About

B# is a dynamically typed multi-paradigm programming language designed for writing high-level modules (like machine learning pipeline using ML.NET) in .NET applications.

C# 11

B#

usingSystem;usingtorch=System.AI.torch;usingnn=System.AI.torch.nn;usingF=System.AI.torch.nn.functional;usingoptim=System.AI.torch.optim;publicsealedclassLinearRegression:nn.Module{publicnn.Parameterweight;publicnn.Parameterbias;publicLinearRegression(){this.weight=newnn.Parameter(torch.randn());this.bias=newnn.Parameter(torch.randn());}publictorch.Tensorforward(torch.Tensorx){returnx*this.weight+this.bias;}}static(torch.Tensor,torch.Tensor)GetData(intbatch_size){return(torch.randn(batch_size),torch.randn(batch_size));}staticvoidMain(){varregr=newLinearRegression();varopt=newoptim.SGD(regr.parameters());for(intep=0;ep<epoch;++ep){(varinput,vartarget)=GetData();y=regr.op_Invocation(input);varloss=F.l1_loss(y,target);loss.backward();opt.step();Console.WriteLine(loss);}}

using*fromSystem;usingtorchfrom System.AI;usingnnfromSystem.AI.torch;usingfunctionalfromSystem.AI.torch.nnasF;usingoptimfromSystem.AI.torch;publicsealedclassLinearRegression:nn.Module{public:weight;bias;constructor(){this.weight=newnn.Parameter(torch.randn());this.bias=newnn.Parameter(torch.randn());}forward(x){returnx*this.weight+this.bias;}}get_data(batch_size){return(torch.randn(batch_size),torch.randn(batch_size));}main(){varregr=newLinearRegression();varopt=newoptim.SGD(regr.parameters());for(varep=0;ep<epoch;++ep){(varinput,vartarget)=GetData();y=regr(input);varloss=F.l1_loss(y,target);loss.backward();opt.step();Console.WriteLine(loss);}}

For more examples see B# vs C# vs Python

Compatibility

B# is compatible with given operating systems and .NET implementations:

OS	.NET
MS Windows XP and higer	Mono 2.0, .NET Framework 4.0 and higer, .NET Core, .NET
Apple macOS 12.0 and higer	Mono 2.0, .NET Core, .NET
Linux (Mint, Ubuntu)	Mono 2.0, .NET Core, .NET
Android 10 and higer	Mono 2.0

Development technology stack

• Windows 7 x64
• .NET Framework 4.5.2
• C# 5
• SharpDevelop 5.1 (IDE)
• Jay (Parser generator)