python - 如何复制和保存 keras adabound 模型？

Question

我正在使用 Kfold 进行 ML，我没有成功复制和保存 adabound keras 模型，但是，相同的代码对 adam 和 sgd 都很好，任何关于如何复制、保存和重用 adabound 模型的帮助都会很棒.

另外，我使用 model1 = deepcopy(model) 而不是使用 model1.save('my_class.h5') 保存它，并使用 new_model = models.load_model('my_class.h5') 预测加载它

for train, test in kfold.split(X, Y):
    # create model
    model = Sequential()
    print('model',type(model))    
    model.add(Dense(250, input_dim=train_inputs.shape[1], 
activation='relu'))
    model.add(Dropout(0.))
    model.add(Dense(250, activation='relu'))
    model.add(Dense(1, activation='linear'))
    model.compile(loss='mean_squared_error', optimizer=opti_Bound, 
 metrics=['accuracy'])
    history = model.fit(X[train], Y[train], \
            validation_data=(X[test], Y[test]), \
            epochs=2,batch_size=1000, verbose=2)
    _, train_acc = model.evaluate(X[train], Y[train], verbose=0)
    _, test_acc = model.evaluate(X[test], Y[test], verbose=0)
    print('Train: %.3f, Test: %.3f' % (train_acc, test_acc))  
    if test_acc>=accc_tst and train_acc>=accc_trn:
        accc_tst=test_acc
        accc_trn=train_acc
        model1=deepcopy(model) 

    plt.plot(history.history['acc'], label='train')
    plt.plot(history.history['val_acc'], label='test')
    plt.legend()
    plt.show()  
    cvscores.append(test_acc * 100)  

 print("%.2f%% (+/- %.2f%%)" % (np.mean(cvscores),np.std(cvscores))) 
 model1.save('my_class.h5') 

不久我得到了这个错误：

   ValueError                                Traceback (most recent 
call last)
<ipython-input-4-43690e87e00a> in <module>()
     34         accc_tst=test_acc
     35         accc_trn=train_acc
---> 36         model1=deepcopy(model)
     37 
     38     # plot history

/usr/lib/python3.6/copy.py in deepcopy(x, memo, _nil)
    178                     y = x
    179                 else:
--> 180                     y = _reconstruct(x, memo, *rv)
    181 
    182     # If is its own copy, don't memoize.

/usr/lib/python3.6/copy.py in _reconstruct(x, memo, func, args, 
state, listiter, dictiter, deepcopy)
    280             state = deepcopy(state, memo)
    281         if hasattr(y, '__setstate__'):
--> 282             y.__setstate__(state)
    283         else:
    284             if isinstance(state, tuple) and len(state) == 
2:

/usr/local/lib/python3.6/dist-packages/keras/engine/network.py in 
__setstate__(self, state)
   1264 
   1265     def __setstate__(self, state):
 -> 1266         model = saving.unpickle_model(state)
   1267         self.__dict__.update(model.__dict__)
   1268 

/usr/local/lib/python3.6/dist-packages/keras/engine/saving.py in 
unpickle_model(state)
    433 def unpickle_model(state):
    434     f = h5dict(state, mode='r')
--> 435     return _deserialize_model(f)
    436 
    437 

/usr/local/lib/python3.6/dist-packages/keras/engine/saving.py in 
_deserialize_model(f, custom_objects, compile)
   297         optimizer_config = 
training_config['optimizer_config']
    298         optimizer = 
optimizers.deserialize(optimizer_config,
--> 299                                            
custom_objects=custom_objects)
    300 
    301         # Recover loss functions and metrics.

/usr/local/lib/python3.6/dist-packages/keras/optimizers.py in 
deserialize(config, custom_objects)
    766                                     
 module_objects=all_classes,
    767                                     
 custom_objects=custom_objects,
--> 768                                     
 printable_module_name='optimizer')
    769 
    770 

/usr/local/lib/python3.6/dist-packages/keras/utils/generic_utils.py 
in deserialize_keras_object(identifier, module_objects, 
custom_objects, printable_module_name)
    136             if cls is None:
    137                 raise ValueError('Unknown ' + 
    printable_module_name +
--> 138                                  ': ' + class_name)
    139         if hasattr(cls, 'from_config'):
    140             custom_objects = custom_objects or {}

ValueError: Unknown optimizer: AdaBound

我怀疑，如果我尝试保存和加载该优化器，我也会得到错误。所以我实际上正在寻找使用 keras 复制、保存和重用\重新加载 adabound 的方法。

标签： pythonkeras