Fix #20429 — Attention layer hardcodes 3-D shape assumptions, breaking N-D inputs
#10757
actions.yml
on: pull_request
Check the code format
2m 15s
Matrix: build