NiloofarMomeni/distilhubert-finetuned-PQVD

distilhubert-finetuned-breathiness

This model is a fine-tuned version of ntu-spml/distilhubert on the PQVD dataset. It achieves the following results on the evaluation set:

• Loss: 1.2978
• Accuracy: 0.8163

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6428	1.0	74	0.6271	0.7279
0.5684	2.0	148	0.4996	0.7891
0.4783	3.0	222	0.4684	0.7823
0.2441	4.0	296	0.4405	0.8027
0.6583	5.0	370	0.4966	0.7891
0.3304	6.0	444	0.5008	0.8231
0.4209	7.0	518	0.5070	0.8435
0.2269	8.0	592	0.6253	0.8027
0.2272	9.0	666	0.9433	0.7755
0.0023	10.0	740	0.8471	0.8231
0.0776	11.0	814	1.0535	0.7891
0.0007	12.0	888	1.0649	0.8299
0.0004	13.0	962	1.0674	0.8231
0.0004	14.0	1036	1.1414	0.8231
0.0003	15.0	1110	1.3342	0.8095
0.0003	16.0	1184	1.2617	0.8163
0.0003	17.0	1258	1.2684	0.8231
0.0002	18.0	1332	1.2787	0.8231
0.0002	19.0	1406	1.2923	0.8163
0.0002	20.0	1480	1.2978	0.8163

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0