[Bugfix] The special_tokens in tokenizer should also be controlled by do_lower_case in encoder_config.

[noooop]

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan, such as providing test command.
• The test results, such as pasting the results comparison before and after, or e2e results
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model.

Purpose

Bert model special_tokens are uppercase, such as [UNK], but "do_lower_case": true in
sentence_bert_config.json

from vllm import LLM

model_name = "BAAI/bge-base-en"

model = LLM(model=model_name)


prompts = ['[UNK]' * 510]
outputs = model.embed(prompts)
for prompt, output in zip(prompts, outputs):
    pass
# ValueError: The decoder prompt (length 2042) is longer than the maximum model length of 512. Make sure that `max_model_len` is no smaller than the number of text tokens.

Test Plan

tests/tokenization/test_do_lower_case.py

Test Result

pass

(Optional) Documentation Update

Fix #20744
Fix #19798

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.

🚀

[gemini-code-assist]

Summary of Changes

Hello @noooop, 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 resolves a critical tokenization bug that caused special tokens to be improperly lowercased, particularly impacting models configured with do_lower_case: true. The fix ensures that these essential tokens retain their original casing, allowing models to process inputs correctly and preventing errors during embedding generation. This directly addresses issues #20744 and #19798.

Highlights

• Bugfix: Special Token Lowercasing: This pull request addresses a bug where special tokens (e.g., [UNK]) were incorrectly lowercased when the do_lower_case option was enabled in the tokenizer configuration. This issue specifically affected models like Bert, which expect these special tokens to retain their original casing.
• Tokenization Logic Enhancement: Introduced a post-processing step within the _tokenize_prompt function in vllm/inputs/preprocess.py. After the entire prompt is lowercased, this new logic iterates through all known special tokens and, if their lowercase form differs from their original form, it replaces the lowercased special token occurrences in the prompt back to their original casing. This ensures proper recognition by the tokenizer.

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 is currently in preview and 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 to provide feedback.

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 fixes an issue with special tokens being lowercased when do_lower_case is true. The approach is to revert the lowercasing for special tokens after lowercasing the whole prompt. My review identifies a potential correctness issue with the implementation where overlapping special tokens might not be handled correctly, and also a performance concern. I've provided a suggestion to fix the correctness issue by sorting tokens before replacement.

[noooop]

cc @DarkLight1337

[mergify]

This pull request has merge conflicts that must be resolved before it can be
merged. Please rebase the PR, @noooop.

https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork

[mergify]

This pull request has merge conflicts that must be resolved before it can be
merged. Please rebase the PR, @noooop.

https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork

[noooop]

@DarkLight1337

Please take a look at this thread

The method for fixing it may already be efficient enough now.

[DarkLight1337]

LGTM then, thanks