roberta_agnews_padding10model

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: 0.5337
• Accuracy: 0.9503

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
0.1966	1.0	7500	0.2068	0.9404
0.1632	2.0	15000	0.1954	0.9457
0.1432	3.0	22500	0.2422	0.9478
0.1223	4.0	30000	0.2275	0.9486
0.0994	5.0	37500	0.2442	0.9486
0.079	6.0	45000	0.3053	0.9486
0.0759	7.0	52500	0.3104	0.9463
0.0506	8.0	60000	0.3757	0.9472
0.0436	9.0	67500	0.3468	0.9470
0.025	10.0	75000	0.4170	0.9468
0.0303	11.0	82500	0.4168	0.9462
0.0273	12.0	90000	0.4173	0.9486
0.024	13.0	97500	0.4305	0.9476
0.0139	14.0	105000	0.4549	0.9480
0.0111	15.0	112500	0.4961	0.9483
0.0102	16.0	120000	0.4733	0.9488
0.0036	17.0	127500	0.5044	0.9493
0.0025	18.0	135000	0.5070	0.95
0.0024	19.0	142500	0.5196	0.9508
0.0018	20.0	150000	0.5337	0.9503

Framework versions

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