auspicious3000 / autovc

AutoVC: Zero-Shot Voice Style Transfer with Only Autoencoder Loss
https://arxiv.org/abs/1905.05879
MIT License
983 stars 207 forks source link

Why the batch size is set to 2? #47

Closed qq547276542 closed 4 years ago

qq547276542 commented 4 years ago

Is it because of memory limitations, or is it better than the result of a larger batch size (such as 16, 32, 64)?

auspicious3000 commented 4 years ago

small batch size leads to better generalization