N_distilbert_agnews_padding80model

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

• Loss: 0.6373
• Accuracy: 0.9467

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
0.1867	1.0	7500	0.1871	0.9422
0.1412	2.0	15000	0.1944	0.9457
0.1163	3.0	22500	0.2140	0.945
0.0838	4.0	30000	0.2455	0.9476
0.0572	5.0	37500	0.2839	0.9436
0.0433	6.0	45000	0.3780	0.9411
0.035	7.0	52500	0.4167	0.9384
0.0303	8.0	60000	0.4762	0.9401
0.0198	9.0	67500	0.4682	0.9407
0.0186	10.0	75000	0.4855	0.9429
0.0166	11.0	82500	0.5062	0.9416
0.0125	12.0	90000	0.5277	0.9418
0.0061	13.0	97500	0.5436	0.9442
0.0028	14.0	105000	0.5485	0.9420
0.0034	15.0	112500	0.5601	0.9441
0.0029	16.0	120000	0.5588	0.9459
0.0032	17.0	127500	0.6078	0.9436
0.0046	18.0	135000	0.6353	0.9429
0.0003	19.0	142500	0.6454	0.9458
0.0011	20.0	150000	0.6373	0.9467

Framework versions

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