[AIR] Rename DLPredictor.call_model tensor parameter to inputs

python/ray/train/_internal/dl_predictor.py

@@ -52,13 +52,14 @@ def _tensor_to_array(self, tensor: TensorType) -> np.ndarray:
         raise NotImplementedError
 
     @abc.abstractmethod
+    @DeveloperAPI
     def call_model(
-        self, tensor: Union[TensorType, Dict[str, TensorType]]
+        self, inputs: Union[TensorType, Dict[str, TensorType]]
     ) -> Union[TensorType, Dict[str, TensorType]]:
         """Inputs the tensor to the model for this Predictor and returns the result.
 
         Args:
-            tensor: The tensor to input to the model.
+            inputs: The tensor to input to the model.
 
         Returns:
             A tensor or dictionary of tensors containing the model output.

python/ray/train/tensorflow/tensorflow_predictor.py

@@ -10,7 +10,7 @@
 from ray.air._internal.tensorflow_utils import convert_ndarray_batch_to_tf_tensor_batch
 from ray.train._internal.dl_predictor import DLPredictor
 from ray.train.tensorflow.tensorflow_checkpoint import TensorflowCheckpoint
-from ray.util.annotations import PublicAPI
+from ray.util.annotations import DeveloperAPI, PublicAPI
 
 if TYPE_CHECKING:
     from ray.data.preprocessor import Preprocessor
@@ -107,8 +107,9 @@ def from_checkpoint(
             use_gpu=use_gpu,
         )
 
+    @DeveloperAPI
     def call_model(
-        self, tensor: Union[tf.Tensor, Dict[str, tf.Tensor]]
+        self, inputs: Union[tf.Tensor, Dict[str, tf.Tensor]]
     ) -> Union[tf.Tensor, Dict[str, tf.Tensor]]:
         """Runs inference on a single batch of tensor data.
 
@@ -130,8 +131,8 @@ def build_model() -> tf.keras.Model:
 
                 # Use a custom predictor to format model output as a dict.
                 class CustomPredictor(TensorflowPredictor):
-                    def call_model(self, tensor):
-                        model_output = super().call_model(tensor)
+                    def call_model(self, inputs):
+                        model_output = super().call_model(inputs)
                         return {
                             str(i): model_output[i] for i in range(len(model_output))
                         }
@@ -140,18 +141,18 @@ def call_model(self, tensor):
                 predictions = predictor.predict(data_batch)
 
         Args:
-            tensor: A batch of data to predict on, represented as either a single
-                PyTorch tensor or for multi-input models, a dictionary of tensors.
+            inputs: A batch of data to predict on, represented as either a single
+                TensorFlow tensor or for multi-input models, a dictionary of tensors.
 
         Returns:
             The model outputs, either as a single tensor or a dictionary of tensors.
 
         """
         if self.use_gpu:
             with tf.device("GPU:0"):
-                return self._model(tensor)
+                return self._model(inputs)
         else:
-            return self._model(tensor)
+            return self._model(inputs)
 
     def predict(
         self,

python/ray/train/torch/torch_predictor.py

@@ -10,7 +10,7 @@
 from ray.air._internal.torch_utils import convert_ndarray_batch_to_torch_tensor_batch
 from ray.train.torch.torch_checkpoint import TorchCheckpoint
 from ray.train._internal.dl_predictor import DLPredictor
-from ray.util.annotations import PublicAPI
+from ray.util.annotations import DeveloperAPI, PublicAPI
 
 if TYPE_CHECKING:
     from ray.data.preprocessor import Preprocessor
@@ -94,8 +94,9 @@ def from_checkpoint(
         preprocessor = checkpoint.get_preprocessor()
         return cls(model=model, preprocessor=preprocessor, use_gpu=use_gpu)
 
+    @DeveloperAPI
     def call_model(
-        self, tensor: Union[torch.Tensor, Dict[str, torch.Tensor]]
+        self, inputs: Union[torch.Tensor, Dict[str, torch.Tensor]]
     ) -> Union[torch.Tensor, Dict[str, torch.Tensor]]:
         """Runs inference on a single batch of tensor data.
 
@@ -106,7 +107,7 @@ def call_model(
         output.
 
         Args:
-            tensor: A batch of data to predict on, represented as either a single
+            inputs: A batch of data to predict on, represented as either a single
                 PyTorch tensor or for multi-input models, a dictionary of tensors.
 
         Returns:
@@ -124,8 +125,8 @@ def forward(self, input_tensor):
 
                 # Use a custom predictor to format model output as a dict.
                 class CustomPredictor(TorchPredictor):
-                    def call_model(self, tensor):
-                        model_output = super().call_model(tensor)
+                    def call_model(self, inputs):
+                        model_output = super().call_model(inputs)
                         return {
                             str(i): model_output[i] for i in range(len(model_output))
                         }
@@ -142,7 +143,7 @@ def call_model(self, tensor):
                 Predictions: [1 2], [1 2]
         """
         with torch.no_grad():
-            output = self.model(tensor)
+            output = self.model(inputs)
         return output
 
     def predict(