trick for less token usage and less halucination
#19
by
gopi87
- opened
in text generation ui just update the lama cpp manually add this in the chat ui
Start reply with
lets plan the steps and review the steps
for better response
*dont use flash attention
- big thanks to qwen team
even in this way you make model to follow your own custom prompt
This comment has been hidden
Start reply with
lets plan it one by one
very organized response like o1