MCPcopy Create free account
hub / github.com/MegEngine/MegEngine / dot

Function dot

imperative/python/megengine/functional/math.py:829–854  ·  view source on GitHub ↗

r"""Computes dot-product of two vectors ``inp1`` and ``inp2``. inputs must be 1-dimensional or scalar. A scalar input is automatically broadcasted. Refer to :func:`~.matmul` for more general usage. Args: inp1: first vector. inp2: second vector. Returns: outp

(inp1: Tensor, inp2: Tensor)

Source from the content-addressed store, hash-verified

827
828
829def dot(inp1: Tensor, inp2: Tensor) -> Tensor:
830 r"""Computes dot-product of two vectors ``inp1`` and ``inp2``.
831 inputs must be 1-dimensional or scalar. A scalar input is automatically broadcasted.
832 Refer to :func:`~.matmul` for more general usage.
833
834 Args:
835 inp1: first vector.
836 inp2: second vector.
837
838 Returns:
839 output value.
840
841 Examples:
842 >>> import numpy as np
843 >>> data1 = Tensor(np.arange(0, 6, dtype=np.float32))
844 >>> data2 = Tensor(np.arange(0, 6, dtype=np.float32))
845 >>> out = F.dot(data1, data2)
846 >>> out.numpy()
847 array(55., dtype=float32)
848 """
849 op = builtin.Dot()
850 assert (
851 inp1.ndim <= 1 and inp2.ndim <= 1
852 ), "Input tensors for dot must be 1-dimensional or scalar"
853 (result,) = apply(op, inp1, inp2)
854 return result
855
856
857def svd(inp: Tensor, full_matrices=False, compute_uv=True) -> Tensor:

Callers

nothing calls this directly

Calls 2

DotMethod · 0.80
applyFunction · 0.50

Tested by

no test coverage detected