Skip to content

m-torhan/NeuralNetwork

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

68 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

NeuralNetwork

c++/asm implementation of neural network.

The layers works on type Tensor which represents $n$-dimensional array and supports mathematical operations (addition, subtraction, dot product, tensor product, $\ldots$) and other not math realted (adding padding, shuffling, reshaping $\ldots$). Some of the operations are optimized using AVX128 instructions, which made them a lot faster.

Example uses of NeuralNetwork class can be found in:

At this moment there are implemented layers of type:

  • dense,
  • activation (sigmoid, ReLU, leakyReLU),
  • conv2D (not optimized yet),
  • pool2D (max or mean pooling),
  • reshape,
  • flatten,
  • dropout,
  • normal distribution layer (for VAE).

Build scripts:

Other scripts:

About

c++/asm neural network implementation

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published