-
Notifications
You must be signed in to change notification settings - Fork 36
Description
I am encountering an error when sending subsequent requests to task_maistro. The first one is processed properly. However, the 2nd sporadicly throws this error. It happens about 80% of the time:
openai.BadRequestError: Error code: 400 - {'error': {'message': "Invalid parameter: messages with role 'tool' must be a response to a preceding message with 'tool_calls'.", 'type': 'invalid_request_error', 'param': 'messages.[5].role', 'code': None}}
Expected Behavior:
The tool calls should be correctly sequenced, and the API should process the messages without errors.
Actual Behavior:
The API returns a 400 error indicating that messages with the role 'tool' must be a response to a preceding message with 'tool_calls'.
Additional Context:
- The error seems to be related to the sequencing of messages in the graph.
- The traceback indicates the error occurs during the execution of the task_mAIstro function.
- I have attempted to ensure correct sequencing and error handling, but the issue persists.
- Any insights or suggestions for resolving this error would be greatly appreciated.
- I'm using Azure OpenAI GPT-4o
Environment:
Operating System: macOS
Python version: 3.11.11
Langgraph version: 0.2.60
Langchain Core Version: 0.3.29
Langchain Version: 0.3.14
Platform: Linux-6.12.5-orbstack-00287-gf8da5d508983-aarch64-with-glibc2.36
Runtime: Python
SDK: langsmith-py 0.2.10
Run Metadata:
Assistant ID: ea4ebafa-a81d-5063-a5fa-67c755d98a21
Created By: system
Graph ID: task_maistro
Langgraph Host: self-hosted
Langgraph Plan: developer
Langgraph Version: 0.2.60
LANGSMITH_LANGGRAPH_API_REVISION: ed92a0f
LANGSMITH_LANGGRAPH_API_VARIANT: local
Revision ID: e7586c9
Run Attempt: 1
Run ID: 1efce944-fd32-6cdc-94de-929c8772430f
Thread ID: b6f30a04-e4f1-4dc1-9ec5-46a6042ff917
X-Middleware-Subrequest: /api/[..._path]