harygo2

Results 1 issues of harygo2

Creating a Torch tensor with the parameter `device=get_accelerator().current_device()` can result in a crash when using an NPU. This issue arises because the `current_device` API across all accelerators is expected to...