karthink

Could you explain how persistent RAG support is? Do you have to specify the sources for RAG in each chat session, or are the sources/embeddings of sources persistently available to...

There's some disconnect in our understanding of RAG. As I understand it, RAG and embeddings are two independent methods. RAG works as follows: 0. A corpus of (text) documents is...

Closing as it doesn't seem like there's anything to be done on the gptel side.

> The workflow is that when input data is available, it can first find the corresponding information in the vector database and then pass it to the LLM to enhance...

> It's about "pass the corresponding text chunks". And embedding model is use for vector the raw data, and find the similar raw data, before pass the information(text, for example),...

> [@karthink](https://github.com/karthink) Could you tell us a bit more about the bring-your-own-RAG support, and what that will look like from a user perspective? Sure, it's nothing fancy. I'm just carving...

> Your perspective is a bit unusual; embedding is the first step in the RAG > process, but you're looking at it separately. @yibie Perhaps there is some misunderstanding. Are...

There were some recent changes where we set the buffer-file-coding-system in the process buffer for non-streaming requests explicitly to `utf-8-unix`. Could you check if your reported behavior still exists?

@ultronozm Thanks for the PR. I think a better way to do this is to use a processing hook that the user can customize. Then they can include/exclude the Org...

I've added an option `gptel-org-convert-response` that does this, so this PR is no longer required. This is still an interim solution while I figure out how to provide a filtering...