N_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: 0.5342
• Accuracy: 0.9507

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
0.1973	1.0	7500	0.2025	0.9403
0.1674	2.0	15000	0.1876	0.9471
0.1488	3.0	22500	0.2367	0.9446
0.1213	4.0	30000	0.2451	0.9461
0.0942	5.0	37500	0.2545	0.9464
0.0842	6.0	45000	0.3061	0.9446
0.0718	7.0	52500	0.2821	0.9476
0.0562	8.0	60000	0.4124	0.9443
0.0387	9.0	67500	0.4309	0.9409
0.0299	10.0	75000	0.4162	0.9470
0.028	11.0	82500	0.4086	0.9479
0.026	12.0	90000	0.4091	0.9466
0.0205	13.0	97500	0.4481	0.9457
0.0124	14.0	105000	0.4895	0.9453
0.0093	15.0	112500	0.5086	0.9463
0.0075	16.0	120000	0.4911	0.9487
0.0068	17.0	127500	0.4924	0.9496
0.0023	18.0	135000	0.5008	0.9503
0.0014	19.0	142500	0.5251	0.9505
0.0049	20.0	150000	0.5342	0.9507

Framework versions

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