You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
On A100 80G, seting train_batch_size = 18, using fp16, one epoch of training takes 17 days, and I have pre-compute all the T5 and VAE features in advance using the script tools/extract_features.py
The text was updated successfully, but these errors were encountered:
On A100 80G, seting train_batch_size = 18, using fp16, one epoch of training takes 17 days, and I have pre-compute all the T5 and VAE features in advance using the script tools/extract_features.py
The text was updated successfully, but these errors were encountered: