Discamus continentiam augere, luxuriam coercere
Home -> Publications
Home
  Publications
    
edited volumes
  Awards
  Research
  Teaching
  Miscellaneous
  Full CV [pdf]
  BLOG






  Events








  Past Events





Publications of Torsten Hoefler
Bryan A. Plummer, Nikoli Dryden, Julius Frost, Torsten Hoefler, Kate Saenko:

 Neural Parameter Allocation Search

(In Tenth International Conference on Learning Representations, Apr. 2022)

Publisher Reference

Abstract

Training neural networks requires increasing amounts of memory. Parameter sharing can reduce memory and communication costs, but existing methods assume networks have many identical layers and utilize hand-crafted sharing strategies that fail to generalize. We introduce Neural Parameter Allocation Search (NPAS), a novel task where the goal is to train a neural network given an arbitrary, fixed parameter budget. NPAS covers both low-budget regimes, which produce compact networks, as well as a novel high-budget regime, where additional capacity can be added to boost performance without increasing inference FLOPs. To address NPAS, we introduce Shapeshifter Networks (SSNs), which automatically learn where and how to share parameters in a network to support any parameter budget without requiring any changes to the architecture or loss function. NPAS and SSNs provide a complete framework for addressing generalized parameter sharing, and can also be combined with prior work for additional performance gains. We demonstrate the effectiveness of our approach using nine network architectures across four diverse tasks, including ImageNet classification and transformers.

Documents

download article:     
download slides:
 

BibTeX

@inproceedings{,
  author={Bryan A. Plummer and Nikoli Dryden and Julius Frost and Torsten Hoefler and Kate Saenko},
  title={{Neural Parameter Allocation Search}},
  year={2022},
  month={Apr.},
  booktitle={Tenth International Conference on Learning Representations},
  source={http://www.unixer.de/~htor/publications/},
}


serving: 44.200.137.63:46508© Torsten Hoefler