DeepSpeed
DeepSpeed copied to clipboard
[Bug] - summing over DP instead of MP
Hi Deepspeed team!
Should this be summing over the MP group rather than the DP group? The comment says MP and the code does DP.
https://github.com/microsoft/DeepSpeed/blob/b3388e1418e80141a902886dec39127b851a1f0a/deepspeed/runtime/zero/stage_1_and_2.py#L1156