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: 0.6727
• Accuracy: 0.9433

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
0.1828	1.0	7500	0.1902	0.94
0.1398	2.0	15000	0.1989	0.9433
0.1177	3.0	22500	0.2083	0.9459
0.0933	4.0	30000	0.2547	0.9439
0.0648	5.0	37500	0.3024	0.9428
0.0427	6.0	45000	0.3627	0.9401
0.034	7.0	52500	0.4282	0.9362
0.0325	8.0	60000	0.4297	0.9404
0.0217	9.0	67500	0.4508	0.9387
0.0126	10.0	75000	0.4900	0.9397
0.0147	11.0	82500	0.5530	0.9399
0.0103	12.0	90000	0.5293	0.9408
0.0108	13.0	97500	0.5388	0.9413
0.0068	14.0	105000	0.6006	0.9397
0.0028	15.0	112500	0.5974	0.9432
0.005	16.0	120000	0.5617	0.9413
0.0027	17.0	127500	0.6217	0.9433
0.0004	18.0	135000	0.6415	0.9420
0.0011	19.0	142500	0.6566	0.9442
0.0004	20.0	150000	0.6727	0.9433

Framework versions

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