Added to the UGI leaderboard
Requested them to add it, has a slight UGI improvement on this one, though the leaderboard author rated it lower than the original for writing style (not sure if preference has changed since the original was rated though). Rating predicition was the same
Thanks for the model!
Thanks for submitting it. I hadn't looked at that benchmark before. It makes sense that it would have more general knowledge given it's based on Qwen2.5.
not sure if preference has changed since the original was rated though
It's possible. When the original came out, there was nothing like it. The newer Magnum's probably rate a lot better, but it's subjective, and I like the raw craziness of V1.
Thanks for the model!
No problem. I made it for myself but someone on reddit asked me to upload it when I mentioned it.