electra-large-squad / trainer_state.json
Ubuntu
initial upload
b200bde
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 5.0,
"global_step": 27665,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.09,
"learning_rate": 2.9457798662570034e-05,
"loss": 1.4635,
"step": 500
},
{
"epoch": 0.18,
"learning_rate": 2.891559732514007e-05,
"loss": 1.036,
"step": 1000
},
{
"epoch": 0.27,
"learning_rate": 2.83733959877101e-05,
"loss": 0.8904,
"step": 1500
},
{
"epoch": 0.36,
"learning_rate": 2.783119465028014e-05,
"loss": 0.8604,
"step": 2000
},
{
"epoch": 0.45,
"learning_rate": 2.7288993312850172e-05,
"loss": 0.8931,
"step": 2500
},
{
"epoch": 0.54,
"learning_rate": 2.6746791975420206e-05,
"loss": 0.8083,
"step": 3000
},
{
"epoch": 0.63,
"learning_rate": 2.6204590637990243e-05,
"loss": 0.8035,
"step": 3500
},
{
"epoch": 0.72,
"learning_rate": 2.5662389300560273e-05,
"loss": 0.8022,
"step": 4000
},
{
"epoch": 0.81,
"learning_rate": 2.512018796313031e-05,
"loss": 0.8007,
"step": 4500
},
{
"epoch": 0.9,
"learning_rate": 2.4577986625700344e-05,
"loss": 0.7696,
"step": 5000
},
{
"epoch": 0.99,
"learning_rate": 2.4035785288270377e-05,
"loss": 0.8287,
"step": 5500
},
{
"epoch": 1.08,
"learning_rate": 2.3493583950840414e-05,
"loss": 0.5808,
"step": 6000
},
{
"epoch": 1.17,
"learning_rate": 2.2951382613410445e-05,
"loss": 0.5755,
"step": 6500
},
{
"epoch": 1.27,
"learning_rate": 2.2409181275980482e-05,
"loss": 0.5957,
"step": 7000
},
{
"epoch": 1.36,
"learning_rate": 2.1866979938550515e-05,
"loss": 0.5736,
"step": 7500
},
{
"epoch": 1.45,
"learning_rate": 2.132477860112055e-05,
"loss": 0.5806,
"step": 8000
},
{
"epoch": 1.54,
"learning_rate": 2.0782577263690586e-05,
"loss": 0.6211,
"step": 8500
},
{
"epoch": 1.63,
"learning_rate": 2.024037592626062e-05,
"loss": 0.5684,
"step": 9000
},
{
"epoch": 1.72,
"learning_rate": 1.9698174588830653e-05,
"loss": 0.5927,
"step": 9500
},
{
"epoch": 1.81,
"learning_rate": 1.9155973251400687e-05,
"loss": 0.5779,
"step": 10000
},
{
"epoch": 1.9,
"learning_rate": 1.861377191397072e-05,
"loss": 0.5363,
"step": 10500
},
{
"epoch": 1.99,
"learning_rate": 1.8071570576540758e-05,
"loss": 0.5872,
"step": 11000
},
{
"epoch": 2.08,
"learning_rate": 1.752936923911079e-05,
"loss": 0.3776,
"step": 11500
},
{
"epoch": 2.17,
"learning_rate": 1.6987167901680825e-05,
"loss": 0.3865,
"step": 12000
},
{
"epoch": 2.26,
"learning_rate": 1.644496656425086e-05,
"loss": 0.3685,
"step": 12500
},
{
"epoch": 2.35,
"learning_rate": 1.5902765226820892e-05,
"loss": 0.3677,
"step": 13000
},
{
"epoch": 2.44,
"learning_rate": 1.536056388939093e-05,
"loss": 0.3914,
"step": 13500
},
{
"epoch": 2.53,
"learning_rate": 1.4818362551960961e-05,
"loss": 0.3858,
"step": 14000
},
{
"epoch": 2.62,
"learning_rate": 1.4276161214530996e-05,
"loss": 0.3726,
"step": 14500
},
{
"epoch": 2.71,
"learning_rate": 1.373395987710103e-05,
"loss": 0.3758,
"step": 15000
},
{
"epoch": 2.8,
"learning_rate": 1.3191758539671065e-05,
"loss": 0.389,
"step": 15500
},
{
"epoch": 2.89,
"learning_rate": 1.26495572022411e-05,
"loss": 0.3689,
"step": 16000
},
{
"epoch": 2.98,
"learning_rate": 1.2107355864811133e-05,
"loss": 0.3964,
"step": 16500
},
{
"epoch": 3.07,
"learning_rate": 1.1565154527381168e-05,
"loss": 0.2548,
"step": 17000
},
{
"epoch": 3.16,
"learning_rate": 1.1022953189951202e-05,
"loss": 0.2504,
"step": 17500
},
{
"epoch": 3.25,
"learning_rate": 1.0480751852521237e-05,
"loss": 0.2285,
"step": 18000
},
{
"epoch": 3.34,
"learning_rate": 9.938550515091272e-06,
"loss": 0.2209,
"step": 18500
},
{
"epoch": 3.43,
"learning_rate": 9.396349177661304e-06,
"loss": 0.2206,
"step": 19000
},
{
"epoch": 3.52,
"learning_rate": 8.85414784023134e-06,
"loss": 0.2366,
"step": 19500
},
{
"epoch": 3.61,
"learning_rate": 8.311946502801373e-06,
"loss": 0.2187,
"step": 20000
},
{
"epoch": 3.71,
"learning_rate": 7.769745165371408e-06,
"loss": 0.2346,
"step": 20500
},
{
"epoch": 3.8,
"learning_rate": 7.227543827941442e-06,
"loss": 0.2387,
"step": 21000
},
{
"epoch": 3.89,
"learning_rate": 6.685342490511477e-06,
"loss": 0.2274,
"step": 21500
},
{
"epoch": 3.98,
"learning_rate": 6.143141153081511e-06,
"loss": 0.2316,
"step": 22000
},
{
"epoch": 4.07,
"learning_rate": 5.6009398156515455e-06,
"loss": 0.1797,
"step": 22500
},
{
"epoch": 4.16,
"learning_rate": 5.058738478221579e-06,
"loss": 0.1121,
"step": 23000
},
{
"epoch": 4.25,
"learning_rate": 4.516537140791614e-06,
"loss": 0.1107,
"step": 23500
},
{
"epoch": 4.34,
"learning_rate": 3.974335803361649e-06,
"loss": 0.1226,
"step": 24000
},
{
"epoch": 4.43,
"learning_rate": 3.4321344659316826e-06,
"loss": 0.1159,
"step": 24500
},
{
"epoch": 4.52,
"learning_rate": 2.889933128501717e-06,
"loss": 0.1211,
"step": 25000
},
{
"epoch": 4.61,
"learning_rate": 2.3477317910717515e-06,
"loss": 0.1413,
"step": 25500
},
{
"epoch": 4.7,
"learning_rate": 1.8055304536417856e-06,
"loss": 0.1346,
"step": 26000
},
{
"epoch": 4.79,
"learning_rate": 1.26332911621182e-06,
"loss": 0.1065,
"step": 26500
},
{
"epoch": 4.88,
"learning_rate": 7.211277787818543e-07,
"loss": 0.1064,
"step": 27000
},
{
"epoch": 4.97,
"learning_rate": 1.7892644135188868e-07,
"loss": 0.0909,
"step": 27500
},
{
"epoch": 5.0,
"step": 27665,
"total_flos": 3.407082354156503e+17,
"train_runtime": 31051.1493,
"train_samples_per_second": 14.255,
"train_steps_per_second": 0.891
}
],
"max_steps": 27665,
"num_train_epochs": 5,
"total_flos": 3.407082354156503e+17,
"trial_name": null,
"trial_params": null
}