Fix the --allow-auto-truncate argument in tokenizer manager.

[hnyls2002]

For some unknown reason, when enabling the --allow-auto-truncate argument, the scheduler will perform the truncation, but the tokenizer manager will directly abort this request. This is not expected.

[gemini-code-assist]

Summary of Changes

Hello @hnyls2002, 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 addresses an issue where the tokenizer manager was not correctly handling the --allow-auto-truncate argument. Previously, even when auto-truncation was enabled, requests with input or total token counts exceeding the model's context length would result in an error. The changes ensure that the system now properly truncates input tokens or adjusts the requested max_new_tokens to fit within the context length when auto-truncation is allowed, preventing unexpected request abortions and improving the robustness of token length validation.

Highlights

• Corrected Input Truncation Logic: The _validate_one_request method now properly truncates input tokens when their length exceeds the model's context length and the --allow-auto-truncate flag is enabled, preventing unnecessary ValueError exceptions.
• Adjusted Max New Tokens for Truncation: When --allow-auto-truncate is active, the system will now automatically reduce the max_new_tokens parameter if the combined length of input and requested new tokens exceeds the context length, allowing requests to proceed without error.
• Introduced _max_req_len: A new variable _max_req_len is introduced to consistently define the maximum allowed request length, which is self.context_len - 1.

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 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 addresses an issue with the --allow-auto-truncate argument by adding truncation logic to the tokenizer manager. This prevents requests from being aborted when they could be truncated. My review identifies a critical bug where the input token list input_ids is not truncated in-place, rendering the truncation ineffective. I have also found a misleading log message that could cause confusion. I've provided code suggestions to resolve both issues.

[hnyls2002]

Forgot to push the in-place slice in this PR, will add it in #9388

-               input_ids = input_ids[:_max_req_len]
+               del input_ids[_max_req_len:]