ResCNN

This is an unofficial PyTorch implementation by Ignacio Oguiza - oguiza@timeseriesAI.co

from tsai.models.utils import *

---

source

ResCNN

 ResCNN (c_in, c_out, coord=False, separable=False, zero_norm=False)

Same as nn.Module, but no need for subclasses to call super().__init__

xb = torch.rand(16, 3, 10)
test_eq(ResCNN(3,2,coord=True, separable=True)(xb).shape, [xb.shape[0], 2])
test_eq(count_parameters(ResCNN(3,2)), 257283)

ResCNN(3,2,coord=True, separable=True)

ResCNN(
  (block1): _ResCNNBlock(
    (convblock1): ConvBlock(
      (0): AddCoords1d()
      (1): SeparableConv1d(
        (depthwise_conv): Conv1d(4, 4, kernel_size=(7,), stride=(1,), padding=(3,), groups=4, bias=False)
        (pointwise_conv): Conv1d(4, 64, kernel_size=(1,), stride=(1,), bias=False)
      )
      (2): BatchNorm1d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (3): ReLU()
    )
    (convblock2): ConvBlock(
      (0): AddCoords1d()
      (1): SeparableConv1d(
        (depthwise_conv): Conv1d(65, 65, kernel_size=(5,), stride=(1,), padding=(2,), groups=65, bias=False)
        (pointwise_conv): Conv1d(65, 64, kernel_size=(1,), stride=(1,), bias=False)
      )
      (2): BatchNorm1d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (3): ReLU()
    )
    (convblock3): ConvBlock(
      (0): AddCoords1d()
      (1): SeparableConv1d(
        (depthwise_conv): Conv1d(65, 65, kernel_size=(3,), stride=(1,), padding=(1,), groups=65, bias=False)
        (pointwise_conv): Conv1d(65, 64, kernel_size=(1,), stride=(1,), bias=False)
      )
      (2): BatchNorm1d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    )
    (shortcut): ConvBlock(
      (0): AddCoords1d()
      (1): Conv1d(4, 64, kernel_size=(1,), stride=(1,), bias=False)
      (2): BatchNorm1d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    )
    (add): Add
    (act): ReLU()
  )
  (block2): ConvBlock(
    (0): AddCoords1d()
    (1): SeparableConv1d(
      (depthwise_conv): Conv1d(65, 65, kernel_size=(3,), stride=(1,), padding=(1,), groups=65, bias=False)
      (pointwise_conv): Conv1d(65, 128, kernel_size=(1,), stride=(1,), bias=False)
    )
    (2): BatchNorm1d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (3): LeakyReLU(negative_slope=0.2)
  )
  (block3): ConvBlock(
    (0): AddCoords1d()
    (1): SeparableConv1d(
      (depthwise_conv): Conv1d(129, 129, kernel_size=(3,), stride=(1,), padding=(1,), groups=129, bias=False)
      (pointwise_conv): Conv1d(129, 256, kernel_size=(1,), stride=(1,), bias=False)
    )
    (2): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (3): PReLU(num_parameters=1)
  )
  (block4): ConvBlock(
    (0): AddCoords1d()
    (1): SeparableConv1d(
      (depthwise_conv): Conv1d(257, 257, kernel_size=(3,), stride=(1,), padding=(1,), groups=257, bias=False)
      (pointwise_conv): Conv1d(257, 128, kernel_size=(1,), stride=(1,), bias=False)
    )
    (2): BatchNorm1d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (3): ELU(alpha=0.3)
  )
  (gap): AdaptiveAvgPool1d(output_size=1)
  (squeeze): Squeeze(dim=-1)
  (lin): Linear(in_features=128, out_features=2, bias=True)
)

check_weight(ResCNN(3,2, zero_norm=True), is_bn)

(array([1., 1., 0., 1., 1., 1., 1.], dtype=float32),
 array([0., 0., 0., 0., 0., 0., 0.], dtype=float32))