This forum is for reporting errors with the Training process. If you want to get tips, or better understand the Training process, then you should look in the Training Discussion forum.
Please mark any answers that fixed your problems so others can find the solutions.
I first loaded stojo preset in configure settings at Model (at Train-Model-Phaze A tab)
then i created new model and quitted after several iterations so that the model was saved succesfully.
but every time I start to train the same model, information says below. (though I didn't changed anything)
02/13/2023 18:39:10 INFO Loading Model from Phaze_A plugin...
02/13/2023 18:39:10 INFO Config item: 'freeze_layers' has been updated from '['efficientnetb4']' to '['keras_encoder']'
I know that this information is shown when config is changed.. but something's going wrong regarding the message.
maybe 'freeze_layers' of stojo setting cannot be saved as same as preset ?
That's not an error, its standard for Phaze-A models.
It's to do with the way the config system works, and the fact that I didn't want to give an option to freeze every single keras encoder in the plugin (too many options, doesn't make sense). When you select "Keras Encoder" for freeze/load weights behind the scenes it rewrites it to the actual encoder used. The config system complains at this, but it isn't a failure. It is working as intended.
That's not an error, its standard for Phaze-A models.
It's to do with the way the config system works, and the fact that I didn't want to give an option to freeze every single keras encoder in the plugin (too many options, doesn't make sense). When you select "Keras Encoder" for freeze/load weights behind the scenes it rewrites it to the actual encoder used. The config system complains at this, but it isn't a failure. It is working as intended.