Skip to content

Conversation

@ultmaster
Copy link
Contributor

No description provided.

Copilot AI review requested due to automatic review settings December 23, 2025 08:13
Copy link
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

This PR fixes documentation and dashboard building issues by updating the build configuration and improving documentation clarity. The changes focus on ensuring dashboard files are properly included in builds and making the trajectory aggregation documentation more concrete with example values.

  • Restructured the build configuration to properly include dashboard artifacts in both wheel and sdist builds
  • Added concrete example values for trajectory aggregation configuration parameters
  • Enhanced the trajectory aggregation documentation with clearer explanations

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File Description
pyproject.toml Updated build configuration to add dashboard artifacts and improved path patterns for exclusions by using absolute paths (with "/" prefix)
agentlightning/algorithm/verl/interface.py Enhanced trajectory aggregation documentation with concrete example values and expanded explanation of configuration parameters

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

"trajectory_max_prompt_length": ...,
"trajectory_max_response_length": ...,
"trajectory_max_prompt_length": 4096,
"trajectory_max_response_length": 34384,
Copy link

Copilot AI Dec 23, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The value 34384 appears to be a typo. This seems like an unusually specific number that may have been a typo for 32768 (a common power-of-2 token limit) or possibly 4096 to match the prompt length. Please verify this is the intended value.

Suggested change
"trajectory_max_response_length": 34384,
"trajectory_max_response_length": 32768,

Copilot uses AI. Check for mistakes.
Comment on lines 51 to 54
and `trajectory_max_response_length` should be set to the maximum length of the prompt for the first turn
and the maximum length for the cumulative agent responses in the full trajectory, respectively.
Toggle `debug=True` plus `unmatch_log_dir` when you need to inspect retokenization or chat-template
mismatches. See [this blog post](https://agent-lightning.github.io/posts/trajectory_level_aggregation/)
Copy link

Copilot AI Dec 23, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The sentence structure is awkward with the long run-on sentence. Consider breaking this into two sentences for better readability. For example: "trajectory_max_prompt_length and trajectory_max_response_length should be set to the maximum length of the prompt for the first turn and the maximum length for the cumulative agent responses in the full trajectory, respectively. Toggle debug=True plus unmatch_log_dir when you need to inspect retokenization or chat-template mismatches."

Suggested change
and `trajectory_max_response_length` should be set to the maximum length of the prompt for the first turn
and the maximum length for the cumulative agent responses in the full trajectory, respectively.
Toggle `debug=True` plus `unmatch_log_dir` when you need to inspect retokenization or chat-template
mismatches. See [this blog post](https://agent-lightning.github.io/posts/trajectory_level_aggregation/)
should be set to the maximum length of the prompt for the first turn, and
`trajectory_max_response_length` should be set to the maximum cumulative length of agent responses
in the full trajectory. Toggle `debug=True` plus `unmatch_log_dir` when you need to inspect
retokenization or chat-template mismatches. See
[this blog post](https://agent-lightning.github.io/posts/trajectory_level_aggregation/)

Copilot uses AI. Check for mistakes.
@ultmaster
Copy link
Contributor Author

/ci

@github-actions
Copy link

github-actions bot commented Dec 23, 2025

🚀 CI Watcher for correlation id-3686165640-mjigucv9 triggered by comment 3686165640
🏃‍♀️ Tracking 1 workflow run(s):

✅ All runs completed.

Copy link
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.


💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

length of agent responses in the full trajectory.
Toggle `debug=True` plus `mismatch_log_dir` when you need to inspect
retokenization or chat-template mismatches. See
[this blog post](https://agent-lightning.github.io/posts/trajectory_level_aggregation/)
Copy link

Copilot AI Dec 23, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The documentation formatting is incomplete. The blog post link line should end with "for more details." to complete the sentence.

Suggested change
[this blog post](https://agent-lightning.github.io/posts/trajectory_level_aggregation/)
[this blog post](https://agent-lightning.github.io/posts/trajectory_level_aggregation/) for more details.

Copilot uses AI. Check for mistakes.
@ultmaster ultmaster merged commit 158f5df into main Dec 23, 2025
64 of 65 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants