SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size

Forrest N. IandolaSong HanMatthew W. MoskewiczKhalid AshrafWilliam J. DallyKurt Keutzer

   Papers with code   Abstract  PDF

Recent research on deep neural networks has focused primarily on improving accuracy. For a given accuracy level, it is typically possible to identify multiple DNN architectures that achieve that accuracy level... (read more)

Benchmarked Models

RANK
MODEL
REPO
CODE RESULT
PAPER RESULT
ε-REPRODUCED
BUILD
1
SqueezeNet v1.0
60.7%
--
2
SqueezeNet v1.1
60.7%
--
3
SqueezeResNet v1.0
60.2%
--
4
SqueezeResNet v1.1
59.9%
--
5
SqueezeNet 1.1
57.9%
--
6
SqueezeNet
57.8%
57.5%