Prompt Template?
#2
by
maywell
- opened
What prompt template should I use? and what merge is this model based on?
If I may also ask what's the best max context length?
In my own experiment, this model always respond with safety output.
It feels like as if it's trained on GOODY-2's output.
I tried ChatML, Alpaca, Mistral.
I hope saltlux release non dpo version of it :>