Validate parameters for keras.Model.fit()

PiperOrigin-RevId: 595627079

tf_keras/engine/data_adapter.py

@@ -1259,20 +1259,25 @@ def __init__(
             `Model` should always set this to `True`.
           pss_evaluation_shards: See `Model.fit`.
         """
+        if batch_size is not None:
+            _check_positive("batch_size", batch_size)
+        if steps_per_epoch not in (None, -1) and steps_per_epoch <= 0:
+            raise ValueError(
+                "steps_per_epoch must be positive, None or -1. Received "
+                f"{steps_per_epoch}. See `Model.fit`."
+            )
+        self._initial_epoch = _check_non_negative(
+            "initial_epoch", initial_epoch
+        )
+        _check_positive("max_queue_size", max_queue_size)
+        _check_positive("workers", workers)
+        if steps_per_execution is not None:
+            _check_positive("steps_per_execution", steps_per_execution)
 
-        self._initial_epoch = initial_epoch
         self._initial_step = 0
-        self._epochs = epochs
+        self._epochs = _check_positive("epochs", epochs)
         self._insufficient_data = False
         self._model = model
-
-        if steps_per_epoch == 0:
-            raise ValueError(
-                "Unexpected value for `steps_per_epoch`. Received value is 0. "
-                "Please check the docstring for `model.fit()` for supported "
-                "values."
-            )
-
         self._steps_per_epoch = steps_per_epoch
 
         # `steps_per_execution_value` is the cached initial value.
@@ -1954,6 +1959,22 @@ def _check_data_cardinality(data):
         raise ValueError(msg)
 
 
+def _check_non_negative(name, value):
+    if value < 0:
+        raise ValueError(
+            f"Expected {name} to be non-negative. Received is {value}."
+        )
+    return value
+
+
+def _check_positive(name, value):
+    if value <= 0:
+        raise ValueError(
+            f"Expected {name} to be positive. Received is {value}."
+        )
+    return value
+
+
 def _get_tensor_types():
     if pd is None:
         return (tf.Tensor, np.ndarray)

tf_keras/engine/data_adapter_test.py

@@ -1447,7 +1447,7 @@ def test_error_if_zero_steps_per_epoch(self):
 
         with self.assertRaisesRegex(
             ValueError,
-            "Unexpected value for `steps_per_epoch`. Received value is 0.",
+            "steps_per_epoch must be positive, None or -1. Received 0.",
         ):
             data_adapter.DataHandler(
                 data, initial_epoch=0, epochs=2, steps_per_epoch=0

tf_keras/engine/training_generator_test.py

@@ -130,7 +130,7 @@ def test_fit_generator_method(self):
             steps_per_epoch=5,
             validation_data=custom_generator(),
             validation_steps=1,
-            workers=0,
+            workers=1,
         )
 
     @test_combinations.run_with_all_model_types
@@ -166,7 +166,7 @@ def test_evaluate_generator_method(self):
             steps=5,
             max_queue_size=10,
             use_multiprocessing=False,
-            workers=0,
+            workers=1,
         )
 
     @test_combinations.run_with_all_model_types
@@ -192,7 +192,7 @@ def test_predict_generator_method(self):
             use_multiprocessing=False,
         )
         model.predict_generator(
-            custom_generator(), steps=5, max_queue_size=10, workers=0
+            custom_generator(), steps=5, max_queue_size=10, workers=1
         )
         # Test generator with just inputs (no targets)
         model.predict_generator(
@@ -209,7 +209,7 @@ def test_predict_generator_method(self):
             use_multiprocessing=False,
         )
         model.predict_generator(
-            custom_generator(mode=1), steps=5, max_queue_size=10, workers=0
+            custom_generator(mode=1), steps=5, max_queue_size=10, workers=1
         )
 
     @test_combinations.run_with_all_model_types
@@ -453,7 +453,7 @@ def __len__(self):
             validation_data=custom_generator(),
             validation_steps=1,
             max_queue_size=10,
-            workers=0,
+            workers=1,
             use_multiprocessing=True,
         )
         model.fit_generator(
@@ -462,7 +462,7 @@ def __len__(self):
             validation_data=custom_generator(),
             validation_steps=1,
             max_queue_size=10,
-            workers=0,
+            workers=1,
             use_multiprocessing=False,
         )