Hey, that's a really nice implementation =)
I'm using tf2.3.1 and training works perfectly but when I save and load a model I get this error message:
File "D:\virtualenv\tensorflow2\lib\site-packages\tensorflow\python\ops\init_ops_v2.py", line 90, in from_config
return cls(**config)
TypeError: init() got an unexpected keyword argument 'scale'
Before loading the model I'm already setting up the custom objects with:
tf.keras.utils.get_custom_objects().update({
'Sine': Sine,
'SIRENInitializer': SIRENInitializer,
'SIRENFirstLayerInitializer': SIRENFirstLayerInitializer,
'SinusodialRepresentationDense': SinusodialRepresentationDense,
'ScaledSinusodialRepresentationDense': ScaledSinusodialRepresentationDense,
})
model = tf.keras.models.load_model(str(modelPath))
I'm only using SinusodialRepresentationDense in my model and there is indeed a scale variable passed to tf.keras.initializers.VarianceScaling in the SIRENInitializer, but I'm not sure where the problem is =(
Hey, that's a really nice implementation =) I'm using tf2.3.1 and training works perfectly but when I save and load a model I get this error message:
File "D:\virtualenv\tensorflow2\lib\site-packages\tensorflow\python\ops\init_ops_v2.py", line 90, in from_config return cls(**config) TypeError: init() got an unexpected keyword argument 'scale'
Before loading the model I'm already setting up the custom objects with: tf.keras.utils.get_custom_objects().update({ 'Sine': Sine, 'SIRENInitializer': SIRENInitializer, 'SIRENFirstLayerInitializer': SIRENFirstLayerInitializer, 'SinusodialRepresentationDense': SinusodialRepresentationDense, 'ScaledSinusodialRepresentationDense': ScaledSinusodialRepresentationDense, }) model = tf.keras.models.load_model(str(modelPath))
I'm only using SinusodialRepresentationDense in my model and there is indeed a scale variable passed to tf.keras.initializers.VarianceScaling in the SIRENInitializer, but I'm not sure where the problem is =(
I hoped you could help me out here. Cheers, Flo