How to save the best training model?

[zhoushuyi-22]

Dear professor,

How to save the best training model?
I tried to add the following code like Keras during training, but it got an error
('SciKernelInitializer' object has no attribute 'bias')

model_checkpoint = ModelCheckpoint(location, monitor='loss', save_best_only=True, mode='auto')
model.train([x1, x2], [y, ly], epochs=200, batch_size=64, callbacks=[model_checkpoint])

kind regards,
shuyi zhou

[ehsanhaghighat]

Hi Shuyi, The keyword is save_weights and it automatically adds the right callback to the stack. Best, Ehsan

…

On Sep 6, 2022, at 11:45 PM, gouzi1209 ***@***.***> wrote: Dear professor, How to save the best training model? I tried to add the following code like Keras during training, but it got an error ('SciKernelInitializer' object has no attribute 'bias') model_checkpoint = ModelCheckpoint(location, monitor='loss', save_best_only=True, mode='auto') model.train([x1, x2], [y, ly], epochs=200, batch_size=64, callbacks=[model_checkpoint]) kind regards, shuyi zhou — Reply to this email directly, view it on GitHub <https://github.com/sciann/sciann/issues/52>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHMBIZ5E6GUYDJWMSUUUU2DV5A22HANCNFSM6AAAAAAQGOPIAE>. You are receiving this because you are subscribed to this thread.

[zhoushuyi-22]

Dear professor,

I made the following changes in the model.train, but the following error occurs.

(model.train([x1, x2], [y, ly], epochs=200, batch_size=64, save_weights={'best': True, 'path': '/home/zhousy/'}))

local variable 'model_check_point' referenced before assignment

I hope you can give me some guidance.

best regards,
shuyi zhou