tree: 8f597c542eaafe987e93729f65e7b46688939d8b [path history] [tgz]
  1. doc/
  2. generic/
  3. ClassNLLCriterion.cu
  4. CMakeLists.txt
  5. common.h
  6. GatedLinearUnit.cu
  7. LogSigmoid.cu
  8. MultiLabelMarginCriterion.cu
  9. MultiMarginCriterion.cu
  10. README.md
  11. RReLU.cu
  12. SharedMem.cuh
  13. SpatialClassNLLCriterion.cu
  14. SpatialConvolutionMM.cu
  15. SpatialDepthwiseConvolution.cu
  16. THCHalfAutoNumerics.cuh
  17. THCUNN.h
aten/src/THCUNN/README.md

THCUNN

THCUNN is a library that gathers nn‘s C implementations of neural network modules. It’s entirely free of Lua dependency and therefore can be used in any application that has a C FFI. Please note that it only contains quite low level functions; most users will want to use ATen, which provides a C++ wrapper around these functions.

Looking to add an implementation? Consider writing an ATen native function instead! See ../ATen/native.

Links

API

THCUNN is a purely functional library. It provides 2-3 functions for each module, that perform the most important operations:

  • updateOutput - applies the module to an input
  • updateGradInput - accepts gradient w.r.t. output and previous module input, and computes a gradient w.r.t. that input
  • accGradParameters - (optional, only modules with parameters) accepts gradient w.r.t. output and previous module input, and computes gradient w.r.t. the parameters

For information on argument types please check the API reference.

Developer docs