chenhaoxing / HDNet

This is the implementation of paper 'Hierarchical Dynamic Image Harmonization' (ACM MM'2023, Oral).

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

PSA Do not train with multiple gpus

mlinmg opened this issue · comments

Just for anyone interested in training this model. Do not use multiple gpus, it shows massive performance dump.
For instance, I was getting 1.49s/i with two gpus at batch size 12, and with one gpu I get 5.53it/s

I thought It might help someone in the future