Tensor.conv2d()

Returns the convolution of the tensor by the weights and bias.

Usage

from tinygrad.tensor import Tensor
 
a1 = Tensor([
    [[1, 2], [4, 5]],
    [[10, 11], [13, 14]]
])
weights = Tensor([
    [[1, 2], [4, 5]],
    [[10, 11], [13, 14]]
])
 
bias = Tensor([1, 2])
 
tensor = a1.conv2d(weights, bias)
 
print(tensor.numpy())

Return value

[[[ 47.]
  [156.]]
 
 [[155.]
  [588.]]]

conv_transpose2d cos

;