community: tongyi multimodal response format fix to support langchain

[achenglike]

• Description: The multimodal(tongyi) response format "message": {"role": "assistant", "content": [{"text": "图像"}]}}]} is not compatible with LangChain.
• Dependencies: No

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

1 Skipped Deployment

Name	Status	Preview	Comments	Updated (UTC)
langchain	⬜️ Ignored (Inspect)			Nov 19, 2024 7:52am

[efriis]

at a minimum let's make this a hidden _fix... method so it doesn't show in the api docs. Can we handle this inline instead though? I feel like message parsing should maybe also happen in other places than streaming...

[achenglike]

at a minimum let's make this a hidden _fix... method so it doesn't show in the api docs. Can we handle this inline instead though? I feel like message parsing should maybe also happen in other places than streaming...

Thank you for your advice. I have created a new PR #28645 (using inline). By the way, I only encountered this issue with the streaming implementation, as Tongyi uses different packages for streaming and one-shot responses.