whenxuan: init the pyramid squeeze attention

Author: wwhenxuan

channel_attention/pyramid_squeeze_attention.py

@@ -0,0 +1,17 @@
+import torch
+from torch import nn
+
+
+class PyramidSqueezeAttention(nn.Module):
+    """
+    The Pyramid Squeeze Attention Module for Time Series (1D) or Image (2D) Analysis.
+    This module implements a pyramid squeeze-and-excitation attention mechanism.
+
+    References: "EPSANet: An Efficient Pyramid Squeeze Attention Block on Convolutional Neural Network" by Hu Zhang, et al.
+
+    URL: https://arxiv.org/abs/2105.14447
+    """
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        pass