mindspore.frombuffer

mindspore.frombuffer(buffer, dtype, count=- 1, offset=0)

Create a 1-D tensor from an object implements the buffer interface. The returned tensor shares the same memory space as the buffer.

Parameters

  • buffer (buffer) – An object that exposes the buffer interface.

  • dtype (mindspore.dtype) – The desired data type for the tensor.

  • count (int, optional) – The number of items to read. Default: -1 .

  • offset (int, optional) – The offset in bytes. Default: 0 .

Returns

Tensor, a 1-D tensor containing the data from the buffer.

Raises

  • RuntimeError – If count is a negative number less than -1.

  • RuntimeError – If the buffer length is less than or equal to 0, or if count is 0.

  • RuntimeError – If offset is not within the range [0, buffer length - 1].

  • RuntimeError – When count is -1, if (buffer length - offset) is not divisible by the element size.

  • RuntimeError – If the remaining buffer size after offset is insufficient to accommodate the requested number of elements.

Examples

>>> import mindspore
>>> data = bytearray([1, 2, 3, 4, 5, 6])
>>> tensor = mindspore.frombuffer(data, dtype=mindspore.uint8)
>>> print(tensor)
[1 2 3 4 5 6]