Closed hoalarious closed 10 months ago
This seems like a crucial part of the paper. It would be helpful if we can build on your existing work to optimize for PGR.
there already is code for that step
look for label="weak2strong" in train_weak_to_strong.py
label="weak2strong"
This seems like a crucial part of the paper. It would be helpful if we can build on your existing work to optimize for PGR.