LexLLMv0.0.0.x.10.16a / trainer_state.json
andrealexroom's picture
Upload folder using huggingface_hub
7250468 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.07904721753794267,
"eval_steps": 750,
"global_step": 3750,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"grad_norm": 13.5,
"learning_rate": 1e-05,
"loss": 1.8758,
"step": 1
},
{
"epoch": 0.02,
"grad_norm": 7.3125,
"learning_rate": 1e-05,
"loss": 1.7401,
"step": 750
},
{
"epoch": 0.02,
"eval_validation_privacy_sources_loss": 1.0377193689346313,
"eval_validation_privacy_sources_runtime": 210.7816,
"eval_validation_privacy_sources_samples_per_second": 1.817,
"eval_validation_privacy_sources_steps_per_second": 0.911,
"step": 750
},
{
"epoch": 0.02,
"eval_validation_agenda_digitale_loss": 1.533129334449768,
"eval_validation_agenda_digitale_runtime": 186.9056,
"eval_validation_agenda_digitale_samples_per_second": 1.819,
"eval_validation_agenda_digitale_steps_per_second": 0.91,
"step": 750
},
{
"epoch": 0.02,
"eval_validation_leggepertutti_loss": 1.3645591735839844,
"eval_validation_leggepertutti_runtime": 35.7426,
"eval_validation_leggepertutti_samples_per_second": 1.819,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 750
},
{
"epoch": 0.02,
"eval_validation_altalexprivacy_loss": 1.3818964958190918,
"eval_validation_altalexprivacy_runtime": 58.2767,
"eval_validation_altalexprivacy_samples_per_second": 1.819,
"eval_validation_altalexprivacy_steps_per_second": 0.909,
"step": 750
},
{
"epoch": 0.02,
"eval_validation_mc4_it_loss": 1.948660135269165,
"eval_validation_mc4_it_runtime": 78.0751,
"eval_validation_mc4_it_samples_per_second": 1.819,
"eval_validation_mc4_it_steps_per_second": 0.909,
"step": 750
},
{
"epoch": 0.02,
"eval_validation_dolma_loss": 1.9987527132034302,
"eval_validation_dolma_runtime": 53.8799,
"eval_validation_dolma_samples_per_second": 1.819,
"eval_validation_dolma_steps_per_second": 0.909,
"step": 750
},
{
"epoch": 0.03,
"grad_norm": 7.0625,
"learning_rate": 1e-05,
"loss": 1.681,
"step": 1500
},
{
"epoch": 0.03,
"eval_validation_privacy_sources_loss": 0.9826437830924988,
"eval_validation_privacy_sources_runtime": 210.5002,
"eval_validation_privacy_sources_samples_per_second": 1.819,
"eval_validation_privacy_sources_steps_per_second": 0.912,
"step": 1500
},
{
"epoch": 0.03,
"eval_validation_agenda_digitale_loss": 1.4970210790634155,
"eval_validation_agenda_digitale_runtime": 186.8812,
"eval_validation_agenda_digitale_samples_per_second": 1.819,
"eval_validation_agenda_digitale_steps_per_second": 0.91,
"step": 1500
},
{
"epoch": 0.03,
"eval_validation_leggepertutti_loss": 1.3328404426574707,
"eval_validation_leggepertutti_runtime": 35.7381,
"eval_validation_leggepertutti_samples_per_second": 1.819,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 1500
},
{
"epoch": 0.03,
"eval_validation_altalexprivacy_loss": 1.3369431495666504,
"eval_validation_altalexprivacy_runtime": 58.2702,
"eval_validation_altalexprivacy_samples_per_second": 1.819,
"eval_validation_altalexprivacy_steps_per_second": 0.91,
"step": 1500
},
{
"epoch": 0.03,
"eval_validation_mc4_it_loss": 1.9100874662399292,
"eval_validation_mc4_it_runtime": 78.0367,
"eval_validation_mc4_it_samples_per_second": 1.82,
"eval_validation_mc4_it_steps_per_second": 0.91,
"step": 1500
},
{
"epoch": 0.03,
"eval_validation_dolma_loss": 1.9980002641677856,
"eval_validation_dolma_runtime": 53.8563,
"eval_validation_dolma_samples_per_second": 1.82,
"eval_validation_dolma_steps_per_second": 0.91,
"step": 1500
},
{
"epoch": 0.05,
"grad_norm": 5.9375,
"learning_rate": 1e-05,
"loss": 1.6532,
"step": 2250
},
{
"epoch": 0.05,
"eval_validation_privacy_sources_loss": 0.9487292170524597,
"eval_validation_privacy_sources_runtime": 210.5366,
"eval_validation_privacy_sources_samples_per_second": 1.819,
"eval_validation_privacy_sources_steps_per_second": 0.912,
"step": 2250
},
{
"epoch": 0.05,
"eval_validation_agenda_digitale_loss": 1.4758782386779785,
"eval_validation_agenda_digitale_runtime": 186.8673,
"eval_validation_agenda_digitale_samples_per_second": 1.819,
"eval_validation_agenda_digitale_steps_per_second": 0.91,
"step": 2250
},
{
"epoch": 0.05,
"eval_validation_leggepertutti_loss": 1.3132394552230835,
"eval_validation_leggepertutti_runtime": 35.7296,
"eval_validation_leggepertutti_samples_per_second": 1.819,
"eval_validation_leggepertutti_steps_per_second": 0.924,
"step": 2250
},
{
"epoch": 0.05,
"eval_validation_altalexprivacy_loss": 1.3129243850708008,
"eval_validation_altalexprivacy_runtime": 58.2737,
"eval_validation_altalexprivacy_samples_per_second": 1.819,
"eval_validation_altalexprivacy_steps_per_second": 0.91,
"step": 2250
},
{
"epoch": 0.05,
"eval_validation_mc4_it_loss": 1.8821855783462524,
"eval_validation_mc4_it_runtime": 78.0531,
"eval_validation_mc4_it_samples_per_second": 1.819,
"eval_validation_mc4_it_steps_per_second": 0.91,
"step": 2250
},
{
"epoch": 0.05,
"eval_validation_dolma_loss": 1.9904078245162964,
"eval_validation_dolma_runtime": 53.8818,
"eval_validation_dolma_samples_per_second": 1.819,
"eval_validation_dolma_steps_per_second": 0.909,
"step": 2250
},
{
"epoch": 0.06,
"grad_norm": 7.1875,
"learning_rate": 1e-05,
"loss": 1.6405,
"step": 3000
},
{
"epoch": 0.06,
"eval_validation_privacy_sources_loss": 0.9250594973564148,
"eval_validation_privacy_sources_runtime": 210.437,
"eval_validation_privacy_sources_samples_per_second": 1.82,
"eval_validation_privacy_sources_steps_per_second": 0.912,
"step": 3000
},
{
"epoch": 0.06,
"eval_validation_agenda_digitale_loss": 1.4562047719955444,
"eval_validation_agenda_digitale_runtime": 186.7992,
"eval_validation_agenda_digitale_samples_per_second": 1.82,
"eval_validation_agenda_digitale_steps_per_second": 0.91,
"step": 3000
},
{
"epoch": 0.06,
"eval_validation_leggepertutti_loss": 1.2950125932693481,
"eval_validation_leggepertutti_runtime": 35.7393,
"eval_validation_leggepertutti_samples_per_second": 1.819,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 3000
},
{
"epoch": 0.06,
"eval_validation_altalexprivacy_loss": 1.291795015335083,
"eval_validation_altalexprivacy_runtime": 58.2417,
"eval_validation_altalexprivacy_samples_per_second": 1.82,
"eval_validation_altalexprivacy_steps_per_second": 0.91,
"step": 3000
},
{
"epoch": 0.06,
"eval_validation_mc4_it_loss": 1.8612309694290161,
"eval_validation_mc4_it_runtime": 78.0155,
"eval_validation_mc4_it_samples_per_second": 1.82,
"eval_validation_mc4_it_steps_per_second": 0.91,
"step": 3000
},
{
"epoch": 0.06,
"eval_validation_dolma_loss": 1.9900999069213867,
"eval_validation_dolma_runtime": 53.8387,
"eval_validation_dolma_samples_per_second": 1.82,
"eval_validation_dolma_steps_per_second": 0.91,
"step": 3000
},
{
"epoch": 0.08,
"grad_norm": 7.21875,
"learning_rate": 1e-05,
"loss": 1.6146,
"step": 3750
},
{
"epoch": 0.08,
"eval_validation_privacy_sources_loss": 0.9070245623588562,
"eval_validation_privacy_sources_runtime": 210.504,
"eval_validation_privacy_sources_samples_per_second": 1.819,
"eval_validation_privacy_sources_steps_per_second": 0.912,
"step": 3750
},
{
"epoch": 0.08,
"eval_validation_agenda_digitale_loss": 1.4407680034637451,
"eval_validation_agenda_digitale_runtime": 186.8241,
"eval_validation_agenda_digitale_samples_per_second": 1.82,
"eval_validation_agenda_digitale_steps_per_second": 0.91,
"step": 3750
},
{
"epoch": 0.08,
"eval_validation_leggepertutti_loss": 1.2863850593566895,
"eval_validation_leggepertutti_runtime": 35.7293,
"eval_validation_leggepertutti_samples_per_second": 1.819,
"eval_validation_leggepertutti_steps_per_second": 0.924,
"step": 3750
},
{
"epoch": 0.08,
"eval_validation_altalexprivacy_loss": 1.270463228225708,
"eval_validation_altalexprivacy_runtime": 58.2503,
"eval_validation_altalexprivacy_samples_per_second": 1.82,
"eval_validation_altalexprivacy_steps_per_second": 0.91,
"step": 3750
},
{
"epoch": 0.08,
"eval_validation_mc4_it_loss": 1.8474267721176147,
"eval_validation_mc4_it_runtime": 78.069,
"eval_validation_mc4_it_samples_per_second": 1.819,
"eval_validation_mc4_it_steps_per_second": 0.909,
"step": 3750
},
{
"epoch": 0.08,
"eval_validation_dolma_loss": 1.9887909889221191,
"eval_validation_dolma_runtime": 53.8833,
"eval_validation_dolma_samples_per_second": 1.819,
"eval_validation_dolma_steps_per_second": 0.909,
"step": 3750
}
],
"logging_steps": 750,
"max_steps": 94880,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 750,
"total_flos": 3.93191060668416e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}