N_distilbert_agnews_padding100model

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: 0.6461
• Accuracy: 0.9453

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
0.1812	1.0	7500	0.1853	0.9424
0.1417	2.0	15000	0.1940	0.9437
0.1201	3.0	22500	0.2239	0.9425
0.0895	4.0	30000	0.2896	0.9422
0.0604	5.0	37500	0.2957	0.9401
0.0471	6.0	45000	0.3845	0.9389
0.032	7.0	52500	0.4266	0.9393
0.0284	8.0	60000	0.4621	0.9420
0.0211	9.0	67500	0.4691	0.9384
0.0158	10.0	75000	0.4800	0.9417
0.0179	11.0	82500	0.5048	0.9422
0.0105	12.0	90000	0.4962	0.9453
0.0102	13.0	97500	0.5280	0.9437
0.0039	14.0	105000	0.5401	0.9442
0.0037	15.0	112500	0.5675	0.9441
0.0052	16.0	120000	0.5934	0.9454
0.003	17.0	127500	0.6308	0.9426
0.0014	18.0	135000	0.6194	0.9436
0.0007	19.0	142500	0.6454	0.945
0.0004	20.0	150000	0.6461	0.9453

Framework versions

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