ValueError: Calculated available context size -203 was not non-negative.

why am i getting this error

how can i set my prompt template to be in a sepcific language?

Is there a "dumb" model that is super fast and just finds matching vectors from a vectorstore database. like i give a prompt about something i want and it only does the query part without the inference.

what is the advantage to using LlamaParse over just building a Vectorstore and loading from that

I hate the overly dependence on ollama and openai stuff in the docs these days. half a year ago everything could be setup locally, now llamaindex docs is bloated with this 3rd party software and always download embeddings and every little thing online