Hi, please review this one. Looks like the insert function did not actually properly insert the documents, since it didn't set the doc_hash. This affected the refresh function, which relies on the doc_hash to be properly updated.
Thanks! By the way, any idea on how the GPTListIndex does tree_summarize? I thought it normally goes through each node one by one. Conceptually TreeIndex makes sense and how it would do tree_summarize, but i have no idea how a list index would do that
Any way to inject an additional prompt / description onto agent? I know i can add descriptions to tools. Perhaps any way I can modify the conversational chat agent a little bit.
tool_config = IndexToolConfig( index=index, name=f"Specific-Details-GPT", description=f"Useful tool for answering specific detailed questions using data from documents.", index_query_kwargs={"similarity_top_k": 3}, tool_kwargs={"return_direct": True} )
I'm using chat-conversational-react-description langchain agent to use the tool. I asked it to tell me about events.
However, I get this output from the agent: { "action": "Specific-Details-GPT", "action_input": "What kind of events are you interested in? Please provide more specific details." }
Observation:
I am interested in events...
Then, it returned the observation itself, rather than telling me about events
Is there a way to view the existing documents in an index? I'd like to periodically check to see if the existing documents in the directory match the ones loaded in the index. If they don't remove from the index. The docs don't seem to be loaded in index.storage_context.docstore.docs
Are there any local LLM models that llama-index can run on? Thinking of using this for work, but the workplace might be skeptical of making API calls to OpenAI. Any Local LLM's that exist and can integrate?