I follow the Nicholas Renotte tutorial "Build a Deep Facial Recognition App"(Python). But at the part 4 I faced a problem, here is the code:

Siamese L1 Distance class

class L1Dist(Layer):    # Init method - inheritance    def __init__(self, **kwargs):        super().__init__()    # Magic happens here - similarity calculation    def call(self, input_embedding, validation_embedding):        return tf.math.abs(input_embedding - validation_embedding)

TypeError: unsupported operand type(s) for -: 'list' and 'list'

In the video all is fine, but in my case function can't do the substraction (input_embedding - validation_embedding)

Arguments received by L1Dist.call():args=(['<KerasTensor shape=(None, 4096), dtype=float32, sparse=False, name=keras_tensor_18>'], ['<KerasTensor shape=(None, 4096), dtype=float32, sparse=False, name=keras_tensor_19>'])

Tried to modify:

def call(self, input_embedding, validation_embedding):        input_embedding = tf.convert_to_tensor(input_embedding)        validation_embedding = tf.convert_to_tensor(validation_embedding)        input_embedding = tf.squeeze(input_embedding, axis=0)  # Remove potential first dimension        validation_embedding = tf.squeeze(validation_embedding, axis=0)        return tf.math.abs(input_embedding - validation_embedding)

But failed

line 108, in convert_to_eager_tensor    return ops.EagerTensor(value, ctx.device_name, dtype)ValueError: TypeError: object of type 'KerasTensor' has no len()

Tried tf.keras.layers.Subtract()([input_embedding, validation_embedding])But AttributeError: Exception encountered when calling Subtract.call().

'list' object has no attribute 'shape'

With keras.ops.subtract(input_embedding, validation_embedding)faced: ValueError(f"Invalid dtype: {dtype}")ValueError: Invalid dtype: list