feat: add design spec for additional prompt classification #358
Conversation
Signed-off-by: Huamin Chen <[email protected]>
✅ Deploy Preview for vllm-semantic-router ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
👥 vLLM Semantic Team NotificationThe following members have been identified for the changed files in this PR and have been automatically assigned: 📁
|
There was a problem hiding this comment.
Pull Request Overview
This PR introduces a comprehensive design specification for additional prompt classification routing capabilities in the vLLM Semantic Router. The proposal outlines a unified content scanning and routing framework that extends beyond the current BERT-based classification with three new signal sources: keyword-based routing, regex content scanning, and embedding similarity scanning.
Key changes include:
- Design specification for a multi-signal routing system with both in-tree and out-of-tree execution paths
- Detailed architecture for signal fusion layer that combines multiple classification approaches
- Comprehensive configuration schema for the new routing capabilities
Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.
| - Use BERT results: `bert.category == 'computer science'` | ||
| - Combine signals: `keyword.security.matched && bert.category == 'security'` | ||
|
|
||
| ## Configuration Schema |
There was a problem hiding this comment.
we can consider adding a configuration schema example for MCP
|
|
||
| ## Integration with Existing Router | ||
|
|
||
| ### Request Processing Flow |
There was a problem hiding this comment.
If possible, we could add a practical example here for how to integrate with a well recognized router for proposed prompt classification feature.
|
Looks good, thanks for adding it! |
What type of PR is this?
Design doc for #313 #200
What this PR does / why we need it:
Which issue(s) this PR fixes:
Fixes #
Release Notes: Yes/No