* Upgrade presets for PyTorch 1.9.0

CHANGELOG.md

@@ -9,7 +9,7 @@
  * Build FFmpeg with libxml2, enabling support for DASH demuxing ([pull #1033](https://github.com/bytedeco/javacpp-presets/pull/1033)), and libsrt for SRT protocol support ([pull #1036](https://github.com/bytedeco/javacpp-presets/pull/1036))
  * Add `@MemberGetter` for `av_log_default_callback()` in presets for FFmpeg ([issue #812](https://github.com/bytedeco/javacpp-presets/issues/812))
  * Include `cudaGL.h` and `cuda_gl_interop.h` header files in presets for CUDA ([pull #1027](https://github.com/bytedeco/javacpp-presets/pull/1027))
- * Add presets for libffi 3.3 ([issue #833](https://github.com/bytedeco/javacpp-presets/issues/833)), NVIDIA Video Codec SDK 11.0.10 ([pull #1020](https://github.com/bytedeco/javacpp-presets/pull/1020)), PyTorch 1.8.1 ([issue #623](https://github.com/bytedeco/javacpp-presets/issues/623)), TensorFlow Lite 2.5.0, DepthAI 2.5.0, ModSecurity ([pull #1012](https://github.com/bytedeco/javacpp-presets/pull/1012))
+ * Add presets for libffi 3.3 ([issue #833](https://github.com/bytedeco/javacpp-presets/issues/833)), NVIDIA Video Codec SDK 11.0.10 ([pull #1020](https://github.com/bytedeco/javacpp-presets/pull/1020)), PyTorch 1.9.0 ([issue #623](https://github.com/bytedeco/javacpp-presets/issues/623)), TensorFlow Lite 2.5.0, DepthAI 2.5.0, ModSecurity ([pull #1012](https://github.com/bytedeco/javacpp-presets/pull/1012))
  * Map `std::vector<cv::Range>` to `RangeVector` in `opencv_core.Mat` for convenience ([issue bytedeco/javacv#1607](https://github.com/bytedeco/javacv/issues/1607))
  * Include `genericaliasobject.h`, `context.h`, `tracemalloc.h`, and `datetime.h` for CPython ([issue #1017](https://github.com/bytedeco/javacpp-presets/issues/1017))
  * Add samples using LLVM modules to deal with bitcode and object files ([pull #1016](https://github.com/bytedeco/javacpp-presets/pull/1016))

README.md

@@ -210,7 +210,7 @@ Each child module in turn relies by default on the included [`cppbuild.sh` scrip
  * NVIDIA Video Codec SDK 11.0.x  https://developer.nvidia.com/nvidia-video-codec-sdk
  * OpenCL 3.0  https://github.com/KhronosGroup/OpenCL-ICD-Loader
  * MXNet 1.8.0  https://github.com/apache/incubator-mxnet
- * PyTorch 1.8.x  https://github.com/pytorch/pytorch
+ * PyTorch 1.9.x  https://github.com/pytorch/pytorch
  * TensorFlow 1.15.x  https://github.com/tensorflow/tensorflow
  * TensorFlow Lite 2.5.x  https://github.com/tensorflow/tensorflow
  * TensorRT 7.x  https://developer.nvidia.com/tensorrt

platform/pom.xml

@@ -276,7 +276,7 @@
     <dependency>
       <groupId>org.bytedeco</groupId>
       <artifactId>pytorch-platform</artifactId>
-      <version>1.8.1-${project.version}</version>
+      <version>1.9.0-${project.version}</version>
     </dependency>
     <dependency>
       <groupId>org.bytedeco</groupId>

pytorch/README.md

@@ -9,7 +9,7 @@ Introduction
 ------------
 This directory contains the JavaCPP Presets module for:
 
- * PyTorch 1.8.1  https://pytorch.org/
+ * PyTorch 1.9.0  https://pytorch.org/
 
 Please refer to the parent README.md file for more detailed information about the JavaCPP Presets.
 
@@ -46,14 +46,14 @@ We can use [Maven 3](http://maven.apache.org/) to download and install automatic
         <dependency>
             <groupId>org.bytedeco</groupId>
             <artifactId>pytorch-platform</artifactId>
-            <version>1.8.1-1.5.6-SNAPSHOT</version>
+            <version>1.9.0-1.5.6-SNAPSHOT</version>
         </dependency>
 
         <!-- Additional dependencies required to use CUDA, cuDNN, and NCCL -->
         <dependency>
             <groupId>org.bytedeco</groupId>
             <artifactId>pytorch-platform-gpu</artifactId>
-            <version>1.8.1-1.5.6-SNAPSHOT</version>
+            <version>1.9.0-1.5.6-SNAPSHOT</version>
         </dependency>
 
         <!-- Additional dependencies to use bundled CUDA, cuDNN, and NCCL -->

pytorch/cppbuild.sh

@@ -19,7 +19,7 @@ if [[ "$EXTENSION" == *gpu ]]; then
     export TORCH_CUDA_ARCH_LIST="3.5+PTX"
 fi
 
-PYTORCH_VERSION=1.8.1
+PYTORCH_VERSION=1.9.0
 
 mkdir -p "$PLATFORM$EXTENSION"
 cd "$PLATFORM$EXTENSION"

pytorch/platform/gpu/pom.xml

@@ -12,7 +12,7 @@
 
   <groupId>org.bytedeco</groupId>
   <artifactId>pytorch-platform-gpu</artifactId>
-  <version>1.8.1-${project.parent.version}</version>
+  <version>1.9.0-${project.parent.version}</version>
   <name>JavaCPP Presets Platform GPU for PyTorch</name>
 
   <properties>

pytorch/platform/pom.xml

@@ -12,7 +12,7 @@
 
   <groupId>org.bytedeco</groupId>
   <artifactId>pytorch-platform</artifactId>
-  <version>1.8.1-${project.parent.version}</version>
+  <version>1.9.0-${project.parent.version}</version>
   <name>JavaCPP Presets Platform for PyTorch</name>
 
   <properties>

pytorch/pom.xml

@@ -11,7 +11,7 @@
 
   <groupId>org.bytedeco</groupId>
   <artifactId>pytorch</artifactId>
-  <version>1.8.1-${project.parent.version}</version>
+  <version>1.9.0-${project.parent.version}</version>
   <name>JavaCPP Presets for PyTorch</name>
 
   <dependencies>

pytorch/samples/pom.xml

@@ -12,14 +12,14 @@
         <dependency>
             <groupId>org.bytedeco</groupId>
             <artifactId>pytorch-platform</artifactId>
-            <version>1.8.1-1.5.6-SNAPSHOT</version>
+            <version>1.9.0-1.5.6-SNAPSHOT</version>
         </dependency>
 
         <!-- Additional dependencies required to use CUDA, cuDNN, and NCCL -->
         <dependency>
             <groupId>org.bytedeco</groupId>
             <artifactId>pytorch-platform-gpu</artifactId>
-            <version>1.8.1-1.5.6-SNAPSHOT</version>
+            <version>1.9.0-1.5.6-SNAPSHOT</version>
         </dependency>
 
         <!-- Additional dependencies to use bundled CUDA, cuDNN, and NCCL -->

pytorch/src/gen/java/org/bytedeco/pytorch/AdagradOptions.java

@@ -34,4 +34,7 @@ public class AdagradOptions extends OptimizerCloneableAdagradOptions {
 
 
 
+  // NOLINTNEXTLINE(modernize-use-override)
+  public native double get_lr();
+  public native void set_lr(double lr);
 }

pytorch/src/gen/java/org/bytedeco/pytorch/AdagradParamState.java

@@ -39,4 +39,5 @@ public class AdagradParamState extends OptimizerCloneableAdagradParamState {
 
 
 
+  // NOLINTNEXTLINE(modernize-use-override)
 }

pytorch/src/gen/java/org/bytedeco/pytorch/AdamOptions.java

@@ -34,4 +34,7 @@ public class AdamOptions extends OptimizerCloneableAdamOptions {
 
 
 
+  // NOLINTNEXTLINE(modernize-use-override)
+  public native double get_lr();
+  public native void set_lr(double lr);
 }

pytorch/src/gen/java/org/bytedeco/pytorch/AdamParamState.java

@@ -41,4 +41,5 @@ public class AdamParamState extends OptimizerCloneableAdamParamState {
 
 
 
+  // NOLINTNEXTLINE(modernize-use-override)
 }

pytorch/src/gen/java/org/bytedeco/pytorch/AdamWOptions.java

@@ -34,4 +34,7 @@ public class AdamWOptions extends OptimizerCloneableAdamWOptions {
 
 
 
+  // NOLINTNEXTLINE(modernize-use-override)
+  public native double get_lr();
+  public native void set_lr(double lr);
 }

pytorch/src/gen/java/org/bytedeco/pytorch/AdamWParamState.java

@@ -41,4 +41,5 @@ public class AdamWParamState extends OptimizerCloneableAdamWParamState {
 
 
 
+  // NOLINTNEXTLINE(modernize-use-override)
 }

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveAvgPool1dImpl.java

@@ -29,6 +29,7 @@
  *  <pre>{@code
  *  AdaptiveAvgPool1d model(AdaptiveAvgPool1dOptions(5));
  *  }</pre> */
+// NOLINTNEXTLINE(bugprone-exception-escape)
 @Namespace("torch::nn") @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
 public class AdaptiveAvgPool1dImpl extends AdaptiveAvgPool1dImplBase {
     static { Loader.load(); }

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveAvgPool1dImplCloneable.java

@@ -25,6 +25,8 @@ public class AdaptiveAvgPool1dImplCloneable extends Module {
   /** {@code reset()} must perform initialization of all members with reference
    *  semantics, most importantly parameters, buffers and submodules. */
   public native void reset();
+  @Override public Module asModule() { return asModule(this); }
+  @Namespace public static native @Name("static_cast<torch::nn::Module*>") Module asModule(AdaptiveAvgPool1dImplCloneable module);
 
   /** Performs a recursive "deep copy" of the {@code Module}, such that all parameters
    *  and submodules in the cloned module are different from those in the

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveAvgPool2dImpl.java

@@ -29,6 +29,7 @@
  *  <pre>{@code
  *  AdaptiveAvgPool2d model(AdaptiveAvgPool2dOptions({3, 2}));
  *  }</pre> */
+// NOLINTNEXTLINE(bugprone-exception-escape)
 @Namespace("torch::nn") @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
 public class AdaptiveAvgPool2dImpl extends AdaptiveAvgPool2dImplBase {
     static { Loader.load(); }

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveAvgPool2dImplCloneable.java

@@ -25,6 +25,8 @@ public class AdaptiveAvgPool2dImplCloneable extends Module {
   /** {@code reset()} must perform initialization of all members with reference
    *  semantics, most importantly parameters, buffers and submodules. */
   public native void reset();
+  @Override public Module asModule() { return asModule(this); }
+  @Namespace public static native @Name("static_cast<torch::nn::Module*>") Module asModule(AdaptiveAvgPool2dImplCloneable module);
 
   /** Performs a recursive "deep copy" of the {@code Module}, such that all parameters
    *  and submodules in the cloned module are different from those in the

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveAvgPool3dImpl.java

@@ -29,6 +29,7 @@
  *  <pre>{@code
  *  AdaptiveAvgPool3d model(AdaptiveAvgPool3dOptions(3));
  *  }</pre> */
+// NOLINTNEXTLINE(bugprone-exception-escape)
 @Namespace("torch::nn") @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
 public class AdaptiveAvgPool3dImpl extends AdaptiveAvgPool3dImplBase {
     static { Loader.load(); }

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveAvgPool3dImplCloneable.java

@@ -25,6 +25,8 @@ public class AdaptiveAvgPool3dImplCloneable extends Module {
   /** {@code reset()} must perform initialization of all members with reference
    *  semantics, most importantly parameters, buffers and submodules. */
   public native void reset();
+  @Override public Module asModule() { return asModule(this); }
+  @Namespace public static native @Name("static_cast<torch::nn::Module*>") Module asModule(AdaptiveAvgPool3dImplCloneable module);
 
   /** Performs a recursive "deep copy" of the {@code Module}, such that all parameters
    *  and submodules in the cloned module are different from those in the

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveLogSoftmaxWithLossImpl.java

@@ -31,6 +31,7 @@
  *  <pre>{@code
  *  AdaptiveLogSoftmaxWithLoss model(AdaptiveLogSoftmaxWithLossOptions(8, 10, {4, 8}).div_value(2.).head_bias(true));
  *  }</pre> */
+// NOLINTNEXTLINE(bugprone-exception-escape)
 @Namespace("torch::nn") @NoOffset @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
 public class AdaptiveLogSoftmaxWithLossImpl extends AdaptiveLogSoftmaxWithLossImplCloneable {
     static { Loader.load(); }
@@ -39,7 +40,7 @@ public class AdaptiveLogSoftmaxWithLossImpl extends AdaptiveLogSoftmaxWithLossIm
 
    public AdaptiveLogSoftmaxWithLossImpl(@Cast("int64_t") long in_features, @Cast("int64_t") long n_classes, @ByVal @Cast("std::vector<int64_t>*") LongVector cutoffs) { super((Pointer)null); allocate(in_features, n_classes, cutoffs); }
    @NoDeallocator private native void allocate(@Cast("int64_t") long in_features, @Cast("int64_t") long n_classes, @ByVal @Cast("std::vector<int64_t>*") LongVector cutoffs);
-     
+
   public AdaptiveLogSoftmaxWithLossImpl(@ByVal AdaptiveLogSoftmaxWithLossOptions options_) { super((Pointer)null); allocate(options_); }
   @NoDeallocator private native void allocate(@ByVal AdaptiveLogSoftmaxWithLossOptions options_);
 
@@ -64,12 +65,12 @@ public class AdaptiveLogSoftmaxWithLossImpl extends AdaptiveLogSoftmaxWithLossIm
   /** The options with which this {@code Module} was constructed */
   public native @ByRef AdaptiveLogSoftmaxWithLossOptions options(); public native AdaptiveLogSoftmaxWithLossImpl options(AdaptiveLogSoftmaxWithLossOptions setter);
 
-  /** Cutoffs used to assign targets to their buckets. It should be an ordered Sequence 
+  /** Cutoffs used to assign targets to their buckets. It should be an ordered Sequence
    *  of integers sorted in the increasing order */
   public native @ByRef @Cast("std::vector<int64_t>*") LongVector cutoffs(); public native AdaptiveLogSoftmaxWithLossImpl cutoffs(LongVector setter);
 
   public native @Cast("int64_t") long shortlist_size(); public native AdaptiveLogSoftmaxWithLossImpl shortlist_size(long setter);
-  
+
   /** Number of clusters */
   public native @Cast("int64_t") long n_clusters(); public native AdaptiveLogSoftmaxWithLossImpl n_clusters(long setter);
 

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveLogSoftmaxWithLossImplCloneable.java

@@ -25,6 +25,8 @@ public class AdaptiveLogSoftmaxWithLossImplCloneable extends Module {
   /** {@code reset()} must perform initialization of all members with reference
    *  semantics, most importantly parameters, buffers and submodules. */
   public native void reset();
+  @Override public Module asModule() { return asModule(this); }
+  @Namespace public static native @Name("static_cast<torch::nn::Module*>") Module asModule(AdaptiveLogSoftmaxWithLossImplCloneable module);
 
   /** Performs a recursive "deep copy" of the {@code Module}, such that all parameters
    *  and submodules in the cloned module are different from those in the

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveMaxPool1dImpl.java

@@ -29,6 +29,7 @@
  *  <pre>{@code
  *  AdaptiveMaxPool1d model(AdaptiveMaxPool1dOptions(3));
  *  }</pre> */
+// NOLINTNEXTLINE(bugprone-exception-escape)
 @Namespace("torch::nn") @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
 public class AdaptiveMaxPool1dImpl extends AdaptiveMaxPool1dImplBase {
     static { Loader.load(); }

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveMaxPool1dImplCloneable.java

@@ -25,6 +25,8 @@ public class AdaptiveMaxPool1dImplCloneable extends Module {
   /** {@code reset()} must perform initialization of all members with reference
    *  semantics, most importantly parameters, buffers and submodules. */
   public native void reset();
+  @Override public Module asModule() { return asModule(this); }
+  @Namespace public static native @Name("static_cast<torch::nn::Module*>") Module asModule(AdaptiveMaxPool1dImplCloneable module);
 
   /** Performs a recursive "deep copy" of the {@code Module}, such that all parameters
    *  and submodules in the cloned module are different from those in the

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveMaxPool2dImpl.java

@@ -29,6 +29,7 @@
  *  <pre>{@code
  *  AdaptiveMaxPool2d model(AdaptiveMaxPool2dOptions({3, 2}));
  *  }</pre> */
+// NOLINTNEXTLINE(bugprone-exception-escape)
 @Namespace("torch::nn") @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
 public class AdaptiveMaxPool2dImpl extends AdaptiveMaxPool2dImplBase {
     static { Loader.load(); }

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveMaxPool2dImplCloneable.java

@@ -25,6 +25,8 @@ public class AdaptiveMaxPool2dImplCloneable extends Module {
   /** {@code reset()} must perform initialization of all members with reference
    *  semantics, most importantly parameters, buffers and submodules. */
   public native void reset();
+  @Override public Module asModule() { return asModule(this); }
+  @Namespace public static native @Name("static_cast<torch::nn::Module*>") Module asModule(AdaptiveMaxPool2dImplCloneable module);
 
   /** Performs a recursive "deep copy" of the {@code Module}, such that all parameters
    *  and submodules in the cloned module are different from those in the

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveMaxPool3dImpl.java

@@ -29,6 +29,7 @@
  *  <pre>{@code
  *  AdaptiveMaxPool3d model(AdaptiveMaxPool3dOptions(3));
  *  }</pre> */
+// NOLINTNEXTLINE(bugprone-exception-escape)
 @Namespace("torch::nn") @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
 public class AdaptiveMaxPool3dImpl extends AdaptiveMaxPool3dImplBase {
     static { Loader.load(); }

pytorch/src/gen/java/org/bytedeco/pytorch/AdaptiveMaxPool3dImplCloneable.java

@@ -25,6 +25,8 @@ public class AdaptiveMaxPool3dImplCloneable extends Module {
   /** {@code reset()} must perform initialization of all members with reference
    *  semantics, most importantly parameters, buffers and submodules. */
   public native void reset();
+  @Override public Module asModule() { return asModule(this); }
+  @Namespace public static native @Name("static_cast<torch::nn::Module*>") Module asModule(AdaptiveMaxPool3dImplCloneable module);
 
   /** Performs a recursive "deep copy" of the {@code Module}, such that all parameters
    *  and submodules in the cloned module are different from those in the

pytorch/src/gen/java/org/bytedeco/pytorch/AlphaDropoutImpl.java

@@ -29,6 +29,7 @@
  *  <pre>{@code
  *  AlphaDropout model(AlphaDropoutOptions(0.2).inplace(true));
  *  }</pre> */
+// NOLINTNEXTLINE(bugprone-exception-escape)
 @Namespace("torch::nn") @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
 public class AlphaDropoutImpl extends AlphaDropoutImplBase {
     static { Loader.load(); }

pytorch/src/gen/java/org/bytedeco/pytorch/AlphaDropoutImplCloneable.java

@@ -25,6 +25,8 @@ public class AlphaDropoutImplCloneable extends Module {
   /** {@code reset()} must perform initialization of all members with reference
    *  semantics, most importantly parameters, buffers and submodules. */
   public native void reset();
+  @Override public Module asModule() { return asModule(this); }
+  @Namespace public static native @Name("static_cast<torch::nn::Module*>") Module asModule(AlphaDropoutImplCloneable module);
 
   /** Performs a recursive "deep copy" of the {@code Module}, such that all parameters
    *  and submodules in the cloned module are different from those in the

pytorch/src/gen/java/org/bytedeco/pytorch/AnyModule.java

@@ -128,9 +128,9 @@ public class AnyModule extends Pointer {
 
   /** Move construction and assignment is allowed, and follows the default
    *  behavior of move for {@code std::unique_ptr}. */
-  public AnyModule(@ByVal AnyModule arg0) { super((Pointer)null); allocate(arg0); }
-  private native void allocate(@ByVal AnyModule arg0);
-  public native @ByRef @Name("operator =") AnyModule put(@ByVal AnyModule arg0);
+  public AnyModule(@ByRef(true) AnyModule arg0) { super((Pointer)null); allocate(arg0); }
+  private native void allocate(@ByRef(true) AnyModule arg0);
+  public native @ByRef @Name("operator =") AnyModule put(@ByRef(true) AnyModule arg0);
 
   /** Creates a shallow copy of an {@code AnyModule}. */
 

pytorch/src/gen/java/org/bytedeco/pytorch/AutoGradMode.java

@@ -18,7 +18,7 @@
 
 // A RAII, thread local (!) guard that enables or disables grad mode upon
 // construction, and sets it back to the original value upon destruction.
-@Namespace("at") @NoOffset @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
+@Namespace("c10") @NoOffset @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
 public class AutoGradMode extends Pointer {
     static { Loader.load(); }
     /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */

pytorch/src/gen/java/org/bytedeco/pytorch/AutogradMeta.java

@@ -57,7 +57,7 @@ public class AutogradMeta extends AutogradMetaInterface {
   public native @SharedPtr ForwardGrad fw_grad_(); public native AutogradMeta fw_grad_(ForwardGrad setter);
 
   public native @ByRef FunctionPreVector hooks_(); public native AutogradMeta hooks_(FunctionPreVector setter);
-  public native @Cast("torch::autograd::hooks_list*") @StdVector @SharedPtr PointerPointer cpp_hooks_list(); public native AutogradMeta cpp_hooks_list(PointerPointer setter);
+  public native @Cast("torch::autograd::hooks_list*") @StdVector @SharedPtr PointerPointer cpp_hooks_list_(); public native AutogradMeta cpp_hooks_list_(PointerPointer setter);
 
   // Only meaningful on leaf variables (must be false otherwise)
   public native @Cast("bool") boolean requires_grad_(); public native AutogradMeta requires_grad_(boolean setter);

pytorch/src/gen/java/org/bytedeco/pytorch/AutogradMetaInterface.java

@@ -22,10 +22,16 @@ public class AutogradMetaInterface extends Pointer {
     /** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
     public AutogradMetaInterface(Pointer p) { super(p); }
 
-  public native void set_requires_grad(@Cast("bool") boolean requires_grad, TensorImpl self_impl);
+  public native void set_requires_grad(
+        @Cast("bool") boolean requires_grad,
+        TensorImpl self_impl);
   public native @Cast("bool") boolean requires_grad();
   public native @ByRef Tensor mutable_grad();
   public native @Const @ByRef Tensor grad();
   public native @Const @ByRef Tensor fw_grad(@Cast("uint64_t") long level, @Const @ByRef Tensor self);
-  public native void set_fw_grad(@Const @ByRef Tensor new_grad, @Const @ByRef Tensor self, @Cast("uint64_t") long level, @Cast("bool") boolean is_inplace_op);
+  public native void set_fw_grad(
+        @Const @ByRef Tensor new_grad,
+        @Const @ByRef Tensor self,
+        @Cast("uint64_t") long level,
+        @Cast("bool") boolean is_inplace_op);
 }

pytorch/src/gen/java/org/bytedeco/pytorch/AvgPool1dImpl.java

@@ -29,6 +29,7 @@
  *  <pre>{@code
  *  AvgPool1d model(AvgPool1dOptions(3).stride(2));
  *  }</pre> */
+// NOLINTNEXTLINE(bugprone-exception-escape)
 @Namespace("torch::nn") @Properties(inherit = org.bytedeco.pytorch.presets.torch.class)
 public class AvgPool1dImpl extends AvgPool1dImplBase {
     static { Loader.load(); }

pytorch/src/gen/java/org/bytedeco/pytorch/AvgPool1dImplCloneable.java

@@ -25,6 +25,8 @@ public class AvgPool1dImplCloneable extends Module {
   /** {@code reset()} must perform initialization of all members with reference
    *  semantics, most importantly parameters, buffers and submodules. */
   public native void reset();
+  @Override public Module asModule() { return asModule(this); }
+  @Namespace public static native @Name("static_cast<torch::nn::Module*>") Module asModule(AvgPool1dImplCloneable module);
 
   /** Performs a recursive "deep copy" of the {@code Module}, such that all parameters
    *  and submodules in the cloned module are different from those in the