Back to explore

Jev Eval: Comparable to monoBert on MS Macro

The author evals Jev on the MS Macro IR benchmark and finds it comparable to trained monoBert, depending on whether you care more about MAP or nDCG.

Incredible work, I'm going to test this theory of the score levels. I evaled Jev against MS Macro and it was comparable to trained monoBert (depending if you cared more about MAP or nDCG)

Jev Eval: Comparable to monoBert on MS Macro 1
· 0 likesOpen on X