When multiWay are selected and set to weighted_score, it still requires configuring a rerank model in workflow

langgenius / dify

Dify is an open-source LLM app development platform. Dify's intuitive interface combines AI workflow, RAG pipeline, agent capabilities, model management, observability features and more, letting you quickly go from prototype to production.

Other

45.07k stars 6.33k forks source link

Self Checks

[X] This is only for bug report, if you would like to ask a question, please head to Discussions.
[X] I have searched for existing issues search for existing issues, including closed ones.
[X] I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
[X] [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:）
[X] Please do not modify this template :) and fill in all the required fields.

Dify version

0.6.15

Cloud or Self Hosted

Self Hosted (Source)

Steps to reproduce

✔️ Expected Behavior

When weighted_score is selected, there is no need to configure a rerank model.

❌ Actual Behavior

When multiWay are selected and set to weighted_score, it still requires configuring a rerank model in workflow

The system still requires configuring a rerank model when the multiWay option is selected and set to weighted_score because the Multi-Path Recall mode involves querying related text fragments from multiple knowledge bases and then reordering the retrieved documents semantically. This reordering step, known as re-ranking, requires a rerank model to calculate the relevance score between the user question and each candidate document, ensuring the best result is selected ^[1]^[2]^[3].

If you want to bypass the rerank model requirement, you can set the reranking_model to undefined in the multiple_retrieval_config. However, it is generally recommended to configure a rerank model for better performance ^[4]^[5].

To continue talking to Dosu, mention @dosu.

langgenius / dify