PyTorch Release 2.12.1

Hi everyone,

We are pleased to announce the patch release PyTorch 2.12.1. This release is being made following the standard PyTorch patch release process. The milestones for this release are:

  • M1: Patch Release Announcement — June 9
  • M2: Cherry-pick post deadline — June 12 (End of day, 5PM PST)
  • M3: All RC binaries ready for testing — June 15
  • M4: General Availability — June 17

Targeted issues and the cherry-pick tracker for this release are tracked on GitHub:

How to submit a cherry-pick

Land your fix in main first, then request the cherry-pick directly on the merged PR:

  @pytorchbot cherry-pick --onto release/2.12 -c <classification>

Only low-risk changes will be cherry-picked from main. All cherry-picks must be submitted before the M2 deadline above.

Criteria for cherry-picks

Accepted:

  • Low-risk, critical fixes for: silent correctness, backwards compatibility, crashes, deadlocks, (large) memory leaks
  • Fixes to regressions against the most recent release (2.12.0)
  • Documentation improvements
  • Release-branch-specific changes (e.g., blocking CI fixes, version identifiers)
  • Critical fixes to new features introduced in the 2.12.0 release

Not accepted:

  • Any feature work
  • Anything risky or large in scope
  • Numerical/perf changes that aren’t fixing a regression

Cheers,
Team PyTorch