Repository navigation
Feature request to customize language model #166
Description
I use Ollama to improve Whisper's output, but it seems that dictation is really slow.
For reference, I have a iMac 2021 with 16gb of RAM, and running Gemma2:e2b on it is blazing fast. But when I select that model in the Ollama provider in this app, dictation is really slow. I know it's not my Whisper mode because without the language model, it is fast.
So, I'm requesting that there is a way to customize the prompt given to the model, so that it can be faster. I should also be able to disable thinking there as well.
Activity