network_set_gradient_tensor_memory#

cuquantum.bindings.cutensornet.network_set_gradient_tensor_memory(
intptr_t handle,
intptr_t network_desc,
int64_t corresponding_tensor_id,
intptr_t buffer,
strides,
)[source]#

Provide memory buffer and strides of the gradient corresponding to tensorId of the network to be used for data writing.

Parameters:

  • handle (intptr_t) – Opaque handle holding cuTensorNet’s library context.

  • network_desc (intptr_t) – The network descriptor.

  • corresponding_tensor_id (int64_t) – The tensorId as returned by network_append_tensor() for the tensor whose gradient is to be computed.

  • buffer (intptr_t) – Pointer to memory bugger in device memory.

  • strides (object) –

    Array of size the number of modes of the corresponding tensor; strides[j] corresponding to the linearized offset – in physical memory – between two logically-neighboring elements w.r.t the j-th mode of the tensor. It can be:

    • an int as the pointer address to the array, or

    • a Python sequence of int64_t.

See also

cutensornetNetworkSetGradientTensorMemory()