Relay and Reuse: The Dual Engine Behind PyTorch Out-of-Tree Release Readiness
PyTorch Conference North America 2026
Oct 20, 2026 · 11:10–11:35 · Timezone not confirmed
For heterogeneous accelerators, keeping pace with PyTorch's fast bi-monthly release cycle is a constant uphill battle. This session shares how we sustain high quality while shipping stable releases within a month of each upstream update. Test Refactoring & Device-Agnostic Reuse: We show how instantiate device type_tests and dynamic skipping break the tie between tests and hardware. The payoff: 580K+ community test cases become reusable out of the box, giving full API semantic parity at almost no extra cost. Cross-Repo CI Relay (CRCR) & Quality Gates: CRCR wires every PyTorch PR directly to accelerator code, co-compiling and testing both in real time. By validating upstream changes against the backend before they merge, it catches regressions at the source, easing adaptation and locking in post-release stability. Together, reuse and cross-repo CI let an out-of-tree backend ship high-quality releases within 30 days. Benefits: this blueprint lowers the barrier to hardware integration, while CRCR doubles as an early-warning system for PyTorch Core — flagging regressions before they spread, so every user gets the newest features at once, with no fragmentation.
Session status: scheduled
Room: LL21DEF
Track: Core PyTorch
Level: Any
Saving a session is not registration. Check the official agenda for availability, eligibility and any additional fees.
Speakers and their event affiliations
- Jiahao Chen
Affiliation at this event: Huawei
- Jiahao Tan
Affiliation at this event: Huawei
Parent event
In person · San Jose Convention Center, San Jose, California, United States
This describes the parent event. Confirm the session’s attendance requirements with the organizer.
Companies participating in the parent event
These are event-level relationships, not confirmed sponsorship of this session.
- Adaption · Speaker company Source
- Agentic AI Foundation · Speaker company Source
- Amazon · Speaker company Source
- Amazon Web Services · Speaker company Source
- Amazon Web Services · Sponsor Source
- Amazon.com · Speaker company Source
- AMD · Speaker company Source
- AMD · Sponsor Source
- Annapurna Labs · Speaker company Source
- Annapurna ML, Amazon · Speaker company Source
- Anyscale · Sponsor Source
- Apple · Speaker company Source
- Argonne National Laboratory · Speaker company Source
- Arm · Speaker company Source
- Arm · Sponsor Source
- AWS Annapurna Labs · Speaker company Source
- Baseten · Speaker company Source
- Baseten · Sponsor Source
- Beijing Academy of Artificial Intelligence · Speaker company Source
- Bird of Paradise AI · Speaker company Source
- Boson AI Inc. · Speaker company Source
- BosonAI · Speaker company Source
- ByteDance · Speaker company Source
- Capital One · Speaker company Source
Showing 24 of 130 company relationships. View all on the event page.