search

theolepage / sslsv

Framework for training and evaluating self-supervised learning methods for speaker verification.
MIT License
19 stars 4 forks source link

sorry to bother you #3

Closed liyunlongaaa closed 1 year ago

liyunlongaaa commented 1 year ago

Hi, I would like to know the training time of your entire experiment with specific equipment, can you tell me?

theolepage commented 1 year ago

Hi @liyunlongaaa,

For example: training VICReg with the provided configuration takes ~10 minutes per epoch with 2x NVIDIA Titan X (Pascal) 12 GB.