Fix for dtype and None dataset

sentence_transformers/data_collator.py

@@ -59,7 +59,7 @@ def __call__(self, features: list[dict[str, Any]]) -> dict[str, torch.Tensor]:
             for key, value in tokenized.items():
                 batch[f"{column_name}_{key}"] = value
             if prompt_len is not None:
-                batch[f"{column_name}_prompt_length"] = torch.Tensor(
+                batch[f"{column_name}_prompt_length"] = torch.tensor(
                     [prompt_len] * len(values), device=batch[f"{column_name}_input_ids"].device, dtype=torch.long
                 )
         return batch

sentence_transformers/trainer.py

@@ -878,10 +878,14 @@ def _load_from_checkpoint(self, checkpoint_path: str) -> None:
         loaded_model = SentenceTransformer(checkpoint_path, trust_remote_code=self.model.trust_remote_code)
         self.model.load_state_dict(loaded_model.state_dict())
 
-    def maybe_add_dataset_name_column(self, dataset_dict: DatasetDict | Dataset) -> DatasetDict | Dataset:
+    def maybe_add_dataset_name_column(
+        self, dataset_dict: DatasetDict | Dataset | None
+    ) -> DatasetDict | Dataset | None:
         """
         Check if the the dataset_name should be added to the dataset. True if the dataset and loss are Dict or the prompts are mapping to dataset names.
         """
+        if dataset_dict is None:
+            return None
         loss_is_dict = isinstance(self.loss, dict)
         dataset_is_dict = isinstance(dataset_dict, DatasetDict)