This model was developed using Self-Play Preference Optimization at iteration 3, based on the google/gemma-2-9b-it architecture as starting point.

765 3 months ago

c827688e53c8 · 137B
<start_of_turn>user
{{ if .System }}{{ .System }} {{ end }}{{ .Prompt }}<end_of_turn>
<start_of_turn>model
{{ .Response }}<end_of_turn>