Vietnamese_Wav2Vec_Finetune_round4

This model is a fine-tuned version of pdabo1607/Vietnamese_Wav2Vec_Finetune_round3 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5299
• Wer: 0.2952

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: 3e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.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_steps: 500
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.6354	0.9434	400	0.5751	0.3284
0.605	1.8868	800	0.5745	0.3228
0.6128	2.8302	1200	0.5783	0.3196
0.594	3.7736	1600	0.5680	0.3148
0.585	4.7170	2000	0.5495	0.3163
0.5651	5.6604	2400	0.5499	0.3102
0.5789	6.6038	2800	0.5518	0.3088
0.5571	7.5472	3200	0.5376	0.3084
0.565	8.4906	3600	0.5444	0.3033
0.564	9.4340	4000	0.5450	0.3012
0.5701	10.3774	4400	0.5336	0.3008
0.5608	11.3208	4800	0.5334	0.2981
0.5765	12.2642	5200	0.5292	0.2958
0.5655	13.2075	5600	0.5328	0.2953
0.5857	14.1509	6000	0.5299	0.2952

Framework versions

• Transformers 4.46.3
• Pytorch 2.5.1+cu121
• Datasets 4.4.1
• Tokenizers 0.20.3