mirror of
https://github.com/langgenius/dify-official-plugins.git
synced 2026-07-21 17:45:23 -04:00
OpenAI-API-compatible add model Qwen3-Reranker-8B error 'results' #726
Reference in New Issue
Block a user
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @Richie666888 on GitHub (Oct 16, 2025).
Self Checks
Dify version
1.9.0.1
Cloud or Self Hosted
Self Hosted (Docker)
Steps to reproduce
模型名称:Qwen/Qwen3-Reranker-8B
模型类型:Rerank
API Key:xxxxxxxxxxxxxx
API endpoint URL:http://10.10.10.10:8000/v1
✔️ Expected Behavior
Add success
❌ Actual Behavior
report an error “results”
@dosubot[bot] commented on GitHub (Oct 16, 2025):
The error occurs because Dify expects the reranker model API response to include a top-level field named
docs(a list of reranked documents), but the Qwen3-Reranker-8B model returns a field namedresultsinstead. Dify does not have built-in support to mapresultstodocs, so it fails to parse the response and throws the error you see. The expected schema for a reranker response is:To resolve this, you need to ensure that your API endpoint or any adapter/plugin you use returns the field as
docsinstead ofresults. You can do this by modifying the API response or adding a lightweight proxy that renames the field before it reaches Dify. There is currently no official plugin or patch that handles this mapping automatically in Dify or its official plugins—this adaptation must happen at the API or plugin layer you control.Reference: Dify reranker model schema
To reply, just mention @dosu.
How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other

@Richie666888 commented on GitHub (Oct 16, 2025):
It would be great if the OpenAI-API-compatible plugin could handle things automatically.
@dosubot[bot] commented on GitHub (Nov 2, 2025):
Hi, @Richie666888. I'm Dosu, and I'm helping the dify-official-plugins team manage their backlog and am marking this issue as stale.
Issue Summary:
What's Next:
Thank you for your understanding and contribution!