bartowski

/

gemma-2-27b-it-SimPO-37K-GGUF

like 10

You should check the original model for details like that, but it's trained on this dataset: https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized

Which is basically just aimed at improving overall generation and chat quality.

Ah, OK, thanks! I recognize your name; I've enjoyed a lot of your models :)