diff --git a/.clang-tidy b/.clang-tidy
index ddcbd2e45648e2..eea5c737d0c496 100644
--- a/.clang-tidy
+++ b/.clang-tidy
@@ -42,7 +42,6 @@ modernize-*,
 -modernize-use-trailing-return-type,
 -modernize-use-nodiscard,
 performance-*,
--performance-unnecessary-value-param,
 readability-container-size-empty,
 '
 HeaderFilterRegex: '^(c10/(?!test)|torch/csrc/(?!deploy/interpreter/cpython)).*$'
diff --git a/aten/src/ATen/core/IListRef_test.cpp b/aten/src/ATen/core/IListRef_test.cpp
index d86b9c4d2f0cc0..2875d05c392b2d 100644
--- a/aten/src/ATen/core/IListRef_test.cpp
+++ b/aten/src/ATen/core/IListRef_test.cpp
@@ -30,15 +30,10 @@ static std::vector<at::OptionalTensorRef> get_unboxed_opt_tensor_vector() {
   static std::vector<at::Tensor> tensors;
   std::vector<at::OptionalTensorRef> optional_tensors;
   constexpr size_t SIZE = 5;
-  for (size_t i = 0; i < SIZE * 2; i++) {
-    if (i % 2 == 0) {
-      if (tensors.size() + 1 < i / 2) {
-        tensors.push_back(at::empty({0}));
-      }
-      optional_tensors.emplace_back(tensors[i / 2]);
-    } else {
-      optional_tensors.emplace_back();
-    }
+  for (size_t i = 0; i < SIZE; i++) {
+    tensors.push_back(at::empty({0}));
+    optional_tensors.emplace_back(tensors[i]);
+    optional_tensors.emplace_back();
   }
   return optional_tensors;
 }
diff --git a/aten/src/ATen/core/List.h b/aten/src/ATen/core/List.h
index 011ecf160b8172..d1271dadec2ac2 100644
--- a/aten/src/ATen/core/List.h
+++ b/aten/src/ATen/core/List.h
@@ -78,7 +78,7 @@ class ListElementReference final {
   ListElementReference& operator=(const T& new_value) &&;
 
   // assigning another ref to this assigns the underlying value
-  ListElementReference& operator=(ListElementReference&& rhs) &&;
+  ListElementReference& operator=(ListElementReference&& rhs) && noexcept;
 
   const IValue& get() const& {
     return *iterator_;
@@ -124,7 +124,7 @@ class ListIterator final {
   ListIterator(const ListIterator&) = default;
   ListIterator(ListIterator&&) noexcept = default;
   ListIterator& operator=(const ListIterator&) = default;
-  ListIterator& operator=(ListIterator&&) = default;
+  ListIterator& operator=(ListIterator&&) noexcept = default;
 
   ListIterator& operator++() {
       ++iterator_;
diff --git a/aten/src/ATen/core/List_inl.h b/aten/src/ATen/core/List_inl.h
index 97bafd95a0672d..110ef5747e5b0d 100644
--- a/aten/src/ATen/core/List_inl.h
+++ b/aten/src/ATen/core/List_inl.h
@@ -140,7 +140,7 @@ ListElementReference<T, Iterator>& ListElementReference<T, Iterator>::operator=(
 }
 
 template<class T, class Iterator>
-ListElementReference<T, Iterator>& ListElementReference<T, Iterator>::operator=(ListElementReference<T, Iterator>&& rhs) && {
+ListElementReference<T, Iterator>& ListElementReference<T, Iterator>::operator=(ListElementReference<T, Iterator>&& rhs) && noexcept {
   *iterator_ = *rhs.iterator_;
   return *this;
 }
diff --git a/aten/src/ATen/core/ivalue.cpp b/aten/src/ATen/core/ivalue.cpp
index 9c4fba46841502..d2936c5d73eed1 100644
--- a/aten/src/ATen/core/ivalue.cpp
+++ b/aten/src/ATen/core/ivalue.cpp
@@ -488,9 +488,9 @@ template <class T>
 std::ostream& printList(
     std::ostream& out,
     const T& list,
-    const std::string start,
-    const std::string finish,
-    IValueFormatter formatter) {
+    const std::string& start,
+    const std::string& finish,
+    const IValueFormatter& formatter) {
   out << start;
   for (const auto i : c10::irange(list.size())) {
     if (i > 0) {
@@ -506,16 +506,16 @@ std::ostream& printList(
 std::ostream& printMaybeAnnotatedList(
     std::ostream& out,
     const IValue& the_list,
-    IValueFormatter formatter) {
+    const IValueFormatter& formatter) {
   auto list_elem_type = the_list.type()->containedType(0);
   if (the_list.toListRef().empty() ||
       !elementTypeCanBeInferredFromMembers(list_elem_type)) {
     out << "annotate(" << the_list.type<c10::Type>()->annotation_str() << ", ";
-    printList(out, the_list.toListRef(), "[", "]", std::move(formatter));
+    printList(out, the_list.toListRef(), "[", "]", formatter);
     out << ")";
     return out;
   } else {
-    return printList(out, the_list.toListRef(), "[", "]", std::move(formatter));
+    return printList(out, the_list.toListRef(), "[", "]", formatter);
   }
 }
 
@@ -523,7 +523,7 @@ template <typename Dict>
 std::ostream& printDict(
     std::ostream& out,
     const Dict& v,
-    IValueFormatter formatter) {
+    const IValueFormatter& formatter) {
   out << "{";
 
   bool first = true;
@@ -547,14 +547,14 @@ std::ostream& printDict(
 static std::ostream& printMaybeAnnotatedDict(
     std::ostream& out,
     const IValue& the_dict,
-    IValueFormatter formatter) {
+    const IValueFormatter& formatter) {
   auto value_type = the_dict.type()->castRaw<DictType>()->getValueType();
   if (the_dict.toGenericDict().empty() ||
       !elementTypeCanBeInferredFromMembers(value_type)) {
     out << "annotate(" << the_dict.type<c10::Type>()->annotation_str() << ",";
-    printDict(out, the_dict.toGenericDict(), std::move(formatter)) << ")";
+    printDict(out, the_dict.toGenericDict(), formatter) << ")";
   } else {
-    return printDict(out, the_dict.toGenericDict(), std::move(formatter));
+    return printDict(out, the_dict.toGenericDict(), formatter);
   }
   return out;
 }
diff --git a/aten/src/ATen/core/ivalue_inl.h b/aten/src/ATen/core/ivalue_inl.h
index 3b57b5bfd781a6..80faacc8c8b1b1 100644
--- a/aten/src/ATen/core/ivalue_inl.h
+++ b/aten/src/ATen/core/ivalue_inl.h
@@ -1919,7 +1919,7 @@ template <
             std::is_lvalue_reference<Args>...,
             guts::negation<std::is_constructible<IValue, Args>>...>::value,
         std::nullptr_t> = nullptr>
-std::tuple<Args...> generic_to(IValue ivalue, _fake_type<std::tuple<Args...>>) {
+std::tuple<Args...> generic_to(const IValue& ivalue, _fake_type<std::tuple<Args...>>) {
   const auto& vals = ivalue.toTupleRef().elements();
   TORCH_CHECK(vals.size() == sizeof...(Args));
   return detail::generic_to_tuple_impl<std::tuple<Args...>>(vals, Indices{});
diff --git a/c10/core/StorageImpl.h b/c10/core/StorageImpl.h
index c42ddb905fcf25..c3471b8bcd0c12 100644
--- a/c10/core/StorageImpl.h
+++ b/c10/core/StorageImpl.h
@@ -55,7 +55,7 @@ struct C10_API StorageImpl : public c10::intrusive_ptr_target {
 
   StorageImpl(
       use_byte_size_t /*use_byte_size*/,
-      SymInt size_bytes,
+      const SymInt& size_bytes,
       at::Allocator* allocator,
       bool resizable)
       : StorageImpl(
diff --git a/c10/core/TensorImpl.h b/c10/core/TensorImpl.h
index 4db02bf0c2b928..8c0cdd489da59b 100644
--- a/c10/core/TensorImpl.h
+++ b/c10/core/TensorImpl.h
@@ -260,10 +260,11 @@ struct C10_API ExtraMeta {
       std::unique_ptr<c10::SymbolicShapeMeta> symbolic_shape_meta,
       std::unique_ptr<c10::NamedTensorMetaInterface> named_tensor_meta,
       intrusive_ptr<c10::BackendMeta> backend_meta,
-      c10::optional<std::string> custom_data_ptr_error_msg_ = c10::nullopt)
+      c10::optional<std::string> custom_data_ptr_error_msg = c10::nullopt)
       : symbolic_shape_meta_(std::move(symbolic_shape_meta)),
         named_tensor_meta_(std::move(named_tensor_meta)),
-        backend_meta_(std::move(backend_meta)) {}
+        backend_meta_(std::move(backend_meta)),
+        custom_data_ptr_error_msg_(std::move(custom_data_ptr_error_msg)) {}
 
   std::unique_ptr<ExtraMeta> clone() const {
     return std::make_unique<ExtraMeta>(*this);
diff --git a/functorch/csrc/dim/dim.cpp b/functorch/csrc/dim/dim.cpp
index 62dfcb599eee08..af6a1aee158f86 100644
--- a/functorch/csrc/dim/dim.cpp
+++ b/functorch/csrc/dim/dim.cpp
@@ -880,7 +880,7 @@ mpy::obj<Tensor> Tensor::create_delayed(mpy::object op, mpy::vector_args args, S
     mpy::obj<Tensor> self = Tensor::create();
     self->capture_levels(levels);
     self->has_device_ = has_device;
-    self->delayed_ = std::make_unique<DelayedOperator>(op, args);
+    self->delayed_ = std::make_unique<DelayedOperator>(std::move(op), args);
     return self;
 }
 
@@ -1082,7 +1082,7 @@ PyObject* py_tree_flatten(PyObject *self,
 
 
 
-mpy::object tree_map(Arena& A, std::function<mpy::handle(mpy::handle)> fn, mpy::handle agg) {
+mpy::object tree_map(Arena& A, const std::function<mpy::handle(mpy::handle)>& fn, mpy::handle agg) {
     Slice<mpy::handle> elements;
     auto unflatten = tree_flatten(A, agg, elements);
     for (auto i : elements.enumerate()) {
diff --git a/torch/csrc/jit/api/module.cpp b/torch/csrc/jit/api/module.cpp
index 319ff292163774..4686014394b574 100644
--- a/torch/csrc/jit/api/module.cpp
+++ b/torch/csrc/jit/api/module.cpp
@@ -471,7 +471,7 @@ IValue Module::create_class(const c10::QualifiedName& name, Stack stack) const {
 
 Module freeze(
     const Module& module,
-    c10::optional<std::vector<std::string>> preserved_attrs,
+    const c10::optional<std::vector<std::string>>& preserved_attrs,
     bool optimize_numerics) {
   TORCH_CHECK(
       !module.hasattr("training") || !module.is_training(),
diff --git a/torch/csrc/jit/api/module.h b/torch/csrc/jit/api/module.h
index 1e5c408602a504..6224ae7fc9ec1c 100644
--- a/torch/csrc/jit/api/module.h
+++ b/torch/csrc/jit/api/module.h
@@ -90,6 +90,8 @@ struct TORCH_API Module : public Object {
   explicit Module(c10::QualifiedName class_name);
   Module(std::shared_ptr<CompilationUnit> cu, const c10::ClassTypePtr& type);
   Module() = default;
+  Module(const Module&) = default;
+  Module& operator=(const Module&) = default;
   Module(
       c10::QualifiedName,
       std::shared_ptr<CompilationUnit> cu,
@@ -268,7 +270,7 @@ struct TORCH_API Module : public Object {
   }
 
   void set_delete_memory(std::shared_ptr<char> delete_mem) {
-    mem_to_delete_ = delete_mem;
+    mem_to_delete_ = std::move(delete_mem);
   }
 
   // A set of functions to maintain input shapes through torch.jit.save and
@@ -279,11 +281,11 @@ struct TORCH_API Module : public Object {
       return;
     }
     auto c10_inputs = c10::impl::GenericList(AnyType::get());
-    for (const IValue& value : inputs) {
+    for (IValue& value : inputs) {
       // Not checking whether this is traceable type as that is already checked
       // higher up in the stack and changing that would require a larger
       // restructuring.
-      c10_inputs.push_back(value);
+      c10_inputs.emplace_back(std::move(value));
     }
     traced_inputs_.insert_or_assign(func_name, c10_inputs);
   }
@@ -326,7 +328,8 @@ struct TORCH_API Module : public Object {
 // details.
 TORCH_API Module freeze(
     const Module& module,
-    c10::optional<std::vector<std::string>> preserved_attrs = c10::nullopt,
+    const c10::optional<std::vector<std::string>>& preserved_attrs =
+        c10::nullopt,
     bool optimize_numerics = true);
 
 // C++ equivalent api of `torch.jit.optimize_for_inference`. See documentation
@@ -400,7 +403,7 @@ struct slot_iterator_impl {
                     // slots of root
       bool return_module) // if true include root itself as the first thing
                           // visited (used in modules())
-      : cursors_({SlotCursor{root, return_module ? -1 : 0}}),
+      : cursors_({SlotCursor{std::move(root), return_module ? -1 : 0}}),
         recurse_(recurse) {
     // advance iterator to first valid element (or the end, if empty)
     while_not_valid_next();
@@ -543,7 +546,7 @@ struct slot_list_impl {
   }
 
   slot_list_impl(Module module, bool recurse, bool return_module)
-      : module_(module),
+      : module_(std::move(module)),
         recurse_(recurse),
         return_module_(return_module),
         size_(c10::nullopt) {
diff --git a/torch/csrc/jit/api/object.h b/torch/csrc/jit/api/object.h
index b32ad1f1d35d0d..15da200de26065 100644
--- a/torch/csrc/jit/api/object.h
+++ b/torch/csrc/jit/api/object.h
@@ -22,10 +22,10 @@ class ObjectAttributeError : public std::runtime_error {
   ObjectAttributeError(const std::string& what) : std::runtime_error(what) {}
 };
 
-// NOLINTNEXTLINE(cppcoreguidelines-pro-type-member-init)
 struct TORCH_API Object {
   Object() = default;
-  // NOLINTNEXTLINE(cppcoreguidelines-pro-type-member-init)
+  Object(const Object&) = default;
+  Object& operator=(const Object&) = default;
   Object(ObjectPtr _ivalue) : _ivalue_(std::move(_ivalue)) {}
   Object(std::shared_ptr<CompilationUnit> cu, const c10::ClassTypePtr& type);
   Object(
@@ -146,7 +146,9 @@ struct TORCH_API Object {
         setter = Method(_ivalue(), prop.setter);
       }
       return Property{
-          prop.name, Method(_ivalue(), prop.getter), std::move(setter)};
+          std::move(prop.name),
+          Method(_ivalue(), prop.getter),
+          std::move(setter)};
     });
   }
 
diff --git a/torch/csrc/profiler/collection.cpp b/torch/csrc/profiler/collection.cpp
index 3e6c06a7c26163..179ee69da86362 100644
--- a/torch/csrc/profiler/collection.cpp
+++ b/torch/csrc/profiler/collection.cpp
@@ -465,7 +465,7 @@ void materialize_vulkan(
     std::vector<std::shared_ptr<Result>>& out,
     AppendOnlyList<ExtraFields<EventType::Vulkan>::raw_event_t, BlockSize>&
         raw_events,
-    const std::function<time_t(approx_time_t)> time_converter,
+    const std::function<time_t(approx_time_t)>& time_converter,
     const uint64_t tid,
     const kineto::DeviceAndResource& kineto_info) {
   for (const auto& i : raw_events) {
diff --git a/torch/csrc/utils/init.cpp b/torch/csrc/utils/init.cpp
index d1c94b6629cd51..0f7ab000b467c4 100644
--- a/torch/csrc/utils/init.cpp
+++ b/torch/csrc/utils/init.cpp
@@ -40,13 +40,15 @@ void initThroughputBenchmarkBindings(PyObject* module) {
             // the GIL or not further down in the stack
             return self.runOnce(std::move(args), std::move(kwargs));
           })
-      .def("benchmark", [](ThroughputBenchmark& self, BenchmarkConfig config) {
-        // The benchmark always runs without the GIL. GIL will be used where
-        // needed. This will happen only in the nn.Module mode when manipulating
-        // inputs and running actual inference
-        pybind11::gil_scoped_release no_gil_guard;
-        return self.benchmark(config);
-      });
+      .def(
+          "benchmark",
+          [](ThroughputBenchmark& self, const BenchmarkConfig& config) {
+            // The benchmark always runs without the GIL. GIL will be used where
+            // needed. This will happen only in the nn.Module mode when
+            // manipulating inputs and running actual inference
+            pybind11::gil_scoped_release no_gil_guard;
+            return self.benchmark(config);
+          });
 }
 
 } // namespace throughput_benchmark
diff --git a/torch/csrc/utils/throughput_benchmark.cpp b/torch/csrc/utils/throughput_benchmark.cpp
index dbd89b9f5368ec..93b95024810b9f 100644
--- a/torch/csrc/utils/throughput_benchmark.cpp
+++ b/torch/csrc/utils/throughput_benchmark.cpp
@@ -39,7 +39,7 @@ py::object ThroughputBenchmark::runOnce(py::args&& args, py::kwargs&& kwargs) {
   }
 }
 
-ThroughputBenchmark::ThroughputBenchmark(jit::Module script_module)
+ThroughputBenchmark::ThroughputBenchmark(const jit::Module& script_module)
     : script_module_(script_module) {}
 
 ThroughputBenchmark::ThroughputBenchmark(py::object module)
diff --git a/torch/csrc/utils/throughput_benchmark.h b/torch/csrc/utils/throughput_benchmark.h
index a01d9e63c7c5b9..e540022e0f50ff 100644
--- a/torch/csrc/utils/throughput_benchmark.h
+++ b/torch/csrc/utils/throughput_benchmark.h
@@ -171,7 +171,7 @@ void ModuleBenchmark::addInput(py::args&& args, py::kwargs&& kwargs);
  */
 class C10_HIDDEN ThroughputBenchmark {
  public:
-  explicit ThroughputBenchmark(jit::Module module);
+  explicit ThroughputBenchmark(const jit::Module& module);
   explicit ThroughputBenchmark(py::object module);
 
   // Add one more input example. This input example should be in the exact