How to perform validation while performing Sciann training

[ssingh-ipa]

Hello,

Unlike keras, sciann does not have an option for validation_split. Additionally the argument validation_data gives the following error:

_> C:\Users\Anaconda3\lib\site-packages\keras\engine\training_v1.py:2045: UserWarning: Model.state_updates will be removed in a future version. This property should not be used in TensorFlow 2.0, as updates are applied automatically.

updates = self.state_updates

InvalidArgumentError Traceback (most recent call last)
Input In [30], in <cell line: 2>()
1 mod = sn.SciModel([x, tao], [PDE1_neg,IC,BC1,BC2], loss_func="mse", optimizer="adam")
----> 2 hist = mod.train([x_in,tao_in],
3 4*['zero'],
4 learning_rate=0.001,
5 epochs=500,
6 callbacks=[my_callback],
7 stop_loss_value=1e-10,
8 reduce_lr_after=10,
9 stop_lr_value=1e-8,
10 verbose=1,
11 batch_size=256,
12 shuffle=True,
13 validation_data=[x_test,t_test],
14 )
InvalidArgumentError: You must feed a value for placeholder tensor 'mul_5_target_2' with dtype double and shape [?,?]
[[{{node mul_5_target_2}}]]_

Please let me know if it is possible to perform validation and plot the model accuracy along with the validation loss using SciANN. This is necessary in order to avoid overfitting.