bert-base-uncased_crows_pairs_finetuned

This model is a fine-tuned version of bert-base-uncased on the crows_pairs dataset. It achieves the following results on the evaluation set:

• Loss: 2.1731
• Accuracy: 0.7649
• Tp: 0.3344
• Tn: 0.4305
• Fp: 0.1126
• Fn: 0.1225

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Tp	Tn	Fp	Fn
0.703	1.05	20	0.6889	0.5430	0.0	0.5430	0.0	0.4570
0.6884	2.11	40	0.6886	0.5	0.3013	0.1987	0.3444	0.1556
0.5388	3.16	60	0.6347	0.7053	0.1821	0.5232	0.0199	0.2748
0.2228	4.21	80	0.9802	0.6987	0.1887	0.5099	0.0331	0.2682
0.1644	5.26	100	0.7523	0.7583	0.3675	0.3907	0.1523	0.0894
0.0478	6.32	120	1.5712	0.6954	0.2053	0.4901	0.0530	0.2517
0.0465	7.37	140	1.2587	0.7351	0.2781	0.4570	0.0861	0.1788
0.0313	8.42	160	1.5825	0.7450	0.3775	0.3675	0.1755	0.0795
0.0137	9.47	180	1.3570	0.7318	0.2815	0.4503	0.0927	0.1755
0.04	10.53	200	2.1377	0.6921	0.1887	0.5033	0.0397	0.2682
0.0041	11.58	220	1.6776	0.7351	0.3278	0.4073	0.1358	0.1291
0.0042	12.63	240	1.8873	0.7086	0.2980	0.4106	0.1325	0.1589
0.0009	13.68	260	2.2464	0.6987	0.3543	0.3444	0.1987	0.1026
0.014	14.74	280	1.9753	0.7252	0.3245	0.4007	0.1424	0.1325
0.0026	15.79	300	1.8852	0.7417	0.2914	0.4503	0.0927	0.1656
0.0147	16.84	320	2.0273	0.7351	0.3113	0.4238	0.1192	0.1457
0.0009	17.89	340	1.7328	0.7483	0.3278	0.4205	0.1225	0.1291
0.0085	18.95	360	2.0146	0.7450	0.2815	0.4636	0.0795	0.1755
0.0001	20.0	380	2.0808	0.7450	0.3113	0.4338	0.1093	0.1457
0.0001	21.05	400	2.2655	0.7417	0.3609	0.3808	0.1623	0.0960
0.0034	22.11	420	2.0298	0.7583	0.3079	0.4503	0.0927	0.1490
0.0082	23.16	440	2.0650	0.7550	0.3344	0.4205	0.1225	0.1225
0.0001	24.21	460	2.2472	0.7450	0.2748	0.4702	0.0728	0.1821
0.0001	25.26	480	2.3655	0.7351	0.3709	0.3642	0.1788	0.0861
0.0004	26.32	500	2.1407	0.7550	0.3510	0.4040	0.1391	0.1060
0.0001	27.37	520	2.1168	0.7450	0.3642	0.3808	0.1623	0.0927
0.0002	28.42	540	2.2050	0.7517	0.3775	0.3742	0.1689	0.0795
0.0	29.47	560	2.0560	0.7682	0.3212	0.4470	0.0960	0.1358
0.0	30.53	580	2.0859	0.7715	0.3179	0.4536	0.0894	0.1391
0.0	31.58	600	2.0958	0.7715	0.3179	0.4536	0.0894	0.1391
0.0	32.63	620	2.1039	0.7715	0.3179	0.4536	0.0894	0.1391
0.0	33.68	640	2.1113	0.7715	0.3179	0.4536	0.0894	0.1391
0.0	34.74	660	2.1180	0.7715	0.3179	0.4536	0.0894	0.1391
0.0	35.79	680	2.1127	0.7715	0.3278	0.4437	0.0993	0.1291
0.0	36.84	700	2.1376	0.7682	0.3377	0.4305	0.1126	0.1192
0.0	37.89	720	2.1460	0.7616	0.3377	0.4238	0.1192	0.1192
0.0	38.95	740	2.1507	0.7649	0.3377	0.4272	0.1159	0.1192
0.0	40.0	760	2.1548	0.7682	0.3377	0.4305	0.1126	0.1192
0.0	41.05	780	2.1586	0.7682	0.3377	0.4305	0.1126	0.1192
0.0	42.11	800	2.1620	0.7682	0.3377	0.4305	0.1126	0.1192
0.0	43.16	820	2.1649	0.7682	0.3377	0.4305	0.1126	0.1192
0.0	44.21	840	2.1674	0.7682	0.3377	0.4305	0.1126	0.1192
0.0	45.26	860	2.1690	0.7682	0.3377	0.4305	0.1126	0.1192
0.0	46.32	880	2.1705	0.7682	0.3377	0.4305	0.1126	0.1192
0.0	47.37	900	2.1717	0.7649	0.3344	0.4305	0.1126	0.1225
0.0	48.42	920	2.1726	0.7649	0.3344	0.4305	0.1126	0.1225
0.0	49.47	940	2.1731	0.7649	0.3344	0.4305	0.1126	0.1225

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1
• Datasets 2.10.1
• Tokenizers 0.13.2