[Bug] [miniMax response] MiniMax-M2.7 No correct response found. No tool found for execution.

Search before asking

• I had searched in the issues and found no similar issues.

Operating system information

Linux

Python version information

=3.11

DB-GPT version

main

Related scenes

• Chat Data
• Chat Excel
• Chat DB
• Chat Knowledge
• Model Management
• Dashboard
• Plugins

Installation Information

• Installation From Source

• Docker Installation

• Docker Compose Installation

• Cluster Installation

• AutoDL Image

• Other

Device information

CPU 48C 64G

Models information

LLM: MiniMax-M2.7

What happened

using MiniMax-M2.7 (It seems that the output format does not match ， but Qwen-plus can respond to answers normally): the backend logs output query data and thought process normally, but the frontend result output fails and returns: No correct response found. Please check your response, which must be in the format indicated in the system prompt.

This is the newly added configuration for configs/dbgpt-proxy-mini-mysql.toml (copy from dbgpt-proxy-openai.toml): [models] [[models.llms]] name = "${env:LLM_MODEL_NAME:-MiniMax-M2.7}" provider = "${env:LLM_MODEL_PROVIDER:-proxy/minimax}" api_base = "${env:OPENAI_API_BASE:- https://api.minimaxi.com/v1 }" api_key = "${env:MINIMAX_API_KEY:-sk-xxx}"

Sometimes finally returns : It seems like all tool calls are failing with the same error. Let me try a simpler approach and just use the terminate tool directly as it might be the only working tool. ( No tool found for execution)

What you expected to happen

1. Has the system prompt been organized into a format that MiniMax does not accept
2. The original JSON returned by MiniMax ：log.txt

log.txt

How to reproduce

Configure to MiniMax, restart, Q&A

Additional context

Or, May I ask what the reason is? What other configurations do I need to do?

Are you willing to submit PR?

• Yes I am willing to submit a PR!