Commit 7ad1f3d4 authored by Yuefeng Zhou's avatar Yuefeng Zhou Committed by TensorFlower Gardener
Browse files

Refactor cross_device_ops so that when indexed slices are present in a batch,... 

Refactor cross_device_ops so that when indexed slices are present in a batch, batched all-reduce can still be done on dense gradients.

PiperOrigin-RevId: 233695231
parent 4a0bbf36
Show whitespace changes
Inline Side-by-side
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please to comment