eval_samples_per_second": 2.387,
"eval_steps_per_second": 2.387,
"step": 14
},
{
"epoch": 5.565217391304348,
"grad_norm": 0.8688008785247803,
"learning_rate": 9.890738003669029e-05,
"loss": 1.234,
"step": 16
},
{
"epoch": 5.913043478260869,
"eval_loss": 1.0325278043746948,
"eval_runtime": 5.0226,
"eval_samples_per_second": 2.389,
"eval_steps_per_second": 2.389,
"step": 17
},
{
"epoch": 6.260869565217392,
"grad_norm": 0.7937119007110596,
"learning_rate": 9.806308479691595e-05,
"loss": 1.021,
"step": 18
},
{
"epoch": 6.956521739130435,
"grad_norm": 0.5331211090087891,
"learning_rate": 9.698463103929542e-05,
"loss": 0.9378,
"step": 20
},
{
"epoch": 6.956521739130435,
"eval_loss": 0.9270639419555664,
"eval_runtime": 5.0239,
"eval_samples_per_second": 2.389,
"eval_steps_per_second": 2.389,
"step": 20
},
{
"epoch": 7.6521739130434785,
"grad_norm": 0.393255352973938,
"learning_rate": 9.567727288213005e-05,
"loss": 0.8782,
"step": 22
},
{
"epoch": 8.0,
"eval_loss": 0.8919858932495117,
"eval_runtime": 5.0253,
"eval_samples_per_second": 2.388,
"eval_steps_per_second": 2.388,
"step": 23
},
{
"epoch": 8.347826086956522,
"grad_norm": 0.3337656855583191,
"learning_rate": 9.414737964294636e-05,
"loss": 0.8394,
"step": 24
},
{
"epoch": 8.695652173913043,
"eval_loss": 0.8783631920814514,
"eval_runtime": 5.0242,
"eval_samples_per_second": 2.388,
"eval_steps_per_second": 2.388,
"step": 25
},
{
"epoch": 9.043478260869565,
"grad_norm": 0.3544324040412903,
"learning_rate": 9.24024048078213e-05,
"loss": 0.838,
"step": 26
},
{
"epoch": 9.73913043478261,
"grad_norm": 0.34565308690071106,
"learning_rate": 9.045084971874738e-05,
"loss": 0.7845,
"step": 28
},
{
"epoch": 9.73913043478261,
"eval_loss": 0.8646710515022278,
"eval_runtime": 5.0272,
"eval_samples_per_second": 2.387,
"eval_steps_per_second": 2.387,
"step": 28
},
{
"epoch": 10.434782608695652,
"grad_norm": 0.3286716938018799,
"learning_rate": 8.83022221559489e-05,
"loss": 0.7863,
"step": 30
},
{
"epoch": 10.782608695652174,
"eval_loss": 0.8503349423408508,
"eval_runtime": 5.0295,
"eval_samples_per_second": 2.386,
"eval_steps_per_second": 2.386,
"step": 31
},
{
"epoch": 11.130434782608695,
"grad_norm": 0.34742411971092224,
"learning_rate": 8.596699001693255e-05,
"loss": 0.782,
"step": 32
},
{
"epoch": 11.826086956521738,
"grad_norm": 0.3356003165245056,
"learning_rate": 8.345653031794292e-05,
"loss": 0.7261,
"step": 34
},
{
"epoch": 11.826086956521738,
"eval_loss": 0.841669499874115,
"eval_runtime": 5.0281,
"eval_samples_per_second": 2.387,
"eval_steps_per_second": 2.387,
"step": 34
},
{
"epoch": 12.521739130434783,
"grad_norm": 0.32681888341903687,
"learning_rate": 8.07830737662829e-05,
"loss": 0.7333,
"step": 36
},
{
"epoch": 12.869565217391305,
"eval_loss": 0.8337127566337585,
"eval_runtime": 5.0271,
"eval_samples_per_second": 2.387,
"eval_steps_per_second": 2.387,
"step": 37
},
{
"epoch": 13.217391304347826,
"grad_norm": 0.40883421897888184,
"learning_rate": 7.795964517353735e-05,
"loss": 0.6964,
"step": 38
},
{
"epoch": 13.91304347826087,
"grad_norm": 0.38553422689437866,
"learning_rate": 7.500000000000001e-05,
"loss": 0.6709,
"step": 40
},
{
"epoch": 13.91304347826087,
"eval_loss": 0.8288694024085999,
"eval_runtime": 5.0387,
"eval_samples_per_second": 2.382,
"eval_steps_per_second": 2.382,
"step": 40
},
{
"epoch": 14.608695652173914,
"grad_norm": 0.39506804943084717,
"learning_rate": 7.191855733945387e-05,
"loss": 0.6612,
"step": 42
},
{
"epoch": 14.956521739130435,
"eval_loss": 0.8270444869995117,
"eval_runtime": 5.0342,
"eval_samples_per_second": 2.384,
"eval_steps_per_second": 2.384,
"step": 43
},
{
"epoch": 15.304347826086957,
"grad_norm": 0.41430413722991943,
"learning_rate": 6.873032967079561e-05,
"loss": 0.6248,
"step": 44
},
{
"epoch": 16.0,
"grad_norm": 0.5096813440322876,
"learning_rate": 6.545084971874738e-05,
"loss": 0.6253,
"step": 46
},
{
"epoch": 16.0,
"eval_loss": 0.8288677334785461,
"eval_runtime": 5.0231,
"eval_samples_per_second": 2.389,
"eval_steps_per_second": 2.389,
"step": 46
},
{
"epoch": 16.695652173913043,
"grad_norm": 0.5710862874984741,
"learning_rate": 6.209609477998338e-05,
"loss": 0.6012,
"step": 48
},
{
"epoch": 16.695652173913043,
"eval_loss": 0.8323108553886414,
"eval_runtime": 5.0271,
"eval_samples_per_second": 2.387,
"eval_steps_per_second": 2.387,
"step": 48
},
{
"epoch": 17.391304347826086,
"grad_norm": 0.5608723759651184,
"learning_rate": 5.868240888334653e-05,
"loss": 0.5792,
"step": 50
},
{
"epoch": 17.73913043478261,
"eval_loss": 0.83847576379776,
"eval_runtime": 5.0311,
"eval_samples_per_second": 2.385,
"eval_steps_per_second": 2.385,
"step": 51
},
{
"epoch": 18.08695652173913,
"grad_norm": 0.6590730547904968,
"learning_rate": 5.522642316338268e-05,
"loss": 0.5196,
"step": 52
},
{
"epoch": 18.782608695652176,
"grad_norm": 0.6217765212059021,
"learning_rate": 5.174497483512506e-05,
"loss": 0.5162,
"step": 54
},
{
"epoch": 18.782608695652176,
"eval_loss": 0.8560710549354553,
"eval_runtime": 5.0295,
"eval_samples_per_second": 2.386,
"eval_steps_per_second": 2.386,
"step": 54
},
{
"epoch": 19.47826086956522,
"grad_norm": 0.6448721289634705,
"learning_rate": 4.825502516487497e-05,
"loss": 0.5219,
"step": 56
},
{
"epoch": 19.82608695652174,
"eval_loss": 0.8603215217590332,
"eval_runtime": 5.0413,
"eval_samples_per_second": 2.38,
"eval_steps_per_second": 2.38,
"step": 57
},
{
"epoch": 20.17391304347826,
"grad_norm": 0.6627869009971619,
"learning_rate": 4.477357683661734e-05,
"loss": 0.4583,
"step": 58
},
{
"epoch": 20.869565217391305,
"grad_norm": 0.675813615322113,
"learning_rate": 4.131759111665349e-05,
"loss": 0.445,
"step": 60
},
{
"epoch": 20.869565217391305,
"eval_loss": 0.8802428245544434,
"eval_runtime": 5.0363,
"eval_samples_per_second": 2.383,
"eval_steps_per_second": 2.383,
"step": 60
},
{
"epoch": 21.565217391304348,
"grad_norm": 0.5834102034568787,
"learning_rate": 3.790390522001662e-05,
"loss": 0.4396,
"step": 62
},
{
"epoch": 21.91304347826087,
"eval_loss": 0.9046290516853333,
"eval_runtime": 5.0326,
"eval_samples_per_second": 2.384,
"eval_steps_per_second": 2.384,
"step": 63
},
{
"epoch": 21.91304347826087,
"step": 63,
"total_flos": 1.6617387570757632e+16,
"train_loss": 0.9590450813845982,
"train_runtime": 1220.0386,
"train_samples_per_second": 1.885,
"train_steps_per_second": 0.082
}
],
"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": 1.6617387570757632e+16,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}