Use reduce-scatter coalescing for FSDP (#6024) #324

	name: Build upstream image
	on:
	push:
	branches:
	- master
	- r[0-9]+.[0-9]+
	paths-ignore:
	- 'experimental/torch_xla2/**'
	workflow_dispatch:
	jobs:
	build:
	runs-on: linux.12xlarge
	timeout-minutes: 30
	env:
	ECR_DOCKER_IMAGE_BASE: 308535385114.dkr.ecr.us-east-1.amazonaws.com/pytorch/xla_base
	BAZEL_JOBS: 16
	steps:
	# See https://github.com/actions/checkout/issues/1014#issuecomment-1906802802
	- name: Clean up workspace
	run: \|
	ls -la
	sudo rm -rvf ${GITHUB_WORKSPACE}/*
	- name: Setup Linux
	uses: pytorch/test-infra/.github/actions/setup-linux@main
	- name: Checkout repo
	uses: actions/checkout@v3
	- name: Build Docker image
	shell: bash
	run: \|
	docker build -t "${ECR_DOCKER_IMAGE_BASE}:v1.2-lite" .github/upstream
	- name: Stage image to ECR
	shell: bash
	run: \|
	# This is to stage PyTorch/XLA base image for use in the upstream.
	# To allow the upstream workflow to access PyTorch/XLA build images, we
	# need to have them in the ECR. This is not expensive, and only pushes it
	# if image layers are not present in the repo.
	# Note: disable the following line while testing a new image, so we do not
	# push to the upstream.
	docker push "${ECR_DOCKER_IMAGE_BASE}:v1.2-lite"

Provide feedback