Save the date! Google I/O returns May 18-20 Register now


Returns the dlpack capsule representing the tensor.

This operation ensures the underlying data memory is ready when returns.

  a = tf.tensor([1, 10])
  dlcapsule = tf.experimental.dlpack.to_dlpack(a)
  # dlcapsule represents the dlpack data structure

tf_tensor Tensorflow eager tensor, to be converted to dlpack capsule.

A PyCapsule named as dltensor, which shares the underlying memory to other framework. This PyCapsule can be consumed only once.