just noticed that sometimes mimi responds weirdly with double quotes at the end, i know where the problem is, i think (parsing malformed json coming from llm, easily fixable) https://i.imgur.com/cpOn0E7.png

I noticed starcoder doing two sets of quotes, when I was running it in text-generation-webui.

Philosopher, voluntarist, crypto builder.

It was not working right in the text-generation-webui. 

I could not get it to give me any code. 

Probably something off about the way I was running it.

Wow damn, that's practically unheard of in the GPT-3.5 or 4. Is the model good enough in helping you program?