distilbert-base-uncased-finetuned-ag-news-v4

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

• Loss: 1.7194

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: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.012	1.0	1384	1.9480
1.978	2.0	2768	1.9294
1.9524	3.0	4152	1.9103
1.9195	4.0	5536	1.8960
1.8815	5.0	6920	1.8661
1.8658	6.0	8304	1.8516
1.8454	7.0	9688	1.8258
1.8268	8.0	11072	1.8317
1.8066	9.0	12456	1.8303
1.7911	10.0	13840	1.8111
1.7781	11.0	15224	1.8049
1.7658	12.0	16608	1.7922
1.7453	13.0	17992	1.7785
1.74	14.0	19376	1.7689
1.721	15.0	20760	1.7608
1.7096	16.0	22144	1.7563
1.706	17.0	23528	1.7515
1.6961	18.0	24912	1.7531
1.684	19.0	26296	1.7046
1.6762	20.0	27680	1.7193
1.6698	21.0	29064	1.7498
1.6638	22.0	30448	1.7177
1.6606	23.0	31832	1.7075
1.6558	24.0	33216	1.7324
1.6444	25.0	34600	1.7121
1.6453	26.0	35984	1.7068
1.6337	27.0	37368	1.7010
1.6376	28.0	38752	1.7065
1.6322	29.0	40136	1.7107
1.6293	30.0	41520	1.6902

Framework versions

• Transformers 4.42.4
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1