Add None return type to init (#132335)

Pull Request resolved: https://github.com/pytorch/pytorch/pull/132335
Approved by: https://github.com/albanD

test/dynamo/test_minifier.py

@@ -119,7 +119,7 @@ inner(torch.randn(20, 20, requires_grad=True) + 1)
         backend_name = "relu_compile_error_TESTING_ONLY"
         run_code = f"""\
 class CpuCudaModule(torch.nn.Module):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.m_x = torch.nn.Linear(20, 20).cuda()
         self.m_y = torch.nn.Linear(20, 20)
@@ -149,7 +149,7 @@ inner(torch.randn(20, 20).cuda(), torch.randn(20, 20))
             res.minifier_module(),
             """\
 class Repro(torch.nn.Module):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.G__mod___m_x = Linear(in_features=20, out_features=20, bias=True).cuda()
         self.G__mod___m_y = Linear(in_features=20, out_features=20, bias=True)
@@ -204,7 +204,7 @@ inner(torch.randn(20, 20))
             res.repro_module(),
             """\
 class Repro(torch.nn.Module):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
 
     def forward(self, x_19):

test/inductor/test_minifier.py

@@ -122,7 +122,7 @@ inner(torch.randn(20))
             res.repro_module(),
             """\
 class Repro(torch.nn.Module):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
 
     def forward(self, arg0_1):
@@ -138,7 +138,7 @@ class Repro(torch.nn.Module):
             res.repro_module(),
             """\
 class Repro(torch.nn.Module):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
 
     def forward(self, arg0_1):

torch/_classes.py

@@ -19,7 +19,7 @@ class _ClassNamespace(types.ModuleType):
 class _Classes(types.ModuleType):
     __file__ = "_classes.py"
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("torch.classes")
 
     def __getattr__(self, name):

torch/_decomp/decompositions_for_rng.py

@@ -71,7 +71,7 @@ class PhiloxState:
     trace time.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.reset()
 
     def reset(self):

torch/_dynamo/backends/distributed.py

@@ -247,7 +247,7 @@ class SubmodCompiler(torch.fx.interpreter.Interpreter):
             # This gives us the appropriately strided outputs here which will reflect runtime strides.
 
             class FakeifyFirstAOTInvocationGuard:
-                def __init__(self):
+                def __init__(self) -> None:
                     self.tc = torch._guards.TracingContext.try_get()
                     assert self.tc
                     torch._guards.TracingContext.try_get().fakify_first_call = True

torch/_dynamo/code_context.py

@@ -5,7 +5,7 @@ from .utils import ExactWeakKeyDictionary
 
 
 class CodeContextDict:
-    def __init__(self):
+    def __init__(self) -> None:
         self.code_context = ExactWeakKeyDictionary()
 
     def has_context(self, code: types.CodeType):

torch/_dynamo/debug_utils.py

@@ -170,7 +170,7 @@ class NNModuleToString:
             """
             from torch.nn import *
             class Repro(torch.nn.Module):
-                def __init__(self):
+                def __init__(self) -> None:
                     super().__init__()
             """
         )
@@ -491,7 +491,7 @@ _is_leaf_or_default = _mk_defaulter(False)
 
 
 class NopInputReader:
-    def __init__(self):
+    def __init__(self) -> None:
         self.total = 0
 
     def storage(self, storage_hash, nbytes, *, device=None, dtype_hint=None):

torch/_dynamo/eval_frame.py

@@ -497,7 +497,7 @@ class _TorchDynamoContext:
                         wrapper function.
 
                         >> class CallableClass:
-                        >>     def __init__(self):
+                        >>     def __init__(self) -> None:
                         >>         super().__init__()
                         >>         self.relu = torch.nn.ReLU()
                         >>
@@ -578,7 +578,7 @@ class OptimizeContext(_TorchDynamoContext):
 
 
 class RunOnlyContext(_TorchDynamoContext):
-    def __init__(self):
+    def __init__(self) -> None:
         # cudagraph trees relies on generation increment
         def on_enter():
             torch._dynamo.mutation_guard.GenerationTracker.generation += 1
@@ -590,7 +590,7 @@ class RunOnlyContext(_TorchDynamoContext):
 
 
 class DisableContext(_TorchDynamoContext):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__(callback=None)
 
     def __call__(self, fn):

torch/_dynamo/exc.py

@@ -74,7 +74,7 @@ class InvalidBackend(TorchDynamoException):
 
 
 class ResetRequired(TorchDynamoException):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__(
             textwrap.dedent(
                 """

torch/_dynamo/profiler.py

@@ -92,7 +92,7 @@ def print_missing(stack):
 class Profiler:
     unique_graphs = 0
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.prof = torch.profiler.profile(
             activities=[torch.profiler.ProfilerActivity.CPU],
             with_stack=should_print_missing(),

torch/_dynamo/variables/base.py

@@ -70,7 +70,7 @@ class MutableLocal(MutableLocalBase):
     state.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__(MutableLocalSource.Local)
 
     def __hash__(self):

torch/_dynamo/variables/builder.py

@@ -274,7 +274,7 @@ class GraphArg:
 
 
 class BackwardStateGraphArg(GraphArg):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__(
             source=None,
             _example=BackwardState(),
@@ -2646,7 +2646,7 @@ class SourcelessBuilder:
     if/else type->VariableTracker trees that were cropping up all over dynamo.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         raise AssertionError("Use SourcelessBuilder.create()")
 
     @staticmethod

torch/_export/db/examples/class_method.py

@@ -10,7 +10,7 @@ class ClassMethod(torch.nn.Module):
     def method(cls, x):
         return x + 1
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.linear = torch.nn.Linear(4, 2)
 

torch/_export/db/examples/cond_branch_class_method.py

@@ -26,7 +26,7 @@ class CondBranchClassMethod(torch.nn.Module):
     NOTE: If the `pred` is test on a dim with batch size < 2, it will be specialized.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.subm = MySubModule()
 

torch/_export/db/examples/model_attr_mutation.py

@@ -8,7 +8,7 @@ class ModelAttrMutation(torch.nn.Module):
     Attribute mutation is not supported.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.attr_list = [torch.randn(3, 2), torch.randn(3, 2)]
 

torch/_export/db/examples/scalar_output.py

@@ -11,7 +11,7 @@ class ScalarOutput(torch.nn.Module):
     Returning scalar values from the graph is supported, in addition to Tensor
     outputs. Symbolic shapes are captured and rank is specialized.
     """
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
 
     def forward(self, x):

torch/_export/db/examples/specialized_attribute.py

@@ -11,7 +11,7 @@ class SpecializedAttribute(torch.nn.Module):
     Model attributes are specialized.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.a = "moo"
         self.b = 4

torch/_export/passes/lift_constants_pass.py

@@ -24,7 +24,7 @@ class ConstantAttrMap(collections.abc.MutableMapping):
     if that's the case).
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         # Underlying dict that we use to implement this mapping.
         self._constant_attrs: Dict[
             Union[int, torch.Tensor, FakeScriptObject], List[Any]

torch/_export/serde/serialize.py

@@ -1413,7 +1413,7 @@ class GraphModuleDeserializer(metaclass=Final):
         constants: Dict[str, Union[torch.Tensor, FakeScriptObject, torch.ScriptObject]]
         example_inputs: Optional[Tuple[Tuple[torch.Tensor, ...], Dict[str, Any]]]
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.serialized_name_to_node: Dict[str, torch.fx.Node] = {}
         self.serialized_name_to_meta: Dict[str, MetaType] = {}
         self.graph = torch.fx.Graph()

torch/_functorch/_aot_autograd/schemas.py

@@ -602,7 +602,7 @@ class SubclassMeta:
     # Optional field because we don't compute for inference graphs
     grad_input_metas: Optional[List[Union[int, SubclassCreationMeta]]] = None
 
-    def __init__(self):
+    def __init__(self) -> None:
         # The fields in this class get set after its construction.
         pass
 

torch/_functorch/aot_autograd.py

@@ -878,7 +878,7 @@ def aot_module(mod: nn.Module, *args, **kwargs) -> nn.Module:
     )
 
     class AOTModule(nn.Module):
-        def __init__(self):
+        def __init__(self) -> None:
             super().__init__()
             self.orig_module = mod
 

torch/_functorch/autograd_function.py

@@ -30,7 +30,7 @@ from torch.autograd.forward_ad import _set_fwd_grad_enabled
 # We do this by using creating a custom HigherOrderOperator that only functorch
 # dispatches specially.
 class CustomFunctionHigherOrderOperator(HigherOrderOperator):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("custom_function_call")
 
     def __call__(self, autograd_function, *args, **kwargs):
@@ -713,7 +713,7 @@ def autograd_function_forward_rewritten(original_forward, original_setup_context
 
 
 class AutogradFunctionApply(HigherOrderOperator):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("autograd_function_apply")
 
     def __call__(self, fwd, bwd, *fwd_args, **fwd_kwargs):

torch/_guards.py

@@ -427,7 +427,7 @@ class ModuleContextCheckpointState:
 
 
 class ModuleContext(Checkpointable[ModuleContextCheckpointState]):
-    def __init__(self):
+    def __init__(self) -> None:
         self.nn_modules: Dict[str, Any] = {}
 
     def copy_graphstate(self):
@@ -476,7 +476,7 @@ class GlobalContext(Checkpointable[GlobalContextCheckpointState]):
         "autocast_cache_enabled",
     }
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.global_state: Dict[str, Tuple[Callable, ...]] = {}
 
     def copy_graphstate(self):
@@ -544,7 +544,7 @@ class GuardsSet:
 
 
 class GuardsContext(Checkpointable[GuardsCheckpointState]):
-    def __init__(self):
+    def __init__(self) -> None:
         self.dynamo_guards: GuardsSet = GuardsSet()
         self.aotautograd_guards: List[GuardEnvExpr] = []
 

torch/_higher_order_ops/auto_functionalize.py

@@ -54,7 +54,7 @@ class AutoFunctionalized(HigherOrderOperator):
     underscore is to prevent collisions with kwarg names in **kwargs.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("auto_functionalized")
 
     def __call__(

torch/_higher_order_ops/effects.py

@@ -55,7 +55,7 @@ class WithEffects(HigherOrderOperator):
     per "effect type", which are enumerated in the _EffectType enum.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("with_effects")
 
     def __call__(

torch/_higher_order_ops/flex_attention.py

@@ -38,7 +38,7 @@ class TransformGetItemToIndex(TorchFunctionMode):
 
 
 class FlexAttentionHOP(HigherOrderOperator):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("flex_attention")
 
     def __call__(
@@ -74,7 +74,7 @@ flex_attention.__module__ = "torch.ops.higher_order"
 
 
 class FlexAttentionBackwardHOP(HigherOrderOperator):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("flex_attention_backward")
 
     def __call__(

torch/_higher_order_ops/out_dtype.py

@@ -45,7 +45,7 @@ class OutDtypeOperator(HigherOrderOperator):
         3. Cast the output to `out_dtype`
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("out_dtype")
         # TODO(ydwu4): Subclassing HigherOrderOperator causes __module__ to
         # become different (torch._higher_order_ops.out_dtype) which will result

torch/_higher_order_ops/triton_kernel_wrap.py

@@ -519,7 +519,7 @@ def identify_mutated_tensors(kernel, kwargs):
 
 # Used for wrapping a Triton Kernel
 class TritonKernelWrapperMutation(HigherOrderOperator):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("triton_kernel_wrapper_mutation")
 
 
@@ -528,7 +528,7 @@ triton_kernel_wrapper_mutation = TritonKernelWrapperMutation()
 
 # Used for wrapping a Triton Kernel in a functional manner
 class TritonKernelWrapperFunctional(HigherOrderOperator):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("triton_kernel_wrapper_functional")
 
 

torch/_higher_order_ops/while_loop.py

@@ -18,7 +18,7 @@ from torch.fx.experimental.proxy_tensor import ProxyTorchDispatchMode, track_ten
 
 
 class WhileLoopOp(HigherOrderOperator):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("while_loop")
 
     def __call__(

torch/_higher_order_ops/wrap.py

@@ -15,7 +15,7 @@ uid = itertools.count(1)
 
 # Used for testing the HigherOrderOperator mechanism
 class Wrap(HigherOrderOperator):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("wrap")
 
     def __call__(self, func, *args, **kwargs):
@@ -36,7 +36,7 @@ wrap = Wrap()
 
 
 class WrapWithSetGradEnabled(HigherOrderOperator):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("wrap_with_set_grad_enabled")
 
     def __call__(self, enable_grad, wrapped_func, *args, **kwargs):
@@ -74,7 +74,7 @@ class WrapActivationCheckpoint(HigherOrderOperator):
     partitioners. See TagActivationCheckpoint for more information.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("wrap_activation_checkpoint")
 
     def __call__(self, function, *args, **kwargs):
@@ -113,7 +113,7 @@ class TagActivationCheckpoint(HigherOrderOperator):
     the forward and recomputed forward in backward.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__("tag_activation_checkpoint")
 
     @staticmethod

torch/_inductor/codegen/common.py

@@ -1560,7 +1560,7 @@ class CSE:
 
 
 class CodeGen:
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.exit_stack = contextlib.ExitStack()
 

torch/_inductor/codegen/cpp_wrapper_cuda.py

@@ -29,7 +29,7 @@ class CppWrapperCuda(CppWrapperCpu):
     Generates cpp wrapper for running on GPU and calls CUDA kernels
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.device = "cuda"
         super().__init__()
         self.grid_id = count()

torch/_inductor/codegen/triton.py

@@ -1113,7 +1113,7 @@ class HelperFunctions:
     _templates_seen: Dict[str, str]  # Template code to function name
     finalized_helpers: List[str]
 
-    def __init__(self):
+    def __init__(self) -> None:
         self._templates_seen = {}
         self.finalized_helpers = []
 

torch/_inductor/dependencies.py

@@ -589,7 +589,7 @@ def canonicalization_prefix():
 class FreeUnbackedSymbolsOpsHandler:
     symbols: OrderedSet[sympy.Symbol]
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.symbols = OrderedSet()
 
     def __getattr__(self, name: str) -> Callable[..., Any]:

torch/_inductor/exc.py

@@ -65,7 +65,7 @@ class SubgraphLoweringException(RuntimeError):
 
 
 class InvalidCxxCompiler(RuntimeError):
-    def __init__(self):
+    def __init__(self) -> None:
         from . import config
 
         super().__init__(

torch/_inductor/fx_passes/misc_patterns.py

@@ -79,7 +79,7 @@ class NumpyCompatNormalization:
     inverse_mapping: Dict[str, str]
     cache: Dict["torch.fx.graph.Target", Set[str]]
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.cache = {}  # callable -> tuple of replaceable args e.g. ["axis"]
         self.inverse_mapping = {}
         for actual_kwarg, numpy_kwargs in self.numpy_compat.items():

torch/_inductor/fx_passes/mkldnn_fusion.py

@@ -1207,7 +1207,7 @@ if torch._C._has_mkldnn:
         Combine packed weight nodes with the same inputs to reduce memory usage.
         for example:
         class Model(nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
                 self.linear = nn.Linear(32, 32, bias=True)
 

torch/_inductor/metrics.py

@@ -99,7 +99,7 @@ class CachedMetricsHelper:
     apply on a cache hit.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.cached_metrics = {}
         for metric in get_metric_fields():
             self.cached_metrics[metric] = globals()[metric]

torch/_inductor/utils.py

@@ -940,7 +940,7 @@ class IndentedBuffer:
 
 
 class FakeIndentedBuffer(IndentedBuffer):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
 
     def __getattribute__(self, name):
@@ -1219,7 +1219,7 @@ class DebugDirManager:
     counter = itertools.count(0)
     prev_debug_name: str
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.id = next(DebugDirManager.counter)
 
     def __enter__(self):
@@ -1268,7 +1268,7 @@ def get_code(fn, *args, **kwargs):
         class DummyModule:
             """This is empty to replace the generated triton module"""
 
-            def __init__(self):
+            def __init__(self) -> None:
                 pass
 
             def call(self, *args, **kwargs):

torch/_lazy/closure.py

@@ -7,7 +7,7 @@ from torch._lazy.device_context import get_device_context
 
 
 class ClosureHandler:
-    def __init__(self):
+    def __init__(self) -> None:
         pass
 
     def run(self, closure):

torch/_library/fake_class_registry.py

@@ -42,7 +42,7 @@ class HasStaticMethodFromReal(Protocol):
 
 
 class FakeClassRegistry:
-    def __init__(self):
+    def __init__(self) -> None:
         self._registered_class: Dict[str, Any] = {}
 
     def has_impl(self, full_qualname: str) -> bool:

torch/_python_dispatcher.py

@@ -70,7 +70,7 @@ class PythonDispatcher:
     ]
     supported_keys = runtime_keys + alias_keys
 
-    def __init__(self):
+    def __init__(self) -> None:
         C._dispatch_check_invariants(self.name)  # type: ignore[attr-defined]
         self.ref = C._dispatch_library("FRAGMENT", self.namespace, "")
         self.ref.def_("foo(Tensor x) -> Tensor")

torch/_subclasses/schema_check_mode.py

@@ -60,7 +60,7 @@ def clone_inputs(args):
 
 
 class SchemaCheckMode(TorchDispatchMode):
-    def __init__(self):
+    def __init__(self) -> None:
         # Information recorded for testing purposes. For example:
         #  - incorrect schemas
         #  - overly conservative schemas

torch/ao/nn/quantized/modules/functional_modules.py

@@ -36,7 +36,7 @@ class FloatFunctional(torch.nn.Module):
         - mul_scalar
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.activation_post_process = torch.nn.Identity()
 
@@ -190,7 +190,7 @@ class QFunctional(torch.nn.Module):
         - mul_scalar
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.scale = 1.0
         self.zero_point = 0

torch/ao/ns/fx/qconfig_multi_mapping.py

@@ -72,7 +72,7 @@ class QConfigMultiMapping:
 
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         # initialize this with 1 QConfigMapping to avoid corner cases
         self.qconfig_mappings_list: List[QConfigMapping] = [QConfigMapping()]
 

torch/ao/pruning/_experimental/pruner/README.md

@@ -99,7 +99,7 @@ from torch.ao.pruning._experimental.pruner import SaliencyPruner
 
 # Define model
 class Model(nn.Module):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.seq = nn.Sequential(
             nn.Linear(700, 500, bias=True),

torch/ao/quantization/fake_quantize.py

@@ -85,7 +85,7 @@ class FakeQuantizeBase(ABC, Module):
     fake_quant_enabled: torch.Tensor
     observer_enabled: torch.Tensor
 
-    def __init__(self):
+    def __init__(self) -> None:
         """Set fake_quant_enabled and observer_enabled."""
         super().__init__()
         # fake_quant_enabled and observer_enabled are buffers to support their

torch/ao/quantization/fx/README.md

@@ -70,7 +70,7 @@ In the following, I’ll first have a detailed description for each step, and th
 
 ```
 class LinearReLUModule(torch.nn.Module):
-   def __init__(self):
+   def __init__(self) -> None:
        super().__init__()
        self.linear = torch.nn.Linear(5, 10).float()
        self.relu = torch.nn.ReLU()

torch/ao/quantization/fx/_model_report/detector.py

@@ -137,7 +137,7 @@ class DetectorBase(ABC):
         - Should return a str-based report and dict info in Tuple[str,Dict] format
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.detector_config_info = None
 

torch/ao/quantization/fx/custom_config.py

@@ -63,7 +63,7 @@ class PrepareCustomConfig:
             .set_preserved_attributes(["attr1", "attr2"])
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.standalone_module_names: Dict[str, StandaloneModuleConfigEntry] = {}
         self.standalone_module_classes: Dict[Type, StandaloneModuleConfigEntry] = {}
         self.float_to_observed_mapping: Dict[QuantType, Dict[Type, Type]] = {}
@@ -382,7 +382,7 @@ class ConvertCustomConfig:
             .set_preserved_attributes(["attr1", "attr2"])
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.observed_to_quantized_mapping: Dict[QuantType, Dict[Type, Type]] = {}
         self.preserved_attributes: List[str] = []
 
@@ -477,7 +477,7 @@ class FuseCustomConfig:
         fuse_custom_config = FuseCustomConfig().set_preserved_attributes(["attr1", "attr2"])
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.preserved_attributes: List[str] = []
 
     def __repr__(self):

torch/ao/quantization/observer.py

@@ -1568,7 +1568,7 @@ class ReuseInputObserver(ObserverBase):
     Note: this is only enabled in FX Graph Mode Quantization
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__(torch.quint8, is_dynamic=False)
 
     def forward(self, x):

torch/ao/quantization/qconfig_mapping.py

@@ -229,7 +229,7 @@ class QConfigMapping:
 
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         # In increasing match priority:
         self.global_qconfig: QConfigAny = None
         self.object_type_qconfigs: OrderedDict[

torch/ao/quantization/quantize_fx.py

@@ -289,7 +289,7 @@ def prepare_fx(
         from torch.ao.quantization.quantize_fx import prepare_fx
 
         class Submodule(torch.nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
                 self.linear = torch.nn.Linear(5, 5)
             def forward(self, x):
@@ -297,7 +297,7 @@ def prepare_fx(
                 return x
 
         class M(torch.nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
                 self.linear = torch.nn.Linear(5, 5)
                 self.sub = Submodule()
@@ -427,7 +427,7 @@ def prepare_qat_fx(
         from torch.ao.quantization.quantize_fx import prepare_qat_fx
 
         class Submodule(torch.nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
                 self.linear = torch.nn.Linear(5, 5)
             def forward(self, x):
@@ -435,7 +435,7 @@ def prepare_qat_fx(
                 return x
 
         class M(torch.nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
                 self.linear = torch.nn.Linear(5, 5)
                 self.sub = Submodule()

torch/ao/quantization/quantize_pt2e.py

@@ -56,7 +56,7 @@ def prepare_pt2e(
         )
 
         class M(torch.nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
                 self.linear = torch.nn.Linear(5, 10)
 
@@ -129,7 +129,7 @@ def prepare_qat_pt2e(
         )
 
         class M(torch.nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
                 self.linear = torch.nn.Linear(5, 10)
 

torch/ao/quantization/quantizer/embedding_quantizer.py

@@ -42,7 +42,7 @@ def get_embedding_operators_config() -> OperatorConfig:
 
 
 class EmbeddingQuantizer(Quantizer):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
 
     @classmethod

torch/ao/quantization/quantizer/x86_inductor_quantizer.py

@@ -436,7 +436,7 @@ class X86InductorQuantizer(Quantizer):
     supported_config_and_operators = _get_supported_config_and_operators()
     module_function_to_aten_operator_type = _map_module_function_to_aten_operator_type()
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.global_config: Optional[QuantizationConfig] = None
         self.operator_type_qconfig: Dict[

torch/ao/quantization/quantizer/xnnpack_quantizer.py

@@ -268,7 +268,7 @@ class XNNPACKQuantizer(Quantizer):
         "linear",
     ]
 
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.global_config: Optional[QuantizationConfig] = None
         self.operator_type_config: Dict[

torch/ao/quantization/utils.py

@@ -513,7 +513,7 @@ def _get_path_of_module(
     Example::
 
     >> class M(torch.nn.Module):
-           def __init__(self):
+           def __init__(self) -> None:
                self.linear = torch.nn.Linear(5, 5)
            def forward(self, x):
                return self.linear(x)

torch/autograd/profiler_util.py

@@ -645,7 +645,7 @@ class FunctionEvent(FormattedTimesMixin):
 class FunctionEventAvg(FormattedTimesMixin):
     """Used to average stats over multiple FunctionEvent objects."""
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.key: Optional[str] = None
         self.count: int = 0
         self.node_id: int = 0

torch/backends/xeon/run_cpu.py

@@ -266,7 +266,7 @@ class _Launcher:
 or /.local/lib/ or /usr/local/lib/ or /usr/local/lib64/ or /usr/lib or /usr/lib64 or \
 {expanduser('~')}/.local/lib/ so the LD_PRELOAD environment variable will not be set."
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.cpuinfo = _CPUinfo()
 
     def add_lib_preload(self, lib_type):

torch/csrc/jit/backends/backend_debug_handler.h

@@ -77,17 +77,17 @@ namespace jit {
  *
  *  So why does debug handle map to DebugInfoTuple = {source range and inlined
  *  cs}? {debug_handle, source_range_tag, serialized_callstack} Take this
- *  example: class L(nn.Module): def __init__(self):
+ *  example: class L(nn.Module): def __init__(self) -> None:
  *      ...
  *    def forward(self, x):
  *      return x * 5
  *  class M(nn.Module):
- *    def __init__(self):
+ *    def __init__(self) -> None:
  *      ...
  *    def forward(self, x):
  *      return x - 2
  *  class N(nn.Module):
- *    def __init__(self):
+ *    def __init__(self) -> None:
  *      self.m = M()
  *    def forward(self, x):
  *      return self.m(x) + 3

torch/csrc/jit/docs/serialization.md

@@ -328,7 +328,7 @@ For example:
 
 ```
 class M(torch.nn.Module):
-    def __init__(self):
+    def __init__(self) -> None:
         self.a = torch.rand(2, 3)
         self.b = torch.nn.Linear(10, 10)
 

torch/csrc/jit/operator_upgraders/README.md

@@ -37,7 +37,7 @@ When making changes to the operators, the first thing to identify is if it's BC/
     1. Add a test module in `test/jit/fixtures_srcs/fixtures_src.py`. In `test/jit/fixtures_srcs/generate_models.py`,
   ```
   class TestVersionedLinspaceV7(torch.nn.Module):
-      def __init__(self):
+      def __init__(self) -> None:
           super().__init__()
 
       def forward(self, a: Union[int, float, complex], b: Union[int, float, complex]):
@@ -163,7 +163,7 @@ When making changes to the operators, the first thing to identify is if it's BC/
 
             # Step 2. Write down how current module should look like
             class MyModuleFloat(torch.nn.Module):
-                def __init__(self):
+                def __init__(self) -> None:
                     super().__init__()
 
                 def forward(self, a, b: float):

torch/csrc/jit/passes/onnx/function_extraction.h

@@ -25,7 +25,7 @@ namespace onnx {
 //
 // clang-format off
 // class M(torch.nn.Module):
-//     def __init__(self):
+//     def __init__(self) -> None:
 //         super().__init__()
 //         self.lns = torch.nn.ModuleList([torch.nn.LayerNorm(3, eps = i) for i in range(2)])
 //         self.celu1 = torch.nn.CELU(1.0)

torch/csrc/lazy/test_mnist.py

@@ -17,7 +17,7 @@ torch._lazy.ts_backend.init()
 
 
 class Net(nn.Module):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.conv1 = nn.Conv2d(1, 32, 3, 1)
         self.conv2 = nn.Conv2d(32, 64, 3, 1)

torch/csrc/lazy/tutorial.md

@@ -135,7 +135,7 @@ Here's our model definition:
 
 ```python
 class Net(nn.Module):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.conv1 = nn.Conv2d(1, 32, 3, 1)
         self.conv2 = nn.Conv2d(32, 64, 3, 1)

torch/cuda/_sanitizer.py

@@ -163,7 +163,7 @@ class TensorInfo:
 
 
 class _TensorsAccessed:
-    def __init__(self):
+    def __init__(self) -> None:
         self.accesses: Dict[DataPtr, TensorInfo] = {}
 
     def ensure_tensor_exists(self, data_ptr: DataPtr) -> None:
@@ -218,7 +218,7 @@ class _TensorsAccessed:
 
 
 class StreamSynchronizations:
-    def __init__(self):
+    def __init__(self) -> None:
         self.current_sync_states: Dict[StreamId, Dict[StreamId, SeqNum]] = {}
         self.recorded_sync_states: Dict[EventId, Dict[StreamId, SeqNum]] = {}
         self.host_sync_state: Dict[StreamId, SeqNum] = {}
@@ -338,7 +338,7 @@ class EventHandler:
     data race.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.tensors_accessed = _TensorsAccessed()
         self.syncs = StreamSynchronizations()
         self.seq_num: SeqNum = 0
@@ -478,7 +478,7 @@ def zip_arguments(
 
 
 class ArgumentHandler:
-    def __init__(self):
+    def __init__(self) -> None:
         self.dataptrs_read: Set[DataPtr] = set()
         self.dataptrs_written: Set[DataPtr] = set()
         self.tensor_aliases: Dict[DataPtr, List[str]] = {}
@@ -527,7 +527,7 @@ class ArgumentHandler:
 
 
 class CUDASanitizerDispatchMode(TorchDispatchMode):
-    def __init__(self):
+    def __init__(self) -> None:
         self.event_handler = EventHandler()
         torch._C._activate_gpu_trace()
         gpu_trace.register_callback_for_event_creation(
@@ -596,7 +596,7 @@ class CUDASanitizer:
     This approach was deemed more elegant than using the atexit module.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.dispatch = CUDASanitizerDispatchMode()
         self.enabled = False
 

torch/distributed/_composable/checkpoint_activation.py

@@ -49,7 +49,7 @@ def checkpoint(module: nn.Module, **kwargs) -> nn.Module:
         >>> import torch.nn as nn
         >>>
         >>> class MyModel(nn.Module):
-        >>>     def __init__(self):
+        >>>     def __init__(self) -> None:
         >>>         super().__init__()
         >>>         self.l1 = nn.Linear(10, 10)
         >>>         self.l2 = nn.Linear(10, 10)

torch/distributed/_composable/contract.py

@@ -47,7 +47,7 @@ def contract(state_cls: Type[_State] = _State):
         >>> import torch.nn as nn
         >>>
         >>> class MyModel(nn.Module):
-        >>>     def __init__(self):
+        >>>     def __init__(self) -> None:
         >>>         super().__init__()
         >>>         self.l1 = nn.Linear(10, 10)
         >>>         self.l2 = nn.Linear(10, 10)

torch/distributed/_composable/fsdp/_fsdp_state.py

@@ -43,7 +43,7 @@ logger = logging.getLogger("torch.distributed._composable.fsdp")
 class FSDPStateContext:
     """This has state shared across FSDP states."""
 
-    def __init__(self):
+    def __init__(self) -> None:
         # All FSDP states in the root state's module tree
         self.all_states: List[FSDPState] = []
         # Iteration's forward root runs the once-per-forward logic; this root
@@ -71,7 +71,7 @@ def disable_if_config_true(func):
 
 
 class FSDPState(_State):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self._fsdp_param_group: Optional[FSDPParamGroup] = None
         self._is_root: Optional[bool] = None  # root set during lazy init

torch/distributed/_shard/sharding_plan/api.py

@@ -38,7 +38,7 @@ class ShardingPlan:
 
         >>> # xdoctest: +REQUIRES(module:torch._C._distributed_c10d)
         >>> class MyModule(nn.Module):
-        >>>     def __init__(self):
+        >>>     def __init__(self) -> None:
         >>>        super().__init__()
         >>>        self.fc1 = nn.Linear()
         >>>        self.gelu = nn.GELU()

torch/distributed/_tensor/README.md

@@ -117,7 +117,7 @@ import torch.nn as nn
 from torch.distributed._tensor import Shard, distribute_tensor, distribute_module, init_device_mesh
 
 class MyModule(nn.Module):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.fc1 = nn.Linear(8, 8)
         self.fc2 = nn.Linear(8, 8)

torch/distributed/_tensor/examples/checkpoint_example.py

@@ -25,7 +25,7 @@ from torch.distributed.tensor.parallel import ColwiseParallel, parallelize_modul
 
 
 class SimpleMLP(torch.nn.Module):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.net1 = torch.nn.Linear(5, 128)
         self.relu = torch.nn.ReLU()

torch/distributed/algorithms/join.py

@@ -55,7 +55,7 @@ class Joinable(ABC):
     """
 
     @abstractmethod
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self._join_config = _JoinConfig.construct_disabled_join_config()
 

torch/distributed/checkpoint/examples/async_checkpointing_example.py

@@ -31,7 +31,7 @@ class InjectedException(Exception):
 
 
 class Model(torch.nn.Module):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.net1 = nn.Linear(8, 32)
         self.net2 = nn.Linear(32, 128)

torch/distributed/checkpoint/examples/stateful_example.py

@@ -22,7 +22,7 @@ CHECKPOINT_DIR = f"~/{os.environ['LOGNAME']}/checkpoint"
 
 
 class Model(torch.nn.Module):
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         torch.manual_seed(0)
         self.net1 = nn.Sequential(nn.Linear(8, 16), nn.ReLU())

torch/distributed/distributed_c10d.py

@@ -434,7 +434,7 @@ class _reduce_op:
     :class:`~torch.distributed.ReduceOp` is recommended to use instead.
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         # __members__ is a dict storing key-value pairs for enum classes
         for k, v in ReduceOp.RedOpType.__members__.items():
             setattr(self, k, v)
@@ -568,7 +568,7 @@ class _World:
        of c10d and is subject to change..
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         self._default_pg = None
         self._pg_coalesce_state: Dict[ProcessGroup, List[_CollOp]] = {}
         self._pg_default_device: Dict[ProcessGroup, torch.device] = {}
@@ -2194,7 +2194,7 @@ class _IllegalWork(Work):
 
 
 class _CoalescingManager:
-    def __init__(self):
+    def __init__(self) -> None:
         self.works: List[Work] = []
 
     def append(self, work: Work):

torch/distributed/fsdp/_common_utils.py

@@ -106,7 +106,7 @@ class _FSDPDeviceHandle:
 
 
 class _UninitializedDeviceHandle(_FSDPDeviceHandle):
-    def __init__(self):
+    def __init__(self) -> None:
         pass
 
     def __getattribute__(self, __name: str) -> Any:

torch/distributed/nn/api/remote_module.py

@@ -156,7 +156,7 @@ class _RemoteModule(nn.Module):
         created outside of remote modules, rather than as submodules of any remote module (by calling ``add_module``).
         Hybrid Example:
                 >>> class HybridModel(nn.Module):
-                >>>     def __init__(self):
+                >>>     def __init__(self) -> None:
                 >>>         nn.Module.__init__(self)
                 >>>         self.remote_embedding = RemoteModule(...)
                 >>>         self.local_linear = nn.Linear(...)

torch/export/graph_signature.py

@@ -248,7 +248,7 @@ class ExportGraphSignature:
     e.g. If following module is exported::
 
         class CustomModule(nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super(CustomModule, self).__init__()
 
                 # Define a parameter

torch/fx/README.md

@@ -45,7 +45,7 @@ FX’s front-end makes use of the dynamic nature of Python to intercept call-sit
 import torch
 
 class MyModule(torch.nn.Module):
-  def __init__(self):
+  def __init__(self) -> None:
     super().__init__()
     self.param = torch.nn.Parameter(
         torch.rand(3, 4))

torch/fx/__init__.py

@@ -9,7 +9,7 @@ demonstration of these components in action:
     import torch
     # Simple module for demonstration
     class MyModule(torch.nn.Module):
-        def __init__(self):
+        def __init__(self) -> None:
             super().__init__()
             self.param = torch.nn.Parameter(torch.rand(3, 4))
             self.linear = torch.nn.Linear(4, 5)

torch/fx/_symbolic_trace.py

@@ -1012,7 +1012,7 @@ class _PatchedFnSetAttr(_PatchedFn):
 
 
 class _Patcher:
-    def __init__(self):
+    def __init__(self) -> None:
         super().__init__()
         self.patches_made: List[_PatchedFn] = []
         self.visited: Set[int] = set()

torch/fx/experimental/migrate_gradual_types/constraint.py

@@ -63,7 +63,7 @@ class T(Constraint):
     """
     True
     """
-    def __init__(self):
+    def __init__(self) -> None:
         pass
 
     def __eq__(self, other):
@@ -76,7 +76,7 @@ class F(Constraint):
     """
     False
     """
-    def __init__(self):
+    def __init__(self) -> None:
         pass
 
     def __eq__(self, other):

torch/fx/passes/graph_drawer.py

@@ -117,7 +117,7 @@ if HAS_PYDOT:
                 >>> # xdoctest: +REQUIRES(module:ubelt)
                 >>> # define module
                 >>> class MyModule(torch.nn.Module):
-                >>>     def __init__(self):
+                >>>     def __init__(self) -> None:
                 >>>         super().__init__()
                 >>>         self.linear = torch.nn.Linear(4, 5)
                 >>>     def forward(self, x):

torch/fx/passes/split_module.py

@@ -83,7 +83,7 @@ def split_module(
             from torch.fx.passes.split_module import split_module
 
             class MyModule(torch.nn.Module):
-                def __init__(self):
+                def __init__(self) -> None:
                     super().__init__()
                     self.param = torch.nn.Parameter(torch.rand(3, 4))
                     self.linear = torch.nn.Linear(4, 5)

torch/fx/passes/split_utils.py

@@ -83,7 +83,7 @@ def split_by_tags(
     Given the following module def:
 
     class SimpleModule(torch.nn.Module):
-        def __init__(self):
+        def __init__(self) -> None:
             super().__init__()
             self.linear1 = torch.nn.Linear(...)
             self.linear2 = torch.nn.Linear(...)

torch/fx/proxy.py

@@ -38,7 +38,7 @@ class Scope:
                 return x.transpose(1, 2)
 
         class M(torch.nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 self.sub = Sub()
 
             def forward(self, x):

torch/fx/subgraph_rewriter.py

@@ -118,7 +118,7 @@ def replace_pattern(
         from torch.fx import symbolic_trace, subgraph_rewriter
 
         class M(torch.nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
 
             def forward(self, x, w1, w2):

torch/fx/tensor_type.py

@@ -38,7 +38,7 @@ class _DynType:
     """
     _DynType defines a type which stands for the absence of type information.
     """
-    def __init__(self):
+    def __init__(self) -> None:
         self.__name__ = '_DynType'
 
     def __eq__(self, other):

torch/jit/__init__.py

@@ -219,7 +219,7 @@ def isinstance(obj, target_type):
         from typing import Any, Dict, List
 
         class MyModule(torch.nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
 
             def forward(self, input: Any): # note the Any type
@@ -255,7 +255,7 @@ class strict_fusion:
 
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         if not torch._jit_internal.is_scripting():
             warnings.warn("Only works in script mode")
         pass

torch/jit/_async.py

@@ -73,7 +73,7 @@ def fork(func, *args, **kwargs):
             def forward(self, a: Tensor, b : int):
                 return a + b
         class Mod(torch.nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super(self).__init__()
                 self.mod = AddMod()
             def forward(self, input):

torch/jit/_check.py

@@ -39,7 +39,7 @@ class AttributeTypeIsSupportedChecker(ast.NodeVisitor):
                 def fn(self):
                     return []
 
-                def __init__(self):
+                def __init__(self) -> None:
                     super().__init__()
                     self.x: List[int] = []
 

torch/jit/_freeze.py

@@ -65,7 +65,7 @@ def freeze(
     .. testcode::
         import torch
         class MyModule2(torch.nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
                 self.modified_tensor = torch.tensor(10.)
                 self.version = 1

torch/jit/_monkeytype_config.py

@@ -89,7 +89,7 @@ if _IS_MONKEYTYPE_INSTALLED:
             self.traces.append(trace)
 
     class JitTypeTraceStore(CallTraceStore):
-        def __init__(self):
+        def __init__(self) -> None:
             super().__init__()
             # A dictionary keeping all collected CallTrace
             # key is fully qualified name of called function
@@ -159,15 +159,15 @@ else:
     # When MonkeyType is not installed, we provide dummy class definitions
     # for the below classes.
     class JitTypeTraceStoreLogger:  # type:  ignore[no-redef]
-        def __init__(self):
+        def __init__(self) -> None:
             pass
 
     class JitTypeTraceStore:  # type:  ignore[no-redef]
-        def __init__(self):
+        def __init__(self) -> None:
             self.trace_records = None
 
     class JitTypeTraceConfig:  # type:  ignore[no-redef]
-        def __init__(self):
+        def __init__(self) -> None:
             pass
 
     monkeytype_trace = None  # type: ignore[assignment]  # noqa: F811

torch/jit/_recursive.py

@@ -426,7 +426,7 @@ class ConcreteTypeStore:
     type_store: Dict[Type[Module], List[torch._C.ConcreteModuleType]]
     methods_compiled: Set[torch._C.ConcreteModuleType]
 
-    def __init__(self):
+    def __init__(self) -> None:
         # Python module type => List[ConcreteModuleType)]
         self.type_store = {}
         # ConcreteTypes that have had their methods already compiled

torch/jit/_script.py

@@ -107,7 +107,7 @@ Attribute.__doc__ = """
         from typing import Dict
 
         class AttributeModule(torch.jit.ScriptModule):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
                 self.foo = torch.jit.Attribute(0.1, float)
 
@@ -138,7 +138,7 @@ Attribute.__doc__ = """
         class AttributeModule(torch.nn.Module):
             names: Dict[str, int]
 
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
                 self.names = {}
 
@@ -522,7 +522,7 @@ if _enabled:
             "original_name",
         ]
 
-        def __init__(self):
+        def __init__(self) -> None:
             super().__init__()
 
         forward: Callable[..., Any] = _CachedForward()  # type: ignore[assignment]
@@ -1351,7 +1351,7 @@ def script(
             import torch.nn.functional as F
 
             class MyModule(nn.Module):
-                def __init__(self):
+                def __init__(self) -> None:
                     super().__init__()
                     # torch.jit.trace produces a ScriptModule's conv1 and conv2
                     self.conv1 = torch.jit.trace(nn.Conv2d(1, 20, 5), torch.rand(1, 1, 16, 16))
@@ -1374,7 +1374,7 @@ def script(
             import torch.nn as nn
 
             class MyModule(nn.Module):
-                def __init__(self):
+                def __init__(self) -> None:
                     super().__init__()
 
                 @torch.jit.export
@@ -1547,7 +1547,7 @@ def interface(obj):
                 return x.relu()
 
         class Impl2(torch.nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
                 self.val = torch.rand(())
 
@@ -1671,7 +1671,7 @@ class _ScriptProfileTable:
 
 
 class _ScriptProfile:
-    def __init__(self):
+    def __init__(self) -> None:
         self.profile = classes.profiling._ScriptProfile()
 
     def enable(self):

torch/jit/_state.py

@@ -19,7 +19,7 @@ class EnabledProxy:
     This is just a wrapper for a bool, so that we get reference semantics
     """
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.enabled = self.parse_env(
             "PYTORCH_JIT", True, "> Using PyTorch JIT", "> PyTorch JIT DISABLED"
         )

torch/jit/_trace.py

@@ -966,7 +966,7 @@ def trace(
         import torch.nn as nn
 
         class Net(nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
                 self.conv = nn.Conv2d(1, 1, 3)
 
@@ -1182,7 +1182,7 @@ def trace_module(
         import torch.nn as nn
 
         class Net(nn.Module):
-            def __init__(self):
+            def __init__(self) -> None:
                 super().__init__()
                 self.conv = nn.Conv2d(1, 1, 3)
 

torch/multiprocessing/reductions.py

@@ -61,7 +61,7 @@ class StorageWeakRef:
 class SharedCache(dict):
     """Dictionary from multiprocessing handles to StorageWeakRef."""
 
-    def __init__(self):
+    def __init__(self) -> None:
         # free_dead_references() is called if the len exceeds the current
         # limit. The limit scales with the number of remaining live objects.
         self.limit = 128