N_bert_agnews_padding20model

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: 0.5675
• Accuracy: 0.9482

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
0.178	1.0	7500	0.2016	0.9387
0.1359	2.0	15000	0.1994	0.9463
0.1199	3.0	22500	0.2296	0.9439
0.0893	4.0	30000	0.2822	0.9433
0.0632	5.0	37500	0.2953	0.9384
0.0441	6.0	45000	0.3583	0.9458
0.0337	7.0	52500	0.3966	0.9433
0.0287	8.0	60000	0.4296	0.9434
0.0241	9.0	67500	0.4442	0.9414
0.0118	10.0	75000	0.5066	0.9405
0.0166	11.0	82500	0.4644	0.94
0.0118	12.0	90000	0.4789	0.9409
0.0115	13.0	97500	0.5151	0.9443
0.0075	14.0	105000	0.4855	0.9458
0.007	15.0	112500	0.5377	0.9430
0.0058	16.0	120000	0.5308	0.9458
0.0024	17.0	127500	0.5328	0.9451
0.0014	18.0	135000	0.5569	0.9462
0.0023	19.0	142500	0.5646	0.9480
0.0019	20.0	150000	0.5675	0.9482

Framework versions

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