what are the system requirements for compiling this code?
how much time it will take to compile?

Hello, i'm trying the code and i found an error that i can not fix in file "Train CNN IMDB.ipynb":

Traceback (most recent call last):
File "/home/ch//Sandbox/keras/kaggle_sentAnalysis/cnn_kaggle.py", line 136, in
output = model.predict_proba(val_X, batch_size=10, verbose=1)
File "/usr/local/lib/python2.7/dist-packages/keras/models.py", line 791, in predict_proba
if preds.min() < 0. or preds.max() > 1.:
AttributeError: 'list' object has no attribute 'min'

Thank you.

@vsl9

I am getting this error when I was trying to load 'GoogleNews-vectors-negative300-SLIM.bin'. (Code given below)

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-21-b567896a8aa6> in <module>
     14 print('vocab size: ' + str(len(vocab)))
     15 print('max sentence length: ' + str(max_l))
---> 16 w2v = load_bin_vec(wv_from_bin, vocab)
     17 print(w2v)
     18 print('word2vec loaded!')

<ipython-input-20-59822c213c28> in load_bin_vec(fname, vocab)
     49     """
     50     word_vecs = {}
---> 51     with open(fname, 'rb') as f:
     52         header = f.readline()
     53         vocab_size, layer1_size = map(int, header.split())

TypeError: expected str, bytes or os.PathLike object, not Word2VecKeyedVectors

Code is

w2v_file = 'GoogleNews-vectors-negative300-SLIM.bin'
revs, vocab = build_data_train_test(data_train, train_ratio=0.6, clean_string=True)
max_l = np.max(pd.DataFrame(revs)['num_words'])
print('data loaded!')
print('number of sentences: ' + str(len(revs)))
print('vocab size: ' + str(len(vocab)))
print('max sentence length: ' + str(max_l))
w2v = load_bin_vec(w2v_file, vocab)
print(w2v)
print('word2vec loaded!')
print('num words already in word2vec: ' + str(len(w2v)))

add_unknown_words(w2v, vocab)
W, word_idx_map = get_W(w2v)
cPickle.dump([revs, W, word_idx_map, vocab], open('imdb-train-val-testN.pickle', 'wb'))
print('dataset created successfully!')

Any help or guidance is highly appreciated.

Hey, the output file currently gives values ranging between 0.0 and 1.0. Is it not supposed to be binary?

How do I get the sentiment value to be binary?

0 being negative, and 1 being positive.

Hi,
which accuracy did the Yoon Kim CNN achieve with these settings?

Seems like the line "model.add(Reshape(1, conv_input_height, conv_input_width))" no longer works. It throws an error:

TypeError Traceback (most recent call last)
in ()
26 # Reshape word vectors from Embedding to tensor format suitable for Convolutional layer
27 # reshape from three dimensional to four dimensional
---> 28 model.add(Reshape(1,conv_input_height, conv_input_width) )
29
30 # first convolutional layer

TypeError: init() takes exactly 2 arguments (4 given)

Based on the Keras documentation, I changed it to:
model.add(Reshape((1, conv_input_height, conv_input_width)))

And yet it throws another error now:

 34 def _prod(a, axis=None, dtype=None, out=None, keepdims=False):

---> 35 return umr_prod(a, axis, dtype, out, keepdims)
36
37 def _any(a, axis=None, dtype=None, out=None, keepdims=False):

TypeError: long() argument must be a string or a number, not 'NoneType'

I have upgraded both keras/theano to the bleeding edge version. It's been very hard to find any alternative working example of embedding followed by 2d convolution with keras. Any idea why this happens and how we may fix it?

Thanks a lot!!!

Hello there,

I was going through the way you constructed the sentence matrix and then I didn't understand one things. Why did you take the indices of the words in the vocabulary while creating the sentence matrices? I would have imagined that as input for the convolutional neural network, we would be constructing matrices out of the word vectors for each word in the review. Or am i missing something here?

And I am sorry to have posted this question as an issue. I would have contacted you by email but I couldn't find any contact information.

Thanks for your help

Hi there,

I was just trying to implement your model and I ran into trouble trying to run the CNN using Keras. When I run model.compile with the parameters you provide i get the following error

 TypeError                                 Traceback (most recent call last)
<ipython-input-24-8079df85a0aa> in <module>()
----> 1 model.compile(loss='categorical_crossentropy', optimizer='adadelta')

D:\Users\Dhruv.Sharma\AppData\Local\Continuum\Anaconda\lib\site-packages\keras-0.1.2-py2.7.egg\keras\models.pyc in compile(self, optimizer, loss, class_mode, theano_mode)
332         for r in self.regularizers:
333             train_loss = r(train_loss)
--> 334         updates = self.optimizer.get_updates(self.params, self.constraints, train_loss)
335 
336         if type(self.X_train) == list:

D:\Users\Dhruv.Sharma\AppData\Local\Continuum\Anaconda\lib\site-packages\keras-0.1.2-py2.7.egg\keras\optimizers.pyc in get_updates(self, params, constraints, loss)
138 
139             new_p = p - self.lr * update
--> 140             updates.append((p, c(new_p))) # apply constraints
141 
142             # update delta_accumulator

TypeError: object() takes no parameters

Seems keras changed a lot, need to change the code as well.

Hi,
Do you still have the trained weights of the model? This would save me some computational time.
Thx in advance and cheers.