Skip to content

Pin jax to 0.11.0 in CI — jax 0.11.1 hangs the fori_loop cells on CPU - #37

Merged
mmcky merged 1 commit into
mainfrom
pin-jax-0110
Aug 19, 2026
Merged

Pin jax to 0.11.0 in CI — jax 0.11.1 hangs the fori_loop cells on CPU#37
mmcky merged 1 commit into
mainfrom
pin-jax-0110

Conversation

@mmcky

@mmcky mmcky commented Aug 19, 2026

Copy link
Copy Markdown
Contributor

jax/jaxlib 0.11.1 (released 2026-08-17 20:31 UTC) regresses lax.fori_loop on the CPU execution path: numpy_vs_numba_vs_jax executes in seconds under 0.11.0 and exceeds even an 1800 s cell timeout under 0.11.1 (locally bisected — the same jitted fori_loop hangs with and without the deprecated device=cpu argument). This repo had no workflow-level jax install at all, so the lecture pip cell resolved the broken release on every build — this is what failed the publish-2026aug19 run twice, at both 600 s and 1800 s ceilings. After merge, the on-push cache rebuild should finally carry the executed lecture, and the failed publish run can be re-run.

The full evidence chain, family exposure map, and the unpin condition are recorded in QuantEcon/workspace-lectures#49. One PR per repo with this same change; with jax pre-installed, the lecture's unpinned !pip install quantecon jax cell is satisfied and does not upgrade.

🤖 Generated with Claude Code

jax 0.11.1 (2026-08-17) regresses lax.fori_loop on the CPU path:
numpy_vs_numba_vs_jax executes in seconds under 0.11.0 and exceeds even
the new 1800s cell timeout under 0.11.1. This repo had no workflow-level
jax install, so the lecture's unpinned '!pip install quantecon jax' cell
resolved the broken release on every build. Full evidence and the
family-wide exposure map are recorded in
QuantEcon/workspace-lectures#49.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Copilot AI lite review requested due to automatic review settings August 19, 2026 02:17

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

This PR updates the repository’s GitHub Actions workflows to proactively install a known-good JAX version in CI/cache/publish runs, preventing a regression in lax.fori_loop CPU execution from breaking (or timing out) lecture builds.

Changes:

  • Add a dedicated workflow step to pip install "jax==0.11.0" (CPU) after the conda environment is created/activated.
  • Apply the same pin consistently across CI, cache rebuild, and publish workflows to keep execution/cache behavior aligned.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File Description
.github/workflows/publish.yml Pre-installs pinned JAX before building the site for GitHub Pages publishing.
.github/workflows/ci.yml Pre-installs pinned JAX to keep PR preview builds from pulling the regressed release.
.github/workflows/cache.yml Pre-installs pinned JAX so the main-branch cache rebuild executes lectures under the stable version.

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

@github-actions

Copy link
Copy Markdown

📖 Netlify Preview Ready!

Preview URL: https://pr-37--verdant-toffee-3261a8.netlify.app

Commit: 78d2d0c


Build Info

@mmcky
mmcky merged commit a3e1293 into main Aug 19, 2026
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants