Enable embedding_function in RetrieveUserProxyAgent to create initial…

[bryant-nn]

… vector_db

Why are these changes needed?

Current RetrieveUserProxyAgent doesn't create initial vector_db with embedding_function, which always cause "No content embedding is provided. Will use the VectorDB's embedding function to generate the content embedding." wanring.

Related issue number

Checks

• I've included any doc changes needed for https://microsoft.github.io/autogen/. See https://microsoft.github.io/autogen/docs/Contribute#documentation to build and test documentation locally.
• I've added tests (if relevant) corresponding to the changes introduced in this PR.
• I've made sure all auto checks have passed.

[bryant-nn]

@microsoft-github-policy-service agree

[thinkall]

Hi @bryant-nn , thank you for the PR. The embeddings will be computed by the vector db. Could you share a complete code snippet for reproducing the issue? Thanks.

[bryant-nn]

Hi @bryant-nn , thank you for the PR. The embeddings will be computed by the vector db. Could you share a complete code snippet for reproducing the issue? Thanks.

I'm very sorry. I initially thought the sentence 'No content embedding is provided. Will use the VectorDB's embedding function to generate the content embedding.' meant that the VectorDB's default embedding function (all-MiniLM-L6-v2) would be used to create the vector_db. However, I found that in Collection.py, it actually utilizes _embedding_function. Sorry for the confusion.