Fix: added backend validation for dataset adapters across backends

keras/src/trainers/data_adapters/py_dataset_adapter.py

@@ -9,6 +9,7 @@
 
 import numpy as np
 
+from keras import backend
 from keras.src.api_export import keras_export
 from keras.src.trainers.data_adapters import data_adapter_utils
 from keras.src.trainers.data_adapters.data_adapter import DataAdapter
@@ -94,6 +95,21 @@ def __init__(self, workers=1, use_multiprocessing=False, max_queue_size=10):
         self._workers = workers
         self._use_multiprocessing = use_multiprocessing
         self._max_queue_size = max_queue_size
+        backend_name = backend.backend()
+        if backend_name not in ("torch", "jax", "tensorflow", "numpy"):
+            raise ValueError(
+                f"PyDataset supports tf, torch, jax, numpy backend"
+                f"Received unsupported backend: '{backend_name}'."
+            )
+        # Optionally warn if using TF (since tf.data.Dataset is better)
+        if backend_name == "tensorflow":
+            import warnings
+
+            warnings.warn(
+                "You are using PyDataset with the TensorFlow backend. "
+                "Consider using `tf.data.Dataset` for better performance.",
+                stacklevel=2,
+            )
 
     def _warn_if_super_not_called(self):
         warn = False

keras/src/trainers/data_adapters/tf_dataset_adapter.py

@@ -17,15 +17,28 @@ def __init__(self, dataset, class_weight=None, distribution=None):
                 shard the input dataset into per worker/process dataset
                 instance.
         """
+        import keras
         from keras.src.utils.module_utils import tensorflow as tf
 
+        # --- ✅ Backend compatibility check ---
+        backend = keras.backend.backend()
+        if backend not in ("tensorflow", "numpy", "torch", "jax"):
+            raise ValueError(
+                f"Incompatible backend '{backend}' for TFDatasetAdapter. "
+                "This adapter only supports the TensorFlow , numpy , torch ,"
+                " jax backend."
+            )
+
+        # --- ✅ Dataset type validation ---
         if not isinstance(
             dataset, (tf.data.Dataset, tf.distribute.DistributedDataset)
         ):
             raise ValueError(
-                "Expected argument `dataset` to be a tf.data.Dataset. "
+                "Expected argument `dataset` to be a tf.data.Dataset or "
+                "tf.distribute.DistributedDataset. "
                 f"Received: {dataset}"
             )
+
         if class_weight is not None:
             dataset = dataset.map(
                 make_class_weight_map_fn(class_weight)

keras/src/trainers/data_adapters/torch_data_loader_adapter.py

@@ -11,8 +11,19 @@ class TorchDataLoaderAdapter(DataAdapter):
     """Adapter that handles `torch.utils.data.DataLoader`."""
 
     def __init__(self, dataloader):
+        # --- ✅ Backend compatibility check ---
         import torch
 
+        import keras
+
+        backend = keras.backend.backend()
+        if backend not in ("torch", "tensorflow", "numpy", "jax"):
+            raise ValueError(
+                f"Incompatible backend '{backend}' for TorchDataLoaderAdapter. "
+                "This adapter only supports the PyTorch, tensorflow, jax, numpy"
+                " backend. "
+            )
+
         if not isinstance(dataloader, torch.utils.data.DataLoader):
             raise ValueError(
                 f"Expected argument `dataloader` to be an instance of"