MULTI GPU error #169
Description
when i train it, all GPUs are running but it works independently. when i set batchsize=4 and use 4 A6000 gpus, CUDA out of memory still show. why is this happening?
{{ message }}
when i train it, all GPUs are running but it works independently. when i set batchsize=4 and use 4 A6000 gpus, CUDA out of memory still show. why is this happening?