Skip to content

Latest commit

 

History

History
10 lines (6 loc) · 508 Bytes

200214 AutoLR.md

File metadata and controls

10 lines (6 loc) · 508 Bytes

https://arxiv.org/abs/2002.06048

AutoLR: Layer-wise Pruning and Auto-tuning of Learning Rates in Fine-tuning of Deep Networks (Youngmin Ro, Jin Young Choi)

Layer-wise Pruning and Auto-tuning of Layer-wise Learning Rates in Fine-tuning of Deep Networks (Youngmin Ro, Jin Young Choi)

finetuning 상황에서 레이어를 pruning 하는 것의 효과성과 레이어별로 learning rate를 자동으로 설정할 수 있는 알고리즘을 제안. 둘 다 흥미로운 방법인 듯.

#pruning #finetuning