distilbert_agnews_padding100model

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: 0.6590
• Accuracy: 0.9449

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
0.1885	1.0	7500	0.1952	0.9407
0.1438	2.0	15000	0.1912	0.9442
0.1323	3.0	22500	0.2133	0.9442
0.0953	4.0	30000	0.2650	0.9442
0.0595	5.0	37500	0.2934	0.9404
0.0444	6.0	45000	0.3532	0.9439
0.0431	7.0	52500	0.3903	0.9368
0.0324	8.0	60000	0.4585	0.94
0.0241	9.0	67500	0.4216	0.9426
0.0208	10.0	75000	0.4646	0.9442
0.015	11.0	82500	0.5329	0.9426
0.0115	12.0	90000	0.5237	0.9424
0.0109	13.0	97500	0.5406	0.9426
0.009	14.0	105000	0.5572	0.9421
0.0046	15.0	112500	0.5948	0.9428
0.0039	16.0	120000	0.5682	0.9436
0.0025	17.0	127500	0.6096	0.9454
0.0008	18.0	135000	0.6312	0.9447
0.0014	19.0	142500	0.6435	0.9439
0.0002	20.0	150000	0.6590	0.9449

Framework versions

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