broadcastAddVectorToMatrixGPU function

GPUTensor<Matrix> broadcastAddVectorToMatrixGPU(
  1. GPUTensor<Matrix> m,
  2. GPUTensor<Vector> v,
  3. CommandBuffer tape, {
  4. GPUTensor<Matrix>? outTensor,
})

Appends commands for addition of a 1D bias vector b to a 2D matrix m via broadcasting over the matrix to the tape. This command is inference only! No gradient path is calculated.

Implementation

GPUTensor<Matrix> broadcastAddVectorToMatrixGPU(
    GPUTensor<Matrix> m, GPUTensor<Vector> v, CommandBuffer tape,
    {GPUTensor<Matrix>? outTensor}) {
  GPUTensor<Matrix> out = outTensor ?? GPUTensor<Matrix>.empty(m.shape);

  tape.putInt(OP_BROADCAST_ADD);
  tape.putString(m.id);
  tape.putString(v.id);
  tape.putString(out.id);

  out.creator = GPUNode(
    <GPUTensor>[m, v],
    (CommandBuffer bTape) {
      // Backward pass logic (stubbed for inference)
    },
    opName: 'broadcastAddVectorToMatrixGPU',
    cost: m.shape[0] * m.shape[1],
  );

  return out;
}