Fix keras model saving error into TF SavedModel format.

Now that keras 3 doesn't support model saving format TF SavedModel directly via `model.save()`, we need to use `tf.saved_model.save()` to save the model. PiperOrigin-RevId: 634985205
tensorflow · May 18, 2024 · 12dfa7e · 12dfa7e
1 parent 52da3cb
commit 12dfa7e
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/tensorflow/lite/python/util_test.py b/tensorflow/lite/python/util_test.py
@@ -274,7 +274,7 @@ def _generate_integer_tflite_model(quantization_type=dtypes.int8,
     # Convert TF Model to an Integer Quantized TFLite Model
     converter = tf.lite.TFLiteConverter.from_keras_model(model)
   else:
-    model.save(saved_model_dir)
+    tf.saved_model.save(model, saved_model_dir)
     converter = tf.lite.TFLiteConverter.from_saved_model(saved_model_dir)
   converter.optimizations = {tf.lite.Optimize.DEFAULT}