Squeezenext baseline architecture implementation in Pytorch with live loss and accuracy update using livelossplot package.
Software used :
Python 3.6
Pytorch 1.0.0
Spyder 3.3.1
Packages needed when running it for first time :
Livelossplot package (https://github.com/stared/livelossplot)
Torch (https://pytorch.org/get-started/locally/)
- Choose all the options according to your system specification.
- Recommendations
OS: Linux
BUILD : 1.0
Package : Pip
Cuda Installation; it depends on your graphic card. Please refer your label
|-----Model-------|Kernel size-|-epochs-|test accuracy||Model Size(MB)||Model speed||Optimizer||Learning rate|
|sqnxt_baseline_23| 7x7 | 200 |--84.69%-----|-----2.617-----|-----22-----|----sgd---|-----0.1------|
|sqnxt_baseline_23| 3x3 | 200 |--87.63%-----|-----2.586-----|-----23-----|----sgd---|-----0.1------|
