Open-Assistant
Open-Assistant copied to clipboard
Needs a "coding mode" preset with temp lowered to 0.01 and max tokens at 1024
I have thoroughly tested this and the lowered temperature seems to do wonders for the logic and reasoning needed to edit and formulate code
Every other parameter can be the same as the default, unless you think it should be adjusted more
Maybe it could be called "Precise" or "Deterministic"