[12/N] Apply clang-tidy and fix warnings in headers of torch/csrc (#116486)

Fixes #ISSUE_NUMBER

Pull Request resolved: https://github.com/pytorch/pytorch/pull/116486
Approved by: https://github.com/albanD

torch/csrc/Exceptions.h

@@ -145,7 +145,7 @@ extern PyObject *THPException_FatalError, *THPException_LinAlgError,
 // Throwing this exception means that the python error flags have been already
 // set and control should be immediately returned to the interpreter.
 struct python_error : public std::exception {
-  python_error() : type(nullptr), value(nullptr), traceback(nullptr) {}
+  python_error() {}
 
   python_error(const python_error& other)
       : type(other.type),
@@ -244,9 +244,9 @@ struct python_error : public std::exception {
     PyErr_Restore(type, value, traceback);
   }
 
-  PyObject* type;
+  PyObject* type{nullptr};
-  PyObject* value;
+  PyObject* value{nullptr};
-  PyObject* traceback;
+  PyObject* traceback{nullptr};
 
   // Message to return to the user when 'what()' is invoked.
   std::string message;

torch/csrc/autograd/FunctionsManual.h

@@ -12,10 +12,7 @@
 #include <ATen/ATen.h>
 #include <torch/csrc/autograd/generated/Functions.h>
 
-namespace torch {
+namespace torch::autograd::generated::details {
-namespace autograd {
-namespace generated {
-namespace details {
 
 extern const char* kCudnnDoubleBackwardMsg;
 
@@ -1101,7 +1098,4 @@ mkldnn_rnn_layer_differentiable_backward(
 
 Tensor values_backward(const Tensor& grad, const Tensor& self);
 
-} // namespace details
+} // namespace torch::autograd::generated::details
-} // namespace generated
-} // namespace autograd
-} // namespace torch

torch/csrc/autograd/InferenceMode.h

@@ -3,10 +3,8 @@
 #include <c10/core/InferenceMode.h>
 #include <torch/csrc/Export.h>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 using InferenceMode = c10::InferenceMode;
 
 }
-} // namespace torch

torch/csrc/autograd/VariableTypeUtils.h

@@ -17,14 +17,9 @@
 #include <torch/csrc/autograd/jit_decomp_interface.h>
 #include <torch/csrc/utils/variadic.h>
 
-#include <array>
 #include <cstddef>
 #include <functional>
-#include <initializer_list>
 #include <memory>
-#include <stdexcept>
-#include <string>
-#include <tuple>
 #include <utility>
 #include <vector>
 
@@ -117,8 +112,8 @@ inline void rebase_history(Variable& var, std::shared_ptr<Node> grad_fn) {
 }
 
 inline void rebase_history(
-    std::vector<Variable>&& vars,
+    const std::vector<Variable>& vars,
-    std::shared_ptr<Node> grad_fn) {
+    const std::shared_ptr<Node>& grad_fn) {
   if (grad_fn) {
     for (auto& var : vars) {
       if (var.defined()) {
@@ -137,6 +132,7 @@ inline void increment_version(const at::Tensor& t) {
 
 struct Flatten : IterArgs<Flatten> {
   Flatten(variable_list& out) : out(out) {}
+  // NOLINTNEXTLINE(cppcoreguidelines-avoid-const-or-ref-data-members)
   variable_list& out;
   void operator()(const at::Tensor& x) {
     out.emplace_back(x);

torch/csrc/autograd/anomaly_mode.h

@@ -4,8 +4,7 @@
 #include <memory>
 #include <string>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 // forward declaration of Node from function.h
 struct Node;
@@ -69,5 +68,4 @@ struct TORCH_API AnomalyMetadata {
   std::shared_ptr<Node> parent_;
 };
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/autograd.h

@@ -2,8 +2,7 @@
 
 #include <torch/csrc/autograd/variable.h>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 /// Computes the sum of gradients of given tensors with respect to graph leaves.
 ///
@@ -102,5 +101,4 @@ TORCH_API uint64_t enter_dual_level();
 TORCH_API void exit_dual_level(uint64_t level);
 
 } // namespace forward_ad
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/autograd_not_implemented_fallback.h

@@ -2,8 +2,7 @@
 
 #include <torch/library.h>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 // Default DispatchKey::Autograd fallback for built-in operators.
 // Can be registered for custom operators.
@@ -30,5 +29,4 @@ enum class AutogradFallbackMode {
 TORCH_API void setAutogradFallbackMode(AutogradFallbackMode mode);
 TORCH_API AutogradFallbackMode getAutogradFallbackMode();
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/cpp_hook.h

@@ -3,8 +3,7 @@
 #include <functional>
 #include <memory>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 using hooks_list =
     std::vector<std::function<at::TensorBase(const at::TensorBase&)>>;
@@ -27,5 +26,4 @@ struct CppFunctionSingleTensorPreHook : public FunctionPreHook {
   size_t value_idx_;
 };
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/custom_function.h

@@ -8,8 +8,7 @@
 #include <torch/csrc/autograd/variable.h>
 #include <vector>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 using optional_variable_list = std::vector<c10::optional<Variable>>;
 using _jvp_fn_t = std::function<variable_list(variable_list, variable_list)>;
@@ -97,7 +96,7 @@ struct TORCH_API Function {
   // the parameter X.
   template <typename X = T, typename... Args>
   static auto apply(Args&&... args)
-      -> std::enable_if_t<std::is_same<X, T>::value, forward_t<X, Args...>>;
+      -> std::enable_if_t<std::is_same_v<X, T>, forward_t<X, Args...>>;
 };
 
 /// Context to save information during `forward` that can be accessed in
@@ -228,8 +227,8 @@ inline void extract_vars(
 }
 
 template <typename T>
-typename std::enable_if<std::is_same<T, variable_list>::value, T>::type
+std::enable_if_t<std::is_same_v<T, variable_list>, T> to_output_type(
-to_output_type(std::vector<c10::optional<Variable>>& output_list) {
+    std::vector<c10::optional<Variable>>& output_list) {
   // NOLINTNEXTLINE(cppcoreguidelines-init-variables)
   variable_list result;
   std::transform(
@@ -241,8 +240,8 @@ to_output_type(std::vector<c10::optional<Variable>>& output_list) {
 }
 
 template <typename T>
-typename std::enable_if<std::is_same<T, Variable>::value, T>::type
+std::enable_if_t<std::is_same_v<T, Variable>, T> to_output_type(
-to_output_type(std::vector<c10::optional<Variable>>& output_list) {
+    std::vector<c10::optional<Variable>>& output_list) {
   return *output_list[0];
 }
 
@@ -264,7 +263,7 @@ inline std::vector<c10::optional<Variable>> to_optional(variable_list& output) {
 template <class T>
 template <typename X, typename... Args>
 auto Function<T>::apply(Args&&... args)
-    -> std::enable_if_t<std::is_same<X, T>::value, forward_t<X, Args...>> {
+    -> std::enable_if_t<std::is_same_v<X, T>, forward_t<X, Args...>> {
   const auto& functorch_tls = at::functorch::functorchTLSAccessor();
   if (functorch_tls) {
     // Function support for functorch is handled in Python.
@@ -434,5 +433,4 @@ void CppNode<T>::set_ctx_grad_fn(const std::shared_ptr<Node>& node) {
   ctx_.grad_fn_ = node;
 }
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/edge.h

@@ -6,8 +6,7 @@
 
 #include <c10/util/hash.h>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 struct Node;
 
@@ -38,8 +37,7 @@ struct Edge {
   /// The identifier of a particular input to the function.
   uint32_t input_nr;
 };
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch
 
 // The idiomatic way of enabling use of a custom type as the key of hash
 // containers in C++11. This method removes the requirement of having to pass

torch/csrc/autograd/engine.h

@@ -27,14 +27,11 @@
 #include <utility>
 #include <vector>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 struct ReadyQueue;
 }
-} // namespace torch
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 // Maximum reentrant backward depth before switching to a new thread
 // This limit is based on the TSAN's deadlock detector, where it will
@@ -291,5 +288,4 @@ struct TORCH_API Engine {
 using EngineStub = Engine& (*)();
 TORCH_API void set_default_engine_stub(EngineStub stub);
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/forward_grad.h

@@ -3,8 +3,7 @@
 #include <ATen/core/Tensor.h>
 #include <unordered_set>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 // [ Using ForwardGrad ]
 // ForwardGrad needs to be a shared_ptr to satisfy constraints of its inner
@@ -208,5 +207,4 @@ struct TORCH_API ForwardGrad : std::enable_shared_from_this<ForwardGrad> {
   mutable std::mutex mutex_;
 };
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/function.h

@@ -24,8 +24,7 @@
 #include <utility>
 #include <vector>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 struct Edge;
 struct FunctionPostHook;
@@ -757,5 +756,4 @@ struct TypeAndSize {
   at::TensorOptions options;
 };
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/function_hook.h

@@ -12,8 +12,7 @@ class SwapSavedVariables;
 
 // A hook that's called on gradients
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 using Variable = at::Tensor;
 using variable_list = std::vector<Variable>;
@@ -62,5 +61,4 @@ struct TORCH_API PostAccumulateGradHook {
   }
 };
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/grad_mode.h

@@ -3,11 +3,9 @@
 #include <ATen/core/grad_mode.h>
 #include <torch/csrc/Export.h>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 using GradMode = at::GradMode;
 using AutoGradMode = at::AutoGradMode;
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/graph_task.h

@@ -6,8 +6,7 @@
 #include <torch/csrc/autograd/utils/warnings.h>
 #include <vector>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 using edge_list = std::vector<Edge>;
 struct ReadyQueue;
@@ -239,5 +238,4 @@ TORCH_API std::vector<Node*> get_current_graph_task_execution_order();
 TORCH_API int get_current_graph_task_id();
 void add_node_to_current_graph_task_exec_info(Node* fn);
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/input_buffer.h

@@ -13,8 +13,7 @@
 #include <c10/util/Optional.h>
 #include <torch/csrc/autograd/variable.h>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 struct InputBuffer {
   explicit InputBuffer(size_t size) : buffer(size) {}
@@ -44,5 +43,4 @@ struct InputBuffer {
   std::vector<Variable> buffer;
 };
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/input_metadata.h

@@ -22,8 +22,7 @@
 #include <cstdint>
 #include <utility>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 using SymIntSmallVec = c10::SmallVector<c10::SymInt, c10::kDimVectorStaticSize>;
 using MetadataShape = std::variant<SymIntSmallVec, at::Tensor>;
@@ -109,5 +108,4 @@ struct TORCH_API InputMetadata {
   bool is_nested_ = false;
   bool was_default_constructed_ = true;
 };
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/jit_decomp_interface.h

@@ -27,9 +27,7 @@
 // For extra context, see VariableHooksInterface.h, where a similar technique
 // is used
 
-namespace torch {
+namespace torch::autograd::impl {
-namespace autograd {
-namespace impl {
 
 struct TORCH_API JitDecompInterface {
   virtual ~JitDecompInterface() = default;
@@ -49,6 +47,4 @@ struct TORCH_API JitDecompRegisterer {
   }
 };
 
-} // namespace impl
+} // namespace torch::autograd::impl
-} // namespace autograd
-} // namespace torch

torch/csrc/autograd/profiler_kineto.h

@@ -9,16 +9,15 @@
 #include <torch/csrc/profiler/util.h>
 
 namespace torch {
-namespace profiler {
+
-namespace impl {
+namespace profiler::impl {
 struct Result;
 namespace kineto {
 struct ActivityTraceWrapper;
 } // namespace kineto
-} // namespace impl
+} // namespace profiler::impl
-} // namespace profiler
+
-namespace autograd {
+namespace autograd::profiler {
-namespace profiler {
 using experimental_event_t = std::shared_ptr<torch::profiler::impl::Result>;
 using extra_meta_t = std::unordered_map<std::string, std::string>;
 
@@ -177,16 +176,13 @@ TORCH_API void prepareProfiler(
     const torch::profiler::impl::ProfilerConfig& config,
     const std::set<torch::profiler::impl::ActivityType>& activities);
 
-} // namespace profiler
+} // namespace autograd::profiler
-} // namespace autograd
 
-namespace profiler {
+namespace profiler::impl {
-namespace impl {
 
 // Experimental.
 TORCH_API void _reportVulkanEventToProfiler(vulkan_id_t id);
 
-} // namespace impl
+} // namespace profiler::impl
-} // namespace profiler
 
 } // namespace torch

torch/csrc/autograd/profiler_legacy.h

@@ -15,8 +15,7 @@
 #include <torch/csrc/profiler/stubs/base.h>
 #include <torch/csrc/profiler/util.h>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 struct Node;
 
@@ -413,5 +412,4 @@ struct TORCH_API TLSLegacyProfilerGuard {
 };
 
 } // namespace profiler
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/profiler_python.h

@@ -1,13 +1,7 @@
 #pragma once
 
-namespace torch {
+namespace torch::autograd::profiler::python_tracer {
-namespace autograd {
-namespace profiler {
-namespace python_tracer {
 
 void init();
 
 }
-} // namespace profiler
-} // namespace autograd
-} // namespace torch

torch/csrc/autograd/python_autograd.h

@@ -4,13 +4,11 @@
 PyObject* THPAutograd_initExtension(PyObject* _unused, PyObject* unused);
 void THPAutograd_initFunctions();
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 PyMethodDef* python_functions();
 
 }
-} // namespace torch
 
 #include <torch/csrc/autograd/python_engine.h>
 #include <torch/csrc/autograd/python_function.h>

torch/csrc/autograd/python_cpp_function.h

@@ -8,8 +8,7 @@
 #include <torch/csrc/autograd/function.h>
 #include <torch/csrc/utils/object_ptr.h>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 struct THPCppFunction {
   PyObject_HEAD std::shared_ptr<Node> cdata;
@@ -103,5 +102,4 @@ PyObject* functionToPyObject(const std::shared_ptr<Node>& cdata);
 
 bool THPCppFunction_Check(PyObject* obj);
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/python_engine.h

@@ -7,9 +7,7 @@
 
 bool THPEngine_initModule(PyObject* module);
 
-namespace torch {
+namespace torch::autograd::python {
-namespace autograd {
-namespace python {
 
 struct PythonEngine : public Engine {
   static Engine& get_python_engine();
@@ -43,6 +41,4 @@ struct PythonEngine : public Engine {
   PythonEngine();
 };
 
-} // namespace python
+} // namespace torch::autograd::python
-} // namespace autograd
-} // namespace torch

torch/csrc/autograd/python_enum_tag.h

@@ -2,8 +2,6 @@
 
 #include <torch/csrc/python_headers.h>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 void initEnumTag(PyObject* module);
 }
-} // namespace torch

torch/csrc/autograd/python_fft_functions.h

@@ -1,9 +1,7 @@
 #pragma once
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 void initFFTFunctions(PyObject* module);
 
 }
-} // namespace torch

torch/csrc/autograd/python_function.h

@@ -16,13 +16,11 @@
 #include <utility>
 #include <vector>
 
-namespace torch {
+namespace torch::jit {
-namespace jit {
 struct Graph;
 }
-} // namespace torch
+
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 // A Function which is implemented by a Python object (i.e., a THPFunction).
 // Calls to 'apply' are forwarded to the Python method implementation.
@@ -71,8 +69,7 @@ inline bool ensure_tuple(THPObjectPtr& obj) {
   return true;
 }
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch
 
 // NOLINTNEXTLINE(cppcoreguidelines-pro-type-member-init)
 struct THPFunction {

torch/csrc/autograd/python_hook.h

@@ -8,8 +8,7 @@ namespace torch::dynamo::autograd {
 class SwapSavedVariables;
 } // namespace torch::dynamo::autograd
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 struct PyFunctionTensorPreHook : public FunctionPreHook {
   PyFunctionTensorPreHook(PyObject* dict, size_t value_idx);
@@ -53,5 +52,4 @@ struct PyFunctionTensorPostAccGradHooks : public PostAccumulateGradHook {
   PyObject* dict;
 };
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/python_legacy_variable.h

@@ -5,10 +5,8 @@
 
 #include <torch/csrc/python_headers.h>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 void init_legacy_variable(PyObject* module);
 
 }
-} // namespace torch

torch/csrc/autograd/python_linalg_functions.h

@@ -1,9 +1,7 @@
 #pragma once
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 void initLinalgFunctions(PyObject* module);
 
 }
-} // namespace torch

torch/csrc/autograd/python_nested_functions.h

@@ -1,11 +1,9 @@
 #pragma once
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 PyMethodDef* get_nested_functions_manual();
 
 void initNestedFunctions(PyObject* module);
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/python_nn_functions.h

@@ -1,9 +1,7 @@
 #pragma once
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 void initNNFunctions(PyObject* module);
 
 }
-} // namespace torch

torch/csrc/autograd/python_saved_variable_hooks.h

@@ -10,8 +10,7 @@
 
 namespace py = pybind11;
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 struct PySavedVariableHooks : public SavedVariableHooks {
   PySavedVariableHooks(py::function& pack_hook, py::function& unpack_hook);
@@ -31,5 +30,4 @@ struct PyDefaultSavedVariableHooks {
   static std::unique_ptr<SavedVariableHooks> get_hooks();
 };
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/python_sparse_functions.h

@@ -1,9 +1,7 @@
 #pragma once
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 void initSparseFunctions(PyObject* module);
 
 }
-} // namespace torch

torch/csrc/autograd/python_special_functions.h

@@ -1,9 +1,7 @@
 #pragma once
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 void initSpecialFunctions(PyObject* module);
 
 }
-} // namespace torch

torch/csrc/autograd/python_torch_functions.h

@@ -2,8 +2,7 @@
 
 #include <vector>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 extern PyObject* THPVariableFunctionsModule;
 
@@ -25,5 +24,4 @@ inline PyObject* TypeError_to_NotImplemented_(
 
 void initTorchFunctions();
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/python_variable_indexing.h

@@ -6,8 +6,7 @@
 #include <torch/csrc/utils/pybind.h>
 #include <torch/csrc/utils/python_symnode.h>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 struct UnpackedSlice {
   c10::SymInt start;
@@ -100,5 +99,4 @@ Variable valueToTensor(
     PyObject* value,
     const at::Device& device);
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/record_function_ops.h

@@ -3,9 +3,7 @@
 #include <c10/util/Optional.h>
 #include <torch/custom_class.h>
 
-namespace torch {
+namespace torch::autograd::profiler {
-namespace autograd {
-namespace profiler {
 
 struct PythonRecordFunction : public torch::CustomClassHolder {
   at::RecordFunction record;
@@ -26,6 +24,4 @@ TORCH_API c10::intrusive_ptr<c10::ivalue::Future> _call_end_callbacks_on_fut_new
     const c10::intrusive_ptr<PythonRecordFunction>& record,
     const c10::intrusive_ptr<c10::ivalue::Future>& fut);
 
-} // namespace profiler
+} // namespace torch::autograd::profiler
-} // namespace autograd
-} // namespace torch

torch/csrc/autograd/saved_variable.h

@@ -9,8 +9,7 @@
 #include <cstdint>
 #include <memory>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 using Variable = at::Tensor;
 struct Node;
@@ -119,5 +118,4 @@ class TORCH_API SavedVariable {
       std::unique_ptr<SavedVariableHooks>&& hooks,
       const Variable& data);
 };
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/saved_variable_hooks.h

@@ -2,8 +2,7 @@
 
 #include <ATen/core/Tensor.h>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 struct TORCH_API SavedVariableHooks {
   virtual void call_pack_hook(const at::Tensor& tensor) = 0;
@@ -11,5 +10,4 @@ struct TORCH_API SavedVariableHooks {
   virtual ~SavedVariableHooks() = default;
 };
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/symbolic.h

@@ -4,8 +4,7 @@
 #include <torch/csrc/onnx/onnx.h>
 #include <vector>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 struct SymbolicContext {
   jit::Block* block;
@@ -15,5 +14,4 @@ struct symbolic_unconvertible : public std::runtime_error {
   using std::runtime_error::runtime_error;
 };
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch

torch/csrc/autograd/variable.h

@@ -21,8 +21,7 @@
 #include <utility>
 #include <vector>
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 /// `Variable` is exactly the same as `Tensor` (i.e. we have `using Variable =
 /// at::Tensor`). This means you can perform all the usual mathematical and
@@ -33,8 +32,7 @@ namespace autograd {
 /// is to eliminate the `Variable` class in the near future.
 using Variable = at::Tensor;
 
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch
 
 // The following are all internal APIs and should not be shown in libtorch docs.
 // Therefore, we wrap the following code with `#ifndef DOXYGEN_SHOULD_SKIP_THIS
@@ -42,8 +40,7 @@ using Variable = at::Tensor;
 
 #ifndef DOXYGEN_SHOULD_SKIP_THIS
 
-namespace torch {
+namespace torch::autograd {
-namespace autograd {
 
 /// Check if this type is supported by the autograd engine.
 /// If you change this, update the doc at the top of the
@@ -844,7 +841,6 @@ namespace utils {
 TORCH_API bool has_same_meta(const Variable& base, const Variable& other);
 
 } // namespace utils
-} // namespace autograd
+} // namespace torch::autograd
-} // namespace torch
 
 #endif /* DOXYGEN_SHOULD_SKIP_THIS */

torch/csrc/functorch/init.h

@@ -1,12 +1,7 @@
-#include <pybind11/pybind11.h>
+#include <Python.h>
-#include <torch/csrc/utils/pybind.h>
 
-namespace torch {
+namespace torch::functorch::impl {
-namespace functorch {
-namespace impl {
 
 void initFuncTorchBindings(PyObject* module);
 
 }
-} // namespace functorch
-} // namespace torch

torch/csrc/jit/frontend/source_range.h

@@ -11,8 +11,7 @@
 #include <sstream>
 #include <unordered_map>
 
-namespace torch {
+namespace torch::jit {
-namespace jit {
 
 class SourceRangeUnpickler;
 struct SourceRange;
@@ -444,8 +443,7 @@ using SourceRangeRecords = std::vector<TaggedRange>;
 using SourceRangeTagMap =
     std::unordered_map<SourceRange, int64_t, SourceRangeHasher>;
 
-} // namespace jit
+} // namespace torch::jit
-} // namespace torch
 
 namespace std {
 template <>

torch/csrc/jit/frontend/tracer.h

@@ -17,8 +17,7 @@
 #include <unordered_map>
 #include <vector>
 
-namespace torch {
+namespace torch::jit {
-namespace jit {
 struct Node;
 struct Value;
 struct Graph;
@@ -382,13 +381,13 @@ TORCH_API void ensureUniqueIfOutOfPlaced(
 
 template <
     typename T,
-    typename = torch::enable_if_t<(
+    typename = torch::enable_if_t<
-        !std::is_convertible<torch::decay_t<T>, at::TensorList>::value &&
+        (!std::is_convertible_v<torch::decay_t<T>, at::TensorList> &&
-        !std::is_convertible<torch::decay_t<T>, c10::List<at::Tensor>>::value &&
+         !std::is_convertible_v<torch::decay_t<T>, c10::List<at::Tensor>> &&
-        !std::is_convertible<torch::decay_t<T>, at::Tensor>::value &&
+         !std::is_convertible_v<torch::decay_t<T>, at::Tensor> &&
-        !std::is_convertible<
+         !std::is_convertible_v<
-            torch::decay_t<T>,
+             torch::decay_t<T>,
-            c10::intrusive_ptr<c10::ivalue::Object>>::value)>>
+             c10::intrusive_ptr<c10::ivalue::Object>>)>>
 void addOutput(Node* node, T&&) {
   AT_ERROR(
       "Found an unsupported argument type ",
@@ -410,5 +409,4 @@ TORCH_API autograd::Variable getSizeOf(
 TORCH_API autograd::Variable getNumelOf(const autograd::Variable& var);
 
 } // namespace tracer
-} // namespace jit
+} // namespace torch::jit
-} // namespace torch

torch/csrc/utils/pybind.h

@@ -28,8 +28,7 @@ PYBIND11_DECLARE_HOLDER_TYPE(T, c10::intrusive_ptr<T>, true);
 PYBIND11_DECLARE_HOLDER_TYPE(T, c10::SingletonOrSharedTypePtr<T>);
 PYBIND11_DECLARE_HOLDER_TYPE(T, c10::SingletonTypePtr<T>, true);
 
-namespace pybind11 {
+namespace pybind11::detail {
-namespace detail {
 
 // torch.Tensor <-> at::Tensor conversions (without unwrapping)
 template <>
@@ -324,11 +323,9 @@ struct type_caster<c10::complex<T>> {
   }
 };
 
-} // namespace detail
+} // namespace pybind11::detail
-} // namespace pybind11
 
-namespace torch {
+namespace torch::impl {
-namespace impl {
 
 // Use this function if you have a C++ object that is used from both C++
 // and Python contexts, and you need its GIL to be released when you
@@ -384,5 +381,4 @@ inline void destroy_without_gil(T* ptr) {
   }
 }
 
-} // namespace impl
+} // namespace torch::impl
-} // namespace torch

torch/csrc/utils/tensor_memoryformats.h

@@ -4,11 +4,9 @@
 #include <torch/csrc/Export.h>
 #include <torch/csrc/utils/python_stub.h>
 
-namespace torch {
+namespace torch::utils {
-namespace utils {
 
 void initializeMemoryFormats();
 TORCH_PYTHON_API PyObject* getTHPMemoryFormat(c10::MemoryFormat);
 
-} // namespace utils
+} // namespace torch::utils
-} // namespace torch

torch/csrc/utils/tensor_numpy.h

@@ -3,8 +3,7 @@
 #include <ATen/core/Tensor.h>
 #include <torch/csrc/python_headers.h>
 
-namespace torch {
+namespace torch::utils {
-namespace utils {
 
 PyObject* tensor_to_numpy(const at::Tensor& tensor, bool force = false);
 at::Tensor tensor_from_numpy(PyObject* obj, bool warn_if_not_writeable = true);
@@ -23,5 +22,4 @@ at::Tensor tensor_from_cuda_array_interface(PyObject* obj);
 void validate_numpy_for_dlpack_deleter_bug();
 bool is_numpy_dlpack_deleter_bugged();
 
-} // namespace utils
+} // namespace torch::utils
-} // namespace torch

torch/csrc/utils/variadic.h

@@ -73,13 +73,13 @@ struct MakeIndices<0, Is...> {
 //===----------------------------------------------------------------------===//
 
 template <bool value, typename T = void>
-using enable_if_t = typename std::enable_if<value, T>::type;
+using enable_if_t = std::enable_if_t<value, T>;
 
 template <bool value, typename T = void>
 using disable_if_t = enable_if_t<!value, T>;
 
 template <typename T>
-using decay_t = typename std::decay<T>::type;
+using decay_t = std::decay_t<T>;
 
 namespace detail {
 template <bool...>
@@ -112,7 +112,7 @@ using enable_if_all_of_t = enable_if_t<all_of<values...>::value>;
 
 template <typename T, typename... Ts>
 using disable_if_contains_t =
-    enable_if_all_of_t<(!std::is_same<T, decay_t<Ts>>::value)...>;
+    enable_if_all_of_t<(!std::is_same_v<T, decay_t<Ts>>)...>;
 
 template <typename Function, typename... Ts>
 void apply(Function function, Ts&&... ts) {