style(ruff): fix lint

.github/verify_labels.py

@@ -82,9 +82,7 @@ def parse_args():
  )
 
  parser.add_argument("pr", type=int, help="PR number")
- args = parser.parse_args()
-
- return args
+ return parser.parse_args()
 
 
 if __name__ == "__main__":

holocron/models/classification/sknet.py

@@ -110,9 +110,7 @@ def forward(self, x: torch.Tensor) -> torch.Tensor:
  b, m, c = paths.shape[:3]
  z = self.sa(paths.sum(dim=1)).view(b, m, c, 1, 1)
  attention_factors = torch.softmax(z, dim=1)
- out = (attention_factors * paths).sum(dim=1)
-
- return out
+ return (attention_factors * paths).sum(dim=1)
 
 
 class SKBottleneck(_ResBlock):

holocron/models/classification/tridentnet.py

@@ -41,7 +41,7 @@ def forward(self, x: torch.Tensor) -> torch.Tensor:
  dilations = [1] * self.num_branches if self.dilation[0] == 1 else [1 + idx for idx in range(self.num_branches)]
 
  # Use shared weight to apply the convolution
- out = torch.cat(
+ return torch.cat(
  [
  F.conv2d(
  _x,
@@ -57,8 +57,6 @@ def forward(self, x: torch.Tensor) -> torch.Tensor:
  1,
  )
 
- return out
-
 
 class Tridentneck(_ResBlock):
  expansion: int = 4

holocron/models/detection/yolo.py

@@ -339,9 +339,7 @@ def _format_outputs(self, x: Tensor) -> Tuple[Tensor, Tensor, Tensor]:
  def _forward(self, x: Tensor) -> Tensor:
  out = self.backbone(x)
  out = self.block4(out)
- out = self.classifier(out)
-
- return out
+ return self.classifier(out)
 
  def forward(
  self, x: Tensor, target: Optional[List[Dict[str, Tensor]]] = None

holocron/models/detection/yolov2.py

@@ -208,9 +208,7 @@ def _forward(self, x: Tensor) -> Tensor:
  out = torch.cat((passthrough, out), 1)
  out = self.block6(out)
 
- out = self.head(out)
-
- return out
+ return self.head(out)
 
  def forward(
  self, x: Union[Tensor, List[Tensor], Tuple[Tensor, ...]], target: Optional[List[Dict[str, Tensor]]] = None

holocron/models/detection/yolov4.py

@@ -628,15 +628,14 @@ def forward(
  y3 = self.yolo3(o3, target)
 
  if not self.training:
- detections = [
+ return [
  {
  "boxes": torch.cat((det1["boxes"], det2["boxes"], det3["boxes"]), dim=0),
  "scores": torch.cat((det1["scores"], det2["scores"], det3["scores"]), dim=0),
  "labels": torch.cat((det1["labels"], det2["labels"], det3["labels"]), dim=0),
  }
  for det1, det2, det3 in zip(y1, y2, y3)
  ]
- return detections
 
  return {k: y1[k] + y2[k] + y3[k] for k in y1}
 

holocron/models/segmentation/unet.py

@@ -223,8 +223,7 @@ def forward(self, x: Tensor) -> Tensor:
  x = decoder(xs.pop(), x)
 
  # Classifier
- x = self.classifier(x)
- return x
+ return self.classifier(x)
 
 
 class UBlock(nn.Module):
@@ -368,8 +367,7 @@ def forward(self, x: Tensor) -> Tensor:
  x = self.upsample(x)
 
  # Classifier
- x = self.classifier(x)
- return x
+ return self.classifier(x)
 
 
 def _unet(arch: str, pretrained: bool, progress: bool, **kwargs: Any) -> UNet:

holocron/models/segmentation/unet3p.py

@@ -155,8 +155,7 @@ def forward(self, x: Tensor) -> Tensor:
  xs[idx] = self.decoder[idx](xs[:idx], xs[idx], xs[idx + 1 :])
 
  # Classifier
- x = self.classifier(xs[0])
- return x
+ return self.classifier(xs[0])
 
 
 def _unet(arch: str, pretrained: bool, progress: bool, **kwargs: Any) -> nn.Module:

holocron/nn/functional.py

@@ -133,9 +133,7 @@ def concat_downsample2d(x: Tensor, scale_factor: int) -> Tensor:
  # N * C * H * W --> N * C * (H/scale_factor) * scale_factor * (W/scale_factor) * scale_factor
  x = x.view(b, c, h // scale_factor, scale_factor, w // scale_factor, scale_factor)
  x = x.permute(0, 3, 5, 1, 2, 4).contiguous()
- x = x.view(b, int(c * scale_factor**2), h // scale_factor, w // scale_factor)
-
- return x
+ return x.view(b, int(c * scale_factor**2), h // scale_factor, w // scale_factor)
 
 
 def z_pool(x: Tensor, dim: int) -> Tensor:
@@ -364,9 +362,7 @@ def _xcorr2d(
  h = floor((h + (2 * padding[0]) - (dilation[0] * (weight.shape[-2] - 1)) - 1) / stride[0] + 1)
  w = floor((w + (2 * padding[1]) - (dilation[1] * (weight.shape[-1] - 1)) - 1) / stride[1] + 1)
 
- x = x.view(-1, weight.shape[0], h, w)
-
- return x
+ return x.view(-1, weight.shape[0], h, w)
 
 
 def _convNd(x: Tensor, weight: Tensor) -> Tensor:

holocron/nn/modules/activation.py

@@ -22,8 +22,7 @@ def __init__(self, inplace: bool = False) -> None:
  self.inplace = inplace
 
  def extra_repr(self) -> str:
- inplace_str = "inplace=True" if self.inplace else ""
- return inplace_str
+ return "inplace=True" if self.inplace else ""
 
 
 class HardMish(_Activation):
@@ -80,5 +79,4 @@ def __init__(self, in_channels: int, kernel_size: int = 3) -> None:
  def forward(self, x: Tensor) -> Tensor:
  out = self.conv(x)
  out = self.bn(out)
- x = torch.max(x, out)
- return x
+ return torch.max(x, out)

holocron/nn/modules/attention.py

@@ -74,5 +74,4 @@ def forward(self, x: Tensor) -> Tensor:
  x_h = cast(Tensor, self.h_branch(x))
  x_w = cast(Tensor, self.w_branch(x))
 
- out = (x_c + x_h + x_w) / 3
- return out
+ return (x_c + x_h + x_w) / 3

holocron/nn/modules/conv.py

@@ -496,6 +496,4 @@ def forward(self, x: Tensor) -> Tensor:
 
  # Multiply-Add operation
  # --> (N, C, H // s, W // s)
- out = (kernel * x_unfolded).sum(dim=3).view(*x.shape[:2], *kernel.shape[-2:])
-
- return out
+ return (kernel * x_unfolded).sum(dim=3).view(*x.shape[:2], *kernel.shape[-2:])

holocron/nn/modules/downsample.py

@@ -74,8 +74,7 @@ def forward(self, x: Tensor) -> Tensor:
  return x.view(x.size(0), x.size(1), -1).mean(-1).view(x.size(0), x.size(1), 1, 1)
 
  def extra_repr(self) -> str:
- inplace_str = "flatten=True" if self.flatten else ""
- return inplace_str
+ return "flatten=True" if self.flatten else ""
 
 
 class GlobalMaxPool2d(nn.Module):
@@ -97,13 +96,11 @@ def forward(self, x: Tensor) -> Tensor:
  return x.view(x.size(0), x.size(1), -1).max(-1).values.view(x.size(0), x.size(1), 1, 1)
 
  def extra_repr(self) -> str:
- inplace_str = "flatten=True" if self.flatten else ""
- return inplace_str
+ return "flatten=True" if self.flatten else ""
 
 
 def get_padding(kernel_size: int, stride: int = 1, dilation: int = 1) -> int:
- padding = ((stride - 1) + dilation * (kernel_size - 1)) // 2
- return padding
+ return ((stride - 1) + dilation * (kernel_size - 1)) // 2
 
 
 class BlurPool2d(nn.Module):

holocron/nn/modules/lambda_layer.py

@@ -105,5 +105,4 @@ def forward(self, x: torch.Tensor) -> torch.Tensor:
 
  Y = Yc + Yp
  # B x (H * W) x num_heads x dim_v -> B x (num_heads * dim_v) x H x W
- out = Y.permute(0, 2, 3, 1).reshape(b, self.num_heads * v.shape[2], h, w)
- return out
+ return Y.permute(0, 2, 3, 1).reshape(b, self.num_heads * v.shape[2], h, w)

references/clean_checkpoint.py

@@ -27,9 +27,7 @@ def parse_args():
 
  parser.add_argument("checkpoint", type=str, help="path to the training checkpoint")
  parser.add_argument("outfile", type=str, help="model")
- args = parser.parse_args()
-
- return args
+ return parser.parse_args()
 
 
 if __name__ == "__main__":