N_roberta_agnews_padding80model

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: 0.5582
• Accuracy: 0.9478

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
0.1999	1.0	7500	0.2201	0.9363
0.1688	2.0	15000	0.2088	0.9443
0.149	3.0	22500	0.2117	0.9451
0.1244	4.0	30000	0.2460	0.9449
0.1107	5.0	37500	0.2623	0.9453
0.09	6.0	45000	0.2823	0.9454
0.0763	7.0	52500	0.3296	0.9447
0.053	8.0	60000	0.3415	0.9468
0.0473	9.0	67500	0.3770	0.9416
0.0364	10.0	75000	0.4377	0.9442
0.0279	11.0	82500	0.4097	0.9442
0.0252	12.0	90000	0.4395	0.9441
0.0269	13.0	97500	0.4557	0.9471
0.0109	14.0	105000	0.5014	0.9466
0.0102	15.0	112500	0.5357	0.9459
0.0101	16.0	120000	0.5305	0.9454
0.0051	17.0	127500	0.5209	0.9480
0.0069	18.0	135000	0.5314	0.9464
0.0024	19.0	142500	0.5451	0.9478
0.0026	20.0	150000	0.5582	0.9478

Framework versions

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