distilbert_agnews_padding90model

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: 0.6543
• Accuracy: 0.9443

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
0.1824	1.0	7500	0.1893	0.9428
0.1381	2.0	15000	0.2112	0.9401
0.1208	3.0	22500	0.2110	0.9449
0.091	4.0	30000	0.2609	0.9428
0.065	5.0	37500	0.2991	0.9417
0.0447	6.0	45000	0.3947	0.9408
0.0374	7.0	52500	0.4052	0.9354
0.0275	8.0	60000	0.4441	0.9418
0.0198	9.0	67500	0.4479	0.9408
0.0131	10.0	75000	0.5038	0.9387
0.0128	11.0	82500	0.5114	0.9424
0.0098	12.0	90000	0.5130	0.9436
0.0085	13.0	97500	0.5473	0.9414
0.0066	14.0	105000	0.5601	0.9429
0.0054	15.0	112500	0.5617	0.9424
0.0023	16.0	120000	0.5911	0.9433
0.0021	17.0	127500	0.6165	0.9442
0.0038	18.0	135000	0.6105	0.9446
0.0008	19.0	142500	0.6462	0.9422
0.0004	20.0	150000	0.6543	0.9443

Framework versions

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