requirement

.github/workflows/pylint.yml

@@ -17,6 +17,7 @@ jobs:
  - name: Install dependencies
  run: |
  python -m pip install --upgrade pip
+ pip install -r ../../requirements.txt
  pip install pylint
  - name: Analysing the code with pylint
  run: |

models/attention.py

@@ -1,12 +1,8 @@
 """_summary_
-
-Returns:
- _type_: _description_
 """
 import torch
 import torch.nn as nn
 from fancy_einsum import einsum
-from einops import rearrange
 
 
 class R3DAttention(nn.Module):
@@ -34,9 +30,10 @@ def forward(self, x: torch.Tensor, mask = None) -> torch.Tensor:
  Returns:
  torch.Tensor: (batch*view, patch, embedding)
  """
- qkv = self.qkv_proj(x)
+ b, p, _ = x.shape
+ qkv = self.qkv_proj(x).reshape(b, p, 3, self.n_head, self.head_size)
+ q, k, v = qkv.chunk(dim=2)
  #b-batch, p-patch, c-constant(3), n-num_heads, s-head_size
- q, k, v = rearrange(qkv, "b p (c n s)-> c b n p s", c=3, n=self.n_head)
  attn_score = einsum("b n pq s, b n pk s->b n pq pk", q, k) / (
  self.head_size**0.5
  )
@@ -51,8 +48,8 @@ def forward(self, x: torch.Tensor, mask = None) -> torch.Tensor:
  return out
 
 
-if __name__ == "__main__":
- attn = R3DAttention(64, 4)
- def count_par(model):
- return sum(p.numel() for p in model.parameters() if p.requires_grad)
- print(count_par(attn))
+# if __name__ == "__main__":
+#  attn = R3DAttention(64, 4)
+#  def count_par(model):
+#  return sum(p.numel() for p in model.parameters() if p.requires_grad)
+#  print(count_par(attn))

requirements.txt

@@ -0,0 +1,9 @@
+fancy_einsum
+torch
+torchvision
+tqdm
+scikit-learn
+matplotlib
+opencv-python
+scipy
+numpy