here is the code from the Distance layer part because site is not allowing me to upload full code.
embedding = make_embedding() distance layer
class L1Dist(Layer): def __init__(self, **kwargs): super().__init__() def call(self, input_embedding, validation_embedding): return tf.math.abs(input_embedding - validation_embedding) def make_siamese_model(): input_image = Input(name='input_img', shape=(100,100,3)) validation_image = Input(name='validation_img', shape=(100,100,3)) siamese_layer = L1Dist() siamese_layer._name = 'distance' distances = siamese_layer(embedding(input_image), embedding(validation_image)) classifier = Dense(1, activation='sigmoid')(distances) return Model(inputs=[input_image, validation_image], outputs=classifier, name='SiameseNetwork') siamese_model = make_siamese_model() binary_cross_loss = tf.losses.BinaryCrossentropy() opt = tf.keras.optimizers.Adam(1e-4) # 0.0001 checkpoint_dir = './training_checkpoints' checkpoint_prefix = os.path.join(checkpoint_dir, 'ckpt') checkpoint = tf.train.Checkpoint(opt=opt, siamese_model=siamese_model) @tf.function def train_step(batch): with tf.GradientTape() as tape: X = batch[:2] y = batch[2] yhat = siamese_model(X, training=True) loss = binary_cross_loss(y, yhat) print(loss) grad = tape.gradient(loss, siamese_model.trainable_variables) opt.apply_gradients(zip(grad, siamese_model.trainable_variables)) return loss from tensorflow.python.keras.metrics import Precision, Recall def train(data, EPOCHS): for epoch in range(1, EPOCHS+1): print('\n Epoch {}/{}'.format(epoch, EPOCHS)) progbar = tf.keras.utils.Progbar(len(data)) r = Recall() p = Precision() for idx, batch in enumerate(data): # Run train step here loss = train_step(batch) yhat = siamese_model(batch[:2], training=False) r.update_state(batch[2], yhat) p.update_state(batch[2], yhat) progbar.update(idx+1) print(loss.numpy(), r.result().numpy(), p.result().numpy()) if epoch % 1 == 0: checkpoint.save(file_prefix=checkpoint_prefix) siamese_model.compile( optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'] ) EPOCHS = 2 train(train_data, EPOCHS) siamese_model.save('siamese_model.h5') new_model = tf.keras.models.load_model('siamese_model.h5',custom_objects={'L1Dist':L1Dist, 'binary_crossentropy':tf.losses.binary_crossentropy}) while reloading the model it is showing an ValueError: Unexpected object from deserialization, expected a layer or operation, got a <class 'main.L1Dist'>
please suggest changes in the code so that i could reload the siamsese model.
after trackbacking :
ValueError Traceback (most recent call last) Cell In[90], line 1 ----> 1 new_model = tf.keras.models.load_model('siamese_model.h5',custom_objects={'L1Dist':L1Dist(), 'binary_crossentropy':tf.losses.binary_crossentropy}) File ~\Desktop\Mazor Project\.venv\Lib\site-packages\keras\src\saving\saving_api.py:196, in load_model(filepath, custom_objects, compile, safe_mode) 189 return saving_lib.load_model( 190 filepath, 191 custom_objects=custom_objects, 192 compile=compile, 193 safe_mode=safe_mode, 194 ) 195 if str(filepath).endswith((".h5", ".hdf5")): --> 196 return legacy_h5_format.load_model_from_hdf5( 197 filepath, custom_objects=custom_objects, compile=compile 198 ) 199 elif str(filepath).endswith(".keras"): 200 raise ValueError( 201 f"File not found: filepath={filepath}. " 202 "Please ensure the file is an accessible `.keras` " 203 "zip file." 204 ) File ~\Desktop\Mazor Project\.venv\Lib\site-packages\keras\src\legacy\saving\legacy_h5_format.py:133, in load_model_from_hdf5(filepath, custom_objects, compile) 130 model_config = json_utils.decode(model_config) 132 with saving_options.keras_option_scope(use_legacy_config=True): --> 133 model = saving_utils.model_from_config( 134 model_config, custom_objects=custom_objects 135 ) 137 # set weights 138 load_weights_from_hdf5_group(f["model_weights"], model) File ~\Desktop\Mazor Project\.venv\Lib\site-packages\keras\src\legacy\saving\saving_utils.py:88, in model_from_config(config, custom_objects) 84 # TODO(nkovela): Swap find and replace args during Keras 3.0 release 85 # Replace keras refs with keras 86 config = _find_replace_nested_dict(config, "keras.", "keras.") ---> 88 return serialization.deserialize_keras_object( 89 config, 90 module_objects=MODULE_OBJECTS.ALL_OBJECTS, 91 custom_objects=custom_objects, 92 printable_module_name="layer", 93 ) File ~\Desktop\Mazor Project\.venv\Lib\site-packages\keras\src\legacy\saving\serialization.py:495, in deserialize_keras_object(identifier, module_objects, custom_objects, printable_module_name) 490 cls_config = _find_replace_nested_dict( 491 cls_config, "keras.", "keras." 492 ) 494 if "custom_objects" in arg_spec.args: --> 495 deserialized_obj = cls.from_config( 496 cls_config, 497 custom_objects={ 498 **object_registration.GLOBAL_CUSTOM_OBJECTS, 499 **custom_objects, 500 }, 501 ) 502 else: 503 with object_registration.CustomObjectScope(custom_objects): File ~\Desktop\Mazor Project\.venv\Lib\site-packages\keras\src\models\model.py:587, in Model.from_config(cls, config, custom_objects) 582 if is_functional_config and revivable_as_functional: 583 # Revive Functional model 584 # (but not Functional subclasses with a custom __init__) 585 from keras.src.models.functional import functional_from_config --> 587 return functional_from_config( 588 cls, config, custom_objects=custom_objects 589 ) 591 # Either the model has a custom __init__, or the config 592 # does not contain all the information necessary to 593 # revive a Functional model. This happens when the user creates (...) 596 # In this case, we fall back to provide all config into the 597 # constructor of the class. 598 try: File ~\Desktop\Mazor Project\.venv\Lib\site-packages\keras\src\models\functional.py:557, in functional_from_config(cls, config, custom_objects) 555 # First, we create all layers and enqueue nodes to be processed 556 for layer_data in functional_config["layers"]: --> 557 process_layer(layer_data) 559 # Then we process nodes in order of layer depth. 560 # Nodes that cannot yet be processed (if the inbound node 561 # does not yet exist) are re-enqueued, and the process 562 # is repeated until all nodes are processed. 563 while unprocessed_nodes: File ~\Desktop\Mazor Project\.venv\Lib\site-packages\keras\src\models\functional.py:528, in functional_from_config.<locals>.process_layer(layer_data) 524 layer = serialization_lib.deserialize_keras_object( 525 layer_data, custom_objects=custom_objects 526 ) 527 if not isinstance(layer, Operation): --> 528 raise ValueError( 529 "Unexpected object from deserialization, expected a layer or " 530 f"operation, got a {type(layer)}" 531 ) 532 created_layers[layer_name] = layer 534 # Gather layer inputs. ValueError: Unexpected object from deserialization, expected a layer or operation, got a <class '__main__.L1Dist'> ``` 