UGI Leaderboard Writin benchmark
#2
by s1arsky - opened
s1arsky changed discussion status to closed
s1arsky changed discussion status to open
so like... is higher the number the better?
No. For example dialogue the best is most close to 60% , that is 60% is the best, 61 and 59 are 2nd.
4 1/2 months on, and this is still the highest scoring model on Writing under ~110B, though tied with the 106BA12B ArliAI-GLM-4.5-Air-Derestricted (nothink.) It's beaten out by three 123B models if you go up to that size, but even up to 235B, there's only one model beating it. Impressive - that's beating all of the 70B models.