[Question] Can't load a model after training

Hi! Long time didn’t see 😃 I faced some issue: I used custom model for classification task and trained it, saved weights as h5. And now I can’t to load it to test. Try this: model.load_weights("model_weights.h5")

and gets:

AttributeError: 'LSTM_Dropout_Dense_Model' object has no attribute 'load_weights'

class LSTM_Dropout_Dense_Model(BaseClassificationModel):
is based on BaseClassificationModel I found in other issues here to use something like loaded_model.model.load_weights(''') but I get AttributeError: 'LSTM_Dropout_Dense_Model' object has no attribute 'model'

What can I do?

[MNCTTY]

I’ve never done it before, but a bit more experience and I’ll try! I try sometimes to accumulate my exp in small texts in my telegram blog, but it’s really rare and only for me and 2-3 people doing ± same tasks. Creating longreads very challenging for me! Never done it in english especially but I’ll try, in September, right after entry exams in the phd program because you develop really convenient, useful project – and it deserves to be more popular

[BrikerMan]

Got it. You need to build a model first to get the tf_model object, try this,

# save the trained model
model.save('model_path')

# load model
# This includes model struct and model wights
new_model = kashgari.utils.load_model('model_path')
# You can reload model weights now
new_model.tf_model.load_weights('xx.h5')