N_roberta_agnews_padding20model

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

• Loss: 0.5475
• Accuracy: 0.9486

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.1932	1.0	7500	0.1956	0.9442
0.1695	2.0	15000	0.1987	0.9459
0.1531	3.0	22500	0.2017	0.9462
0.1247	4.0	30000	0.2431	0.9479
0.1026	5.0	37500	0.2465	0.9479
0.078	6.0	45000	0.3001	0.9478
0.0752	7.0	52500	0.3209	0.945
0.048	8.0	60000	0.3456	0.9484
0.0408	9.0	67500	0.4119	0.9436
0.0338	10.0	75000	0.3976	0.945
0.0305	11.0	82500	0.4211	0.945
0.022	12.0	90000	0.4607	0.9470
0.0224	13.0	97500	0.4733	0.9464
0.0127	14.0	105000	0.4984	0.9466
0.0135	15.0	112500	0.4808	0.9492
0.0085	16.0	120000	0.4993	0.9497
0.0045	17.0	127500	0.5184	0.9475
0.0046	18.0	135000	0.5259	0.9479
0.0037	19.0	142500	0.5419	0.9483
0.0017	20.0	150000	0.5475	0.9486

Framework versions

• Transformers 4.33.2
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3