N_bert_agnews_padding10model

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: 0.5739
• Accuracy: 0.9457

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
0.176	1.0	7500	0.1846	0.9449
0.1315	2.0	15000	0.1942	0.9455
0.1116	3.0	22500	0.2441	0.9434
0.0804	4.0	30000	0.3103	0.9421
0.0548	5.0	37500	0.2975	0.9429
0.0417	6.0	45000	0.3861	0.9413
0.0299	7.0	52500	0.4010	0.9388
0.0325	8.0	60000	0.4365	0.9428
0.0232	9.0	67500	0.4431	0.9429
0.0173	10.0	75000	0.4699	0.9386
0.0139	11.0	82500	0.4937	0.9412
0.0121	12.0	90000	0.4899	0.9439
0.0047	13.0	97500	0.5263	0.9449
0.0106	14.0	105000	0.5317	0.9436
0.0028	15.0	112500	0.5426	0.9426
0.0052	16.0	120000	0.5332	0.9472
0.0025	17.0	127500	0.5458	0.9464
0.0023	18.0	135000	0.5433	0.9442
0.0003	19.0	142500	0.5707	0.9461
0.0009	20.0	150000	0.5739	0.9457

Framework versions

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