Think this is deprecated? The QA template does work I think but that just uses a regular prompt right?

I mean it starts printing out/streaming the response nodes but they're nonsensical, it just keep repeating one source node and doesn't function the same way it does when not streaming

How would you save this to disk with the newest update?

Did you get the gpt_index version working? I've only managed to get the langchain one working

Could llamaindex be used for building and configuring the plugins?

Hey, I have an application that allows the user to upload financial reports (analyst briefs on specific stocks) PDFs which then automatically turn into GPTSimpleVectorIndex embeddings.

How could I improve the performance since currently when querying the index the results it returns tend to be filled with filler information such as disclaimers and warnings?

Is there a way to filter out this information using the GPT-index libraries or has someone experimented with another method like fine-tuning for this purpose?

Ok I think I found the issue. When creating the embeddings it looks like the encoding doesn't allow for certain characters such as 'ä' which is causing errors. hmm

This program is so vast I don't even know where to start. I currently have a totally separate bot that uses embeddings ada 002 to create embeddings (from a long and large text document) and then I have a python bot that answers those using a davinci model.

How would I go about recreating this using GPT Index? For my use case I would need the bot to answer extremely specifically (think legal statue, very fine detail- specific). What avenue should I start with?

What's the most up to date refine_template or import class for the chat turbo model? Been using the Chat_Refine_Prompt but it started doing the thing where the completions have mentions of the refine/context information.

What's the best way to approach it with GPT index?

Does llama-index have any multimodal implementations that can do image to text?

Just me or is the FLAREInstructQueryEngine still very much experimental?

Ah alright, no worries. I was wondering since none of the imports seemed to work 😅

Even this example snippet (I might be misunderstanding the formatted sources module) https://gpt-index.readthedocs.io/en/latest/guides/primer/usage_pattern.html#parsing-the-response

Is there a way to pass chunk_size_limit to a streamlit app that uses the data loader widget for creating embeddings from PDFs? My load from disk function in the app has the chunk_size_limit defined but it's not applying it.

Is there another way to apply the chunk_size_limit?

I haven't really had time yet, do you know the command for changing the prompt with gpt index?

Is there a way to increase response length with GPTSimpleVectorIndex? I think the standard max is 256 tokens?

Thank you for responding. I read through all the docs and I just can't get it to work