distilbert_agnews_padding40model

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: 0.6171
• Accuracy: 0.9468

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
0.1804	1.0	7500	0.1955	0.9395
0.1406	2.0	15000	0.1970	0.9429
0.1206	3.0	22500	0.2108	0.9467
0.0878	4.0	30000	0.2626	0.9429
0.0605	5.0	37500	0.3047	0.9417
0.0472	6.0	45000	0.3698	0.9397
0.0331	7.0	52500	0.4269	0.9367
0.0251	8.0	60000	0.4326	0.9416
0.0247	9.0	67500	0.4525	0.9428
0.0151	10.0	75000	0.4580	0.9462
0.0164	11.0	82500	0.5027	0.9455
0.0074	12.0	90000	0.5040	0.9437
0.0054	13.0	97500	0.5347	0.9449
0.0031	14.0	105000	0.5753	0.9451
0.0065	15.0	112500	0.5445	0.9453
0.0012	16.0	120000	0.5966	0.9461
0.0028	17.0	127500	0.5994	0.9445
0.0006	18.0	135000	0.5948	0.9455
0.0002	19.0	142500	0.6115	0.9471
0.0008	20.0	150000	0.6171	0.9468

Framework versions

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