search

facebookresearch / fairseq

Facebook AI Research Sequence-to-Sequence Toolkit written in Python.
MIT License
30.27k stars 6.39k forks source link

Topic classification about XLSR-53，did I do something wrong? #3406

Open Rambo06 opened 3 years ago

Rambo06 commented 3 years ago

❓ Questions and Help

Hello， I used XLSR-53 for topic classification on the Chinese dataset and the accuracy rate can reach 80%. But when I use 1200 hours of Chinese data for self-supervised training after 50 epochs. The accuracy becomes 50%.

clipboard

I would like to ask, how many hours of data does self-training XLSR-53 need? How many epochs need to be trained ？ Did I do something wrong?

I use the following code： https://github.com/mailong25/self-supervised-speech-recognition python3 pretrain.py --fairseq_path path/to/libs/fairseq --audio_path path/to/audio_directory --init_model path/to/xlsr_53_56k.pt

Thanks for your answer!

stale[bot] commented 3 years ago

This issue has been automatically marked as stale. If this issue is still affecting you, please leave any comment (for example, "bump"), and we'll keep it open. We are sorry that we haven't been able to prioritize it yet. If you have any new additional information, please include it with your comment!