[Frontend] OpenAI Responses API supports Tool/Function calling #20874
Conversation
|
👋 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 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 🚀 |
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Summary of Changes
Hello @chaunceyjiang, 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 significantly extends the functionality of the OpenAI Responses API by adding support for tool and function calling. The changes involve modifying the request processing pipeline to interpret tool definitions, dynamically generating guided decoding constraints for model output, and restructuring the API response to include detailed function call information. This enables the system to leverage external tools and APIs through the language model's capabilities.
Highlights
- Tool/Function Calling Support: Implemented comprehensive support for OpenAI-style tool/function calling within the Responses API, enabling the model to generate structured tool calls based on provided tool definitions and user prompts.
- Dynamic Guided Decoding for Tools: Enhanced the guided decoding mechanism to dynamically generate JSON schemas for tool calls. This allows the model to be constrained to output valid JSON for function arguments, based on the
tool_choiceandtoolsparameters in the request. - Extended Response Format: Modified the
ResponsesResponseoutput type to includeResponseFunctionToolCallobjects. This ensures that the API can return structured information about the function calls generated by the model. - Tool Parsing Integration: Integrated a
ToolParserto extract and format function calls from the model's generated output. This supports varioustool_choicemodes, including 'auto', 'required', and specific named function calls. - New Configuration Options: Introduced new configuration options (
enable_auto_tools,expand_tools_even_if_tool_choice_none) to provide granular control over tool parsing behavior and how tool definitions are handled in the prompt.
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
-
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. ↩
|
This pull request has merge conflicts that must be resolved before it can be |
There was a problem hiding this comment.
Code Review
This pull request introduces Tool/Function calling support to the OpenAI Responses API. The changes involve protocol modifications, serving engine updates, and response serving logic enhancements. The review identifies a critical bug related to empty responses, a high-severity issue concerning data mutation, and some medium-severity issues regarding debug statements and type checking.
mergify
bot
added
deepseek
|
@WoosukKwon @aarnphm PTAL. |
| self, request: ChatCompletionRequest) -> ChatCompletionRequest: | ||
| self, request: Union[ChatCompletionRequest, ResponsesRequest] | ||
| ) -> Union[ChatCompletionRequest, ResponsesRequest]: | ||
| if not isinstance(request, ChatCompletionRequest): |
There was a problem hiding this comment.
This will come back to bite us when we introduce sampling_params support for ResponsesRequest but forget to update these lines
There was a problem hiding this comment.
This is because ResponsesRequest does not have the skip_special_tokens attribute, which forces me to add this check.
Should I add a skip_special_tokens attribute to ResponsesRequest and mark it as not yet implemented for now? What do you think?
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com> [Frontend] OpenAI Responses API supports Tool/Function calling Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
OpenAI Responses API supports Tool/Function calling
Follow up #20504