diff --git a/aten/src/ATen/MatrixRef.h b/aten/src/ATen/MatrixRef.h
index 74a20a687090..21c010e66db5 100644
--- a/aten/src/ATen/MatrixRef.h
+++ b/aten/src/ATen/MatrixRef.h
@@ -92,6 +92,7 @@ class MatrixRef {
   /// The declaration here is extra complicated so that "arrayRef = {}"
   /// continues to select the move assignment operator.
   template <typename U>
+  // NOLINTNEXTLINE(cppcoreguidelines-missing-std-forward)
   std::enable_if_t<std::is_same_v<U, T>, MatrixRef<T>>& operator=(
       U&& Temporary) = delete;
 
diff --git a/aten/src/ATen/core/ivalue.cpp b/aten/src/ATen/core/ivalue.cpp
index 2f55e7448005..5a49b16387aa 100644
--- a/aten/src/ATen/core/ivalue.cpp
+++ b/aten/src/ATen/core/ivalue.cpp
@@ -574,12 +574,7 @@ static std::ostream& printMaybeAnnotatedDict(
 static std::ostream& printComplex(std::ostream & out, const IValue & v) {
   c10::complex<double> d = v.toComplexDouble();
   IValue real(d.real()), imag(std::abs(d.imag()));
-  auto sign = "";
-  if (d.imag() >= 0) {
-    sign = "+";
-  } else {
-    sign = "-";
-  }
+  auto sign = d.imag() >= 0 ? '+' : '-';
   return out << real << sign << imag << "j";
 }
 
diff --git a/aten/src/ATen/cuda/EmptyTensor.cpp b/aten/src/ATen/cuda/EmptyTensor.cpp
index 108b7be47de1..5b96cdaaf77f 100644
--- a/aten/src/ATen/cuda/EmptyTensor.cpp
+++ b/aten/src/ATen/cuda/EmptyTensor.cpp
@@ -68,9 +68,11 @@ TensorBase empty_strided_cuda(
     std::optional<Device> device_opt,
     std::optional<bool> pin_memory_opt) {
   TORCH_CHECK(!pin_memory_opt.has_value() || !*pin_memory_opt, "Only dense CPU tensors can be pinned");
+#ifndef NDEBUG
   // TODO: remove check for jagged, see https://github.com/pytorch/pytorch/issues/130073
   const auto layout = layout_or_default(layout_opt);
   TORCH_INTERNAL_ASSERT_DEBUG_ONLY(layout == Layout::Strided || layout == Layout::Jagged);
+#endif
 
   const auto dtype = dtype_or_default(dtype_opt);
   return at::detail::empty_strided_cuda(size, stride, dtype, device_opt);
diff --git a/aten/src/ATen/cuda/tunable/GemmCommon.h b/aten/src/ATen/cuda/tunable/GemmCommon.h
index 6f2ff5583a73..8e44014d756a 100644
--- a/aten/src/ATen/cuda/tunable/GemmCommon.h
+++ b/aten/src/ATen/cuda/tunable/GemmCommon.h
@@ -12,6 +12,7 @@
 #include <string>
 
 #include <ATen/cuda/tunable/TunableOp.h>
+#include <ATen/cuda/CUDABlas.h>
 #include <ATen/cuda/Exceptions.h>
 #include <c10/util/StringUtil.h>
 
@@ -22,6 +23,7 @@
 #include <ATen/ops/allclose.h>
 #include <ATen/ops/from_blob.h>
 #endif
+#include <ATen/OpMathType.h>
 #include <fmt/printf.h>
 
 namespace at::cuda::tunable {
@@ -150,19 +152,19 @@ struct GemmParams : OpParams {
     return detail::NumericalCheck(c_dtype, c, other->c, GetSizeC()/sizeof(T)) ? OK : FAIL;
   }
 
-  char transa;
-  char transb;
-  int64_t m;
-  int64_t n;
-  int64_t k;
+  char transa{};
+  char transb{};
+  int64_t m{};
+  int64_t n{};
+  int64_t k{};
   at::opmath_type<T> alpha;
-  const T* a;
-  int64_t lda;
-  const T* b;
-  int64_t ldb;
+  const T* a{};
+  int64_t lda{};
+  const T* b{};
+  int64_t ldb{};
   at::opmath_type<T> beta;
-  T* c;
-  int64_t ldc;
+  T* c{};
+  int64_t ldc{};
 private:
   bool duplicate_inputs_{false};
 };
@@ -223,7 +225,9 @@ struct GemmAndBiasParams : OpParams {
   void Delete() {
     c10::cuda::CUDACachingAllocator::raw_delete(c);
     if (duplicate_inputs_) {
+      // NOLINTNEXTLINE(*const-cast)
       c10::cuda::CUDACachingAllocator::raw_delete(const_cast<T*>(a));
+      // NOLINTNEXTLINE(*const-cast)
       c10::cuda::CUDACachingAllocator::raw_delete(const_cast<T*>(b));
     }
   }
@@ -233,30 +237,26 @@ struct GemmAndBiasParams : OpParams {
     return detail::NumericalCheck(c_dtype, c, other->c, GetSizeC()/sizeof(T)) ? OK : FAIL;
   }
 
-  char transa;
-  char transb;
-  int64_t m;
-  int64_t n;
-  int64_t k;
-  at::opmath_type<T> alpha;
-  const T* a;
-  int64_t lda;
-  const T* b;
-  int64_t ldb;
-  T* c;
-  int64_t ldc;
-  const T* bias;
-  at::cuda::blas::GEMMAndBiasActivationEpilogue activation;
+  char transa{};
+  char transb{};
+  int64_t m{};
+  int64_t n{};
+  int64_t k{};
+  at::opmath_type<T> alpha{};
+  const T* a{};
+  int64_t lda{};
+  const T* b{};
+  int64_t ldb{};
+  T* c{};
+  int64_t ldc{};
+  const T* bias{};
+  at::cuda::blas::GEMMAndBiasActivationEpilogue activation{};
 private:
   bool duplicate_inputs_{false};
 };
 
 template <typename T>
 struct GemmStridedBatchedParams : OpParams {
-  GemmStridedBatchedParams() = default;
-  GemmStridedBatchedParams(const GemmStridedBatchedParams&) = default;
-  GemmStridedBatchedParams& operator=(const GemmStridedBatchedParams&) = default;
-
   std::string Signature() const override {
     return fmt::sprintf("%c%c_%ld_%ld_%ld_B_%ld", transa, transb, m, n, k, batch);
   }
@@ -325,23 +325,23 @@ struct GemmStridedBatchedParams : OpParams {
     return detail::NumericalCheck(c_dtype, c, other->c, GetSizeC()/sizeof(T)) ? OK : FAIL;
   }
 
-  char transa;
-  char transb;
-  int64_t m;
-  int64_t n;
-  int64_t k;
-  at::opmath_type<T> alpha;
-  const T* a;
-  int64_t lda;
-  int64_t stride_a;
-  const T* b;
-  int64_t ldb;
-  int64_t stride_b;
+  char transa{};
+  char transb{};
+  int64_t m{};
+  int64_t n{};
+  int64_t k{};
+  at::opmath_type<T> alpha{};
+  const T* a{};
+  int64_t lda{};
+  int64_t stride_a{};
+  const T* b{};
+  int64_t ldb{};
+  int64_t stride_b{};
   at::opmath_type<T> beta;
-  T* c;
-  int64_t ldc;
-  int64_t stride_c;
-  int64_t batch;
+  T* c{};
+  int64_t ldc{};
+  int64_t stride_c{};
+  int64_t batch{};
 private:
   bool duplicate_inputs_{false};
 };
@@ -415,27 +415,27 @@ struct ScaledGemmParams : OpParams {
     return detail::NumericalCheck(c_dtype, c, other->c, GetSizeC()/sizeof(T)) ? OK : FAIL;
   }
 
-  char transa;
-  char transb;
-  int64_t m;
-  int64_t n;
-  int64_t k;
-  const void* a;
-  const void* a_scale_ptr;
-  int64_t lda;
-  ScalarType a_dtype;
-  const void* b;
-  const void* b_scale_ptr;
-  int64_t ldb;
-  ScalarType b_dtype;
-  const void* bias_ptr;
-  ScalarType bias_dtype;
-  void* c;
-  const void* c_scale_ptr;
-  int64_t ldc;
-  ScalarType c_dtype;
-  void* amax_ptr;
-  bool use_fast_accum;
+  char transa{};
+  char transb{};
+  int64_t m{};
+  int64_t n{};
+  int64_t k{};
+  const void* a{};
+  const void* a_scale_ptr{};
+  int64_t lda{};
+  ScalarType a_dtype{};
+  const void* b{};
+  const void* b_scale_ptr{};
+  int64_t ldb{};
+  ScalarType b_dtype{};
+  const void* bias_ptr{};
+  ScalarType bias_dtype{};
+  void* c{};
+  const void* c_scale_ptr{};
+  int64_t ldc{};
+  ScalarType c_dtype{};
+  void* amax_ptr{};
+  bool use_fast_accum{};
 private:
   bool duplicate_inputs_{false};
 };
diff --git a/aten/src/ATen/cuda/tunable/StreamTimer.cpp b/aten/src/ATen/cuda/tunable/StreamTimer.cpp
index 1407c32dbb35..ed24a29d9919 100644
--- a/aten/src/ATen/cuda/tunable/StreamTimer.cpp
+++ b/aten/src/ATen/cuda/tunable/StreamTimer.cpp
@@ -9,9 +9,10 @@
 //
 #include <cuda_runtime.h>
 
-#include <c10/cuda/CUDAStream.h>
 #include <ATen/cuda/Exceptions.h>
 #include <ATen/cuda/tunable/StreamTimer.h>
+#include <c10/cuda/CUDAStream.h>
+#include <cmath>
 
 namespace at::cuda::tunable {
 
@@ -20,8 +21,7 @@ StreamTimer::StreamTimer() {
   AT_CUDA_CHECK(cudaEventCreate(&end_));
 }
 
-StreamTimer::~StreamTimer() {
-}
+StreamTimer::~StreamTimer() = default;
 
 void StreamTimer::Start() {
   AT_CUDA_CHECK(cudaDeviceSynchronize());
@@ -34,7 +34,7 @@ void StreamTimer::End() {
 }
 
 float StreamTimer::Duration() {
-  float time;
+  auto time = std::numeric_limits<float>::quiet_NaN();
   // time is in ms with a resolution of 1 us
   AT_CUDA_CHECK(cudaEventElapsedTime(&time, start_, end_));
   return time;
diff --git a/aten/src/ATen/cuda/tunable/StreamTimer.h b/aten/src/ATen/cuda/tunable/StreamTimer.h
index 36b8d72a4953..c83291d1b0e5 100644
--- a/aten/src/ATen/cuda/tunable/StreamTimer.h
+++ b/aten/src/ATen/cuda/tunable/StreamTimer.h
@@ -27,8 +27,8 @@ class StreamTimer : public ITimer {
     float Duration() override;
 
   private:
-    cudaEvent_t start_;
-    cudaEvent_t end_;
+    cudaEvent_t start_{};
+    cudaEvent_t end_{};
 };
 
 } // namespace at::cuda::tunable
diff --git a/aten/src/ATen/cuda/tunable/TunableOp.h b/aten/src/ATen/cuda/tunable/TunableOp.h
index 74d49f49e575..b1c607c72e0c 100644
--- a/aten/src/ATen/cuda/tunable/TunableOp.h
+++ b/aten/src/ATen/cuda/tunable/TunableOp.h
@@ -26,8 +26,6 @@ namespace at::cuda::tunable {
 template <typename ParamsT>
 class Callable {
   public:
-    Callable() = default;
-    Callable(Callable&&) = default;
     virtual ~Callable() = default;
     virtual TuningStatus Call(const ParamsT*) {
       return FAIL;
@@ -40,8 +38,6 @@ class Callable {
 template <typename ParamsT, typename TimerT>
 class TunableOp {
   public:
-    TunableOp() = default;
-    TunableOp(TunableOp&&) = default;
     virtual ~TunableOp() = default;
 
     TuningStatus operator()(const ParamsT* params) {
diff --git a/aten/src/ATen/functorch/TensorWrapper.cpp b/aten/src/ATen/functorch/TensorWrapper.cpp
index 53111ea98d08..f9fa6ee60d00 100644
--- a/aten/src/ATen/functorch/TensorWrapper.cpp
+++ b/aten/src/ATen/functorch/TensorWrapper.cpp
@@ -126,7 +126,7 @@ c10::intrusive_ptr<TensorImpl> TensorWrapper::shallow_copy_and_detach(
     c10::VariableVersion&& version_counter,
     bool allow_tensor_metadata_change) const {
   auto dest_impl = makeTensorWrapperPtr(value(), level_, is_alive_);
-  dest_impl->set_version_counter(version_counter);
+  dest_impl->set_version_counter(std::move(version_counter));
 
   // TODO: is this even right?
   dest_impl->set_allow_tensor_metadata_change(allow_tensor_metadata_change);
diff --git a/aten/src/ATen/native/nested/NestedTensorUtils.h b/aten/src/ATen/native/nested/NestedTensorUtils.h
index e36ae8a372f9..427840b6e0e0 100644
--- a/aten/src/ATen/native/nested/NestedTensorUtils.h
+++ b/aten/src/ATen/native/nested/NestedTensorUtils.h
@@ -264,8 +264,8 @@ class _map;
 template <class F, class A, class... Args>
 class _map<F, A, c10::guts::typelist::typelist<Args...>> {
  public:
-  static A function_one(F&& fn, const Args&... nested_node) {
-    return std::forward<F>(fn)(nested_node...);
+  static A function_one(const F& fn, const Args&... nested_node) {
+    return fn(nested_node...);
   }
   static NestedNode<A> function(
       const F& fn,
diff --git a/c10/core/AutogradState.h b/c10/core/AutogradState.h
index f98de71a83b6..ad168b8c0598 100644
--- a/c10/core/AutogradState.h
+++ b/c10/core/AutogradState.h
@@ -66,6 +66,7 @@ struct C10_API AutogradState {
   bool inference_mode_ : 1;
   bool fw_grad_mode_ : 1;
   bool multithreading_enabled_ : 1;
+  // NOLINTNEXTLINE(cppcoreguidelines-use-default-member-init)
   bool view_replay_enabled_ : 1;
 };
 
diff --git a/torch/csrc/Generator.cpp b/torch/csrc/Generator.cpp
index 60e83ed1c3f1..dfbffa7ba68f 100644
--- a/torch/csrc/Generator.cpp
+++ b/torch/csrc/Generator.cpp
@@ -30,7 +30,7 @@ using namespace torch;
 
 PyObject* THPGeneratorClass = nullptr;
 
-PyObject* THPGenerator_initDefaultGenerator(at::Generator cdata) {
+PyObject* THPGenerator_initDefaultGenerator(const at::Generator& cdata) {
   auto type = (PyTypeObject*)THPGeneratorClass;
   auto self = THPObjectPtr{type->tp_alloc(type, 0)};
   if (!self)
@@ -391,7 +391,7 @@ PyObject* pyobj(const Generator& self) {
   return self.pyobj();
 }
 
-PyObject* THPGenerator_Wrap(Generator gen) {
+PyObject* THPGenerator_Wrap(const Generator& gen) {
   if (!gen.defined()) {
     Py_RETURN_NONE;
   }
diff --git a/torch/csrc/Generator.h b/torch/csrc/Generator.h
index 4fef5911bab0..b5f72cb47b76 100644
--- a/torch/csrc/Generator.h
+++ b/torch/csrc/Generator.h
@@ -14,7 +14,7 @@ struct THPGenerator {
 // is borrowed. The caller should ensure that the at::Generator object lifetime
 // last at least as long as the Python wrapper.
 TORCH_PYTHON_API PyObject* THPGenerator_initDefaultGenerator(
-    at::Generator cdata);
+    const at::Generator& cdata);
 
 #define THPGenerator_Check(obj) PyObject_IsInstance(obj, THPGeneratorClass)
 
@@ -22,7 +22,7 @@ TORCH_PYTHON_API extern PyObject* THPGeneratorClass;
 
 bool THPGenerator_init(PyObject* module);
 
-TORCH_PYTHON_API PyObject* THPGenerator_Wrap(at::Generator gen);
+TORCH_PYTHON_API PyObject* THPGenerator_Wrap(const at::Generator& gen);
 
 TORCH_PYTHON_API at::Generator THPGenerator_Unwrap(PyObject* state);
 
diff --git a/torch/csrc/export/pybind.cpp b/torch/csrc/export/pybind.cpp
index 458b08c3f361..71464be05cac 100644
--- a/torch/csrc/export/pybind.cpp
+++ b/torch/csrc/export/pybind.cpp
@@ -7,6 +7,7 @@ void initExportBindings(PyObject* module) {
   auto rootModule = py::handle(module).cast<py::module>();
   auto m = rootModule.def_submodule("_export");
 
+  // NOLINTNEXTLINE(bugprone-unused-raii)
   py::class_<ExportedProgram>(m, "CppExportedProgram");
 
   m.def("deserialize_exported_program", [](const std::string& serialized) {
diff --git a/torch/csrc/lazy/backend/backend_interface.h b/torch/csrc/lazy/backend/backend_interface.h
index 064a578a39fd..19b1403cd285 100644
--- a/torch/csrc/lazy/backend/backend_interface.h
+++ b/torch/csrc/lazy/backend/backend_interface.h
@@ -7,7 +7,6 @@
 #include <torch/csrc/lazy/core/lazy_graph_executor.h>
 #include <torch/csrc/lazy/core/shape.h>
 #include <torch/csrc/lazy/core/tensor.h>
-#include <atomic>
 
 namespace torch::lazy {