LexLLMv0.0.0.x.10.19b / trainer_state.json
andrealexroom's picture
Upload folder using huggingface_hub
24a8f12 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.4089421935197287,
"eval_steps": 187,
"global_step": 2992,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"grad_norm": 4.8125,
"learning_rate": 1e-05,
"loss": 1.2442,
"step": 1
},
{
"epoch": 0.03,
"grad_norm": 5.46875,
"learning_rate": 1e-05,
"loss": 1.1455,
"step": 187
},
{
"epoch": 0.03,
"eval_validation_privacy_sources_loss": 0.9909348487854004,
"eval_validation_privacy_sources_runtime": 210.9899,
"eval_validation_privacy_sources_samples_per_second": 1.815,
"eval_validation_privacy_sources_steps_per_second": 0.91,
"step": 187
},
{
"epoch": 0.03,
"eval_validation_agenda_digitale_loss": 1.5858619213104248,
"eval_validation_agenda_digitale_runtime": 187.2047,
"eval_validation_agenda_digitale_samples_per_second": 1.816,
"eval_validation_agenda_digitale_steps_per_second": 0.908,
"step": 187
},
{
"epoch": 0.03,
"eval_validation_leggepertutti_loss": 1.3584009408950806,
"eval_validation_leggepertutti_runtime": 35.7769,
"eval_validation_leggepertutti_samples_per_second": 1.817,
"eval_validation_leggepertutti_steps_per_second": 0.922,
"step": 187
},
{
"epoch": 0.03,
"eval_validation_altalexprivacy_loss": 1.3655166625976562,
"eval_validation_altalexprivacy_runtime": 58.3249,
"eval_validation_altalexprivacy_samples_per_second": 1.817,
"eval_validation_altalexprivacy_steps_per_second": 0.909,
"step": 187
},
{
"epoch": 0.03,
"eval_validation_stack_loss": 0.5954318642616272,
"eval_validation_stack_runtime": 190.5195,
"eval_validation_stack_samples_per_second": 1.816,
"eval_validation_stack_steps_per_second": 0.908,
"step": 187
},
{
"epoch": 0.05,
"grad_norm": 5.21875,
"learning_rate": 1e-05,
"loss": 1.0615,
"step": 374
},
{
"epoch": 0.05,
"eval_validation_privacy_sources_loss": 0.9419633746147156,
"eval_validation_privacy_sources_runtime": 210.7968,
"eval_validation_privacy_sources_samples_per_second": 1.817,
"eval_validation_privacy_sources_steps_per_second": 0.911,
"step": 374
},
{
"epoch": 0.05,
"eval_validation_agenda_digitale_loss": 1.561063528060913,
"eval_validation_agenda_digitale_runtime": 187.1864,
"eval_validation_agenda_digitale_samples_per_second": 1.816,
"eval_validation_agenda_digitale_steps_per_second": 0.908,
"step": 374
},
{
"epoch": 0.05,
"eval_validation_leggepertutti_loss": 1.3387541770935059,
"eval_validation_leggepertutti_runtime": 35.7704,
"eval_validation_leggepertutti_samples_per_second": 1.817,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 374
},
{
"epoch": 0.05,
"eval_validation_altalexprivacy_loss": 1.32594895362854,
"eval_validation_altalexprivacy_runtime": 58.3358,
"eval_validation_altalexprivacy_samples_per_second": 1.817,
"eval_validation_altalexprivacy_steps_per_second": 0.909,
"step": 374
},
{
"epoch": 0.05,
"eval_validation_stack_loss": 0.5792369246482849,
"eval_validation_stack_runtime": 190.4571,
"eval_validation_stack_samples_per_second": 1.817,
"eval_validation_stack_steps_per_second": 0.908,
"step": 374
},
{
"epoch": 0.08,
"grad_norm": 3.421875,
"learning_rate": 1e-05,
"loss": 1.0133,
"step": 561
},
{
"epoch": 0.08,
"eval_validation_privacy_sources_loss": 0.9125503897666931,
"eval_validation_privacy_sources_runtime": 210.7361,
"eval_validation_privacy_sources_samples_per_second": 1.817,
"eval_validation_privacy_sources_steps_per_second": 0.911,
"step": 561
},
{
"epoch": 0.08,
"eval_validation_agenda_digitale_loss": 1.5489054918289185,
"eval_validation_agenda_digitale_runtime": 187.1329,
"eval_validation_agenda_digitale_samples_per_second": 1.817,
"eval_validation_agenda_digitale_steps_per_second": 0.908,
"step": 561
},
{
"epoch": 0.08,
"eval_validation_leggepertutti_loss": 1.3227838277816772,
"eval_validation_leggepertutti_runtime": 35.7601,
"eval_validation_leggepertutti_samples_per_second": 1.818,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 561
},
{
"epoch": 0.08,
"eval_validation_altalexprivacy_loss": 1.305092453956604,
"eval_validation_altalexprivacy_runtime": 58.3163,
"eval_validation_altalexprivacy_samples_per_second": 1.818,
"eval_validation_altalexprivacy_steps_per_second": 0.909,
"step": 561
},
{
"epoch": 0.08,
"eval_validation_stack_loss": 0.569362461566925,
"eval_validation_stack_runtime": 190.4309,
"eval_validation_stack_samples_per_second": 1.817,
"eval_validation_stack_steps_per_second": 0.908,
"step": 561
},
{
"epoch": 0.1,
"grad_norm": 2.828125,
"learning_rate": 1e-05,
"loss": 0.9878,
"step": 748
},
{
"epoch": 0.1,
"eval_validation_privacy_sources_loss": 0.8928124904632568,
"eval_validation_privacy_sources_runtime": 210.6755,
"eval_validation_privacy_sources_samples_per_second": 1.818,
"eval_validation_privacy_sources_steps_per_second": 0.911,
"step": 748
},
{
"epoch": 0.1,
"eval_validation_agenda_digitale_loss": 1.5373064279556274,
"eval_validation_agenda_digitale_runtime": 187.1042,
"eval_validation_agenda_digitale_samples_per_second": 1.817,
"eval_validation_agenda_digitale_steps_per_second": 0.909,
"step": 748
},
{
"epoch": 0.1,
"eval_validation_leggepertutti_loss": 1.3180450201034546,
"eval_validation_leggepertutti_runtime": 35.7671,
"eval_validation_leggepertutti_samples_per_second": 1.817,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 748
},
{
"epoch": 0.1,
"eval_validation_altalexprivacy_loss": 1.2893434762954712,
"eval_validation_altalexprivacy_runtime": 58.3327,
"eval_validation_altalexprivacy_samples_per_second": 1.817,
"eval_validation_altalexprivacy_steps_per_second": 0.909,
"step": 748
},
{
"epoch": 0.1,
"eval_validation_stack_loss": 0.555778980255127,
"eval_validation_stack_runtime": 190.4189,
"eval_validation_stack_samples_per_second": 1.817,
"eval_validation_stack_steps_per_second": 0.909,
"step": 748
},
{
"epoch": 0.13,
"grad_norm": 3.0,
"learning_rate": 1e-05,
"loss": 0.9849,
"step": 935
},
{
"epoch": 0.13,
"eval_validation_privacy_sources_loss": 0.8825064301490784,
"eval_validation_privacy_sources_runtime": 210.6896,
"eval_validation_privacy_sources_samples_per_second": 1.818,
"eval_validation_privacy_sources_steps_per_second": 0.911,
"step": 935
},
{
"epoch": 0.13,
"eval_validation_agenda_digitale_loss": 1.5294857025146484,
"eval_validation_agenda_digitale_runtime": 187.007,
"eval_validation_agenda_digitale_samples_per_second": 1.818,
"eval_validation_agenda_digitale_steps_per_second": 0.909,
"step": 935
},
{
"epoch": 0.13,
"eval_validation_leggepertutti_loss": 1.3130334615707397,
"eval_validation_leggepertutti_runtime": 35.7713,
"eval_validation_leggepertutti_samples_per_second": 1.817,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 935
},
{
"epoch": 0.13,
"eval_validation_altalexprivacy_loss": 1.2819935083389282,
"eval_validation_altalexprivacy_runtime": 58.2912,
"eval_validation_altalexprivacy_samples_per_second": 1.818,
"eval_validation_altalexprivacy_steps_per_second": 0.909,
"step": 935
},
{
"epoch": 0.13,
"eval_validation_stack_loss": 0.5473493337631226,
"eval_validation_stack_runtime": 190.3335,
"eval_validation_stack_samples_per_second": 1.818,
"eval_validation_stack_steps_per_second": 0.909,
"step": 935
},
{
"epoch": 0.15,
"grad_norm": 2.84375,
"learning_rate": 1e-05,
"loss": 0.949,
"step": 1122
},
{
"epoch": 0.15,
"eval_validation_privacy_sources_loss": 0.8628236651420593,
"eval_validation_privacy_sources_runtime": 210.6338,
"eval_validation_privacy_sources_samples_per_second": 1.818,
"eval_validation_privacy_sources_steps_per_second": 0.912,
"step": 1122
},
{
"epoch": 0.15,
"eval_validation_agenda_digitale_loss": 1.5227819681167603,
"eval_validation_agenda_digitale_runtime": 187.0513,
"eval_validation_agenda_digitale_samples_per_second": 1.818,
"eval_validation_agenda_digitale_steps_per_second": 0.909,
"step": 1122
},
{
"epoch": 0.15,
"eval_validation_leggepertutti_loss": 1.3133653402328491,
"eval_validation_leggepertutti_runtime": 35.7704,
"eval_validation_leggepertutti_samples_per_second": 1.817,
"eval_validation_leggepertutti_steps_per_second": 0.923,
"step": 1122
},
{
"epoch": 0.15,
"eval_validation_altalexprivacy_loss": 1.270833134651184,
"eval_validation_altalexprivacy_runtime": 58.3107,
"eval_validation_altalexprivacy_samples_per_second": 1.818,
"eval_validation_altalexprivacy_steps_per_second": 0.909,
"step": 1122
},
{
"epoch": 0.15,
"eval_validation_stack_loss": 0.5453876852989197,
"eval_validation_stack_runtime": 190.3729,
"eval_validation_stack_samples_per_second": 1.817,
"eval_validation_stack_steps_per_second": 0.909,
"step": 1122
},
{
"epoch": 0.18,
"grad_norm": 3.0,
"learning_rate": 1e-05,
"loss": 0.9454,
"step": 1309
},
{
"epoch": 0.18,
"eval_validation_privacy_sources_loss": 0.8534297347068787,
"eval_validation_privacy_sources_runtime": 211.0071,
"eval_validation_privacy_sources_samples_per_second": 1.815,
"eval_validation_privacy_sources_steps_per_second": 0.91,
"step": 1309
},
{
"epoch": 0.18,
"eval_validation_agenda_digitale_loss": 1.5153428316116333,
"eval_validation_agenda_digitale_runtime": 187.3358,
"eval_validation_agenda_digitale_samples_per_second": 1.815,
"eval_validation_agenda_digitale_steps_per_second": 0.907,
"step": 1309
},
{
"epoch": 0.18,
"eval_validation_leggepertutti_loss": 1.3075683116912842,
"eval_validation_leggepertutti_runtime": 35.8327,
"eval_validation_leggepertutti_samples_per_second": 1.814,
"eval_validation_leggepertutti_steps_per_second": 0.921,
"step": 1309
},
{
"epoch": 0.18,
"eval_validation_altalexprivacy_loss": 1.2597737312316895,
"eval_validation_altalexprivacy_runtime": 58.4031,
"eval_validation_altalexprivacy_samples_per_second": 1.815,
"eval_validation_altalexprivacy_steps_per_second": 0.907,
"step": 1309
},
{
"epoch": 0.18,
"eval_validation_stack_loss": 0.53725665807724,
"eval_validation_stack_runtime": 190.6517,
"eval_validation_stack_samples_per_second": 1.815,
"eval_validation_stack_steps_per_second": 0.907,
"step": 1309
},
{
"epoch": 0.2,
"grad_norm": 3.171875,
"learning_rate": 1e-05,
"loss": 0.9365,
"step": 1496
},
{
"epoch": 0.2,
"eval_validation_privacy_sources_loss": 0.8417152762413025,
"eval_validation_privacy_sources_runtime": 210.9207,
"eval_validation_privacy_sources_samples_per_second": 1.816,
"eval_validation_privacy_sources_steps_per_second": 0.91,
"step": 1496
},
{
"epoch": 0.2,
"eval_validation_agenda_digitale_loss": 1.5109071731567383,
"eval_validation_agenda_digitale_runtime": 187.2506,
"eval_validation_agenda_digitale_samples_per_second": 1.816,
"eval_validation_agenda_digitale_steps_per_second": 0.908,
"step": 1496
},
{
"epoch": 0.2,
"eval_validation_leggepertutti_loss": 1.3020944595336914,
"eval_validation_leggepertutti_runtime": 35.811,
"eval_validation_leggepertutti_samples_per_second": 1.815,
"eval_validation_leggepertutti_steps_per_second": 0.922,
"step": 1496
},
{
"epoch": 0.2,
"eval_validation_altalexprivacy_loss": 1.2524465322494507,
"eval_validation_altalexprivacy_runtime": 58.3785,
"eval_validation_altalexprivacy_samples_per_second": 1.816,
"eval_validation_altalexprivacy_steps_per_second": 0.908,
"step": 1496
},
{
"epoch": 0.2,
"eval_validation_stack_loss": 0.5389564037322998,
"eval_validation_stack_runtime": 190.5717,
"eval_validation_stack_samples_per_second": 1.816,
"eval_validation_stack_steps_per_second": 0.908,
"step": 1496
},
{
"epoch": 0.23,
"grad_norm": 2.734375,
"learning_rate": 1e-05,
"loss": 0.9186,
"step": 1683
},
{
"epoch": 0.23,
"eval_validation_privacy_sources_loss": 0.8329127430915833,
"eval_validation_privacy_sources_runtime": 211.0554,
"eval_validation_privacy_sources_samples_per_second": 1.815,
"eval_validation_privacy_sources_steps_per_second": 0.91,
"step": 1683
},
{
"epoch": 0.23,
"eval_validation_agenda_digitale_loss": 1.505892038345337,
"eval_validation_agenda_digitale_runtime": 187.3144,
"eval_validation_agenda_digitale_samples_per_second": 1.815,
"eval_validation_agenda_digitale_steps_per_second": 0.908,
"step": 1683
},
{
"epoch": 0.23,
"eval_validation_leggepertutti_loss": 1.296944260597229,
"eval_validation_leggepertutti_runtime": 35.8059,
"eval_validation_leggepertutti_samples_per_second": 1.815,
"eval_validation_leggepertutti_steps_per_second": 0.922,
"step": 1683
},
{
"epoch": 0.23,
"eval_validation_altalexprivacy_loss": 1.241568684577942,
"eval_validation_altalexprivacy_runtime": 58.381,
"eval_validation_altalexprivacy_samples_per_second": 1.816,
"eval_validation_altalexprivacy_steps_per_second": 0.908,
"step": 1683
},
{
"epoch": 0.23,
"eval_validation_stack_loss": 0.5280548334121704,
"eval_validation_stack_runtime": 190.6339,
"eval_validation_stack_samples_per_second": 1.815,
"eval_validation_stack_steps_per_second": 0.907,
"step": 1683
},
{
"epoch": 0.26,
"grad_norm": 2.765625,
"learning_rate": 1e-05,
"loss": 0.9282,
"step": 1870
},
{
"epoch": 0.26,
"eval_validation_privacy_sources_loss": 0.8271979689598083,
"eval_validation_privacy_sources_runtime": 211.0996,
"eval_validation_privacy_sources_samples_per_second": 1.814,
"eval_validation_privacy_sources_steps_per_second": 0.91,
"step": 1870
},
{
"epoch": 0.26,
"eval_validation_agenda_digitale_loss": 1.5014517307281494,
"eval_validation_agenda_digitale_runtime": 187.3966,
"eval_validation_agenda_digitale_samples_per_second": 1.814,
"eval_validation_agenda_digitale_steps_per_second": 0.907,
"step": 1870
},
{
"epoch": 0.26,
"eval_validation_leggepertutti_loss": 1.2852504253387451,
"eval_validation_leggepertutti_runtime": 35.8382,
"eval_validation_leggepertutti_samples_per_second": 1.814,
"eval_validation_leggepertutti_steps_per_second": 0.921,
"step": 1870
},
{
"epoch": 0.26,
"eval_validation_altalexprivacy_loss": 1.231817603111267,
"eval_validation_altalexprivacy_runtime": 58.4131,
"eval_validation_altalexprivacy_samples_per_second": 1.815,
"eval_validation_altalexprivacy_steps_per_second": 0.907,
"step": 1870
},
{
"epoch": 0.26,
"eval_validation_stack_loss": 0.5229911208152771,
"eval_validation_stack_runtime": 190.7066,
"eval_validation_stack_samples_per_second": 1.814,
"eval_validation_stack_steps_per_second": 0.907,
"step": 1870
},
{
"epoch": 0.28,
"grad_norm": 3.1875,
"learning_rate": 1e-05,
"loss": 0.9263,
"step": 2057
},
{
"epoch": 0.28,
"eval_validation_privacy_sources_loss": 0.8199824690818787,
"eval_validation_privacy_sources_runtime": 211.0532,
"eval_validation_privacy_sources_samples_per_second": 1.815,
"eval_validation_privacy_sources_steps_per_second": 0.91,
"step": 2057
},
{
"epoch": 0.28,
"eval_validation_agenda_digitale_loss": 1.4965113401412964,
"eval_validation_agenda_digitale_runtime": 187.3679,
"eval_validation_agenda_digitale_samples_per_second": 1.815,
"eval_validation_agenda_digitale_steps_per_second": 0.907,
"step": 2057
},
{
"epoch": 0.28,
"eval_validation_leggepertutti_loss": 1.2861623764038086,
"eval_validation_leggepertutti_runtime": 35.8184,
"eval_validation_leggepertutti_samples_per_second": 1.815,
"eval_validation_leggepertutti_steps_per_second": 0.921,
"step": 2057
},
{
"epoch": 0.28,
"eval_validation_altalexprivacy_loss": 1.2276028394699097,
"eval_validation_altalexprivacy_runtime": 58.4047,
"eval_validation_altalexprivacy_samples_per_second": 1.815,
"eval_validation_altalexprivacy_steps_per_second": 0.907,
"step": 2057
},
{
"epoch": 0.28,
"eval_validation_stack_loss": 0.5233271718025208,
"eval_validation_stack_runtime": 190.6786,
"eval_validation_stack_samples_per_second": 1.815,
"eval_validation_stack_steps_per_second": 0.907,
"step": 2057
},
{
"epoch": 0.31,
"grad_norm": 3.03125,
"learning_rate": 1e-05,
"loss": 0.9055,
"step": 2244
},
{
"epoch": 0.31,
"eval_validation_privacy_sources_loss": 0.8143637180328369,
"eval_validation_privacy_sources_runtime": 210.9307,
"eval_validation_privacy_sources_samples_per_second": 1.816,
"eval_validation_privacy_sources_steps_per_second": 0.91,
"step": 2244
},
{
"epoch": 0.31,
"eval_validation_agenda_digitale_loss": 1.4932323694229126,
"eval_validation_agenda_digitale_runtime": 187.2867,
"eval_validation_agenda_digitale_samples_per_second": 1.815,
"eval_validation_agenda_digitale_steps_per_second": 0.908,
"step": 2244
},
{
"epoch": 0.31,
"eval_validation_leggepertutti_loss": 1.2840496301651,
"eval_validation_leggepertutti_runtime": 35.7977,
"eval_validation_leggepertutti_samples_per_second": 1.816,
"eval_validation_leggepertutti_steps_per_second": 0.922,
"step": 2244
},
{
"epoch": 0.31,
"eval_validation_altalexprivacy_loss": 1.2231019735336304,
"eval_validation_altalexprivacy_runtime": 58.3774,
"eval_validation_altalexprivacy_samples_per_second": 1.816,
"eval_validation_altalexprivacy_steps_per_second": 0.908,
"step": 2244
},
{
"epoch": 0.31,
"eval_validation_stack_loss": 0.5161340236663818,
"eval_validation_stack_runtime": 190.6053,
"eval_validation_stack_samples_per_second": 1.815,
"eval_validation_stack_steps_per_second": 0.908,
"step": 2244
},
{
"epoch": 0.33,
"grad_norm": 2.5625,
"learning_rate": 1e-05,
"loss": 0.9028,
"step": 2431
},
{
"epoch": 0.33,
"eval_validation_privacy_sources_loss": 0.8075902462005615,
"eval_validation_privacy_sources_runtime": 210.9173,
"eval_validation_privacy_sources_samples_per_second": 1.816,
"eval_validation_privacy_sources_steps_per_second": 0.91,
"step": 2431
},
{
"epoch": 0.33,
"eval_validation_agenda_digitale_loss": 1.4887945652008057,
"eval_validation_agenda_digitale_runtime": 187.2535,
"eval_validation_agenda_digitale_samples_per_second": 1.816,
"eval_validation_agenda_digitale_steps_per_second": 0.908,
"step": 2431
},
{
"epoch": 0.33,
"eval_validation_leggepertutti_loss": 1.2816591262817383,
"eval_validation_leggepertutti_runtime": 35.8013,
"eval_validation_leggepertutti_samples_per_second": 1.816,
"eval_validation_leggepertutti_steps_per_second": 0.922,
"step": 2431
},
{
"epoch": 0.33,
"eval_validation_altalexprivacy_loss": 1.2183953523635864,
"eval_validation_altalexprivacy_runtime": 58.3818,
"eval_validation_altalexprivacy_samples_per_second": 1.816,
"eval_validation_altalexprivacy_steps_per_second": 0.908,
"step": 2431
},
{
"epoch": 0.33,
"eval_validation_stack_loss": 0.5140035152435303,
"eval_validation_stack_runtime": 190.5309,
"eval_validation_stack_samples_per_second": 1.816,
"eval_validation_stack_steps_per_second": 0.908,
"step": 2431
},
{
"epoch": 0.36,
"grad_norm": 3.171875,
"learning_rate": 1e-05,
"loss": 0.886,
"step": 2618
},
{
"epoch": 0.36,
"eval_validation_privacy_sources_loss": 0.802619218826294,
"eval_validation_privacy_sources_runtime": 210.8958,
"eval_validation_privacy_sources_samples_per_second": 1.816,
"eval_validation_privacy_sources_steps_per_second": 0.91,
"step": 2618
},
{
"epoch": 0.36,
"eval_validation_agenda_digitale_loss": 1.4873639345169067,
"eval_validation_agenda_digitale_runtime": 187.2571,
"eval_validation_agenda_digitale_samples_per_second": 1.816,
"eval_validation_agenda_digitale_steps_per_second": 0.908,
"step": 2618
},
{
"epoch": 0.36,
"eval_validation_leggepertutti_loss": 1.278908133506775,
"eval_validation_leggepertutti_runtime": 35.8133,
"eval_validation_leggepertutti_samples_per_second": 1.815,
"eval_validation_leggepertutti_steps_per_second": 0.921,
"step": 2618
},
{
"epoch": 0.36,
"eval_validation_altalexprivacy_loss": 1.217095136642456,
"eval_validation_altalexprivacy_runtime": 58.3847,
"eval_validation_altalexprivacy_samples_per_second": 1.816,
"eval_validation_altalexprivacy_steps_per_second": 0.908,
"step": 2618
},
{
"epoch": 0.36,
"eval_validation_stack_loss": 0.5101712346076965,
"eval_validation_stack_runtime": 190.5834,
"eval_validation_stack_samples_per_second": 1.815,
"eval_validation_stack_steps_per_second": 0.908,
"step": 2618
},
{
"epoch": 0.38,
"grad_norm": 3.78125,
"learning_rate": 1e-05,
"loss": 0.8815,
"step": 2805
},
{
"epoch": 0.38,
"eval_validation_privacy_sources_loss": 0.796984076499939,
"eval_validation_privacy_sources_runtime": 210.8742,
"eval_validation_privacy_sources_samples_per_second": 1.816,
"eval_validation_privacy_sources_steps_per_second": 0.91,
"step": 2805
},
{
"epoch": 0.38,
"eval_validation_agenda_digitale_loss": 1.4829306602478027,
"eval_validation_agenda_digitale_runtime": 187.223,
"eval_validation_agenda_digitale_samples_per_second": 1.816,
"eval_validation_agenda_digitale_steps_per_second": 0.908,
"step": 2805
},
{
"epoch": 0.38,
"eval_validation_leggepertutti_loss": 1.2758851051330566,
"eval_validation_leggepertutti_runtime": 35.8041,
"eval_validation_leggepertutti_samples_per_second": 1.815,
"eval_validation_leggepertutti_steps_per_second": 0.922,
"step": 2805
},
{
"epoch": 0.38,
"eval_validation_altalexprivacy_loss": 1.2099803686141968,
"eval_validation_altalexprivacy_runtime": 58.3594,
"eval_validation_altalexprivacy_samples_per_second": 1.816,
"eval_validation_altalexprivacy_steps_per_second": 0.908,
"step": 2805
},
{
"epoch": 0.38,
"eval_validation_stack_loss": 0.505918562412262,
"eval_validation_stack_runtime": 190.523,
"eval_validation_stack_samples_per_second": 1.816,
"eval_validation_stack_steps_per_second": 0.908,
"step": 2805
},
{
"epoch": 0.41,
"grad_norm": 2.90625,
"learning_rate": 1e-05,
"loss": 0.8852,
"step": 2992
},
{
"epoch": 0.41,
"eval_validation_privacy_sources_loss": 0.7935929894447327,
"eval_validation_privacy_sources_runtime": 210.8218,
"eval_validation_privacy_sources_samples_per_second": 1.817,
"eval_validation_privacy_sources_steps_per_second": 0.911,
"step": 2992
},
{
"epoch": 0.41,
"eval_validation_agenda_digitale_loss": 1.4817965030670166,
"eval_validation_agenda_digitale_runtime": 187.1739,
"eval_validation_agenda_digitale_samples_per_second": 1.816,
"eval_validation_agenda_digitale_steps_per_second": 0.908,
"step": 2992
},
{
"epoch": 0.41,
"eval_validation_leggepertutti_loss": 1.2713061571121216,
"eval_validation_leggepertutti_runtime": 35.7922,
"eval_validation_leggepertutti_samples_per_second": 1.816,
"eval_validation_leggepertutti_steps_per_second": 0.922,
"step": 2992
},
{
"epoch": 0.41,
"eval_validation_altalexprivacy_loss": 1.2076174020767212,
"eval_validation_altalexprivacy_runtime": 58.3429,
"eval_validation_altalexprivacy_samples_per_second": 1.817,
"eval_validation_altalexprivacy_steps_per_second": 0.908,
"step": 2992
},
{
"epoch": 0.41,
"eval_validation_stack_loss": 0.505508542060852,
"eval_validation_stack_runtime": 190.5073,
"eval_validation_stack_samples_per_second": 1.816,
"eval_validation_stack_steps_per_second": 0.908,
"step": 2992
}
],
"logging_steps": 187,
"max_steps": 14632,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 187,
"total_flos": 1.2548561637545607e+19,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}