DuyTa/vi_whisper

Whisper Base Vi - DuyTa

This model is a fine-tuned version of openai/whisper-base on the Vivos dataset. It achieves the following results on the evaluation set:

• Loss: 0.2565
• Wer: 25.0583

Model description

Finetune Whisper model on Vietnamese Dataset

Intended uses & limitations

More information needed

Training and evaluation data

Vivos

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 4000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2096	1.37	1000	0.2949	32.0383
0.1205	2.74	2000	0.2548	26.8583
0.0767	4.12	3000	0.2549	25.3432
0.0532	5.49	4000	0.2565	25.0583

Framework versions

• Transformers 4.31.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.13.1
• Tokenizers 0.13.3