Bump torch from 2.4.0+cu121 to 2.12.0 in /experiments/agentcompany/openhands

[dependabot]

Bumps torch from 2.4.0+cu121 to 2.12.0.

Release notes

Sourced from torch's releases.

PyTorch 2.12.0 Release Notes

• Highlights
• Backwards Incompatible Changes
• Deprecations
• New Features
• Improvements
• Bug fixes
• Performance
• Documentation
• Developers
• Security

Highlights

For more details about these highlighted features, you can look at the release blogpost. Below are the full release notes for this release.

Backwards Incompatible Changes

Build Frontend

• Strengthened SVE compile checks in FindARM.cmake, which may reject previously accepted but incorrect SVE configurations (#176646)

  Source builds that enable SVE now validate the compiler configuration more strictly. If a build previously passed with an incomplete or mismatched SVE setup, it may now fail during CMake configuration instead of later in compilation. Update the compiler/toolchain flags so they accurately describe the target SVE support, or disable SVE for that build.

• Updated the minimum CUDA version required to build PyTorch from source to CUDA 12.6 (#178925)

  Building PyTorch from source with CUDA versions older than 12.6 is no longer supported. Users building custom binaries should install CUDA 12.6 or newer and make sure CUDA_HOME points to that installation.

  Version 2.11:

  CUDA_HOME=/usr/local/cuda-12.4 python setup.py develop

  Version 2.12:

  CUDA_HOME=/usr/local/cuda-12.6 python setup.py develop

• Enforced a C++20 minimum in CMake build files (#178662)

... (truncated)

Commits

• See full diff in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)
  You can disable automated security fix PRs for this repo from the Security Alerts page.

Greptile Summary

This dependabot PR bumps torch from 2.4.0+cu121 to 2.12.0 in the OpenHands experiment requirements, but does not update the tightly-coupled companion packages torchaudio and torchvision which remain pinned to their torch 2.4.x-compatible versions.

• torchaudio==2.4.0+cu121 and torchvision==0.19.0 link against torch's versioned C++ ABI; importing either with torch==2.12.0 will raise a version mismatch error at runtime.
• triton==3.0.0 corresponds to the torch 2.4.x release series and also needs updating.
• The +cu121 CUDA suffix was dropped from the new torch pin while the companion packages still reference CUDA 12.1 binaries, adding further incompatibility.

Confidence Score: 1/5

Not safe to merge — the environment will be broken on install due to version mismatches between torch and its companion packages.

Bumping torch alone without updating torchaudio, torchvision, and triton leaves the dependency set in a state where importing any of those packages raises a RuntimeError due to ABI incompatibility. Any code that imports these packages will fail immediately.

experiments/agentcompany/openhands/requirements.txt — torchaudio, torchvision, and triton all need to be updated to versions compatible with torch 2.12.0.

Important Files Changed

Filename	Overview
experiments/agentcompany/openhands/requirements.txt	Bumps torch from 2.4.0+cu121 to 2.12.0, but leaves torchaudio (2.4.0+cu121), torchvision (0.19.0), and triton (3.0.0) at versions that are only compatible with torch 2.4.x, causing a broken environment.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[torch==2.12.0\nnew pin] -->|ABI incompatible| B[torchaudio==2.4.0+cu121\nunchanged]
    A -->|ABI incompatible| C[torchvision==0.19.0\nunchanged]
    A -->|version mismatch| D[triton==3.0.0\nunchanged]
    B --> E[RuntimeError at import]
    C --> E
    D --> E

Loading

Prompt To Fix All With AI

Fix the following 1 code review issue. Work through them one at a time, proposing concise fixes.

---

### Issue 1 of 1
experiments/agentcompany/openhands/requirements.txt:177-179
**Version mismatch between torch and its companion packages**

`torchaudio==2.4.0+cu121` and `torchvision==0.19.0` are only compatible with `torch==2.4.x`; they link against that specific version of torch's C++ ABI. After this bump to `torch==2.12.0`, importing either package will raise a `RuntimeError` about a torch version mismatch at startup. The CUDA suffix was also dropped from the new torch pin (no `+cu12x`), which may mean a different CUDA variant or the CPU wheel — either way, the companion packages still expect CUDA 12.1 binaries. To align, `torchaudio` and `torchvision` should be bumped to their respective `2.12.0` and `0.22.0` (or whatever the matching release is) compatible versions, and `triton==3.0.0` (which corresponds to torch 2.4.x) likely needs updating as well.

_{Reviews (1): Last reviewed commit: "Bump torch in /experiments/agentcompany/..." | Re-trigger Greptile}

Greptile also left 1 inline comment on this PR.

[greptile-apps]

Version mismatch between torch and its companion packages

torchaudio==2.4.0+cu121 and torchvision==0.19.0 are only compatible with torch==2.4.x; they link against that specific version of torch's C++ ABI. After this bump to torch==2.12.0, importing either package will raise a RuntimeError about a torch version mismatch at startup. The CUDA suffix was also dropped from the new torch pin (no +cu12x), which may mean a different CUDA variant or the CPU wheel — either way, the companion packages still expect CUDA 12.1 binaries. To align, torchaudio and torchvision should be bumped to their respective 2.12.0 and 0.22.0 (or whatever the matching release is) compatible versions, and triton==3.0.0 (which corresponds to torch 2.4.x) likely needs updating as well.

Prompt To Fix With AI

This is a comment left during a code review.
Path: experiments/agentcompany/openhands/requirements.txt
Line: 177-179

Comment:
**Version mismatch between torch and its companion packages**

`torchaudio==2.4.0+cu121` and `torchvision==0.19.0` are only compatible with `torch==2.4.x`; they link against that specific version of torch's C++ ABI. After this bump to `torch==2.12.0`, importing either package will raise a `RuntimeError` about a torch version mismatch at startup. The CUDA suffix was also dropped from the new torch pin (no `+cu12x`), which may mean a different CUDA variant or the CPU wheel — either way, the companion packages still expect CUDA 12.1 binaries. To align, `torchaudio` and `torchvision` should be bumped to their respective `2.12.0` and `0.22.0` (or whatever the matching release is) compatible versions, and `triton==3.0.0` (which corresponds to torch 2.4.x) likely needs updating as well.

How can I resolve this? If you propose a fix, please make it concise.