Conversation
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
🪼 branch checks and previews
|
🦄 no changes detectedThis Pull Request does not include changes to any packages.__No changes detected. __
|
🪼 branch checks and previews
Install Trackio from this PR (includes built frontend) pip install "https://huggingface.co/buckets/trackio/trackio-wheels/resolve/adaeee3b8988f7ee72e419f9d478ceee1a591689/trackio-0.25.1-py3-none-any.whl" |
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
There was a problem hiding this comment.
Pull request overview
This PR adds documentation and a runnable example demonstrating Trackio’s native integration with Hugging Face TRL trainers, and updates the docs sidebar labels to make integration guides easier to discover.
Changes:
- Add a new TRL integration doc page with an
SFTTrainersnippet and configuration example. - Add an end-to-end
examples/trl-integration.pyscript intended to run quickly with a tiny test model. - Rename Integration section sidebar titles for better discoverability and add the TRL page to the toctree.
Reviewed changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
examples/trl-integration.py |
Adds a runnable TRL SFT example configured to report to Trackio. |
docs/source/trl_integration.md |
New TRL integration documentation page mirroring the Transformers guide structure. |
docs/source/_toctree.yml |
Updates Integration section titles and includes the new TRL integration page. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| # "trl>=1.2.0", | ||
| # "datasets>=4.4.0", | ||
| # "transformers[torch]>=5.0.0rc2", | ||
| # "huggingface_hub>=1.0.0", |
There was a problem hiding this comment.
Good catch — removed. Also dropped datasets and transformers[torch] since trl already pulls them in transitively.
| # "datasets>=4.4.0", | ||
| # "transformers[torch]>=5.0.0rc2", |
There was a problem hiding this comment.
I'm not sure this is needed: trl already depends on datasets, transformers and torch
Co-authored-by: Quentin Gallouédec <45557362+qgallouedec@users.noreply.github.com>
Co-authored-by: Quentin Gallouédec <45557362+qgallouedec@users.noreply.github.com>
Co-authored-by: Quentin Gallouédec <45557362+qgallouedec@users.noreply.github.com>
Co-authored-by: Quentin Gallouédec <45557362+qgallouedec@users.noreply.github.com>
Addresses Copilot's note that huggingface_hub is unused, and qgallouedec's note that datasets/transformers/torch are already pulled in via trl. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-authored-by: Quentin Gallouédec <45557362+qgallouedec@users.noreply.github.com>
|
Thanks so much @qgallouedec for the cleanup! |
|
Confirmed example runs locally for me, merging when CI is green to get the docs live |
docs/source/trl_integration.md) mirroring the Transformers guide, with a runnableSFTTrainersnippet and project/space configuration example.examples/trl-integration.py, a complete end-to-end SFT example using a tiny test model that runs in well under a minute. Verified locally.For ML-Intern -- see internal context: https://huggingface.slack.com/archives/C08SW1X12C8/p1777305610753559?thread_ts=1777063926.876859&cid=C08SW1X12C8