N_bert_agnews_padding90model

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: 0.5832
• Accuracy: 0.9462

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
0.1819	1.0	7500	0.1916	0.9392
0.1386	2.0	15000	0.2010	0.9447
0.1191	3.0	22500	0.2364	0.9429
0.0814	4.0	30000	0.2852	0.945
0.0599	5.0	37500	0.3128	0.9428
0.0478	6.0	45000	0.3510	0.9432
0.0333	7.0	52500	0.4077	0.9432
0.0259	8.0	60000	0.4419	0.9442
0.022	9.0	67500	0.4473	0.9428
0.0173	10.0	75000	0.4467	0.9449
0.0147	11.0	82500	0.4510	0.9453
0.0118	12.0	90000	0.4422	0.9432
0.0075	13.0	97500	0.5144	0.9463
0.008	14.0	105000	0.5141	0.9416
0.0066	15.0	112500	0.4913	0.9462
0.0066	16.0	120000	0.5086	0.9436
0.0022	17.0	127500	0.5400	0.9461
0.0007	18.0	135000	0.5426	0.9461
0.0005	19.0	142500	0.5688	0.9453
0.0005	20.0	150000	0.5832	0.9462

Framework versions

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