Add new deep learning functionality

src/Bonsai.ML.Torch/NeuralNets/ActivationFunction/Celu.cs

@@ -0,0 +1,50 @@
+using System;
+using System.ComponentModel;
+using System.Reactive.Linq;
+using static TorchSharp.torch;
+using static TorchSharp.torch.nn;
+
+namespace Bonsai.ML.Torch.NeuralNets.ActivationFunction;
+
+/// <summary>
+/// Represents an operator that creates a continuously differentiable exponential linear unit (CELU) activation function.
+/// </summary>
+/// <remarks>
+/// See <see href="https://pytorch.org/docs/stable/generated/torch.nn.CELU.html"/> for more information.
+/// </remarks>
+[Description("Creates a continuously differentiable exponential linear unit (CELU) activation function.")]
+[DisplayName("CELU")]
+public class Celu
+{
+    /// <summary>
+    /// The alpha value for the CELU activation function.
+    /// </summary>
+    [Description("The alpha value for the CELU activation function.")]
+    public double Alpha { get; set; } = 1D;
+
+    /// <summary>
+    /// If set to true, will do this operation in-place.
+    /// </summary>
+    [Description("If set to true, will do this operation in-place.")]
+    public bool Inplace { get; set; } = false;
+
+    /// <summary>
+    /// Creates a continuously differentiable exponential linear unit (CELU) module.
+    /// </summary>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.CELU> Process()
+    {
+        return Observable.Return(CELU(Alpha, Inplace));
+    }
+
+    /// <summary>
+    /// Creates a continuously differentiable exponential linear unit (CELU) module.
+    /// </summary>
+    /// <typeparam name="T"></typeparam>
+    /// <param name="source"></param>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.CELU> Process<T>(IObservable<T> source)
+    {
+        return source.Select(_ => CELU(Alpha, Inplace));
+    }
+}

src/Bonsai.ML.Torch/NeuralNets/ActivationFunction/Elu.cs

@@ -0,0 +1,50 @@
+using System;
+using System.ComponentModel;
+using System.Reactive.Linq;
+using static TorchSharp.torch;
+using static TorchSharp.torch.nn;
+
+namespace Bonsai.ML.Torch.NeuralNets.ActivationFunction;
+
+/// <summary>
+/// Represents an operator that creates an exponential linear unit (ELU) activation function.
+/// </summary>
+/// <remarks>
+/// See <see href="https://pytorch.org/docs/stable/generated/torch.nn.ELU.html"/> for more information.
+/// </remarks>
+[Description("Creates an exponential linear unit (ELU) activation function.")]
+[DisplayName("ELU")]
+public class Elu
+{
+    /// <summary>
+    /// The alpha value for the ELU activation function.
+    /// </summary>
+    [Description("The alpha value for the ELU activation function")]
+    public double Alpha { get; set; } = 1D;
+
+    /// <summary>
+    /// If set to true, will do this operation in-place.
+    /// </summary>
+    [Description("If set to true, will do this operation in-place")]
+    public bool Inplace { get; set; } = false;
+
+    /// <summary>
+    /// Creates an exponential linear unit (ELU) module.
+    /// </summary>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.ELU> Process()
+    {
+        return Observable.Return(ELU(Alpha, Inplace));
+    }
+
+    /// <summary>
+    /// Creates an exponential linear unit (ELU) module.
+    /// </summary>
+    /// <typeparam name="T"></typeparam>
+    /// <param name="source"></param>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.ELU> Process<T>(IObservable<T> source)
+    {
+        return source.Select(_ => ELU(Alpha, Inplace));
+    }
+}

src/Bonsai.ML.Torch/NeuralNets/ActivationFunction/Gelu.cs

@@ -0,0 +1,44 @@
+using System;
+using System.ComponentModel;
+using System.Reactive.Linq;
+using static TorchSharp.torch;
+using static TorchSharp.torch.nn;
+
+namespace Bonsai.ML.Torch.NeuralNets.ActivationFunction;
+
+/// <summary>
+/// Represents an operator that creates a gaussian error linear unit (GELU) activation function.
+/// </summary>
+/// <remarks>
+/// See <see href="https://pytorch.org/docs/stable/generated/torch.nn.GELU.html"/> for more information.
+/// </remarks>
+[Description("Creates a gaussian error linear unit (GELU) activation function.")]
+[DisplayName("GELU")]
+public class Gelu
+{
+    /// <summary>
+    /// If set to true, will do this operation in-place.
+    /// </summary>
+    [Description("If set to true, will do this operation in-place.")]
+    public bool InPlace { get; set; } = false;
+
+    /// <summary>
+    /// Creates a gaussian error linear unit (GELU) module.
+    /// </summary>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.GELU> Process()
+    {
+        return Observable.Return(GELU(InPlace));
+    }
+
+    /// <summary>
+    /// Creates a gaussian error linear unit (GELU) module.
+    /// </summary>
+    /// <typeparam name="T"></typeparam>
+    /// <param name="source"></param>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.GELU> Process<T>(IObservable<T> source)
+    {
+        return source.Select(_ => GELU(InPlace));
+    }
+}

src/Bonsai.ML.Torch/NeuralNets/ActivationFunction/Glu.cs

@@ -0,0 +1,44 @@
+using System;
+using System.ComponentModel;
+using System.Reactive.Linq;
+using static TorchSharp.torch;
+using static TorchSharp.torch.nn;
+
+namespace Bonsai.ML.Torch.NeuralNets.ActivationFunction;
+
+/// <summary>
+/// Represents an operator that creates a gated linear unit (GLU) module.
+/// </summary>
+/// <remarks>
+/// See <see href="https://pytorch.org/docs/stable/generated/torch.nn.GLU.html"/> for more information.
+/// </remarks>
+[Description("Creates a gated linear unit (GLU) module.")]
+[DisplayName("GLU")]
+public class Glu
+{
+    /// <summary>
+    /// The dimension on which to split the input tensor.
+    /// </summary>
+    [Description("The dimension on which to split the input tensor.")]
+    public long Dim { get; set; } = -1;
+
+    /// <summary>
+    /// Creates a gated linear unit (GLU) module.
+    /// </summary>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.GLU> Process()
+    {
+        return Observable.Return(GLU(Dim));
+    }
+
+    /// <summary>
+    /// Creates a gated linear unit (GLU) module.
+    /// </summary>
+    /// <typeparam name="T"></typeparam>
+    /// <param name="source"></param>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.GLU> Process<T>(IObservable<T> source)
+    {
+        return source.Select(_ => GLU(Dim));
+    }
+}

src/Bonsai.ML.Torch/NeuralNets/ActivationFunction/Hardshrink.cs

@@ -0,0 +1,43 @@
+using System;
+using System.ComponentModel;
+using System.Reactive.Linq;
+using static TorchSharp.torch;
+using static TorchSharp.torch.nn;
+
+namespace Bonsai.ML.Torch.NeuralNets.ActivationFunction;
+
+/// <summary>
+/// Represents an operator that creates a Hardshrink module.
+/// </summary>
+/// <remarks>
+/// See <see href="https://pytorch.org/docs/stable/generated/torch.nn.Hardshrink.html"/> for more information.
+/// </remarks>
+[Description("Creates a Hardshrink module.")]
+public class Hardshrink
+{
+    /// <summary>
+    /// The lambda parameter for the Hardshrink function.
+    /// </summary>
+    [Description("The lambda parameter for the Hardshrink function")]
+    public double Lambda { get; set; } = 0.5D;
+
+    /// <summary>
+    /// Creates a Hardshrink module.
+    /// </summary>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.Hardshrink> Process()
+    {
+        return Observable.Return(Hardshrink(Lambda));
+    }
+
+    /// <summary>
+    /// Creates a Hardshrink module.
+    /// </summary>
+    /// <typeparam name="T"></typeparam>
+    /// <param name="source"></param>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.Hardshrink> Process<T>(IObservable<T> source)
+    {
+        return source.Select(_ => Hardshrink(Lambda));
+    }
+}

src/Bonsai.ML.Torch/NeuralNets/ActivationFunction/Hardsigmoid.cs

@@ -0,0 +1,43 @@
+using System;
+using System.ComponentModel;
+using System.Reactive.Linq;
+using static TorchSharp.torch;
+using static TorchSharp.torch.nn;
+
+namespace Bonsai.ML.Torch.NeuralNets.ActivationFunction;
+
+/// <summary>
+/// Represents an operator that creates a Hardsigmoid module.
+/// </summary>
+/// <remarks>
+/// See <see href="https://pytorch.org/docs/stable/generated/torch.nn.Hardsigmoid.html"/> for more information.
+/// </remarks>
+[Description("Creates a Hardsigmoid module.")]
+public class Hardsigmoid
+{
+    /// <summary>
+    /// If set to true, will do this operation in-place.
+    /// </summary>
+    [Description("If set to true, will do this operation in-place")]
+    public bool Inplace { get; set; } = false;
+
+    /// <summary>
+    /// Creates a Hardsigmoid module.
+    /// </summary>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.Hardsigmoid> Process()
+    {
+        return Observable.Return(Hardsigmoid(Inplace));
+    }
+
+    /// <summary>
+    /// Creates a Hardsigmoid module.
+    /// </summary>
+    /// <typeparam name="T"></typeparam>
+    /// <param name="source"></param>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.Hardsigmoid> Process<T>(IObservable<T> source)
+    {
+        return source.Select(_ => Hardsigmoid(Inplace));
+    }
+}

src/Bonsai.ML.Torch/NeuralNets/ActivationFunction/Hardswish.cs

@@ -0,0 +1,43 @@
+using System;
+using System.ComponentModel;
+using System.Reactive.Linq;
+using static TorchSharp.torch;
+using static TorchSharp.torch.nn;
+
+namespace Bonsai.ML.Torch.NeuralNets.ActivationFunction;
+
+/// <summary>
+/// Represents an operator that creates a Hardswish module.
+/// </summary>
+/// <remarks>
+/// See <see href="https://pytorch.org/docs/stable/generated/torch.nn.Hardswish.html"/> for more information.
+/// </remarks>
+[Description("Creates a Hardswish module.")]
+public class Hardswish
+{
+    /// <summary>
+    /// If set to true, will do this operation in-place.
+    /// </summary>
+    [Description("If set to true, will do this operation in-place")]
+    public bool Inplace { get; set; } = false;
+
+    /// <summary>
+    /// Creates a Hardswish module.
+    /// </summary>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.Hardswish> Process()
+    {
+        return Observable.Return(Hardswish(Inplace));
+    }
+
+    /// <summary>
+    /// Creates a Hardswish module.
+    /// </summary>
+    /// <typeparam name="T"></typeparam>
+    /// <param name="source"></param>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.Hardswish> Process<T>(IObservable<T> source)
+    {
+        return source.Select(_ => Hardswish(Inplace));
+    }
+}

src/Bonsai.ML.Torch/NeuralNets/ActivationFunction/Hardtanh.cs

@@ -0,0 +1,55 @@
+using System;
+using System.ComponentModel;
+using System.Reactive.Linq;
+using static TorchSharp.torch;
+using static TorchSharp.torch.nn;
+
+namespace Bonsai.ML.Torch.NeuralNets.ActivationFunction;
+
+/// <summary>
+/// Represents an operator that creates a Hardtanh module.
+/// </summary>
+/// <remarks>
+/// See <see href="https://pytorch.org/docs/stable/generated/torch.nn.Hardtanh.html"/> for more information.
+/// </remarks>
+[Description("Creates a Hardtanh module.")]
+public class Hardtanh
+{
+    /// <summary>
+    /// The minimum value of the linear region range.
+    /// </summary>
+    [Description("The minimum value of the linear region range.")]
+    public double MinVal { get; set; } = -1D;
+
+    /// <summary>
+    /// The maximum value of the linear region range.
+    /// </summary>
+    [Description("The maximum value of the linear region range.")]
+    public double MaxVal { get; set; } = 1D;
+
+    /// <summary>
+    /// If set to true, will do this operation in-place.
+    /// </summary>
+    [Description("If set to true, will do this operation in-place")]
+    public bool Inplace { get; set; } = false;
+
+    /// <summary>
+    /// Creates a Hardtanh module.
+    /// </summary>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.Hardtanh> Process()
+    {
+        return Observable.Return(Hardtanh(MinVal, MaxVal, Inplace));
+    }
+
+    /// <summary>
+    /// Creates a Hardtanh module.
+    /// </summary>
+    /// <typeparam name="T"></typeparam>
+    /// <param name="source"></param>
+    /// <returns></returns>
+    public IObservable<TorchSharp.Modules.Hardtanh> Process<T>(IObservable<T> source)
+    {
+        return source.Select(_ => Hardtanh(MinVal, MaxVal, Inplace));
+    }
+}