Keras: Error when checking target: expected dense_2 to have shape (20,) but got array with shape (1000,) while running pretrained_word_embeddings.py

[jonanem]

Error when checking target: expected dense_2 to have shape (20,) but got array with shape (1000,) while running pretrained_word_embeddings.py

https://github.com/keras-team/keras/blob/master/examples/pretrained_word_embeddings.py

[chriskoups]

the error is related to your network as far as I can understand.
Can you post your code or the initial part?
I can assume that you did not divided the network as it must be. You have a shape of 20 but you try to pass 1000 without batches.

[jonanem]

Please find the code below

`BASE_DIR = 'C:\Users\'
GLOVE_DIR = os.path.join(BASE_DIR, 'glove.6B')
TEXT_DATA_DIR = os.path.join(BASE_DIR, 'news20\20_newsgroup')
MAX_SEQUENCE_LENGTH = 1000
MAX_NUM_WORDS = 20000
EMBEDDING_DIM = 100
VALIDATION_SPLIT = 0.2
EMBEDDING_DIM = 100

embeddings_index = {}
with open(os.path.join(GLOVE_DIR, 'glove.6B.100d.txt'), encoding="utf8") as f:
for line in f:
values = line.split()
word = values[0]
coefs = np.asarray(values[1:],dtype = 'float32')
embeddings_index[word] = coefs

texts = [] # list of text samples
labels_index = {} # Dictionary mapping of label to label id
labels = [] # List of label ids
for name in sorted(os.listdir(TEXT_DATA_DIR)):
path = os.path.join(TEXT_DATA_DIR, name)
if os.path.isdir(path):
label_id = len(labels_index)
labels_index[name] = label_id

    for fname in sorted(os.listdir(path)):
        if fname.isdigit():
            fpath = os.path.join(path, fname)
            args = {} if sys.version_info < (3,) else {'encoding':'latin-1'}

            with open(fpath, **args) as f:
                t = f.read()
                i = t.find("\n\n") # Skips header
                if 0 < i:
                    t = t[i:]
                texts.append(t)
            labels.append(label_id)

tokenizer = Tokenizer(num_words = MAX_NUM_WORDS)
tokenizer.fit_on_texts(texts)
sequences = tokenizer.texts_to_sequences(texts)
word_index = tokenizer.word_index

Padding sequences

data = pad_sequences(sequences, maxlen = MAX_SEQUENCE_LENGTH)

To convert into binary

labels = to_categorical(np.asarray(labels))

indices = np.arange(data.shape[0])
np.random.shuffle(indices)
data = data[indices]
labels = labels[indices]
num_validation_samples = int(VALIDATION_SPLIT*data.shape[0])

x_train = data[:-num_validation_samples]
y_train = data[:-num_validation_samples]
x_test = data[-num_validation_samples:]
y_test = data[-num_validation_samples:]

num_words = min(MAX_NUM_WORDS, len(word_index)+1)
embedding_matrix = np.zeros((num_words, EMBEDDING_DIM))

for word, i in word_index.items():
if i >= MAX_NUM_WORDS:
continue
embedding_vector = embeddings_index.get(word)
if embedding_vector is not None:
embedding_matrix[i] = embedding_vector

embedding_layer = Embedding(num_words,
EMBEDDING_DIM,
embeddings_initializer=Constant(embedding_matrix),
input_length = MAX_SEQUENCE_LENGTH,
trainable = False)

Training 1D convent with global maxpooling

sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype ='int32')
embedded_sequences = embedding_layer(sequence_input)
x = Conv1D(128, 5, activation = 'relu')(embedded_sequences)
x = MaxPooling1D(5)(x)
x = Conv1D(128, 5, activation = 'relu')(x)
x = MaxPooling1D(5)(x)
x = Conv1D(128, 5, activation = 'relu')(x)
x = GlobalMaxPooling1D()(x)
print(x.shape)
x = Dense(128, activation='relu')(x)
print(x.shape)
preds = Dense(len(labels_index), activation='softmax')(x)

model = Model(sequence_input, preds)
model.compile(loss='categorical_crossentropy',
optimizer='rmsprop',
metrics=['acc'])
model.fit(x_train, y_train, batch_size=128, epochs = 10, validation_data=(x_test, y_test))`

[chriskoups]

I am not sure what is costing that problem.
Wild guess will be that the lables_index is 20 and that is creating the problem. Can you confirm the output of len(labels_index)?

[jonanem]

output is 20 to predict 20 Categories

[tRosenflanz]

Github is for issues in Keras while this is just an implementation error. In future please open stackoverflow questions instead of posting on Github:

x_train = data[:-num_validation_samples]
y_train = data[:-num_validation_samples] .  # supposed to be labels
x_test = data[-num_validation_samples:]
y_test = data[-num_validation_samples:] .  # supposed to be lablels

[jonanem]

Thank you so much for your help

[johnbond2]

I have a similiar error message: "ValueError: Error when checking target: expected dense_2 to have shape (10,) but got array with shape (1,)", and it seems that what is causing this problem is the loss parameter in the model architecture, when I use loss = 'sparse_categorical_crossentropy' gives me no error but when using this one, it does loss = 'categorical_crossentropy'

[RaviChandraMadipadiga]

I have a similiar error message: "ValueError: Error when checking target: expected dense_2 to have shape (10,) but got array with shape (1,)", and it seems that what is causing this problem is the loss parameter in the model architecture, when I use loss = 'sparse_categorical_crossentropy' gives me no error but when using this one, it does loss = 'categorical_crossentropy'

But even after i changed that still i got that error but with less error shape