LexLLMv0.0.0.x.10.16b / trainer_state.json
andrealexroom's picture
Upload folder using huggingface_hub
bdab471 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.25295109612141653,
"eval_steps": 750,
"global_step": 12000,
"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
},
{
"epoch": 0.09,
"grad_norm": 12.9375,
"learning_rate": 1e-05,
"loss": 1.6013,
"step": 4500
},
{
"epoch": 0.09,
"eval_validation_privacy_sources_loss": 0.8938004374504089,
"eval_validation_privacy_sources_runtime": 210.5349,
"eval_validation_privacy_sources_samples_per_second": 1.819,
"eval_validation_privacy_sources_steps_per_second": 0.912,
"step": 4500
},
{
"epoch": 0.09,
"eval_validation_agenda_digitale_loss": 1.430904746055603,
"eval_validation_agenda_digitale_runtime": 186.8995,
"eval_validation_agenda_digitale_samples_per_second": 1.819,
"eval_validation_agenda_digitale_steps_per_second": 0.91,
"step": 4500
},
{
"epoch": 0.09,
"eval_validation_leggepertutti_loss": 1.2800484895706177,
"eval_validation_leggepertutti_runtime": 35.7362,
"eval_validation_leggepertutti_samples_per_second": 1.819,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 4500
},
{
"epoch": 0.09,
"eval_validation_altalexprivacy_loss": 1.2601969242095947,
"eval_validation_altalexprivacy_runtime": 58.267,
"eval_validation_altalexprivacy_samples_per_second": 1.819,
"eval_validation_altalexprivacy_steps_per_second": 0.91,
"step": 4500
},
{
"epoch": 0.09,
"eval_validation_mc4_it_loss": 1.8354917764663696,
"eval_validation_mc4_it_runtime": 78.0368,
"eval_validation_mc4_it_samples_per_second": 1.82,
"eval_validation_mc4_it_steps_per_second": 0.91,
"step": 4500
},
{
"epoch": 0.09,
"eval_validation_dolma_loss": 1.9825700521469116,
"eval_validation_dolma_runtime": 53.8422,
"eval_validation_dolma_samples_per_second": 1.82,
"eval_validation_dolma_steps_per_second": 0.91,
"step": 4500
},
{
"epoch": 0.11,
"grad_norm": 6.5,
"learning_rate": 1e-05,
"loss": 1.6194,
"step": 5250
},
{
"epoch": 0.11,
"eval_validation_privacy_sources_loss": 0.8817969560623169,
"eval_validation_privacy_sources_runtime": 210.5387,
"eval_validation_privacy_sources_samples_per_second": 1.819,
"eval_validation_privacy_sources_steps_per_second": 0.912,
"step": 5250
},
{
"epoch": 0.11,
"eval_validation_agenda_digitale_loss": 1.4225614070892334,
"eval_validation_agenda_digitale_runtime": 186.8844,
"eval_validation_agenda_digitale_samples_per_second": 1.819,
"eval_validation_agenda_digitale_steps_per_second": 0.91,
"step": 5250
},
{
"epoch": 0.11,
"eval_validation_leggepertutti_loss": 1.2730649709701538,
"eval_validation_leggepertutti_runtime": 35.7438,
"eval_validation_leggepertutti_samples_per_second": 1.818,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 5250
},
{
"epoch": 0.11,
"eval_validation_altalexprivacy_loss": 1.2494661808013916,
"eval_validation_altalexprivacy_runtime": 58.261,
"eval_validation_altalexprivacy_samples_per_second": 1.819,
"eval_validation_altalexprivacy_steps_per_second": 0.91,
"step": 5250
},
{
"epoch": 0.11,
"eval_validation_mc4_it_loss": 1.8241732120513916,
"eval_validation_mc4_it_runtime": 78.0351,
"eval_validation_mc4_it_samples_per_second": 1.82,
"eval_validation_mc4_it_steps_per_second": 0.91,
"step": 5250
},
{
"epoch": 0.11,
"eval_validation_dolma_loss": 1.9821105003356934,
"eval_validation_dolma_runtime": 53.8348,
"eval_validation_dolma_samples_per_second": 1.82,
"eval_validation_dolma_steps_per_second": 0.91,
"step": 5250
},
{
"epoch": 0.13,
"grad_norm": 10.4375,
"learning_rate": 1e-05,
"loss": 1.5952,
"step": 6000
},
{
"epoch": 0.13,
"eval_validation_privacy_sources_loss": 0.8718611001968384,
"eval_validation_privacy_sources_runtime": 210.5572,
"eval_validation_privacy_sources_samples_per_second": 1.819,
"eval_validation_privacy_sources_steps_per_second": 0.912,
"step": 6000
},
{
"epoch": 0.13,
"eval_validation_agenda_digitale_loss": 1.4152109622955322,
"eval_validation_agenda_digitale_runtime": 186.9175,
"eval_validation_agenda_digitale_samples_per_second": 1.819,
"eval_validation_agenda_digitale_steps_per_second": 0.909,
"step": 6000
},
{
"epoch": 0.13,
"eval_validation_leggepertutti_loss": 1.272850513458252,
"eval_validation_leggepertutti_runtime": 35.7513,
"eval_validation_leggepertutti_samples_per_second": 1.818,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 6000
},
{
"epoch": 0.13,
"eval_validation_altalexprivacy_loss": 1.2443125247955322,
"eval_validation_altalexprivacy_runtime": 58.2787,
"eval_validation_altalexprivacy_samples_per_second": 1.819,
"eval_validation_altalexprivacy_steps_per_second": 0.909,
"step": 6000
},
{
"epoch": 0.13,
"eval_validation_mc4_it_loss": 1.8142365217208862,
"eval_validation_mc4_it_runtime": 78.0695,
"eval_validation_mc4_it_samples_per_second": 1.819,
"eval_validation_mc4_it_steps_per_second": 0.909,
"step": 6000
},
{
"epoch": 0.13,
"eval_validation_dolma_loss": 1.974460482597351,
"eval_validation_dolma_runtime": 53.8685,
"eval_validation_dolma_samples_per_second": 1.819,
"eval_validation_dolma_steps_per_second": 0.91,
"step": 6000
},
{
"epoch": 0.14,
"grad_norm": 6.71875,
"learning_rate": 1e-05,
"loss": 1.5684,
"step": 6750
},
{
"epoch": 0.14,
"eval_validation_privacy_sources_loss": 0.8657892942428589,
"eval_validation_privacy_sources_runtime": 210.5467,
"eval_validation_privacy_sources_samples_per_second": 1.819,
"eval_validation_privacy_sources_steps_per_second": 0.912,
"step": 6750
},
{
"epoch": 0.14,
"eval_validation_agenda_digitale_loss": 1.4074455499649048,
"eval_validation_agenda_digitale_runtime": 186.8991,
"eval_validation_agenda_digitale_samples_per_second": 1.819,
"eval_validation_agenda_digitale_steps_per_second": 0.91,
"step": 6750
},
{
"epoch": 0.14,
"eval_validation_leggepertutti_loss": 1.2634377479553223,
"eval_validation_leggepertutti_runtime": 35.7287,
"eval_validation_leggepertutti_samples_per_second": 1.819,
"eval_validation_leggepertutti_steps_per_second": 0.924,
"step": 6750
},
{
"epoch": 0.14,
"eval_validation_altalexprivacy_loss": 1.237537145614624,
"eval_validation_altalexprivacy_runtime": 58.2781,
"eval_validation_altalexprivacy_samples_per_second": 1.819,
"eval_validation_altalexprivacy_steps_per_second": 0.909,
"step": 6750
},
{
"epoch": 0.14,
"eval_validation_mc4_it_loss": 1.806999921798706,
"eval_validation_mc4_it_runtime": 78.0496,
"eval_validation_mc4_it_samples_per_second": 1.819,
"eval_validation_mc4_it_steps_per_second": 0.91,
"step": 6750
},
{
"epoch": 0.14,
"eval_validation_dolma_loss": 1.9717859029769897,
"eval_validation_dolma_runtime": 53.877,
"eval_validation_dolma_samples_per_second": 1.819,
"eval_validation_dolma_steps_per_second": 0.909,
"step": 6750
},
{
"epoch": 0.16,
"grad_norm": 6.375,
"learning_rate": 1e-05,
"loss": 1.5669,
"step": 7500
},
{
"epoch": 0.16,
"eval_validation_privacy_sources_loss": 0.8547385334968567,
"eval_validation_privacy_sources_runtime": 210.6656,
"eval_validation_privacy_sources_samples_per_second": 1.818,
"eval_validation_privacy_sources_steps_per_second": 0.911,
"step": 7500
},
{
"epoch": 0.16,
"eval_validation_agenda_digitale_loss": 1.4008606672286987,
"eval_validation_agenda_digitale_runtime": 186.9912,
"eval_validation_agenda_digitale_samples_per_second": 1.818,
"eval_validation_agenda_digitale_steps_per_second": 0.909,
"step": 7500
},
{
"epoch": 0.16,
"eval_validation_leggepertutti_loss": 1.2604777812957764,
"eval_validation_leggepertutti_runtime": 35.7381,
"eval_validation_leggepertutti_samples_per_second": 1.819,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 7500
},
{
"epoch": 0.16,
"eval_validation_altalexprivacy_loss": 1.230379343032837,
"eval_validation_altalexprivacy_runtime": 58.2725,
"eval_validation_altalexprivacy_samples_per_second": 1.819,
"eval_validation_altalexprivacy_steps_per_second": 0.91,
"step": 7500
},
{
"epoch": 0.16,
"eval_validation_mc4_it_loss": 1.799519658088684,
"eval_validation_mc4_it_runtime": 78.0653,
"eval_validation_mc4_it_samples_per_second": 1.819,
"eval_validation_mc4_it_steps_per_second": 0.909,
"step": 7500
},
{
"epoch": 0.16,
"eval_validation_dolma_loss": 1.9693360328674316,
"eval_validation_dolma_runtime": 53.9102,
"eval_validation_dolma_samples_per_second": 1.818,
"eval_validation_dolma_steps_per_second": 0.909,
"step": 7500
},
{
"epoch": 0.17,
"grad_norm": 5.90625,
"learning_rate": 1e-05,
"loss": 1.5739,
"step": 8250
},
{
"epoch": 0.17,
"eval_validation_privacy_sources_loss": 0.8504645228385925,
"eval_validation_privacy_sources_runtime": 210.717,
"eval_validation_privacy_sources_samples_per_second": 1.818,
"eval_validation_privacy_sources_steps_per_second": 0.911,
"step": 8250
},
{
"epoch": 0.17,
"eval_validation_agenda_digitale_loss": 1.3965485095977783,
"eval_validation_agenda_digitale_runtime": 187.041,
"eval_validation_agenda_digitale_samples_per_second": 1.818,
"eval_validation_agenda_digitale_steps_per_second": 0.909,
"step": 8250
},
{
"epoch": 0.17,
"eval_validation_leggepertutti_loss": 1.253336787223816,
"eval_validation_leggepertutti_runtime": 35.7643,
"eval_validation_leggepertutti_samples_per_second": 1.817,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 8250
},
{
"epoch": 0.17,
"eval_validation_altalexprivacy_loss": 1.2244532108306885,
"eval_validation_altalexprivacy_runtime": 58.2913,
"eval_validation_altalexprivacy_samples_per_second": 1.818,
"eval_validation_altalexprivacy_steps_per_second": 0.909,
"step": 8250
},
{
"epoch": 0.17,
"eval_validation_mc4_it_loss": 1.7938719987869263,
"eval_validation_mc4_it_runtime": 78.0965,
"eval_validation_mc4_it_samples_per_second": 1.818,
"eval_validation_mc4_it_steps_per_second": 0.909,
"step": 8250
},
{
"epoch": 0.17,
"eval_validation_dolma_loss": 1.9664684534072876,
"eval_validation_dolma_runtime": 53.9155,
"eval_validation_dolma_samples_per_second": 1.818,
"eval_validation_dolma_steps_per_second": 0.909,
"step": 8250
},
{
"epoch": 0.19,
"grad_norm": 6.46875,
"learning_rate": 1e-05,
"loss": 1.562,
"step": 9000
},
{
"epoch": 0.19,
"eval_validation_privacy_sources_loss": 0.8448366522789001,
"eval_validation_privacy_sources_runtime": 210.5401,
"eval_validation_privacy_sources_samples_per_second": 1.819,
"eval_validation_privacy_sources_steps_per_second": 0.912,
"step": 9000
},
{
"epoch": 0.19,
"eval_validation_agenda_digitale_loss": 1.391615867614746,
"eval_validation_agenda_digitale_runtime": 186.8437,
"eval_validation_agenda_digitale_samples_per_second": 1.82,
"eval_validation_agenda_digitale_steps_per_second": 0.91,
"step": 9000
},
{
"epoch": 0.19,
"eval_validation_leggepertutti_loss": 1.2515813112258911,
"eval_validation_leggepertutti_runtime": 35.7517,
"eval_validation_leggepertutti_samples_per_second": 1.818,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 9000
},
{
"epoch": 0.19,
"eval_validation_altalexprivacy_loss": 1.2215769290924072,
"eval_validation_altalexprivacy_runtime": 58.2739,
"eval_validation_altalexprivacy_samples_per_second": 1.819,
"eval_validation_altalexprivacy_steps_per_second": 0.909,
"step": 9000
},
{
"epoch": 0.19,
"eval_validation_mc4_it_loss": 1.7882682085037231,
"eval_validation_mc4_it_runtime": 78.0639,
"eval_validation_mc4_it_samples_per_second": 1.819,
"eval_validation_mc4_it_steps_per_second": 0.91,
"step": 9000
},
{
"epoch": 0.19,
"eval_validation_dolma_loss": 1.965390682220459,
"eval_validation_dolma_runtime": 53.8824,
"eval_validation_dolma_samples_per_second": 1.819,
"eval_validation_dolma_steps_per_second": 0.909,
"step": 9000
},
{
"epoch": 0.21,
"grad_norm": 9.8125,
"learning_rate": 1e-05,
"loss": 1.5531,
"step": 9750
},
{
"epoch": 0.21,
"eval_validation_privacy_sources_loss": 0.8414725065231323,
"eval_validation_privacy_sources_runtime": 210.4886,
"eval_validation_privacy_sources_samples_per_second": 1.82,
"eval_validation_privacy_sources_steps_per_second": 0.912,
"step": 9750
},
{
"epoch": 0.21,
"eval_validation_agenda_digitale_loss": 1.3873320817947388,
"eval_validation_agenda_digitale_runtime": 186.8465,
"eval_validation_agenda_digitale_samples_per_second": 1.82,
"eval_validation_agenda_digitale_steps_per_second": 0.91,
"step": 9750
},
{
"epoch": 0.21,
"eval_validation_leggepertutti_loss": 1.2437468767166138,
"eval_validation_leggepertutti_runtime": 35.7324,
"eval_validation_leggepertutti_samples_per_second": 1.819,
"eval_validation_leggepertutti_steps_per_second": 0.924,
"step": 9750
},
{
"epoch": 0.21,
"eval_validation_altalexprivacy_loss": 1.2164758443832397,
"eval_validation_altalexprivacy_runtime": 58.2539,
"eval_validation_altalexprivacy_samples_per_second": 1.82,
"eval_validation_altalexprivacy_steps_per_second": 0.91,
"step": 9750
},
{
"epoch": 0.21,
"eval_validation_mc4_it_loss": 1.783717155456543,
"eval_validation_mc4_it_runtime": 78.0228,
"eval_validation_mc4_it_samples_per_second": 1.82,
"eval_validation_mc4_it_steps_per_second": 0.91,
"step": 9750
},
{
"epoch": 0.21,
"eval_validation_dolma_loss": 1.9636856317520142,
"eval_validation_dolma_runtime": 53.8632,
"eval_validation_dolma_samples_per_second": 1.819,
"eval_validation_dolma_steps_per_second": 0.91,
"step": 9750
},
{
"epoch": 0.22,
"grad_norm": 6.125,
"learning_rate": 1e-05,
"loss": 1.5595,
"step": 10500
},
{
"epoch": 0.22,
"eval_validation_privacy_sources_loss": 0.8364436030387878,
"eval_validation_privacy_sources_runtime": 210.449,
"eval_validation_privacy_sources_samples_per_second": 1.82,
"eval_validation_privacy_sources_steps_per_second": 0.912,
"step": 10500
},
{
"epoch": 0.22,
"eval_validation_agenda_digitale_loss": 1.3836082220077515,
"eval_validation_agenda_digitale_runtime": 186.8368,
"eval_validation_agenda_digitale_samples_per_second": 1.82,
"eval_validation_agenda_digitale_steps_per_second": 0.91,
"step": 10500
},
{
"epoch": 0.22,
"eval_validation_leggepertutti_loss": 1.2467608451843262,
"eval_validation_leggepertutti_runtime": 35.73,
"eval_validation_leggepertutti_samples_per_second": 1.819,
"eval_validation_leggepertutti_steps_per_second": 0.924,
"step": 10500
},
{
"epoch": 0.22,
"eval_validation_altalexprivacy_loss": 1.2111377716064453,
"eval_validation_altalexprivacy_runtime": 58.2474,
"eval_validation_altalexprivacy_samples_per_second": 1.82,
"eval_validation_altalexprivacy_steps_per_second": 0.91,
"step": 10500
},
{
"epoch": 0.22,
"eval_validation_mc4_it_loss": 1.7798333168029785,
"eval_validation_mc4_it_runtime": 78.0289,
"eval_validation_mc4_it_samples_per_second": 1.82,
"eval_validation_mc4_it_steps_per_second": 0.91,
"step": 10500
},
{
"epoch": 0.22,
"eval_validation_dolma_loss": 1.9624452590942383,
"eval_validation_dolma_runtime": 53.8564,
"eval_validation_dolma_samples_per_second": 1.82,
"eval_validation_dolma_steps_per_second": 0.91,
"step": 10500
},
{
"epoch": 0.24,
"grad_norm": 6.1875,
"learning_rate": 1e-05,
"loss": 1.5657,
"step": 11250
},
{
"epoch": 0.24,
"eval_validation_privacy_sources_loss": 0.8329368829727173,
"eval_validation_privacy_sources_runtime": 210.663,
"eval_validation_privacy_sources_samples_per_second": 1.818,
"eval_validation_privacy_sources_steps_per_second": 0.911,
"step": 11250
},
{
"epoch": 0.24,
"eval_validation_agenda_digitale_loss": 1.3799829483032227,
"eval_validation_agenda_digitale_runtime": 186.9433,
"eval_validation_agenda_digitale_samples_per_second": 1.819,
"eval_validation_agenda_digitale_steps_per_second": 0.909,
"step": 11250
},
{
"epoch": 0.24,
"eval_validation_leggepertutti_loss": 1.2420663833618164,
"eval_validation_leggepertutti_runtime": 35.7651,
"eval_validation_leggepertutti_samples_per_second": 1.817,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 11250
},
{
"epoch": 0.24,
"eval_validation_altalexprivacy_loss": 1.206700325012207,
"eval_validation_altalexprivacy_runtime": 58.327,
"eval_validation_altalexprivacy_samples_per_second": 1.817,
"eval_validation_altalexprivacy_steps_per_second": 0.909,
"step": 11250
},
{
"epoch": 0.24,
"eval_validation_mc4_it_loss": 1.7747920751571655,
"eval_validation_mc4_it_runtime": 78.1384,
"eval_validation_mc4_it_samples_per_second": 1.817,
"eval_validation_mc4_it_steps_per_second": 0.909,
"step": 11250
},
{
"epoch": 0.24,
"eval_validation_dolma_loss": 1.957626461982727,
"eval_validation_dolma_runtime": 53.9238,
"eval_validation_dolma_samples_per_second": 1.817,
"eval_validation_dolma_steps_per_second": 0.909,
"step": 11250
},
{
"epoch": 0.25,
"grad_norm": 5.90625,
"learning_rate": 1e-05,
"loss": 1.564,
"step": 12000
},
{
"epoch": 0.25,
"eval_validation_privacy_sources_loss": 0.8280984163284302,
"eval_validation_privacy_sources_runtime": 210.585,
"eval_validation_privacy_sources_samples_per_second": 1.819,
"eval_validation_privacy_sources_steps_per_second": 0.912,
"step": 12000
},
{
"epoch": 0.25,
"eval_validation_agenda_digitale_loss": 1.3771744966506958,
"eval_validation_agenda_digitale_runtime": 186.9251,
"eval_validation_agenda_digitale_samples_per_second": 1.819,
"eval_validation_agenda_digitale_steps_per_second": 0.909,
"step": 12000
},
{
"epoch": 0.25,
"eval_validation_leggepertutti_loss": 1.2391917705535889,
"eval_validation_leggepertutti_runtime": 35.7521,
"eval_validation_leggepertutti_samples_per_second": 1.818,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 12000
},
{
"epoch": 0.25,
"eval_validation_altalexprivacy_loss": 1.2027921676635742,
"eval_validation_altalexprivacy_runtime": 58.2673,
"eval_validation_altalexprivacy_samples_per_second": 1.819,
"eval_validation_altalexprivacy_steps_per_second": 0.91,
"step": 12000
},
{
"epoch": 0.25,
"eval_validation_mc4_it_loss": 1.772462248802185,
"eval_validation_mc4_it_runtime": 78.1049,
"eval_validation_mc4_it_samples_per_second": 1.818,
"eval_validation_mc4_it_steps_per_second": 0.909,
"step": 12000
},
{
"epoch": 0.25,
"eval_validation_dolma_loss": 1.957063913345337,
"eval_validation_dolma_runtime": 53.8823,
"eval_validation_dolma_samples_per_second": 1.819,
"eval_validation_dolma_steps_per_second": 0.909,
"step": 12000
}
],
"logging_steps": 750,
"max_steps": 94880,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 750,
"total_flos": 1.2582113941389312e+19,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}