Skip to content

Latest commit

 

History

History

Convert-MM

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 

torch-demo-tools

convertGPUmodelCPU

on GPU run as

th convertGPUmodelCPU.lua './dir/model.net' 

to modify your trained network on GPU to run on CPU.

convertLinearmodelConvMM

on CPU run as

 th convertLinearmodelConvMM.lua './dir/model.net' 

to modify your trained network with Linear layer. By replacing the linear layers with ConvolutionMM, you can feed the network bigger inputs than the ones network trained with. You need to modify the input tensor based on your training.