suppress errors in vision - batch 1

Summary:
This diff is auto-generated to upgrade the Pyre version and suppress errors in vision. The upgrade will affect Pyre local configurations in the following directories:
```
vision/ale/search
vision/fair/fvcore
vision/fair/pytorch3d
vision/ocr/rosetta_hash
vision/vogue/personalization
```

Differential Revision: D21688454

fbshipit-source-id: 1f3c3fee42b6da2e162fd0932742ab8c5c96aa45

pytorch3d/io/mtl_io.py

@@ -51,6 +51,7 @@ def make_mesh_texture_atlas(
 
     # Initialize the per face texture map to a white color.
     # TODO: allow customization of this base color?
+    # pyre-fixme[16]: `Tensor` has no attribute `new_ones`.
     atlas = faces_verts_uvs.new_ones(size=(F, R, R, 3))
 
     # Check for empty materials.
@@ -63,10 +64,13 @@ def make_mesh_texture_atlas(
         # will be ignored and a repeating pattern is formed.
         # Shapenet data uses this format see:
         # https://shapenet.org/qaforum/index.php?qa=15&qa_1=why-is-the-texture-coordinate-in-the-obj-file-not-in-the-range # noqa: B950
+        # pyre-fixme[16]: `ByteTensor` has no attribute `any`.
         if (faces_verts_uvs > 1).any() or (faces_verts_uvs < 0).any():
             msg = "Texture UV coordinates outside the range [0, 1]. \
                 The integer part will be ignored to form a repeating pattern."
             warnings.warn(msg)
+            # pyre-fixme[9]: faces_verts_uvs has type `Tensor`; used as `int`.
+            # pyre-fixme[6]: Expected `int` for 1st param but got `Tensor`.
             faces_verts_uvs = faces_verts_uvs % 1
     elif texture_wrap == "clamp":
         # Clamp uv coordinates to the [0, 1] range.
@@ -257,6 +261,7 @@ def make_material_atlas(
     # Meshgrid returns (row, column) i.e (Y, X)
     # Change order to (X, Y) to make the grid.
     Y, X = torch.meshgrid(rng, rng)
+    # pyre-fixme[28]: Unexpected keyword argument `axis`.
     grid = torch.stack([X, Y], axis=-1)  # (R, R, 2)
 
     # Grid cells below the diagonal: x + y < R.
@@ -269,6 +274,7 @@ def make_material_atlas(
     # w0, w1
     bary[below_diag, slc] = ((grid[below_diag] + 1.0 / 3.0) / R).T
     # w0, w1 for above diagonal grid cells.
+    # pyre-fixme[16]: `float` has no attribute `T`.
     bary[~below_diag, slc] = (((R - 1.0 - grid[~below_diag]) + 2.0 / 3.0) / R).T
     # w2 = 1. - w0 - w1
     bary[..., -1] = 1 - bary[..., :2].sum(dim=-1)

pytorch3d/io/obj_io.py

@@ -213,6 +213,8 @@ def load_obj(
     """
     data_dir = "./"
     if isinstance(f_obj, (str, bytes, os.PathLike)):
+        # pyre-fixme[6]: Expected `_PathLike[Variable[typing.AnyStr <: [str,
+        #  bytes]]]` for 1st param but got `Union[_PathLike[typing.Any], bytes, str]`.
         data_dir = os.path.dirname(f_obj)
     f_obj, new_f = _open_file(f_obj)
     try:
@@ -453,6 +455,8 @@ def _load(
     material_colors, texture_images, texture_atlas = None, None, None
     if load_textures:
         if (len(material_names) > 0) and (f_mtl is not None):
+            # pyre-fixme[6]: Expected `Union[_PathLike[typing.Any], bytes, str]` for
+            #  1st param but got `Optional[str]`.
             if os.path.isfile(f_mtl):
                 # Texture mode uv wrap
                 material_colors, texture_images = load_mtl(

pytorch3d/io/utils.py

@@ -30,6 +30,8 @@ def _read_image(file_name: str, format=None):
     if format not in ["RGB", "BGR"]:
         raise ValueError("format can only be one of [RGB, BGR]; got %s", format)
     with PathManager.open(file_name, "rb") as f:
+        # pyre-fixme[6]: Expected `Union[str, typing.BinaryIO]` for 1st param but
+        #  got `Union[typing.IO[bytes], typing.IO[str]]`.
         image = Image.open(f)
         if format is not None:
             # PIL only supports RGB. First convert to RGB and flip channels

pytorch3d/loss/chamfer.py

@@ -41,6 +41,7 @@ def _handle_pointcloud_input(
         lengths = points.num_points_per_cloud()
         normals = points.normals_padded()  # either a tensor or None
     elif torch.is_tensor(points):
+        # pyre-fixme[16]: `Tensor` has no attribute `ndim`.
         if points.ndim != 3:
             raise ValueError("Expected points to be of shape (N, P, D)")
         X = points
@@ -173,6 +174,7 @@ def chamfer_distance(
         )
 
         if is_x_heterogeneous:
+            # pyre-fixme[16]: `int` has no attribute `__setitem__`.
             cham_norm_x[x_mask] = 0.0
         if is_y_heterogeneous:
             cham_norm_y[y_mask] = 0.0

pytorch3d/loss/point_mesh_distance.py

@@ -65,6 +65,7 @@ class _PointFaceDistance(Function):
         return grad_points, None, grad_tris, None, None
 
 
+# pyre-fixme[16]: `_PointFaceDistance` has no attribute `apply`.
 point_face_distance = _PointFaceDistance.apply
 
 
@@ -115,6 +116,7 @@ class _FacePointDistance(Function):
         return grad_points, None, grad_tris, None, None
 
 
+# pyre-fixme[16]: `_FacePointDistance` has no attribute `apply`.
 face_point_distance = _FacePointDistance.apply
 
 
@@ -165,6 +167,7 @@ class _PointEdgeDistance(Function):
         return grad_points, None, grad_segms, None, None
 
 
+# pyre-fixme[16]: `_PointEdgeDistance` has no attribute `apply`.
 point_edge_distance = _PointEdgeDistance.apply
 
 
@@ -215,6 +218,7 @@ class _EdgePointDistance(Function):
         return grad_points, None, grad_segms, None, None
 
 
+# pyre-fixme[16]: `_EdgePointDistance` has no attribute `apply`.
 edge_point_distance = _EdgePointDistance.apply
 
 

pytorch3d/ops/cubify.py

@@ -229,6 +229,7 @@ def cubify(voxels, thresh, device=None, align: str = "topleft") -> Meshes:
     idlenum = idleverts.cumsum(1)
 
     verts_list = [
+        # pyre-fixme[16]: `Tensor` has no attribute `index_select`.
         grid_verts.index_select(0, (idleverts[n] == 0).nonzero(as_tuple=False)[:, 0])
         for n in range(N)
     ]

pytorch3d/ops/graph_conv.py

@@ -35,6 +35,7 @@ class GraphConv(nn.Module):
         if init == "normal":
             nn.init.normal_(self.w0.weight, mean=0, std=0.01)
             nn.init.normal_(self.w1.weight, mean=0, std=0.01)
+            # pyre-fixme[16]: Optional type has no attribute `data`.
             self.w0.bias.data.zero_()
             self.w1.bias.data.zero_()
         elif init == "zero":
@@ -115,6 +116,7 @@ def gather_scatter_python(input, edges, directed: bool = False):
     idx0 = edges[:, 0].view(num_edges, 1).expand(num_edges, input_feature_dim)
     idx1 = edges[:, 1].view(num_edges, 1).expand(num_edges, input_feature_dim)
 
+    # pyre-fixme[16]: `Tensor` has no attribute `scatter_add`.
     output = output.scatter_add(0, idx0, input.gather(0, idx1))
     if not directed:
         output = output.scatter_add(0, idx1, input.gather(0, idx0))
@@ -167,4 +169,5 @@ class GatherScatter(Function):
         return grad_input, grad_edges, grad_directed
 
 
+# pyre-fixme[16]: `GatherScatter` has no attribute `apply`.
 gather_scatter = GatherScatter.apply

pytorch3d/ops/knn.py

@@ -157,6 +157,7 @@ def knn_points(
     if lengths2 is None:
         lengths2 = torch.full((p1.shape[0],), P2, dtype=torch.int64, device=p1.device)
 
+    # pyre-fixme[16]: `_knn_points` has no attribute `apply`.
     p1_dists, p1_idx = _knn_points.apply(p1, p2, lengths1, lengths2, K, version)
 
     p2_nn = None

pytorch3d/ops/mesh_face_areas_normals.py

@@ -59,4 +59,5 @@ class _MeshFaceAreasNormals(Function):
         return grad_verts, None
 
 
+# pyre-fixme[16]: `_MeshFaceAreasNormals` has no attribute `apply`.
 mesh_face_areas_normals = _MeshFaceAreasNormals.apply

pytorch3d/ops/sample_points_from_meshes.py

@@ -120,4 +120,6 @@ def _rand_barycentric_coords(
     w0 = 1.0 - u_sqrt
     w1 = u_sqrt * (1.0 - v)
     w2 = u_sqrt * v
+    # pyre-fixme[7]: Expected `Tuple[torch.Tensor, torch.Tensor, torch.Tensor]` but
+    #  got `Tuple[float, typing.Any, typing.Any]`.
     return w0, w1, w2

pytorch3d/ops/vert_align.py

@@ -96,6 +96,7 @@ def vert_align(
                 .view(-1, 1)
                 .expand(-1, feats_sampled.shape[-1])
             )
+            # pyre-fixme[16]: `Tensor` has no attribute `gather`.
             feats_sampled = feats_sampled.gather(0, idx)  # (sum(V), C)
 
     return feats_sampled

pytorch3d/renderer/blending.py

@@ -150,7 +150,10 @@ def softmax_rgb_blend(
     # TODO: there may still be some instability in the exponent calculation.
 
     z_inv = (zfar - fragments.zbuf) / (zfar - znear) * mask
+    # pyre-fixme[16]: `Tuple` has no attribute `values`.
+    # pyre-fixme[6]: Expected `Tensor` for 1st param but got `float`.
     z_inv_max = torch.max(z_inv, dim=-1).values[..., None]
+    # pyre-fixme[6]: Expected `Tensor` for 1st param but got `float`.
     weights_num = prob_map * torch.exp((z_inv - z_inv_max) / blend_params.gamma)
 
     # Normalize weights.

pytorch3d/renderer/compositing.py

@@ -95,6 +95,7 @@ def alpha_composite(pointsidx, alphas, pt_clds, blend_params=None) -> torch.Tens
         Combined features: Tensor of shape (N, C, image_size, image_size)
             giving the accumulated features at each point.
     """
+    # pyre-fixme[16]: `_CompositeAlphaPoints` has no attribute `apply`.
     return _CompositeAlphaPoints.apply(pt_clds, alphas, pointsidx)
 
 
@@ -173,6 +174,7 @@ def norm_weighted_sum(pointsidx, alphas, pt_clds, blend_params=None) -> torch.Te
         Combined features: Tensor of shape (N, C, image_size, image_size)
             giving the accumulated features at each point.
     """
+    # pyre-fixme[16]: `_CompositeNormWeightedSumPoints` has no attribute `apply`.
     return _CompositeNormWeightedSumPoints.apply(pt_clds, alphas, pointsidx)
 
 
@@ -244,4 +246,5 @@ def weighted_sum(pointsidx, alphas, pt_clds, blend_params=None) -> torch.Tensor:
         Combined features: Tensor of shape (N, C, image_size, image_size)
             giving the accumulated features at each point.
     """
+    # pyre-fixme[16]: `_CompositeWeightedSumPoints` has no attribute `apply`.
     return _CompositeWeightedSumPoints.apply(pt_clds, alphas, pointsidx)

pytorch3d/renderer/lighting.py

@@ -183,6 +183,7 @@ class DirectionalLights(TensorProperties):
             direction=direction,
         )
         _validate_light_properties(self)
+        # pyre-fixme[16]: `DirectionalLights` has no attribute `direction`.
         if self.direction.shape[-1] != 3:
             msg = "Expected direction to have shape (N, 3); got %r"
             raise ValueError(msg % repr(self.direction.shape))
@@ -196,14 +197,20 @@ class DirectionalLights(TensorProperties):
         # the same for directional and point lights. The call sites should not
         # need to know the light type.
         return diffuse(
-            normals=normals, color=self.diffuse_color, direction=self.direction
+            normals=normals,
+            # pyre-fixme[16]: `DirectionalLights` has no attribute `diffuse_color`.
+            color=self.diffuse_color,
+            # pyre-fixme[16]: `DirectionalLights` has no attribute `direction`.
+            direction=self.direction,
         )
 
     def specular(self, normals, points, camera_position, shininess) -> torch.Tensor:
         return specular(
             points=points,
             normals=normals,
+            # pyre-fixme[16]: `DirectionalLights` has no attribute `specular_color`.
             color=self.specular_color,
+            # pyre-fixme[16]: `DirectionalLights` has no attribute `direction`.
             direction=self.direction,
             camera_position=camera_position,
             shininess=shininess,
@@ -242,6 +249,7 @@ class PointLights(TensorProperties):
             location=location,
         )
         _validate_light_properties(self)
+        # pyre-fixme[16]: `PointLights` has no attribute `location`.
         if self.location.shape[-1] != 3:
             msg = "Expected location to have shape (N, 3); got %r"
             raise ValueError(msg % repr(self.location.shape))
@@ -251,14 +259,18 @@ class PointLights(TensorProperties):
         return super().clone(other)
 
     def diffuse(self, normals, points) -> torch.Tensor:
+        # pyre-fixme[16]: `PointLights` has no attribute `location`.
         direction = self.location - points
+        # pyre-fixme[16]: `PointLights` has no attribute `diffuse_color`.
         return diffuse(normals=normals, color=self.diffuse_color, direction=direction)
 
     def specular(self, normals, points, camera_position, shininess) -> torch.Tensor:
+        # pyre-fixme[16]: `PointLights` has no attribute `location`.
         direction = self.location - points
         return specular(
             points=points,
             normals=normals,
+            # pyre-fixme[16]: `PointLights` has no attribute `specular_color`.
             color=self.specular_color,
             direction=direction,
             camera_position=camera_position,

pytorch3d/renderer/mesh/rasterize_meshes.py

@@ -132,6 +132,7 @@ def rasterize_meshes(
     if max_faces_per_bin is None:
         max_faces_per_bin = int(max(10000, verts_packed.shape[0] / 5))
 
+    # pyre-fixme[16]: `_RasterizeFaceVerts` has no attribute `apply`.
     return _RasterizeFaceVerts.apply(
         face_verts,
         mesh_to_face_first_idx,
@@ -184,6 +185,7 @@ class _RasterizeFaceVerts(torch.autograd.Function):
         perspective_correct: bool = False,
         cull_backfaces: bool = False,
     ):
+        # pyre-fixme[16]: Module `pytorch3d` has no attribute `_C`.
         pix_to_face, zbuf, barycentric_coords, dists = _C.rasterize_meshes(
             face_verts,
             mesh_to_face_first_idx,
@@ -283,6 +285,7 @@ def rasterize_meshes_python(
     )
 
     # Calculate all face bounding boxes.
+    # pyre-fixme[16]: `Tuple` has no attribute `values`.
     x_mins = torch.min(faces_verts[:, :, 0], dim=1, keepdim=True).values
     x_maxs = torch.max(faces_verts[:, :, 0], dim=1, keepdim=True).values
     y_mins = torch.min(faces_verts[:, :, 1], dim=1, keepdim=True).values

pytorch3d/renderer/mesh/utils.py

@@ -63,6 +63,7 @@ def interpolate_face_attributes(
     pix_to_face = pix_to_face.clone()
     pix_to_face[mask] = 0
     idx = pix_to_face.view(N * H * W * K, 1, 1).expand(N * H * W * K, 3, D)
+    # pyre-fixme[16]: `Tensor` has no attribute `gather`.
     pixel_face_vals = face_attributes.gather(0, idx).view(N, H, W, K, 3, D)
     pixel_vals = (barycentric_coords[..., None] * pixel_face_vals).sum(dim=-2)
     pixel_vals[mask] = 0  # Replace masked values in output.

pytorch3d/renderer/points/rasterize_points.py

@@ -89,6 +89,7 @@ def rasterize_points(
 
     if bin_size != 0:
         # There is a limit on the number of points per bin in the cuda kernel.
+        # pyre-fixme[6]: Expected `int` for 1st param but got `Union[int, None, int]`.
         points_per_bin = 1 + (image_size - 1) // bin_size
         if points_per_bin >= kMaxPointsPerBin:
             raise ValueError(
@@ -101,6 +102,7 @@ def rasterize_points(
 
     # Function.apply cannot take keyword args, so we handle defaults in this
     # wrapper and call apply with positional args only
+    # pyre-fixme[16]: `_RasterizePoints` has no attribute `apply`.
     return _RasterizePoints.apply(
         points_packed,
         cloud_to_packed_first_idx,
@@ -138,6 +140,7 @@ class _RasterizePoints(torch.autograd.Function):
             bin_size,
             max_points_per_bin,
         )
+        # pyre-fixme[16]: Module `pytorch3d` has no attribute `_C`.
         idx, zbuf, dists = _C.rasterize_points(*args)
         ctx.save_for_backward(points, idx)
         ctx.mark_non_differentiable(idx)

pytorch3d/structures/meshes.py

@@ -820,6 +820,7 @@ class Meshes(object):
 
             # NOTE: this is already applying the area weighting as the magnitude
             # of the cross product is 2 x area of the triangle.
+            # pyre-fixme[16]: `Tensor` has no attribute `index_add`.
             verts_normals = verts_normals.index_add(
                 0,
                 faces_packed[:, 1],
@@ -1392,6 +1393,7 @@ def join_meshes_as_batch(meshes: List[Meshes], include_textures: bool = True):
         # Meshes objects can be iterated and produce single Meshes. We avoid
         # letting join_meshes_as_batch(mesh1, mesh2) silently do the wrong thing.
         raise ValueError("Wrong first argument to join_meshes_as_batch.")
+    # pyre-fixme[10]: Name `mesh` is used but not defined.
     verts = [v for mesh in meshes for v in mesh.verts_list()]
     faces = [f for mesh in meshes for f in mesh.faces_list()]
     if len(meshes) == 0 or not include_textures:

pytorch3d/structures/textures.py

@@ -63,6 +63,7 @@ def _extend_tensor(input_tensor: torch.Tensor, N: int) -> torch.Tensor:
         input_tensor: torch.Tensor with ndim > 2 representing a batched input.
         N: number of times to extend each element of the batch.
     """
+    # pyre-fixme[16]: `Tensor` has no attribute `ndim`.
     if input_tensor.ndim < 2:
         raise ValueError("Input tensor must have ndimensions >= 2.")
     B = input_tensor.shape[0]
@@ -98,6 +99,7 @@ class Textures(object):
         and the packed/list representations are computed on the fly and
         not cached.
         """
+        # pyre-fixme[16]: `Tensor` has no attribute `ndim`.
         if faces_uvs is not None and faces_uvs.ndim != 3:
             msg = "Expected faces_uvs to be of shape (N, F, 3); got %r"
             raise ValueError(msg % repr(faces_uvs.shape))
@@ -108,8 +110,10 @@ class Textures(object):
             msg = "Expected verts_rgb to be of shape (N, V, 3); got %r"
             raise ValueError(msg % repr(verts_rgb.shape))
         if maps is not None:
+            # pyre-fixme[16]: `List` has no attribute `ndim`.
             if torch.is_tensor(maps) and maps.ndim != 4:
                 msg = "Expected maps to be of shape (N, H, W, 3); got %r"
+                # pyre-fixme[16]: `List` has no attribute `shape`.
                 raise ValueError(msg % repr(maps.shape))
             elif isinstance(maps, list):
                 maps = _pad_texture_maps(maps)
@@ -155,20 +159,27 @@ class Textures(object):
         return other
 
     def faces_uvs_padded(self) -> torch.Tensor:
+        # pyre-fixme[7]: Expected `Tensor` but got `Optional[torch.Tensor]`.
         return self._faces_uvs_padded
 
     def faces_uvs_list(self) -> Union[List[torch.Tensor], None]:
         if self._faces_uvs_padded is None:
             return None
         return padded_to_list(
-            self._faces_uvs_padded, split_size=self._num_faces_per_mesh
+            # pyre-fixme[6]: Expected `Tensor` for 1st param but got
+            #  `Optional[torch.Tensor]`.
+            self._faces_uvs_padded,
+            split_size=self._num_faces_per_mesh,
         )
 
     def faces_uvs_packed(self) -> Union[torch.Tensor, None]:
         if self._faces_uvs_padded is None:
             return None
         return padded_to_packed(
-            self._faces_uvs_padded, split_size=self._num_faces_per_mesh
+            # pyre-fixme[6]: Expected `Tensor` for 1st param but got
+            #  `Optional[torch.Tensor]`.
+            self._faces_uvs_padded,
+            split_size=self._num_faces_per_mesh,
         )
 
     def verts_uvs_padded(self) -> Union[torch.Tensor, None]:
@@ -182,6 +193,8 @@ class Textures(object):
         # each face so the num_verts_uvs_per_mesh
         # may be different from num_verts_per_mesh.
         # Therefore don't use any split_size.
+        # pyre-fixme[6]: Expected `Tensor` for 1st param but got
+        #  `Optional[torch.Tensor]`.
         return padded_to_list(self._verts_uvs_padded)
 
     def verts_uvs_packed(self) -> Union[torch.Tensor, None]:
@@ -192,6 +205,8 @@ class Textures(object):
         # each face so the num_verts_uvs_per_mesh
         # may be different from num_verts_per_mesh.
         # Therefore don't use any split_size.
+        # pyre-fixme[6]: Expected `Tensor` for 1st param but got
+        #  `Optional[torch.Tensor]`.
         return padded_to_packed(self._verts_uvs_padded)
 
     def verts_rgb_padded(self) -> Union[torch.Tensor, None]:
@@ -201,18 +216,26 @@ class Textures(object):
         if self._verts_rgb_padded is None:
             return None
         return padded_to_list(
-            self._verts_rgb_padded, split_size=self._num_verts_per_mesh
+            # pyre-fixme[6]: Expected `Tensor` for 1st param but got
+            #  `Optional[torch.Tensor]`.
+            self._verts_rgb_padded,
+            split_size=self._num_verts_per_mesh,
         )
 
     def verts_rgb_packed(self) -> Union[torch.Tensor, None]:
         if self._verts_rgb_padded is None:
             return None
         return padded_to_packed(
-            self._verts_rgb_padded, split_size=self._num_verts_per_mesh
+            # pyre-fixme[6]: Expected `Tensor` for 1st param but got
+            #  `Optional[torch.Tensor]`.
+            self._verts_rgb_padded,
+            split_size=self._num_verts_per_mesh,
         )
 
     # Currently only the padded maps are used.
     def maps_padded(self) -> Union[torch.Tensor, None]:
+        # pyre-fixme[7]: Expected `Optional[torch.Tensor]` but got `Union[None,
+        #  List[typing.Any], torch.Tensor]`.
         return self._maps_padded
 
     def extend(self, N: int) -> "Textures":
@@ -234,13 +257,21 @@ class Textures(object):
             v is not None
             for v in [self._faces_uvs_padded, self._verts_uvs_padded, self._maps_padded]
         ):
+            # pyre-fixme[6]: Expected `Tensor` for 1st param but got
+            #  `Optional[torch.Tensor]`.
             new_verts_uvs = _extend_tensor(self._verts_uvs_padded, N)
+            # pyre-fixme[6]: Expected `Tensor` for 1st param but got
+            #  `Optional[torch.Tensor]`.
             new_faces_uvs = _extend_tensor(self._faces_uvs_padded, N)
+            # pyre-fixme[6]: Expected `Tensor` for 1st param but got `Union[None,
+            #  List[typing.Any], torch.Tensor]`.
             new_maps = _extend_tensor(self._maps_padded, N)
             return self.__class__(
                 verts_uvs=new_verts_uvs, faces_uvs=new_faces_uvs, maps=new_maps
             )
         elif self._verts_rgb_padded is not None:
+            # pyre-fixme[6]: Expected `Tensor` for 1st param but got
+            #  `Optional[torch.Tensor]`.
             new_verts_rgb = _extend_tensor(self._verts_rgb_padded, N)
             return self.__class__(verts_rgb=new_verts_rgb)
         else:

pytorch3d/structures/utils.py

@@ -48,6 +48,7 @@ def list_to_padded(
     )
     for i, y in enumerate(x):
         if len(y) > 0:
+            # pyre-fixme[16]: `Tensor` has no attribute `ndim`.
             if y.ndim != 2:
                 raise ValueError("Supports only 2-dimensional tensor items")
             x_padded[i, : y.shape[0], : y.shape[1]] = y
@@ -69,6 +70,7 @@ def padded_to_list(x: torch.Tensor, split_size: Union[list, tuple, None] = None)
     Returns:
       x_list: a list of tensors
     """
+    # pyre-fixme[16]: `Tensor` has no attribute `ndim`.
     if x.ndim != 3:
         raise ValueError("Supports only 3-dimensional input tensors")
     x_list = list(x.unbind(0))
@@ -145,6 +147,7 @@ def packed_to_list(x: torch.Tensor, split_size: Union[list, int]):
     Returns:
       x_list: A list of Tensors
     """
+    # pyre-fixme[16]: `Tensor` has no attribute `split`.
     return x.split(split_size, dim=0)
 
 
@@ -174,6 +177,7 @@ def padded_to_packed(
     Returns:
       x_packed: a packed tensor.
     """
+    # pyre-fixme[16]: `Tensor` has no attribute `ndim`.
     if x.ndim != 3:
         raise ValueError("Supports only 3-dimensional input tensors")
 
@@ -189,15 +193,19 @@ def padded_to_packed(
 
     # Convert to packed using pad value
     if pad_value is not None:
+        # pyre-fixme[16]: `ByteTensor` has no attribute `any`.
         mask = x_packed.ne(pad_value).any(-1)
         x_packed = x_packed[mask]
         return x_packed
 
     # Convert to packed using split sizes
+    # pyre-fixme[6]: Expected `Sized` for 1st param but got `Union[None,
+    #  List[typing.Any], typing.Tuple[typing.Any, ...]]`.
     N = len(split_size)
     if x.shape[0] != N:
         raise ValueError("Split size must be of same length as inputs first dimension")
 
+    # pyre-fixme[16]: `None` has no attribute `__iter__`.
     if not all(isinstance(i, int) for i in split_size):
         raise ValueError(
             "Support only 1-dimensional unbinded tensor. \
@@ -207,6 +215,8 @@ def padded_to_packed(
     padded_to_packed_idx = torch.cat(
         [
             torch.arange(v, dtype=torch.int64, device=x.device) + i * M
+            # pyre-fixme[6]: Expected `Iterable[Variable[_T]]` for 1st param but got
+            #  `Union[None, List[typing.Any], typing.Tuple[typing.Any, ...]]`.
             for (i, v) in enumerate(split_size)
         ],
         dim=0,

pytorch3d/transforms/so3.py

@@ -77,6 +77,7 @@ def so3_rotation_angle(R, eps: float = 1e-4, cos_angle: bool = False):
     if cos_angle:
         return phi
     else:
+        # pyre-fixme[16]: `float` has no attribute `acos`.
         return phi.acos()
 
 
@@ -118,6 +119,7 @@ def so3_exponential_map(log_rot, eps: float = 0.0001):
     skews = hat(log_rot)
 
     R = (
+        # pyre-fixme[16]: `float` has no attribute `__getitem__`.
         fac1[:, None, None] * skews
         + fac2[:, None, None] * torch.bmm(skews, skews)
         + torch.eye(3, dtype=log_rot.dtype, device=log_rot.device)[None]

pytorch3d/transforms/transform3d.py

@@ -155,6 +155,7 @@ class Transform3d:
         if matrix is None:
             self._matrix = torch.eye(4, dtype=dtype, device=device).view(1, 4, 4)
         else:
+            # pyre-fixme[16]: `Tensor` has no attribute `ndim`.
             if matrix.ndim not in (2, 3):
                 raise ValueError('"matrix" has to be a 2- or a 3-dimensional tensor.')
             if matrix.shape[-2] != 4 or matrix.shape[-1] != 4: