fix(deps): update dependency torch to v2.6.0 [security]

[renovate-bot]

This PR contains the following updates:

Package	Change	Age	Adoption	Passing	Confidence
torch	==2.2.0 -> ==2.6.0

---

Warning

Some dependencies could not be looked up. Check the Dependency Dashboard for more information.

GitHub Vulnerability Alerts

CVE-2025-32434

Description

I found a Remote Command Execution (RCE) vulnerability in the PyTorch. When load model using torch.load with weights_only=True, it can still achieve RCE.

Background knowledge

https://github.com/pytorch/pytorch/security
As you can see, the PyTorch official documentation considers using torch.load() with weights_only=True to be safe.

Since everyone knows that weights_only=False is unsafe, so they will use the weights_only=True to mitigate the security issue.
But now, I just proved that even if you use weights_only=True, it still can achieve RCE.
So it is time to update your PyTorch version~.

Credit

This vulnerability was found by Ji'an Zhou.

---

Configuration

📅 Schedule: Branch creation - At any time (no schedule defined), Automerge - At any time (no schedule defined).

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Never, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by Mend Renovate. View the repository job log.

[gemini-code-assist]

Hello @renovate-bot, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

Summary of Changes

This pull request updates the torch dependency from version 2.2.0 to 2.6.0 in the pyproject.toml file. This update is classified as a security update, addressing CVE-2025-32434, a Remote Command Execution (RCE) vulnerability. The PR was automatically generated by Mend Renovate.

Highlights

• Dependency Update: Updates the torch dependency from version 2.2.0 to 2.6.0.
• Security Fix: Addresses CVE-2025-32434, a Remote Command Execution vulnerability in torch.

Changelog

• people-and-planet-ai/weather-forecasting/serving/weather-model/pyproject.toml
  • Updates torch dependency from 2.2.0 to 2.6.0 to address a security vulnerability.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in issue comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist is currently in preview and may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments to provide feedback.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

---

A torch update's call,
Stops RCE for all.
New version's bright,
Securing the night,
No commands remotely install.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This PR updates the torch dependency from version 2.2.0 to 2.6.0, addressing a security vulnerability (CVE-2025-32434) that allows for Remote Command Execution (RCE) even when using torch.load with weights_only=True. The update is reflected in the pyproject.toml file. This is a critical update due to the security implications.

Summary of Findings

• Security Vulnerability: The PR addresses a critical security vulnerability (CVE-2025-32434) in the torch library, which could lead to Remote Command Execution (RCE). This vulnerability exists even when using torch.load with weights_only=True, which was previously considered a safe mitigation.
• Dependency Update: The torch dependency is updated from version 2.2.0 to 2.6.0 in the pyproject.toml file. This update is crucial to mitigate the identified security vulnerability.
• Notebook Consistency: The comment in pyproject.toml reminds to ensure the torch version matches the container_uri in notebooks/3-training.ipynb. It's important to verify this consistency to avoid potential compatibility issues.

Merge Readiness

The pull request addresses a critical security vulnerability and updates the torch dependency accordingly. Given the severity of the vulnerability, it is highly recommended to merge this PR as soon as possible after verifying the consistency of the torch version across the project, especially in the notebooks/3-training.ipynb file. I am unable to directly approve this pull request, and strongly recommend that others review and approve this code before merging. The security fix warrants immediate attention and deployment.