20230824002458

This model is a fine-tuned version of bert-large-cased on the super_glue dataset. It achieves the following results on the evaluation set:

• Loss: 0.0768
• Accuracy: 0.7112

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: 0.003
• train_batch_size: 4
• eval_batch_size: 8
• seed: 11
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 60.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.4042	1.0	623	0.3862	0.5271
0.3203	2.0	1246	1.0958	0.4729
0.3087	3.0	1869	0.5979	0.4729
0.2723	4.0	2492	0.1618	0.5271
0.2635	5.0	3115	0.2704	0.5343
0.2826	6.0	3738	0.3245	0.4729
0.2663	7.0	4361	0.2230	0.5957
0.2562	8.0	4984	0.1453	0.6390
0.2259	9.0	5607	0.1312	0.6282
0.1806	10.0	6230	0.1118	0.7148
0.1525	11.0	6853	0.1076	0.6787
0.1509	12.0	7476	0.1241	0.6643
0.149	13.0	8099	0.1158	0.6931
0.1509	14.0	8722	0.1154	0.7040
0.1397	15.0	9345	0.1096	0.6823
0.1311	16.0	9968	0.0999	0.6751
0.13	17.0	10591	0.0986	0.6968
0.1244	18.0	11214	0.1063	0.6895
0.1278	19.0	11837	0.1229	0.6931
0.1228	20.0	12460	0.0905	0.7112
0.1153	21.0	13083	0.0916	0.7004
0.1171	22.0	13706	0.1085	0.7148
0.1179	23.0	14329	0.1101	0.7256
0.1069	24.0	14952	0.0917	0.6895
0.1019	25.0	15575	0.0837	0.7112
0.1017	26.0	16198	0.0832	0.7148
0.1034	27.0	16821	0.0847	0.7220
0.0989	28.0	17444	0.0830	0.7256
0.0969	29.0	18067	0.0817	0.7148
0.0964	30.0	18690	0.0835	0.7112
0.0957	31.0	19313	0.0846	0.7148
0.0937	32.0	19936	0.0827	0.7112
0.0895	33.0	20559	0.0860	0.7220
0.0905	34.0	21182	0.0830	0.7220
0.0875	35.0	21805	0.0796	0.7184
0.0895	36.0	22428	0.0811	0.7076
0.0861	37.0	23051	0.0805	0.7112
0.0868	38.0	23674	0.0786	0.7040
0.0798	39.0	24297	0.0787	0.7148
0.0827	40.0	24920	0.0815	0.7112
0.0798	41.0	25543	0.0790	0.7184
0.079	42.0	26166	0.0813	0.7220
0.0794	43.0	26789	0.0802	0.7112
0.0766	44.0	27412	0.0796	0.7076
0.0766	45.0	28035	0.0813	0.7329
0.0765	46.0	28658	0.0810	0.7112
0.0744	47.0	29281	0.0781	0.7148
0.076	48.0	29904	0.0794	0.7148
0.0728	49.0	30527	0.0780	0.7112
0.0745	50.0	31150	0.0767	0.7256
0.0711	51.0	31773	0.0771	0.7220
0.0726	52.0	32396	0.0772	0.7256
0.0747	53.0	33019	0.0772	0.7184
0.0711	54.0	33642	0.0772	0.7256
0.0676	55.0	34265	0.0767	0.7329
0.0697	56.0	34888	0.0783	0.7220
0.0692	57.0	35511	0.0766	0.7184
0.067	58.0	36134	0.0773	0.7148
0.0676	59.0	36757	0.0774	0.7112
0.0678	60.0	37380	0.0768	0.7112

Framework versions

• Transformers 4.26.1
• Pytorch 2.0.1+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3