Skip to content

Support backend OpenVINO #33

Support backend OpenVINO

Support backend OpenVINO #33

Workflow file for this run

# Cross-platform VLA regression CI.
# Fires when a PR from `dev` targets `main`. Runs on the self-hosted runner
# labelled `vla-ci-orchestrator` (register your orchestrator host with that
# label - the actual hostname stays out of the repo, in ci/config/hosts.env).
# Every platform is a remote server reached over the LAN via vla-ci-agent (no
# SSH); the orchestrator drives one LIBERO client per platform. The three
# platforms are swept CONCURRENTLY in a single job (a self-hosted runner runs one
# job at a time, so the parallelism lives inside the job via `orchestrate.sh all`).
name: vla-ci
on:
pull_request:
branches: [main]
concurrency:
group: vla-ci-${{ github.ref }}
cancel-in-progress: true
jobs:
ci:
# Only PRs whose source branch is `dev`.
if: github.head_ref == 'dev'
runs-on: [self-hosted, vla-ci-orchestrator]
timeout-minutes: 240
steps:
- uses: actions/checkout@v7
- name: Sweep all platforms (parallel) + gate
run: bash ci/orchestrate.sh all
- uses: actions/upload-artifact@v7
if: always()
with:
name: ci-results
path: outputs/ci/