fully_connected

dragon.nn.fully_connected(
  inputs,
  axis=1,
  transpose_w=True,
  **kwargs
)[source]

Compute the dense matrix multiplication along the given axes.

\[y = Wx + b \]

The column of input matrix is determined by:

\[\text{Col} = \text{DimSince}(\text{Input}, \text{Axis}) \]
Parameters:
  • inputs (Sequence[dragon.Tensor]) – The tensor \(x\), \(W\) and \(b\).
  • axis (int, optional, default=1) – The start axis to compute, can be negative.
  • transpose_w (bool, optional, default=True) – True to transpose \(W\) before computation.
Returns:

dragon.Tensor – The output tensor.