Remove default typing argument in PACKAGE file] [batch:79/416] [shard…

…:4/N]

Reviewed By: MaggieMoss

Differential Revision: D64867383

fbshipit-source-id: efc412f93077c0c1dbb0911a10dc051d0309fe06

flowtorch/bijectors/autoregressive.py

@@ -56,11 +56,15 @@ def inverse(
         # NOTE: Inversion is an expensive operation that scales in the
         # dimension of the input
         permutation = (
+            # pyre-fixme[16]: Item `None` of `Union[None, Parameters, ModuleList]`
+            #  has no attribute `permutation`.
             self._params_fn.permutation
         )  # TODO: type-safe named buffer (e.g. "permutation") access
         # TODO: Make permutation, inverse work for other event shapes
         log_detJ: torch.Tensor | None = None
         for idx in cast(torch.LongTensor, permutation):
+            # pyre-fixme[29]: `Union[None, flowtorch.parameters.base.Parameters,
+            #  torch.nn.modules.container.ModuleList]` is not a function.
             _params = self._params_fn(x_new.clone(), context=context)
             x_temp, log_detJ = self._inverse(y, params=_params)
             x_new[..., idx] = x_temp[..., idx]

flowtorch/bijectors/bijective_tensor.py

@@ -22,11 +22,17 @@ def register(
         log_detJ: Tensor | None,
         mode: str,
     ) -> "BijectiveTensor":
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_input`.
         self._input = input
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_output`.
         self._output = output
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_context`.
         self._context = context
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_bijector`.
         self._bijector = bijector
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_log_detJ`.
         self._log_detJ = log_detJ
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_mode`.
         self._mode = mode
 
         if not (self.from_forward() or self.from_inverse()):
@@ -57,12 +63,14 @@ def check_bijector(self, bijector: "Bijector") -> bool:
         return is_bijector
 
     def bijectors(self) -> Iterator["Bijector"]:
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_bijector`.
         yield self._bijector
         for parent in self.parents():
             if isinstance(parent, BijectiveTensor):
                 yield parent._bijector
 
     def get_parent_from_bijector(self, bijector: "Bijector") -> Tensor:
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_bijector`.
         if self._bijector is bijector:
             return self.parent
         for parent in self.parents():
@@ -73,15 +81,20 @@ def get_parent_from_bijector(self, bijector: "Bijector") -> Tensor:
         raise RuntimeError("bijector not found in flow")
 
     def check_context(self, context: Tensor | None) -> bool:
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_context`.
         return self._context is context
 
     def from_forward(self) -> bool:
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_mode`.
         return self._mode == "forward"
 
     def from_inverse(self) -> bool:
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_mode`.
         return self._mode == "inverse"
 
     def detach_from_flow(self) -> Tensor:
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_output`.
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_input`.
         detached_tensor = self._output if self.from_forward() else self._input
         if isinstance(detached_tensor, BijectiveTensor):
             raise RuntimeError("the detached tensor is an instance of BijectiveTensor.")
@@ -90,8 +103,10 @@ def detach_from_flow(self) -> Tensor:
     def has_ancestor(self, tensor: Tensor) -> bool:
         if tensor is self:
             return False  # self is no parent of self
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_input`.
         elif self.from_forward() and self._input is tensor:
             return True
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_output`.
         elif self.from_inverse() and self._output is tensor:
             return True
         elif self.from_forward() and isinstance(self._input, BijectiveTensor):
@@ -103,19 +118,23 @@ def has_ancestor(self, tensor: Tensor) -> bool:
 
     @property
     def log_detJ(self) -> Tensor | None:
+        # pyre-fixme[16]: `BijectiveTensor` has no attribute `_log_detJ`.
         return self._log_detJ
 
     @property
     def parent(self) -> Tensor:
         if self.from_forward():
+            # pyre-fixme[16]: `BijectiveTensor` has no attribute `_input`.
             return self._input
         else:
+            # pyre-fixme[16]: `BijectiveTensor` has no attribute `_output`.
             return self._output
 
     def parents(self) -> Iterator[Tensor]:
         child: Tensor | BijectiveTensor = self
         while True:
             assert isinstance(child, BijectiveTensor)
+            # pyre-fixme[16]: `Tensor` has no attribute `parent`.
             child = parent = child.parent
             yield parent
             if not isinstance(child, BijectiveTensor):

flowtorch/bijectors/compose.py

@@ -58,6 +58,7 @@ def forward(
             y = bijector.forward(x_temp, context)  # type: ignore
             if is_record_flow_graph_enabled() and requires_log_detJ():
                 if isinstance(y, BijectiveTensor) and y.from_forward():
+                    # pyre-fixme[16]: `BijectiveTensor` has no attribute `_log_detJ`.
                     _log_detJ = y._log_detJ
                 elif isinstance(x_temp, BijectiveTensor) and x_temp.from_inverse():
                     _log_detJ = x_temp._log_detJ
@@ -74,11 +75,14 @@ def forward(
         # TODO: Check that this doesn't contain bugs!
         if (
             is_record_flow_graph_enabled()
+            # pyre-fixme[61]: `y` is undefined, or not always defined.
             and not isinstance(y, BijectiveTensor)
+            # pyre-fixme[61]: `y` is undefined, or not always defined.
             and not (isinstance(x, BijectiveTensor) and y in set(x.parents()))
         ):
             # we exclude y that are bijective tensors for Compose
             y = to_bijective_tensor(x, x_temp, context, self, log_detJ, mode="forward")
+        # pyre-fixme[61]: `y` is undefined, or not always defined.
         return y
 
     def inverse(
@@ -93,6 +97,7 @@ def inverse(
             x = bijector.inverse(y_temp, context)  # type: ignore
             if is_record_flow_graph_enabled() and requires_log_detJ():
                 if isinstance(y_temp, BijectiveTensor) and y_temp.from_forward():
+                    # pyre-fixme[16]: `BijectiveTensor` has no attribute `_log_detJ`.
                     _log_detJ = y_temp._log_detJ
                 elif isinstance(x, BijectiveTensor) and x.from_inverse():
                     _log_detJ = x._log_detJ

flowtorch/bijectors/permute.py

@@ -33,6 +33,7 @@ def _forward(
         if self.permutation is None:
             self.permutation = torch.randperm(x.shape[-1])
 
+        # pyre-fixme[6]: For 3rd argument expected `Tensor` but got `Optional[Tensor]`.
         y = torch.index_select(x, -1, self.permutation)
         ladj = self._log_abs_det_jacobian(x, y, params)
         return y, ladj
@@ -54,6 +55,11 @@ def inv_permutation(self) -> torch.Tensor | None:
 
         result = torch.empty_like(self.permutation, dtype=torch.long)
         result[self.permutation] = torch.arange(
-            self.permutation.size(0), dtype=torch.long, device=self.permutation.device
+            # pyre-fixme[16]: `Optional` has no attribute `size`.
+            # pyre-fixme[16]: `Optional` has no attribute `device`.
+            self.permutation.size(0),
+            dtype=torch.long,
+            # pyre-fixme[16]: `Optional` has no attribute `device`.
+            device=self.permutation.device,
         )
         return result

flowtorch/bijectors/tanh.py

@@ -34,4 +34,5 @@ def _inverse(
     def _log_abs_det_jacobian(
         self, x: torch.Tensor, y: torch.Tensor, params: Sequence[torch.Tensor] | None
     ) -> torch.Tensor:
+        # pyre-fixme[7]: Expected `Tensor` but got `float`.
         return 2.0 * (math.log(2.0) - x - F.softplus(-2.0 * x))

flowtorch/bijectors/volume_preserving.py

@@ -17,6 +17,6 @@ def _log_abs_det_jacobian(
         return torch.zeros(
             x.size()[: -self.domain.event_dim],
             dtype=x.dtype,
-            layout=x.layout,  # pyre-ignore[16]
+            layout=x.layout,
             device=x.device,
         )

flowtorch/distributions/flow.py

@@ -28,21 +28,30 @@ def __init__(
         # TODO: Confirm that the following logic works. Shouldn't it use
         # .domain and .codomain?? Infer shape from constructed self.bijector
         shape = (
-            self.base_dist.batch_shape + self.base_dist.event_shape  # pyre-ignore[16]
+            self.base_dist.batch_shape
+            # pyre-fixme[58]: `+` is not supported for operand types `Size` and `Size`.
+            + self.base_dist.event_shape
         )
         event_dim = self.bijector.domain.event_dim  # type: ignore
         event_dim = max(event_dim, len(self.base_dist.event_shape))
         batch_shape = shape[: len(shape) - event_dim]
         event_shape = shape[len(shape) - event_dim :]
 
         dist.Distribution.__init__(
-            self, batch_shape, event_shape, validate_args=validate_args
+            self,
+            # pyre-fixme[6]: For 2nd argument expected `Size` but got `Tuple[int, ...]`.
+            batch_shape,
+            # pyre-fixme[6]: For 3rd argument expected `Size` but got `Tuple[int, ...]`.
+            event_shape,
+            validate_args=validate_args,
         )
 
     def condition(self, context: torch.Tensor) -> "Flow":
         self._context = context
         return self
 
+    # pyre-fixme[14]: `sample` overrides method defined in `Distribution`
+    #  inconsistently.
     def sample(
         self,
         sample_shape: Tensor | torch.Size = _default_sample_shape,
@@ -57,10 +66,14 @@ def sample(
         if context is None:
             context = self._context
         with torch.no_grad():
+            # pyre-fixme[6]: For 1st argument expected `Union[List[int], Size,
+            #  typing.Tuple[int, ...]]` but got `Union[Size, Tensor]`.
             x = self.base_dist.sample(sample_shape)
             x = self.bijector.forward(x, context)  # type: ignore
             return x
 
+    # pyre-fixme[14]: `rsample` overrides method defined in `Distribution`
+    #  inconsistently.
     def rsample(
         self,
         sample_shape: Tensor | torch.Size = _default_sample_shape,
@@ -74,6 +87,8 @@ def rsample(
         """
         if context is None:
             context = self._context
+        # pyre-fixme[6]: For 1st argument expected `Union[List[int], Size,
+        #  typing.Tuple[int, ...]]` but got `Union[Size, Tensor]`.
         x = self.base_dist.rsample(sample_shape)
         x = self.bijector.forward(x, context)  # type: ignore
         return x
@@ -109,7 +124,7 @@ def log_prob(
         """
         if context is None:
             context = self._context
-        event_dim = len(self.event_shape)  # pyre-ignore[16]
+        event_dim = len(self.event_shape)
 
         x = self.bijector.inverse(value, context)  # type: ignore
         log_prob = -_sum_rightmost(
@@ -118,7 +133,7 @@ def log_prob(
         )
         log_prob = log_prob + _sum_rightmost(
             self.base_dist.log_prob(x),
-            event_dim - len(self.base_dist.event_shape),  # pyre-ignore[16]
+            event_dim - len(self.base_dist.event_shape),
         )
 
         return log_prob

flowtorch/distributions/neals_funnel.py

@@ -19,8 +19,11 @@ class NealsFunnel(dist.Distribution):
     def __init__(self, validate_args: Any = None) -> None:
         d = 2
         batch_shape, event_shape = torch.Size([]), (d,)
+        # pyre-fixme[6]: For 2nd argument expected `Size` but got `Tuple[int]`.
         super().__init__(batch_shape, event_shape, validate_args=validate_args)
 
+    # pyre-fixme[14]: `rsample` overrides method defined in `Distribution`
+    #  inconsistently.
     def rsample(
         self,
         sample_shape: torch.Tensor | torch.Size | None = None,

flowtorch/docs.py

@@ -89,6 +89,10 @@ def generate_class_markdown(symbol_name: str, entity: Any) -> str:
         try:
             if hasattr(member_object, "__wrapped__"):
                 # decorators = get_decorators(member_object)
+                # pyre-fixme[16]: Item `BuiltinFunctionType` of
+                #  `Union[BuiltinFunctionType, ClassMethodDescriptorType, FunctionType,
+                #  MethodDescriptorType, MethodType, WrapperDescriptorType]` has no
+                #  attribute `__wrapped__`.
                 member_object = member_object.__wrapped__
         except Exception:
             pass
@@ -267,6 +271,7 @@ def walk_packages(
         else:
             finder = None
 
+        # pyre-fixme[61]: `finder` is undefined, or not always defined.
         if finder is not None:
             module = finder.load_module(this_modname)
 
@@ -278,6 +283,7 @@ def walk_packages(
             modules[this_modname] = (module, documentable_symbols(module))
 
             del module
+            # pyre-fixme[61]: `finder` is undefined, or not always defined.
             del finder
 
         else:

flowtorch/nn/made.py

@@ -116,6 +116,7 @@ def __init__(
         super().__init__(in_features, out_features, bias)
         self.register_buffer("mask", mask.data)
 
+    # pyre-fixme[14]: `forward` overrides method defined in `Linear` inconsistently.
     def forward(self, _input: torch.Tensor) -> torch.Tensor:
         masked_weight = self.weight * self.mask
         return F.linear(_input, masked_weight, self.bias)

flowtorch/ops/__init__.py

@@ -46,6 +46,8 @@ def _select_bins(x: torch.Tensor, idx: torch.Tensor) -> torch.Tensor:
     # Note that by convention, the context variable batch dimensions must broadcast
     # over the input batch dimensions.
     if len(idx.shape) >= len(x.shape):
+        # pyre-fixme[58]: `+` is not supported for operand types `Tuple[int, ...]`
+        #  and `Size`.
         x = x.reshape((1,) * (len(idx.shape) - len(x.shape)) + x.shape)
         x = x.expand(idx.shape[:-2] + (-1,) * 2)
 
@@ -268,6 +270,7 @@ def monotonic_rational_spline(
                 * theta_one_minus_theta
             )
             derivative_numerator = input_delta.pow(2) * (
+                # pyre-fixme[16]: `float` has no attribute `pow`.
                 input_derivatives_plus_one * root.pow(2)
                 + 2 * input_delta * theta_one_minus_theta
                 + input_derivatives * (1 - root).pow(2)

flowtorch/parameters/dense_autoregressive.py

@@ -179,6 +179,7 @@ def _forward(
 
         # results ~ (batch_shape, param_shapes[0]), ...
         result = tuple(
+            # pyre-fixme[58]: `+` is not supported for operand types `Size` and `Size`.
             h_slice.view(batch_shape + p_shape)
             for h_slice, p_shape in zip(result, list(self.param_shapes))
         )

flowtorch/utils.py

@@ -79,6 +79,7 @@ def _walk_packages(
         else:
             finder = None
 
+        # pyre-fixme[61]: `finder` is undefined, or not always defined.
         if finder is not None:
             module = finder.load_module(this_modname)
 
@@ -90,6 +91,7 @@ def _walk_packages(
             classes.extend(this_classes)
 
             del module
+            # pyre-fixme[61]: `finder` is undefined, or not always defined.
             del finder
 
         else: