bugfix deserializer,test_compare_onnx_model

docs/Examples/tf_image_classification_Indoorscenes_dataset.py

@@ -41,15 +41,23 @@
 
 ############################################################################
 
-openml_tensorflow.config.epoch = 10 #  small epoch for test runs
+openml_tensorflow.config.epoch = 1 #  small epoch for test runs
 
 IMG_SIZE = (128, 128)
 IMG_SHAPE = IMG_SIZE + (3,)
 base_learning_rate = 0.0001
 
-# dataset = openml.datasets.get_dataset(45936, download_all_files=True, download_data=True)
-
 # Toy example
+datagen = ImageDataGenerator(
+    rotation_range=25,
+    width_shift_range=0.01,
+    height_shift_range=0.01,
+    brightness_range=(0.9, 1.1),
+    zoom_range=0.1,
+    horizontal_flip=True,
+    vertical_flip=True,
+)
+
 openml_tensorflow.config.datagen = datagen
 openml_tensorflow.config.dir = openml.config.get_cache_directory()+'/datasets/45936/Images/'
 openml_tensorflow.config.x_col = "Filename"
@@ -69,16 +77,6 @@
 ############################################################################
 # Large CNN
 
-datagen = ImageDataGenerator(
-    rotation_range=25,
-    width_shift_range=0.01,
-    height_shift_range=0.01,
-    brightness_range=(0.9, 1.1),
-    zoom_range=0.1,
-    horizontal_flip=True,
-    vertical_flip=True,
-)
-
 IMG_SIZE = 128
 NUM_CLASSES = 67
 
@@ -144,7 +142,7 @@
 # Careful to not call this function when another run_model_on_task is called in between, 
 # as during publish later, only the last trained model (from last run_model_on_task call) is uploaded.   
 run = openml_tensorflow.add_onnx_to_run(run)
-# breakpoint()
+
 run.publish()
 
 print('URL for run: %s/run/%d?api_key=%s' % (openml.config.server, run.run_id, openml.config.apikey))

openml_tensorflow/extension.py

@@ -430,8 +430,8 @@ def _from_parameters(self, parameters: 'OrderedDict[str, Any]') -> Any:
             # Recover loss functions and metrics
             loss = training_config['loss']
             metrics = training_config['metrics']
-            sample_weight_mode = training_config['sample_weight_mode']
-            loss_weights = training_config['loss_weights']
+            sample_weight_mode = training_config.get('sample_weight_mode', None)
+            loss_weights = training_config.get('loss_weights', None)
 
             # Compile model
             model.compile(optimizer=optimizer,

tests/test_compare_onnx_models.py

@@ -1,5 +1,16 @@
+"""
+Below pytest test compares two onnx models for identical structure and parameters.  
+"""
 import onnx
 import numpy as np
+import os
+import pytest
+
+# Paths to ONNX models
+MODEL_PATH1 = "model1.onnx"
+MODEL_PATH2 = "model2.onnx"
+
+# Helper function to compare graph structures
 def compare_graphs(graph1, graph2):
     nodes1 = graph1.node
     nodes2 = graph2.node
@@ -17,12 +28,12 @@ def compare_graphs(graph1, graph2):
     print("Graph structures are identical.")
     return True
 
-def compare_models(model_path1, model_path2):
+# Helper function to compare model parameters
+def compare_parameters(model_path1, model_path2):
     # Load ONNX models
     model1 = onnx.load(model_path1)
     model2 = onnx.load(model_path2)
 
-    # Compare graph structures
     # Compare graph structures
     if not compare_graphs(model1.graph, model2.graph):
         return False
@@ -40,9 +51,19 @@ def compare_models(model_path1, model_path2):
     print("Models are identical.")
     return True
 
-# Paths to ONNX models
-model_path1 = "model1.onnx"
-model_path2 = "model2.onnx"
 
-# Compare models
-compare_models(model_path1, model_path2)
+@pytest.mark.skipif(
+    not (os.path.exists("model1.onnx") and os.path.exists("model2.onnx")),
+    reason="ONNX models are not available"
+)
+def test_compare_onnx_models():
+    # Load ONNX models
+    model1 = onnx.load(MODEL_PATH1)
+    model2 = onnx.load(MODEL_PATH2)
+
+    # Perform comparisons
+    assert compare_graphs(model1.graph, model2.graph), "Graph structures are different."
+    assert compare_parameters(model1, model2), "Model parameters are different."
+
+    print("ONNX models are identical.")
+