Multi Agent Chat with 2 Rag Systems #3062
Unanswered
Jonathanpro
asked this question in
Q&A
Replies: 1 comment
-
you should try to initialize the vectordb for both agents individually , then try again with the correct parameters based on that |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hello everyone I want to create a Groupchat between two agents. Each of them has an assistant that has access to its own RAG System.
Sofar I have created a solution that somehow is related to https://microsoft.github.io/autogen/blog/2023/10/18/RetrieveChat/#integrate-with-other-agents-in-a-group-chat
When the first Agent asks its RAG system I receive a good answer however when the second agent than askes its RAG system i receive a "Error: Collection autogen-docs already exists."
I have noticed that both RAG systems use the "tmp" folder to write the embedding to.
Hence I concluded that there might be a problem that the two RAG system needs to write their embeddings into seperated folders.
For each of the RAG assistants I have written a RetrieveUserProxyAgent
that looks like this
A_aid = RetrieveUserProxyAgent(
name="A_aid",
llm_config=llm_config,
human_input_mode="NEVER",
retrieve_config={
"task": "qa",
"docs_path": "path_to_file",
# "vector_db": vectordb_A,
# "docs_path": None,
"vector_db": "chroma",
"client": chromadb.PersistentClient(path="chromadb_A_aid"),
},
description="Assistent A",
)
for B likewise
I tried to use vector_db instead of docs_path but than the conversation seems to have no access to the knowledge of the RAG
I would welcome any suggestions !
Beta Was this translation helpful? Give feedback.
All reactions