Prevent BatchEncoding from blindly passing casts down to the tensors …

…it contains. Fixes #6582. (#8860)

Update src/transformers/tokenization_utils_base.py with review fix

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

src/transformers/tokenization_utils_base.py

@@ -776,7 +776,16 @@ def to(self, device: Union[str, "torch.device"]) -> "BatchEncoding":
             :class:`~transformers.BatchEncoding`: The same instance of :class:`~transformers.BatchEncoding` after
             modification.
         """
-        self.data = {k: v.to(device) for k, v in self.data.items()}
+
+        # This check catches things like APEX blindly calling "to" on all inputs to a module
+        # Otherwise it passes the casts down and casts the LongTensor containing the token idxs
+        # into a HalfTensor
+        if isinstance(device, str) or isinstance(device, torch.device):
+            self.data = {k: v.to(device=device) for k, v in self.data.items()}
+        else:
+            logger.warning(
+                f"Attempting to cast a BatchEncoding to another type, {str(device)}. This is not supported."
+            )
         return self