Training data?
#1
by
darkstorm2150
- opened
I know this is a long shot, but this model has shown strong signs that it counters Homogenization in AI, it produces rich outputs based on the assigned prompt. I wanted to know on how you trained it, I also saw that it came from gemma3 architecture, which is impressive!
The data was mostly RP conversations from various forums but the part that improved the output the most was RLHF
thank you for the feedback! :D
darkstorm2150
changed discussion status to
closed