[RES-804] Track additional per-user properties

[safaricd]

Change Description

With this change, we track additional properties for fit and predict events:

• torch version
• sklearn version
• Name of the GPU device

On the GDPR side, we are safe with tracking these as they will not allow us to fingerprint users based on these.

[gemini-code-assist]

Summary of Changes

Hello @safaricd, 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!

This pull request expands the telemetry data collected during model operations (fit and predict) to include details about the user's software and hardware environment. By capturing versions of key libraries like PyTorch and scikit-learn, along with GPU information, the system gains richer insights into usage patterns without compromising user privacy, as confirmed by GDPR considerations.

Highlights

• Telemetry Enhancement: Added tracking for torch version, sklearn version, and GPU device name for fit and predict events to gather more comprehensive environment data.
• New Utility Functions: Introduced _get_torch_version, _get_sklearn_version, _get_gpu_type, and a generic _get_package_version to efficiently retrieve software versions and hardware details.
• Event Class Refactoring: Created a new ModelCallEvent base class to centralize common properties and the newly added environment details, which FitEvent and PredictEvent now inherit from.

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 by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

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 pull request 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 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. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

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.

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 pull request adds tracking for torch and sklearn versions, as well as the GPU type, to the telemetry events. The introduction of a ModelCallEvent base class is a nice refactoring that improves code structure by reducing duplication between FitEvent and PredictEvent. My review includes a suggestion for the new _get_gpu_type function to make its return values less ambiguous and to improve its robustness. Overall, these are good additions for richer telemetry.

[noahho]

Very nice! Minor suggestions:

• [med] Add TPU detection?
• [high] add package versions: tabpfn-extensions, numpy, pandas, maybe autogluon (would help us understand if people ran us from AG as well?)
• [med] detect Colab / Kaggle / Notebook context
• [important] torch.version.cuda, torch.backends.cudnn.version()
• [important] python version!
• [important] "os": platform.system(),

[noahho]

see previous feedback, adding more info

[safaricd]

@noahho

[high] add package versions: tabpfn-extensions, numpy, pandas, maybe autogluon (would help us understand if people ran us from AG as well?)

Added these too.

[med] detect Colab / Kaggle / Notebook context

We already have methods for detecting whether someone runs inside a notebook or not - however, how would this information be used after all, from an analytics point of view?

[important] python version!

We track this since day 1.

[important] "os": platform.system(),

How would this information be used after all, from an analytics point of view? I think that having information about e.g. packages/dependencies is very useful, but is information about whether someone runs TabPFN on a Mac vs. Windows relevant to us?

[noahho]

System (windows / Mac / Unix) is very important for development, wea already have a bunch of code to support these and it's good to know how far we should go. For the colab this tells us about the kind of workload (production or experimentation?). Detecting haggle specifically would be great for gtm

[safaricd]

@noahho just added the following additional properties tracked across all events:

• platform/OS
• Runtime kernel (including Kaggle)

With that being said, I'll heard the changes and push a new PyPI version. Anything else you'd like to add?

[noahho]

Perfect, thanks a lot!