Fix env.py in cases where torch is not present (#31113)

* Fix env.py in cases where torch is not present

* Simplify the fix (and avoid some issues)

src/transformers/commands/env.py

@@ -133,6 +133,7 @@ class EnvironmentCommand(BaseTransformersCLICommand):
             "JaxLib version": f"{jaxlib_version}",
             "Using distributed or parallel set-up in script?": "<fill in>",
         }
+        if is_torch_available():
             if pt_cuda_available:
                 info["Using GPU in script?"] = "<fill in>"
                 info["GPU type"] = torch.cuda.get_device_name()