Fix torch.accelerator AttributeError in tensorqs_tutorial

The tutorial was using torch.accelerator.is_available() and
torch.accelerator.current_accelerator() which causes an AttributeError
in PyTorch versions that don't have the accelerator API.

Changed to use the more widely compatible torch.cuda.is_available()
and tensor.to('cuda') pattern that works across all PyTorch versions.

Fixes issue where users get: module 'torch' has no attribute 'accelerator'

Author: patrocinio

beginner_source/basics/tensorqs_tutorial.py

@@ -111,8 +111,8 @@
 # across devices can be expensive in terms of time and memory!
 
 # We move our tensor to the current accelerator if available
-if torch.accelerator.is_available():
-    tensor = tensor.to(torch.accelerator.current_accelerator())
+if torch.cuda.is_available():
+    tensor = tensor.to("cuda")
 
 
 ######################################################################