N_distilbert_agnews_padding60model

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: 0.6714
• Accuracy: 0.945

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
0.1829	1.0	7500	0.1828	0.9445
0.1406	2.0	15000	0.1995	0.9436
0.1229	3.0	22500	0.2025	0.9470
0.0893	4.0	30000	0.2578	0.9464
0.0547	5.0	37500	0.3250	0.9392
0.0394	6.0	45000	0.3882	0.9391
0.0368	7.0	52500	0.4054	0.9383
0.0221	8.0	60000	0.4602	0.9396
0.0179	9.0	67500	0.4535	0.9430
0.015	10.0	75000	0.4745	0.9421
0.0177	11.0	82500	0.5035	0.9439
0.0147	12.0	90000	0.4848	0.9436
0.0077	13.0	97500	0.5222	0.9447
0.0039	14.0	105000	0.5591	0.9432
0.0042	15.0	112500	0.5863	0.9433
0.0017	16.0	120000	0.6075	0.9457
0.003	17.0	127500	0.6207	0.9455
0.0017	18.0	135000	0.6482	0.9454
0.0004	19.0	142500	0.6646	0.9457
0.0017	20.0	150000	0.6714	0.945

Framework versions

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