catch tensor.numel() == 0 in nan detector (#140741) #2

	name: unstable

	on:
	push:
	branches:
	- main
	tags:
	- ciflow/unstable/*
	workflow_dispatch:

	concurrency:
	group: ${{ github.workflow }}-${{ github.event.pull_request.number \|\| github.ref_name }}-${{ github.ref_type == 'branch' && github.sha }}-${{ github.event_name == 'workflow_dispatch' }}
	cancel-in-progress: true

	permissions: read-all

	jobs:
	# There must be at least one job here to satisfy GitHub action workflow syntax
	introduction:
	if: github.repository_owner == 'pytorch'
	runs-on: ubuntu-latest
	continue-on-error: true
	steps:
	- name: Introduce PyTorch unstable workflow
	run: \|
	echo "PyTorch unstable workflow is used to host experimental or flaky jobs"
	echo " that needs to be run for every commit, but doesn't block PR merging"
	echo " as part of the stable pull or trunk workflows."
	echo
	echo "In addition, a new label called ciflow/unstable can be attached to the"
	echo " PR to trigger this workflow. That can be done either manually or"
	echo " automatically using PyTorch auto-label bot."
	echo
	echo "Once the jobs are deemed stable enough (% red signal < 5% and TTS < 3h),"
	echo " they can graduate and move back to pull or trunk."

Provide feedback