Skip to content

Fine-tuning Wav2Vec2 model without reliance on Huggingface's Trainer

Notifications You must be signed in to change notification settings

thangld201/Fine-tuning-Wav2Vec2-Pytorch

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

20 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Fine-tuning-Wav2Vec2-Pytorch

Fine-tuning a Wav2Vec2 model without reliance on Huggingface's Trainer.

base folder contains a list of .txt and .wav files, in which a .wav file's transcript should be contained in the .txt file with the same name. main.sh bash script to fine-tune the model.
eval.sh bash script to evaluate the trained checkpoints on an evaluation dataset.

By default, this repository uses a Vietnamese checkpoint as in src/train_util.py. Feel free to change the original checkpoint depending on the language/model to be fine-tuned.

About

Fine-tuning Wav2Vec2 model without reliance on Huggingface's Trainer

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published