N_distilbert_agnews_padding10model

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: 0.6711
• Accuracy: 0.9438

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
0.1775	1.0	7500	0.1897	0.9399
0.1357	2.0	15000	0.1881	0.9464
0.1124	3.0	22500	0.2236	0.9453
0.0822	4.0	30000	0.2775	0.9439
0.0517	5.0	37500	0.3485	0.9370
0.0405	6.0	45000	0.3514	0.9418
0.0356	7.0	52500	0.4108	0.9409
0.0292	8.0	60000	0.4351	0.9411
0.0191	9.0	67500	0.4604	0.9374
0.0159	10.0	75000	0.4604	0.9429
0.0126	11.0	82500	0.4860	0.9409
0.0092	12.0	90000	0.5452	0.9416
0.0075	13.0	97500	0.5498	0.9442
0.0023	14.0	105000	0.5712	0.9417
0.0015	15.0	112500	0.6107	0.9420
0.0033	16.0	120000	0.5908	0.9428
0.0029	17.0	127500	0.6449	0.9416
0.0009	18.0	135000	0.6534	0.9430
0.0013	19.0	142500	0.6641	0.9442
0.0005	20.0	150000	0.6711	0.9438

Framework versions

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