distilbert_agnews_padding70model

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: 0.6779
• Accuracy: 0.9439

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
0.1805	1.0	7500	0.1957	0.9380
0.1379	2.0	15000	0.1941	0.9458
0.1158	3.0	22500	0.2227	0.9455
0.0883	4.0	30000	0.2490	0.9426
0.067	5.0	37500	0.2904	0.9432
0.0453	6.0	45000	0.3748	0.9374
0.0432	7.0	52500	0.4024	0.9393
0.032	8.0	60000	0.4109	0.9437
0.0151	9.0	67500	0.4531	0.9425
0.0126	10.0	75000	0.4813	0.9418
0.0171	11.0	82500	0.5143	0.9424
0.0122	12.0	90000	0.5222	0.9412
0.0099	13.0	97500	0.5366	0.9438
0.008	14.0	105000	0.5356	0.9437
0.0066	15.0	112500	0.5689	0.9437
0.0026	16.0	120000	0.5860	0.9449
0.0031	17.0	127500	0.6363	0.9434
0.0017	18.0	135000	0.6474	0.9433
0.0007	19.0	142500	0.6755	0.9437
0.0008	20.0	150000	0.6779	0.9439

Framework versions

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