[misc] change default values of topk/topp

[hjh0119]

When computing the rollout importance sampling ratio in GRPO, we set vLLM's logprobs_mode to 'processed_logprobs' to get temperature-scaled log probabilities.

However, these logprobs are also affected by top_k and top_p filtering (which sets unselected token probabilities to zero), causing a significant discrepancy between rollout logprobs and training logprobs.

[gemini-code-assist]

Summary of Changes

Hello, 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 modifies the default behavior of text generation by setting top_k and top_p parameters to values that effectively disable their filtering by default. This change promotes more open-ended and diverse generation unless specific filtering is explicitly configured, and ensures the documentation accurately reflects these new defaults.

Highlights

• Default Generation Parameters: The default values for top_k and top_p generation parameters have been updated. top_k is now -1 (disabling top-k filtering) and top_p is 1.0 (disabling top-p filtering).
• Documentation Updates: The documentation for command-line parameters, in both Chinese and English, has been updated to reflect the new default values and clarify the meaning of top_k and top_p.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

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.

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 updates the default values for top_k and top_p parameters used in rollout sampling for RLHF training. The top_k default is changed from 50 to -1 (disabling top-k filtering), and top_p is changed from 0.9 to 1.0 (disabling top-p filtering). Corresponding documentation in both Chinese and English has been updated to reflect these new defaults and their meanings. There is no feedback to provide.