[v0.5.0] Release Tracker

[SS-JIA]

The v0.5.0 release will be cut from the "release/0.5" branch for critical fixes to the v0.4.0 release.

Branch Cut Date: 12/16/2024
Intended Release Date: 1/29/2025

This issue is for tracking cherry-picks to the release branch.

Cherry-Pick Criteria

Focus on stability and documentation correctness. Ideally only low risk changes may be cherry-picked from main:

1. Critical fixes in core components: build, exir, backends, runtime, sdk
2. Bug fixes in demos/examples. No new features/experiments
3. Critical bug fixes based on community feedback
4. Documentation improvements
5. Test/CI fixes
6. Release branch specific changes (e.g. change version identifiers)

Any core components changes and other change will require special dispensation from the release manager (currently @SS-JIA). If this applies to your change please write "Special Dispensation" in the "Criteria Category:" template below and explain.

Cherry-Pick Process

1. Ensure your PR has landed in master. This does not apply for release-branch specific changes (see Phase 1 criteria).

2. Create (but do not land) a PR against the release branch.
   Option 1: Use pytorchbot to cherry pick a PR that has been committed to the main branch using @pytorchbot cherry-pick command as follows.

   @pytorchbot cherry-pick --onto release/0.5 -c {regression,critical,fixnewfeature,docs,release}
   

   usage: @pytorchbot cherry-pick --onto ONTO [--fixes FIXES] -c
                            {regression,critical,fixnewfeature,docs,release}
   
   Cherry pick a pull request onto a release branch for inclusion in a release
   
   optional arguments:
     --onto ONTO           Branch you would like to cherry pick onto (Example: release/0.4)
     --fixes FIXES         Link to the issue that your PR fixes (i.e. https://github.com/pytorch/executorch/issues/2345)
     -c {regression,critical,fixnewfeature,docs,release}
                     A machine-friendly classification of the cherry-pick reason.
                     
   

   Option 2: Manually create a cherry pick PR using git command as follows.

   # Find the hash of the commit you want to cherry pick
   # (for example, abcdef12345)
   git log
   
   git fetch origin release/0.5
   git checkout release/0.5
   git cherry-pick -x abcdef12345
   
   # Submit a PR based against 'release/0.5' either:
   # via the GitHub UI
   git push my-fork
   
   # via the GitHub CLI
   gh pr create --base release/0.5
   
   

3. Make a request below with the following format:

Link to landed trunk PR:
* 

Link to release branch PR:
* 

Criteria category and reasons:
* 

1. Someone from the release team will reply with approved / denied or ask for more information.
2. If approved, someone from the release team will merge your PR once the tests pass. Do not land the release branch PR yourself.

Please note HUD Link with branch CI status and link to the HUD to be provided here.
HUD

Versions

0.5.0

[SS-JIA]

Link to landed trunk PR:

• N/A

Link to release branch PR:

• [2.6 RELEASE] Update PyTorch pin to 2.6 #7368

Criteria category and reasons:

• Self explanatory; update pytorch pin to match release.

[SS-JIA]

Link to landed trunk PR:

• Fix wheel build CI jobs #7371

Link to release branch PR:

• Fix wheel build CI jobs #7372

Criteria category and reasons:

• Fix CI on release branch

[mergennachin]

Link to landed trunk PR:

• Change status to beta in pyproject.toml file #7520

Link to release branch PR:

• Change status to beta in pyproject.toml file #7548

Criteria category and reasons:

• Fixing forward

---

@SS-JIA: Merged

[pytorchbot]

Link to landed trunk PR (if applicable):

• Upgrade numpy version requirement to 2.0 #7599

Link to release branch PR:

• Upgrade numpy version requirement to 2.0 #7659

Criteria Category:
Release - Coreml tools is not upgrading to latest ET because our version is still using an old numpy version (#6639) which they're not compatible anymore. It's good to have this change in 0.5

---

@SS-JIA: Unfortunately there seems to be a bunch of test failures. The error message may suggest an infra failure, so I triggered a retry.

@SS-JIA: Seems that it was indeed an infra failure, most signals have recovered. However, there are still some lint failures. Mind looking into those?

@mergennachin: Most likely will need this manual cherry-pick: #7727

@SS-JIA: Merged via #7727

[pytorchbot]

Link to landed trunk PR (if applicable):

• Add dim order assert #7561

Link to release branch PR:

• Add dim order assert #7660

Criteria Category:
Fixnewfeature -

---

@SS-JIA: All checks pass on the release branch PR. Merged!

[pytorchbot]

Link to landed trunk PR (if applicable):

• [mps] dim_order Fix #7563

Link to release branch PR:

• [mps] dim_order Fix #7661

Criteria Category:
Fixnewfeature -

---

@SS-JIA: All checks pass on the release branch PR. Merged!

[mergennachin]

@digantdesai @Gasoonjia

What about cherry-picking the original PR #7252? It's not part of the release/0.5 branch (missed it by a few hours)

And the latest one

#7658

@digantdesai : Songhao already made a list yesterday. @Gasoonjia can you please create requests here to be cherry-picked?

@lucylq

also, what about this? #7610

[lucylq]

Thanks @mergennachin. #7610 doesn't need to be in the release.

It doesn't affect any functionality, just changes the dim_order to be represented as int instead of bytes at the higher-level (serialization level is still bytes).

[pytorchbot]

Link to landed trunk PR (if applicable):

• make dim order as default for everywhere in ET #7658

Link to release branch PR:

• make dim order as default for everywhere in ET #7681

Criteria Category:
Fixnewfeature -

---

@SS-JIA: Merged. There are torchtune failures but I believe that should be fixed by #7727.