endolith
Chapter 1, not 2. https://github.com/rlabbe/Kalman-and-Bayesian-Filters-in-Python/blob/9e3d2f6ed023d937587cf2ef2ecfbf7afc3d8054/01-g-h-filter.ipynb#L974
@Barafu Do you know what model it uses? Agreed it should use a cheap model for the title
oh yeah that's wasteful
It also looks like it puts the entire conversation in the system message instead of the "generate a title" prompt in the system message and the conversation in a user...
@Barafu Well I tried to improve the prompt. I did two tests of it with gpt-3.5 and it worked better than the current prompt. If you want, you can try...
I think it just means there is some inaccuracy (red) that is then reduced (green): 
It's fully autonomous, though? I want something interactive like ChatGPT Code Interpreter or Open Interpreter (which is somewhat abandoned but may have a new maintainer now)
I think this was made before reasoning tokens were a thing and they're just not supported. When I've tried reasoning models from OpenRouter they just fail.
I was thinking the reasoning tokens could be streamed into a sliding window and then replaced with "thought for n seconds" when complete? Or there could be an option to...
Well I mean there could be a command line setting or maybe a setting in %commands as to whether you want it to persist or not. But once its written...