N_bert_agnews_padding70model

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: 0.5807
• Accuracy: 0.9464

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
0.1804	1.0	7500	0.1897	0.9418
0.1359	2.0	15000	0.1998	0.9454
0.1189	3.0	22500	0.2309	0.9443
0.0882	4.0	30000	0.2622	0.9458
0.0579	5.0	37500	0.3019	0.9417
0.0389	6.0	45000	0.3608	0.9438
0.0359	7.0	52500	0.3635	0.9426
0.0296	8.0	60000	0.4064	0.9428
0.015	9.0	67500	0.4595	0.9417
0.0191	10.0	75000	0.4516	0.9433
0.0146	11.0	82500	0.5207	0.9409
0.0138	12.0	90000	0.4787	0.9442
0.0108	13.0	97500	0.5216	0.9408
0.0071	14.0	105000	0.5377	0.9455
0.0047	15.0	112500	0.5283	0.9459
0.0035	16.0	120000	0.4928	0.9451
0.0007	17.0	127500	0.5760	0.9454
0.0004	18.0	135000	0.5759	0.9449
0.0002	19.0	142500	0.5637	0.9472
0.0003	20.0	150000	0.5807	0.9464

Framework versions

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