torch.Tensor.share_memory_

Tensor.share_memory_() [source]

Moves the underlying storage to shared memory.

This is a no-op if the underlying storage is already in shared memory and for CUDA tensors. Tensors in shared memory cannot be resized.