1 paper · 1 filter
Krish Sharma, Omar Naim, Soumadeep Saha +3
Recent work has promoted task-aware layer pruning as a way to improve model performance on particular tasks, as shown by TALE. In this paper, we investigate when such improvements…