torch_conv_tbc function

Conv_tbc

torch_conv_tbc(self, weight, bias, pad = 0L)

Arguments

• self: NA input tensor of shape $(\mbox{sequence length} \times batch \times \mbox{in\_channels})$
• weight: NA filter of shape ($\mbox{kernel width} \times \mbox{in\_channels} \times \mbox{out\_channels}$)
• bias: NA bias of shape ($\mbox{out\_channels}$)
• pad: NA number of timesteps to pad. Default: 0

TEST

Applies a 1-dimensional sequence convolution over an input sequence. Input and output dimensions are (Time, Batch, Channels) - hence TBC.