Merge branch 'main' into dispose-implicitly-converted-scalars

Author: ds5678

RELEASENOTES.md

@@ -1,7 +1,11 @@
 ## TorchSharp Release Notes
 
 Releases, starting with 9/2/2021, are listed with the most recent release at the top.
+# NuGet Version 0.105.1
 
+__Bug Fixes__:
+
+#1426 Sequential.eval() does not put model into eval mode<br/>
 # NuGet Version 0.105.0
 
 Move to libtorch 2.5.1. As with the 2.4.0 release, MacOS / Intel is no longer supported by libtorch, so TorchSharp doesn, either.

src/TorchSharp/NN/GenericSequential.cs

@@ -221,6 +221,7 @@ protected override void Dispose(bool disposing)
             /// </remarks>
             public override void train(bool on = true)
             {
+                base.train(on);
                 foreach (var m in _modules) { ((torch.nn.Module)m).train(on); }
             }
 
@@ -232,6 +233,7 @@ public override void train(bool on = true)
             /// </remarks>
             public override void eval()
             {
+                base.eval();
                 foreach (var m in _modules) { ((torch.nn.Module)m).eval(); }
             }
 
@@ -465,6 +467,7 @@ protected override void Dispose(bool disposing)
             /// </remarks>
             public override void train(bool on = true)
             {
+                base.train(on);
                 foreach (var m in _modules) { ((torch.nn.Module)m).train(on); }
             }
 
@@ -476,6 +479,7 @@ public override void train(bool on = true)
             /// </remarks>
             public override void eval()
             {
+                base.eval();
                 foreach (var m in _modules) { ((torch.nn.Module)m).eval(); }
             }
 
@@ -702,6 +706,7 @@ protected override void Dispose(bool disposing)
             /// </remarks>
             public override void train(bool on = true)
             {
+                base.train(on);
                 foreach (var m in _modules) { ((torch.nn.Module)m).train(on); }
             }
 
@@ -713,6 +718,7 @@ public override void train(bool on = true)
             /// </remarks>
             public override void eval()
             {
+                base.eval();
                 foreach (var m in _modules) { ((torch.nn.Module)m).eval(); }
             }
 
@@ -939,6 +945,7 @@ protected override void Dispose(bool disposing)
             /// </remarks>
             public override void train(bool on = true)
             {
+                base.train(on);
                 foreach (var m in _modules) { ((torch.nn.Module)m).train(on); }
             }
 
@@ -950,6 +957,7 @@ public override void train(bool on = true)
             /// </remarks>
             public override void eval()
             {
+                base.eval();
                 foreach (var m in _modules) { ((torch.nn.Module)m).eval(); }
             }
 
@@ -1176,6 +1184,7 @@ protected override void Dispose(bool disposing)
             /// </remarks>
             public override void train(bool on = true)
             {
+                base.train(on);
                 foreach (var m in _modules) { ((torch.nn.Module)m).train(on); }
             }
 
@@ -1187,6 +1196,7 @@ public override void train(bool on = true)
             /// </remarks>
             public override void eval()
             {
+                base.eval();
                 foreach (var m in _modules) { ((torch.nn.Module)m).eval(); }
             }
 
@@ -1413,6 +1423,7 @@ protected override void Dispose(bool disposing)
             /// </remarks>
             public override void train(bool on = true)
             {
+                base.train(on);
                 foreach (var m in _modules) { ((torch.nn.Module)m).train(on); }
             }
 
@@ -1424,6 +1435,7 @@ public override void train(bool on = true)
             /// </remarks>
             public override void eval()
             {
+                base.eval();
                 foreach (var m in _modules) { ((torch.nn.Module)m).eval(); }
             }
 

src/TorchSharp/NN/Sequential.cs

@@ -253,6 +253,7 @@ protected override void Dispose(bool disposing)
             /// </remarks>
             public override void train(bool on = true)
             {
+                base.train(on);
                 foreach (var m in _modules) { ((torch.nn.Module)m).train(on); }
             }
 
@@ -264,6 +265,7 @@ public override void train(bool on = true)
             /// </remarks>
             public override void eval()
             {
+                base.eval();
                 foreach (var m in _modules) { ((torch.nn.Module)m).eval(); }
             }
 

test/TorchSharpTest/TestTraining.cs

@@ -16,6 +16,34 @@ namespace TorchSharp
     public class TestTraining
     {
 
+        // <summary>
+        /// Check if sequential module goes from training to eval mode
+        // </summary>
+        [Fact]
+        public void TestTrainingEvalModes()
+        {
+            var sequential = Sequential(
+                ("lin1", Linear(100, 10)),
+                ("lin2", Linear(10, 5))
+            );
+            sequential.eval();
+            // The entire sequential module and its layers should be in evaluation mode, not in training
+            var firstLayer = (torch.nn.Module)sequential[0];
+            Assert.False(firstLayer.training);
+
+            var secondLayer = (torch.nn.Module)sequential[1];
+            Assert.False(secondLayer.training);
+
+            Assert.False(sequential.training);
+
+            sequential.train();
+            // The entire sequential module and its layers should be in training mode, not in evaluation
+            Assert.True(firstLayer.training);
+            Assert.True(secondLayer.training);
+            Assert.True(sequential.training);
+        }
+
+
         /// <summary>
         /// Fully connected ReLU net with one hidden layer trained using gradient descent.
         /// Taken from <see href="https://pytorch.org/tutorials/beginner/examples_nn/two_layer_net_nn.html"/>.