hi,mousavi,
first, thanks to EQTransformer!
I got an issue : I use eqt on my own dataset without changing model structure or training code
But when I run the code on multiple GPUs, the usage on every GPU is so low, and the time of one epoch is more than using one gpu,
hi,mousavi, first, thanks to EQTransformer! I got an issue : I use eqt on my own dataset without changing model structure or training code But when I run the code on multiple GPUs, the usage on every GPU is so low, and the time of one epoch is more than using one gpu,
can you give any suggestion please?