Ok I think I found the issue When

Which embedding model are you using? Troubling to hear 😅

Just using this command: I managed to find a workaround and I can now generate good responses but the index.json file still doesn't seem to be encoded properly.

I guess this would be using embedding ada 002? I'm still new to GPT Index

Yeah that would be ada by default

If you want to debug you can call the openai embeddings model directly

Actually i'll probably do that 😄

My previous bot had the same problem with the encoding format and letter 'ä' and 'ö'. For some reason I was able to get the GPT Index to ignore this encoding error and it kinda just fills it in properly.

At least openai should handle them properly

Alright, thx. I guess i'll just choose to ignore the issue for now since it isn't causing further issues atm.