N_bert_agnews_padding80model

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

• Loss: 0.5733
• Accuracy: 0.9466

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
0.1814	1.0	7500	0.1946	0.9393
0.1378	2.0	15000	0.1999	0.9443
0.1185	3.0	22500	0.2327	0.9470
0.0766	4.0	30000	0.2848	0.9446
0.057	5.0	37500	0.3384	0.9409
0.0439	6.0	45000	0.3604	0.9425
0.0384	7.0	52500	0.3707	0.9436
0.0312	8.0	60000	0.3830	0.9432
0.0156	9.0	67500	0.4272	0.9443
0.0156	10.0	75000	0.4233	0.9464
0.0092	11.0	82500	0.4810	0.9457
0.0102	12.0	90000	0.5085	0.9447
0.0065	13.0	97500	0.4786	0.9455
0.009	14.0	105000	0.5062	0.9451
0.0049	15.0	112500	0.5219	0.9443
0.0043	16.0	120000	0.5577	0.9447
0.0032	17.0	127500	0.5405	0.9459
0.001	18.0	135000	0.5904	0.9457
0.0003	19.0	142500	0.5733	0.9454
0.0004	20.0	150000	0.5733	0.9466

Framework versions

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