why gpt-4 could change the vector search keywords in the second conversation , by other LLM model can not do this

[xutengfei]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to file this report (我已阅读并同意 Language Policy).

Dify version

0.3.32

Cloud or Self Hosted

Self Hosted (Docker)

Steps to reproduce

why gpt-4 could change the vector search keywords in the second conversation, by other LLM model could not chang the keyword（use whatever that user input）

✔️ Expected Behavior

change the keywords to right intention of user, not the user input

❌ Actual Behavior

search vector with whatever user typed in the second conversation

[guchenhe]

Hi there,

could you provide more info, like embedding model and prompt used?

[xutengfei]

Hi there,

could you provide more info, like embedding model and prompt used?

prompt: you are an analyst ，you will give a report about {{input}}
embedding : text2vec
LLM: baichuan\chatglm\gpt4

[dosubot]

Hi, @xutengfei

I'm helping the team manage their backlog and am marking this issue as stale. From what I understand, you opened this issue regarding the behavior of GPT-4 in changing vector search keywords in the second conversation. There was a request for more information such as the embedding model and prompt used, to which you responded with the details.

Could you please confirm if this issue is still relevant to the latest version of the repository? If it is, kindly let the team know by commenting on the issue. Otherwise, feel free to close the issue yourself, or it will be automatically closed in 7 days. Thank you!