N_distilbert_agnews_padding50model

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: 0.6732
• Accuracy: 0.9418

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
0.1757	1.0	7500	0.1847	0.9428
0.1399	2.0	15000	0.1983	0.9439
0.1196	3.0	22500	0.2251	0.9418
0.0894	4.0	30000	0.2583	0.9436
0.0587	5.0	37500	0.3116	0.9425
0.0404	6.0	45000	0.3567	0.9432
0.0318	7.0	52500	0.4279	0.9392
0.0257	8.0	60000	0.4443	0.9407
0.0212	9.0	67500	0.4974	0.9378
0.0106	10.0	75000	0.4965	0.9417
0.0145	11.0	82500	0.4986	0.9433
0.011	12.0	90000	0.5389	0.9392
0.0121	13.0	97500	0.5671	0.9441
0.0046	14.0	105000	0.6063	0.9396
0.0011	15.0	112500	0.6245	0.9414
0.002	16.0	120000	0.6103	0.9426
0.001	17.0	127500	0.6181	0.9426
0.0017	18.0	135000	0.6401	0.9408
0.0002	19.0	142500	0.6667	0.9422
0.0004	20.0	150000	0.6732	0.9418

Framework versions

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