LexLLMv0.0.0.x.10.14a / trainer_state.json
andrealexroom's picture
Upload folder using huggingface_hub
973adcb verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.13596911990210223,
"eval_steps": 375,
"global_step": 2250,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1e-05,
"loss": 1.8182,
"step": 1
},
{
"epoch": 0.02,
"learning_rate": 1e-05,
"loss": 1.8531,
"step": 375
},
{
"epoch": 0.02,
"eval_validation_privacy_sources_loss": 1.0427056550979614,
"eval_validation_privacy_sources_runtime": 201.1093,
"eval_validation_privacy_sources_samples_per_second": 1.904,
"eval_validation_privacy_sources_steps_per_second": 1.904,
"step": 375
},
{
"epoch": 0.02,
"eval_validation_agenda_digitale_loss": 1.5768671035766602,
"eval_validation_agenda_digitale_runtime": 178.6225,
"eval_validation_agenda_digitale_samples_per_second": 1.903,
"eval_validation_agenda_digitale_steps_per_second": 1.903,
"step": 375
},
{
"epoch": 0.02,
"eval_validation_leggepertutti_loss": 1.41618013381958,
"eval_validation_leggepertutti_runtime": 34.1702,
"eval_validation_leggepertutti_samples_per_second": 1.902,
"eval_validation_leggepertutti_steps_per_second": 1.902,
"step": 375
},
{
"epoch": 0.02,
"eval_validation_altalexprivacy_loss": 1.420487880706787,
"eval_validation_altalexprivacy_runtime": 55.6679,
"eval_validation_altalexprivacy_samples_per_second": 1.904,
"eval_validation_altalexprivacy_steps_per_second": 1.904,
"step": 375
},
{
"epoch": 0.02,
"eval_validation_mc4_it_loss": 2.0156803131103516,
"eval_validation_mc4_it_runtime": 74.5861,
"eval_validation_mc4_it_samples_per_second": 1.904,
"eval_validation_mc4_it_steps_per_second": 1.904,
"step": 375
},
{
"epoch": 0.05,
"learning_rate": 1e-05,
"loss": 1.659,
"step": 750
},
{
"epoch": 0.05,
"eval_validation_privacy_sources_loss": 0.9875672459602356,
"eval_validation_privacy_sources_runtime": 201.0647,
"eval_validation_privacy_sources_samples_per_second": 1.905,
"eval_validation_privacy_sources_steps_per_second": 1.905,
"step": 750
},
{
"epoch": 0.05,
"eval_validation_agenda_digitale_loss": 1.5326132774353027,
"eval_validation_agenda_digitale_runtime": 178.5645,
"eval_validation_agenda_digitale_samples_per_second": 1.904,
"eval_validation_agenda_digitale_steps_per_second": 1.904,
"step": 750
},
{
"epoch": 0.05,
"eval_validation_leggepertutti_loss": 1.3737919330596924,
"eval_validation_leggepertutti_runtime": 34.1417,
"eval_validation_leggepertutti_samples_per_second": 1.904,
"eval_validation_leggepertutti_steps_per_second": 1.904,
"step": 750
},
{
"epoch": 0.05,
"eval_validation_altalexprivacy_loss": 1.3635202646255493,
"eval_validation_altalexprivacy_runtime": 55.6795,
"eval_validation_altalexprivacy_samples_per_second": 1.904,
"eval_validation_altalexprivacy_steps_per_second": 1.904,
"step": 750
},
{
"epoch": 0.05,
"eval_validation_mc4_it_loss": 1.9601805210113525,
"eval_validation_mc4_it_runtime": 74.5661,
"eval_validation_mc4_it_samples_per_second": 1.904,
"eval_validation_mc4_it_steps_per_second": 1.904,
"step": 750
},
{
"epoch": 0.07,
"learning_rate": 1e-05,
"loss": 1.5963,
"step": 1125
},
{
"epoch": 0.07,
"eval_validation_privacy_sources_loss": 0.9522767066955566,
"eval_validation_privacy_sources_runtime": 201.0521,
"eval_validation_privacy_sources_samples_per_second": 1.905,
"eval_validation_privacy_sources_steps_per_second": 1.905,
"step": 1125
},
{
"epoch": 0.07,
"eval_validation_agenda_digitale_loss": 1.4975751638412476,
"eval_validation_agenda_digitale_runtime": 178.5472,
"eval_validation_agenda_digitale_samples_per_second": 1.904,
"eval_validation_agenda_digitale_steps_per_second": 1.904,
"step": 1125
},
{
"epoch": 0.07,
"eval_validation_leggepertutti_loss": 1.3494906425476074,
"eval_validation_leggepertutti_runtime": 34.1339,
"eval_validation_leggepertutti_samples_per_second": 1.904,
"eval_validation_leggepertutti_steps_per_second": 1.904,
"step": 1125
},
{
"epoch": 0.07,
"eval_validation_altalexprivacy_loss": 1.3334871530532837,
"eval_validation_altalexprivacy_runtime": 55.6542,
"eval_validation_altalexprivacy_samples_per_second": 1.905,
"eval_validation_altalexprivacy_steps_per_second": 1.905,
"step": 1125
},
{
"epoch": 0.07,
"eval_validation_mc4_it_loss": 1.9144717454910278,
"eval_validation_mc4_it_runtime": 74.5606,
"eval_validation_mc4_it_samples_per_second": 1.904,
"eval_validation_mc4_it_steps_per_second": 1.904,
"step": 1125
},
{
"epoch": 0.09,
"learning_rate": 1e-05,
"loss": 1.577,
"step": 1500
},
{
"epoch": 0.09,
"eval_validation_privacy_sources_loss": 0.9246706366539001,
"eval_validation_privacy_sources_runtime": 201.0413,
"eval_validation_privacy_sources_samples_per_second": 1.905,
"eval_validation_privacy_sources_steps_per_second": 1.905,
"step": 1500
},
{
"epoch": 0.09,
"eval_validation_agenda_digitale_loss": 1.4834680557250977,
"eval_validation_agenda_digitale_runtime": 178.5955,
"eval_validation_agenda_digitale_samples_per_second": 1.904,
"eval_validation_agenda_digitale_steps_per_second": 1.904,
"step": 1500
},
{
"epoch": 0.09,
"eval_validation_leggepertutti_loss": 1.3424042463302612,
"eval_validation_leggepertutti_runtime": 34.1348,
"eval_validation_leggepertutti_samples_per_second": 1.904,
"eval_validation_leggepertutti_steps_per_second": 1.904,
"step": 1500
},
{
"epoch": 0.09,
"eval_validation_altalexprivacy_loss": 1.3129479885101318,
"eval_validation_altalexprivacy_runtime": 55.6818,
"eval_validation_altalexprivacy_samples_per_second": 1.904,
"eval_validation_altalexprivacy_steps_per_second": 1.904,
"step": 1500
},
{
"epoch": 0.09,
"eval_validation_mc4_it_loss": 1.897855281829834,
"eval_validation_mc4_it_runtime": 74.5687,
"eval_validation_mc4_it_samples_per_second": 1.904,
"eval_validation_mc4_it_steps_per_second": 1.904,
"step": 1500
},
{
"epoch": 0.11,
"learning_rate": 1e-05,
"loss": 1.551,
"step": 1875
},
{
"epoch": 0.11,
"eval_validation_privacy_sources_loss": 0.9161506295204163,
"eval_validation_privacy_sources_runtime": 201.028,
"eval_validation_privacy_sources_samples_per_second": 1.905,
"eval_validation_privacy_sources_steps_per_second": 1.905,
"step": 1875
},
{
"epoch": 0.11,
"eval_validation_agenda_digitale_loss": 1.4714491367340088,
"eval_validation_agenda_digitale_runtime": 178.5046,
"eval_validation_agenda_digitale_samples_per_second": 1.905,
"eval_validation_agenda_digitale_steps_per_second": 1.905,
"step": 1875
},
{
"epoch": 0.11,
"eval_validation_leggepertutti_loss": 1.349782109260559,
"eval_validation_leggepertutti_runtime": 34.1218,
"eval_validation_leggepertutti_samples_per_second": 1.905,
"eval_validation_leggepertutti_steps_per_second": 1.905,
"step": 1875
},
{
"epoch": 0.11,
"eval_validation_altalexprivacy_loss": 1.3038463592529297,
"eval_validation_altalexprivacy_runtime": 55.6592,
"eval_validation_altalexprivacy_samples_per_second": 1.904,
"eval_validation_altalexprivacy_steps_per_second": 1.904,
"step": 1875
},
{
"epoch": 0.11,
"eval_validation_mc4_it_loss": 1.8886678218841553,
"eval_validation_mc4_it_runtime": 74.5616,
"eval_validation_mc4_it_samples_per_second": 1.904,
"eval_validation_mc4_it_steps_per_second": 1.904,
"step": 1875
},
{
"epoch": 0.14,
"learning_rate": 1e-05,
"loss": 1.5334,
"step": 2250
},
{
"epoch": 0.14,
"eval_validation_privacy_sources_loss": 0.8999343514442444,
"eval_validation_privacy_sources_runtime": 201.0719,
"eval_validation_privacy_sources_samples_per_second": 1.905,
"eval_validation_privacy_sources_steps_per_second": 1.905,
"step": 2250
},
{
"epoch": 0.14,
"eval_validation_agenda_digitale_loss": 1.470467448234558,
"eval_validation_agenda_digitale_runtime": 178.5136,
"eval_validation_agenda_digitale_samples_per_second": 1.905,
"eval_validation_agenda_digitale_steps_per_second": 1.905,
"step": 2250
},
{
"epoch": 0.14,
"eval_validation_leggepertutti_loss": 1.3344582319259644,
"eval_validation_leggepertutti_runtime": 34.1414,
"eval_validation_leggepertutti_samples_per_second": 1.904,
"eval_validation_leggepertutti_steps_per_second": 1.904,
"step": 2250
},
{
"epoch": 0.14,
"eval_validation_altalexprivacy_loss": 1.2899404764175415,
"eval_validation_altalexprivacy_runtime": 55.6737,
"eval_validation_altalexprivacy_samples_per_second": 1.904,
"eval_validation_altalexprivacy_steps_per_second": 1.904,
"step": 2250
},
{
"epoch": 0.14,
"eval_validation_mc4_it_loss": 1.8787847757339478,
"eval_validation_mc4_it_runtime": 74.5812,
"eval_validation_mc4_it_samples_per_second": 1.904,
"eval_validation_mc4_it_steps_per_second": 1.904,
"step": 2250
}
],
"logging_steps": 375,
"max_steps": 33094,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 375,
"total_flos": 4.718292728020992e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}