Any way to utilize multiple GPUs on DFP training? i am not sure if dfencode has this option
Ver 23.03 on DGX 4xA100 ,
in order to train using all GPUs does it matter if using DFP jupyter production docker or morpheus-sdk docker?
The current implementation of the DFEncoder targets a specific CUDA device but our engineers are going to take a look at the issue you created as an enhancement:
1 Like