Add pydocstyle to the ruff rules (#52136)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Signed-off-by: Harry Mellor <hej.mellor@gmail.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
Harry Mellor
2026-09-16 16:44:29 +00:00
committed by GitHub
co-authored by Claude Opus 5
parent 804e5377aa
commit 8c1557a79c
2002 changed files with 15024 additions and 14269 deletions
+12 -17
View File
@@ -157,8 +157,7 @@ def _is_layer_none_or_staged(layer: nn.Module) -> bool:
class PixtralImagePixelInputs(TensorSchema):
"""
Dimensions:
"""Dimensions:
- bn: Batch size * number of images
- c: Number of channels (3)
- h: Height of each image
@@ -688,8 +687,7 @@ class VisionEncoderArgs:
def _reshape_for_broadcast(freqs_cis: torch.Tensor, x: torch.Tensor) -> torch.Tensor:
"""
freqs_cis: complex - (seq_len, head_dim / 2)
"""freqs_cis: complex - (seq_len, head_dim / 2)
x: complex - (bsz, seq_len, head_dim / 2)
"""
ndim = x.ndim
@@ -708,9 +706,8 @@ def precompute_freqs_cis_2d(
width: int,
theta: float,
) -> torch.Tensor:
"""
freqs_cis: 2D complex tensor of shape (height, width, dim // 2)
to be indexed by (height, width) position tuples
"""freqs_cis: 2D complex tensor of shape (height, width, dim // 2)
to be indexed by (height, width) position tuples
"""
# (dim / 2) frequency bases
freqs = 1.0 / (theta ** (torch.arange(0, dim, 2).float() / dim))
@@ -1018,13 +1015,14 @@ class VisionTransformer(nn.Module):
self,
images: list[torch.Tensor],
) -> torch.Tensor:
"""
Args:
"""Args:
images: list of N_img images of variable sizes,
each of shape (C, H, W)
Returns:
image_features: tensor of token features for
all tokens of all images of shape (N_toks, D)
"""
# pass images through initial convolution independently
patch_embeds_list = [
@@ -1082,9 +1080,7 @@ class VisionLanguageAdapter(nn.Module):
class PatchMerger(nn.Module):
"""
Learned merging of spatial_merge_size ** 2 patches
"""
"""Learned merging of spatial_merge_size ** 2 patches."""
def __init__(
self,
@@ -1124,8 +1120,7 @@ class PatchMerger(nn.Module):
x: torch.Tensor,
image_sizes: list[tuple[int, int]],
) -> torch.Tensor:
"""
Args:
"""Args:
x: (N, D) where N is flattened and concatenated patch tokens
for all images
image_sizes: list of tuple of (height, width) in tokens for
@@ -1134,8 +1129,8 @@ class PatchMerger(nn.Module):
image_features: reorders patch tokens so each grid of
(spatial_merge_size, spatial_merge_size) is contiguous.
now (N / spatial_merge_size ** 2, D * spatial_merge_size ** 2)
"""
"""
sub_grids = get_sub_grids(
x=x, image_sizes=image_sizes, spatial_merge_size=self.spatial_merge_size
) # list of [d x sub_grid_size x sub_grid_size x n_patches]
@@ -1509,8 +1504,7 @@ class PixtralHFVisionModel(nn.Module):
select_layers: list[int] | None = None,
feature_select_strategy: VisionFeatureSelectStrategy | None = None,
) -> tuple[torch.Tensor, ...]:
"""
Args:
"""Args:
pixel_values: Each image to be processed will be a separate tensor
in pixel_values. This means it will be a list of tensors
because multiple requests batched can have multiple images,
@@ -1522,6 +1516,7 @@ class PixtralHFVisionModel(nn.Module):
Returns:
image_features: tensor of token features for
all tokens of all images of shape (N_toks, D)
"""
# pass images through initial convolution independently
patch_embeds_list = [