roberta_agnews_padding0model

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: 0.5563
• Accuracy: 0.9493

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
0.1901	1.0	7500	0.2099	0.9378
0.1664	2.0	15000	0.2084	0.9446
0.148	3.0	22500	0.2164	0.9479
0.1201	4.0	30000	0.2506	0.9442
0.0999	5.0	37500	0.2447	0.9505
0.0731	6.0	45000	0.3085	0.9463
0.0668	7.0	52500	0.3298	0.9467
0.0577	8.0	60000	0.3703	0.9453
0.0435	9.0	67500	0.3854	0.9462
0.0313	10.0	75000	0.3833	0.945
0.023	11.0	82500	0.4196	0.9459
0.0231	12.0	90000	0.4412	0.9441
0.0207	13.0	97500	0.4519	0.9458
0.0153	14.0	105000	0.4682	0.9463
0.0136	15.0	112500	0.4854	0.9487
0.0118	16.0	120000	0.5146	0.9468
0.0058	17.0	127500	0.5119	0.9487
0.002	18.0	135000	0.5292	0.9495
0.0026	19.0	142500	0.5443	0.9483
0.0015	20.0	150000	0.5563	0.9493

Framework versions

• Transformers 4.32.1
• Pytorch 2.1.1
• Datasets 2.12.0
• Tokenizers 0.13.3