Closed yyhycx closed 1 year ago
When I set score=[1,2,3,4], rw_score = [4,3,2,1], the obtained rrhf loss is 0
I also found this problem. Is there something wrong with this loss calculation
When I set score=[1,2,3,4], rw_score = [4,3,2,1], the obtained rrhf loss is 0