bert-base-uncased-qqp / trainer_state.json
Jeremiah Zhou
End of training
87f8056
{
"best_metric": 0.9099925797674994,
"best_model_checkpoint": "./fine-tune/bert-base-uncased/qqp/checkpoint-34113",
"epoch": 3.0,
"global_step": 34113,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.04,
"learning_rate": 1.9706856623574592e-05,
"loss": 0.4417,
"step": 500
},
{
"epoch": 0.09,
"learning_rate": 1.9413713247149183e-05,
"loss": 0.3623,
"step": 1000
},
{
"epoch": 0.13,
"learning_rate": 1.9120569870723774e-05,
"loss": 0.3469,
"step": 1500
},
{
"epoch": 0.18,
"learning_rate": 1.882742649429836e-05,
"loss": 0.3275,
"step": 2000
},
{
"epoch": 0.22,
"learning_rate": 1.8534283117872952e-05,
"loss": 0.315,
"step": 2500
},
{
"epoch": 0.26,
"learning_rate": 1.8241139741447543e-05,
"loss": 0.3067,
"step": 3000
},
{
"epoch": 0.31,
"learning_rate": 1.7947996365022134e-05,
"loss": 0.3031,
"step": 3500
},
{
"epoch": 0.35,
"learning_rate": 1.7654852988596725e-05,
"loss": 0.2986,
"step": 4000
},
{
"epoch": 0.4,
"learning_rate": 1.7361709612171312e-05,
"loss": 0.2994,
"step": 4500
},
{
"epoch": 0.44,
"learning_rate": 1.7068566235745903e-05,
"loss": 0.2825,
"step": 5000
},
{
"epoch": 0.48,
"learning_rate": 1.6775422859320497e-05,
"loss": 0.2814,
"step": 5500
},
{
"epoch": 0.53,
"learning_rate": 1.6482279482895085e-05,
"loss": 0.2746,
"step": 6000
},
{
"epoch": 0.57,
"learning_rate": 1.6189136106469676e-05,
"loss": 0.2706,
"step": 6500
},
{
"epoch": 0.62,
"learning_rate": 1.5895992730044266e-05,
"loss": 0.2669,
"step": 7000
},
{
"epoch": 0.66,
"learning_rate": 1.5602849353618857e-05,
"loss": 0.264,
"step": 7500
},
{
"epoch": 0.7,
"learning_rate": 1.5309705977193448e-05,
"loss": 0.2746,
"step": 8000
},
{
"epoch": 0.75,
"learning_rate": 1.5016562600768037e-05,
"loss": 0.2598,
"step": 8500
},
{
"epoch": 0.79,
"learning_rate": 1.4723419224342628e-05,
"loss": 0.2585,
"step": 9000
},
{
"epoch": 0.84,
"learning_rate": 1.4430275847917217e-05,
"loss": 0.2528,
"step": 9500
},
{
"epoch": 0.88,
"learning_rate": 1.4137132471491808e-05,
"loss": 0.2501,
"step": 10000
},
{
"epoch": 0.92,
"learning_rate": 1.3843989095066399e-05,
"loss": 0.2445,
"step": 10500
},
{
"epoch": 0.97,
"learning_rate": 1.3550845718640988e-05,
"loss": 0.2511,
"step": 11000
},
{
"epoch": 1.0,
"eval_accuracy": 0.8968587682414049,
"eval_combined_score": 0.8805007987294208,
"eval_f1": 0.8641428292174367,
"eval_loss": 0.24689364433288574,
"eval_runtime": 193.2661,
"eval_samples_per_second": 209.193,
"eval_steps_per_second": 26.15,
"step": 11371
},
{
"epoch": 1.01,
"learning_rate": 1.3257702342215579e-05,
"loss": 0.2299,
"step": 11500
},
{
"epoch": 1.06,
"learning_rate": 1.2964558965790168e-05,
"loss": 0.1843,
"step": 12000
},
{
"epoch": 1.1,
"learning_rate": 1.2671415589364759e-05,
"loss": 0.1884,
"step": 12500
},
{
"epoch": 1.14,
"learning_rate": 1.2378272212939351e-05,
"loss": 0.1784,
"step": 13000
},
{
"epoch": 1.19,
"learning_rate": 1.208512883651394e-05,
"loss": 0.1761,
"step": 13500
},
{
"epoch": 1.23,
"learning_rate": 1.1791985460088531e-05,
"loss": 0.1846,
"step": 14000
},
{
"epoch": 1.28,
"learning_rate": 1.149884208366312e-05,
"loss": 0.188,
"step": 14500
},
{
"epoch": 1.32,
"learning_rate": 1.1205698707237711e-05,
"loss": 0.1926,
"step": 15000
},
{
"epoch": 1.36,
"learning_rate": 1.0912555330812302e-05,
"loss": 0.1854,
"step": 15500
},
{
"epoch": 1.41,
"learning_rate": 1.0619411954386891e-05,
"loss": 0.1867,
"step": 16000
},
{
"epoch": 1.45,
"learning_rate": 1.0326268577961482e-05,
"loss": 0.1839,
"step": 16500
},
{
"epoch": 1.5,
"learning_rate": 1.0033125201536071e-05,
"loss": 0.1803,
"step": 17000
},
{
"epoch": 1.54,
"learning_rate": 9.739981825110662e-06,
"loss": 0.1841,
"step": 17500
},
{
"epoch": 1.58,
"learning_rate": 9.446838448685253e-06,
"loss": 0.1845,
"step": 18000
},
{
"epoch": 1.63,
"learning_rate": 9.153695072259843e-06,
"loss": 0.1825,
"step": 18500
},
{
"epoch": 1.67,
"learning_rate": 8.860551695834433e-06,
"loss": 0.1796,
"step": 19000
},
{
"epoch": 1.71,
"learning_rate": 8.567408319409023e-06,
"loss": 0.1759,
"step": 19500
},
{
"epoch": 1.76,
"learning_rate": 8.274264942983614e-06,
"loss": 0.1739,
"step": 20000
},
{
"epoch": 1.8,
"learning_rate": 7.981121566558205e-06,
"loss": 0.1713,
"step": 20500
},
{
"epoch": 1.85,
"learning_rate": 7.687978190132794e-06,
"loss": 0.1761,
"step": 21000
},
{
"epoch": 1.89,
"learning_rate": 7.394834813707384e-06,
"loss": 0.1774,
"step": 21500
},
{
"epoch": 1.93,
"learning_rate": 7.101691437281976e-06,
"loss": 0.1792,
"step": 22000
},
{
"epoch": 1.98,
"learning_rate": 6.808548060856566e-06,
"loss": 0.1763,
"step": 22500
},
{
"epoch": 2.0,
"eval_accuracy": 0.9070986890922582,
"eval_combined_score": 0.8919997902724007,
"eval_f1": 0.8769008914525432,
"eval_loss": 0.2379385530948639,
"eval_runtime": 192.8495,
"eval_samples_per_second": 209.645,
"eval_steps_per_second": 26.207,
"step": 22742
},
{
"epoch": 2.02,
"learning_rate": 6.515404684431156e-06,
"loss": 0.1462,
"step": 23000
},
{
"epoch": 2.07,
"learning_rate": 6.222261308005746e-06,
"loss": 0.1244,
"step": 23500
},
{
"epoch": 2.11,
"learning_rate": 5.929117931580336e-06,
"loss": 0.118,
"step": 24000
},
{
"epoch": 2.15,
"learning_rate": 5.6359745551549274e-06,
"loss": 0.1237,
"step": 24500
},
{
"epoch": 2.2,
"learning_rate": 5.3428311787295174e-06,
"loss": 0.1216,
"step": 25000
},
{
"epoch": 2.24,
"learning_rate": 5.049687802304107e-06,
"loss": 0.1242,
"step": 25500
},
{
"epoch": 2.29,
"learning_rate": 4.756544425878697e-06,
"loss": 0.1199,
"step": 26000
},
{
"epoch": 2.33,
"learning_rate": 4.463401049453288e-06,
"loss": 0.1209,
"step": 26500
},
{
"epoch": 2.37,
"learning_rate": 4.170257673027878e-06,
"loss": 0.1187,
"step": 27000
},
{
"epoch": 2.42,
"learning_rate": 3.877114296602469e-06,
"loss": 0.1149,
"step": 27500
},
{
"epoch": 2.46,
"learning_rate": 3.583970920177059e-06,
"loss": 0.1144,
"step": 28000
},
{
"epoch": 2.51,
"learning_rate": 3.290827543751649e-06,
"loss": 0.1261,
"step": 28500
},
{
"epoch": 2.55,
"learning_rate": 2.9976841673262398e-06,
"loss": 0.1227,
"step": 29000
},
{
"epoch": 2.59,
"learning_rate": 2.7045407909008298e-06,
"loss": 0.1276,
"step": 29500
},
{
"epoch": 2.64,
"learning_rate": 2.41139741447542e-06,
"loss": 0.1181,
"step": 30000
},
{
"epoch": 2.68,
"learning_rate": 2.1182540380500106e-06,
"loss": 0.1184,
"step": 30500
},
{
"epoch": 2.73,
"learning_rate": 1.825110661624601e-06,
"loss": 0.1253,
"step": 31000
},
{
"epoch": 2.77,
"learning_rate": 1.531967285199191e-06,
"loss": 0.1122,
"step": 31500
},
{
"epoch": 2.81,
"learning_rate": 1.2388239087737813e-06,
"loss": 0.1077,
"step": 32000
},
{
"epoch": 2.86,
"learning_rate": 9.456805323483717e-07,
"loss": 0.1211,
"step": 32500
},
{
"epoch": 2.9,
"learning_rate": 6.525371559229619e-07,
"loss": 0.1197,
"step": 33000
},
{
"epoch": 2.95,
"learning_rate": 3.5939377949755233e-07,
"loss": 0.1148,
"step": 33500
},
{
"epoch": 2.99,
"learning_rate": 6.62504030721426e-08,
"loss": 0.1221,
"step": 34000
},
{
"epoch": 3.0,
"eval_accuracy": 0.9099925797674994,
"eval_combined_score": 0.8944088968565445,
"eval_f1": 0.8788252139455897,
"eval_loss": 0.2828538715839386,
"eval_runtime": 194.3457,
"eval_samples_per_second": 208.031,
"eval_steps_per_second": 26.005,
"step": 34113
},
{
"epoch": 3.0,
"step": 34113,
"total_flos": 7.179892878638592e+16,
"train_loss": 0.19735700893404895,
"train_runtime": 13532.9561,
"train_samples_per_second": 80.658,
"train_steps_per_second": 2.521
}
],
"max_steps": 34113,
"num_train_epochs": 3,
"total_flos": 7.179892878638592e+16,
"trial_name": null,
"trial_params": null
}