2D convolution with separable filters. Arguments: x: input tensor depthwise_kernel: convolution kernel for the depthwise convolution. strides: strides tuple (length 2). padding: string, `"same"` or `"valid"`. data_format: string, `"channels_last"` or `"channels_first
(x,
depthwise_kernel,
strides=(1, 1),
padding='valid',
data_format=None,
dilation_rate=(1, 1))
source not stored for this graph (policy: none)
nothing calls this directly
no test coverage detected