NOTE: This branch is currently undergoing significant development. It has incomplete code, functionality, and design that are likely to change without notice, until we officially release DiffSharp 1.0 and the corresponding documentation.
This is the development branch of DiffSharp 1.0.
You can clone this repository to your machine as follows:
git clone --branch dev https://github.com/DiffSharp/DiffSharp.git
cd DiffSharp
Required:
- Install .NET Core SDK for your system
Use the following command in the r 8000 oot directory of this repository:
dotnet test
Required:
- Install Docker for your system
Build a Docker image called diffsharp. This will work without any local .NET Core installation and build DiffSharp inside the image.
docker build -t diffsharp .
Use the following to instantiate a Docker container from the diffsharp image and run the tests inside:
docker run --rm diffsharp dotnet test
After building, to use DiffSharp in this development branch you must reference an appropriate configuration.
Reference one of
DiffSharp-cpuDiffSharp-cuda-winDiffSharp-cuda-linux
Then use:
dsharp.config(backend=Backend.Torch)
Alternatively use the reference backend via DiffSharp-reference.
To develop libraries built on DiffSharp which are designed for general use (not just in your own code), do the following:
- reference
DiffSharp.Corein your library code - reference
DiffSharp.Backends.Referencein your correctness testing code. - reference
DiffSharp.Backends.Torchandlibtorch-cpuin your CPU performence testing code. - reference
DiffSharp.Backends.Torchandlibtorch-cudain your GPU performence testing code.
To consume CI build packages in a .NET Interactive Jupyter notebook (
#i "nuget:https://ci.appveyor.com/nuget/diffsharp"
In F# 5.0 scripts (with --langversion:preview) do:
#r "nuget:RestoreSources=https://ci.appveyor.com/nuget/diffsharp"
Then add a reference to the version you want, tha package numbers can be found in the "artifacts" tabs of the DiffSharp CI builds.
#r "nuget: DiffSharp-reference,0.9.5-preview-NNNN"
or
#r "nuget: DiffSharp-cpu,0.9.5-preview-NNNN"
dsharp.config(backend=Backend.Torch)
or
#r "nuget: DiffSharp-cuda-linux,0.9.5-preview-NNNN"
dsharp.config(backend=Backend.Torch)
or
#r "nuget: DiffSharp-cuda-windows,0.9.5-preview-NNNN"
dsharp.config(backend=Backend.Torch)
To add features you may have extend TorchSharp to make extra features of LibTorch available.
The build is set up to look for a parallel build of TorchSharp, e.g.
C:\GitHub\dsyme\DiffSharp
C:\GitHub\dsyme\TorchSharp
To build, test and pack TorchSharp in that repo do this:
.\build build
.\build test
.\build pack
You will see something like this
Packing LibTorch.Cuda.10.2.Redist nupkg (takes a long time!)...
Successfully created package 'C:\GitHub\dsyme\TorchSharp\bin/packages/Debug/TorchSharp.0.3.0-local-Debug-20200520.nupkg'.
Successfully created package 'C:\GitHub\dsyme\TorchSharp\bin/packages/Debug/LibTorch.Cuda.10.2.Redist.0.3.0-local-Debug-20200520.nupkg'.
Successfully created package 'C:\GitHub\dsyme\TorchSharp\bin/packages/Debug/LibTorch.Redist.0.3.0-local-Debug-20200520.nupkg'.
with warning:
warning : Packages will be incomplete and unusable on other platforms...
To consume the packages into DiffSharp adust TorchSharpVersion in DIrectory.Build.props.
When rebuilding the TorchSHarp you will need to vlear your package cache to pick up the new nuget package with the same version id, e.g.
rmdir /q /s %USERPROFILE%\.nuget\packages\torchsharp
rmdir /q /s %USERPROFILE%\.nuget\packages\LibTorch.Redist
rmdir /q /s %USERPROFILE%\.nuget\packages\LibTorch.Cuda.10.2.Redist
dotnet restore
The LibTorch packages are quite large and you may need to watch disk space.