N_bert_agnews_padding30model

This model is a fine-tuned version of bert-base-uncased on the ag_news dataset. It achieves the following results on the evaluation set:

• Loss: 0.5638
• Accuracy: 0.9464

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
0.1818	1.0	7500	0.1926	0.9422
0.1395	2.0	15000	0.2087	0.9454
0.1138	3.0	22500	0.2287	0.9446
0.0858	4.0	30000	0.2681	0.9475
0.0569	5.0	37500	0.2953	0.9451
0.0421	6.0	45000	0.3934	0.9408
0.0363	7.0	52500	0.3943	0.9408
0.0283	8.0	60000	0.4069	0.9414
0.0165	9.0	67500	0.4448	0.9433
0.0142	10.0	75000	0.4708	0.9445
0.0134	11.0	82500	0.4708	0.9432
0.0089	12.0	90000	0.5035	0.9414
0.0083	13.0	97500	0.5031	0.9430
0.0064	14.0	105000	0.4990	0.9432
0.0046	15.0	112500	0.5265	0.945
0.0032	16.0	120000	0.5370	0.9449
0.0035	17.0	127500	0.5445	0.9447
0.0018	18.0	135000	0.5548	0.9462
0.0031	19.0	142500	0.5627	0.9454
0.0002	20.0	150000	0.5638	0.9464

Framework versions

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