Cannot serialize socket object

[sskarkhanis]

Hello

I'm trying to use the TQDMNotebookCallback when training a keras model with pipelining. However, I'm unable to get it working.

my code is as follows

========================

from keras_tqdm import TQDMCallback, TQDMNotebookCallback from tqdm import tqdm

seed = 108

Function to create model, required for KerasClassifier

def create_large_wide_model():

# create model
model = Sequential()
model.add(Dense(800, input_dim=784, kernel_initializer ='uniform', activation='relu'))
model.add(Dense(64, kernel_initializer='uniform', activation='relu'))
model.add(Dense(10, kernel_initializer ='uniform', activation='softmax'))

# Compile model
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
return model

from numpy.random import seed seed(1)

from tensorflow import set_random_seed set_random_seed(108)

estimators = [] estimators.append(('standardize', StandardScaler())) estimators.append(('mlp', KerasClassifier(build_fn=create_large_wide_model, nb_epoch=250,
validation_split=0.15,batch_size=25, verbose=0,callbacks=[TQDMNotebookCallback()])))

pipeline = Pipeline(estimators)

%time results = pipeline.fit(X_train, y_train)

========================

the error message is as follows,

C:\Anaconda\lib\socket.py in getstate(self) 183 184 def getstate(self): --> 185 raise TypeError("Cannot serialize socket object") 186 187 def dup(self):

TypeError: Cannot serialize socket object

how do I use TQDMNotebookCallback when pipelining?