Inputs to operation sentence_lstm/while_1/Select_1 of type Select must have the same size and shape

[shreeshaa]

Hi! I was using your project dEFEND code (https://www.dropbox.com/sh/rzczwopo618jyv2/AAA1mI2yvbt6TAfqpcxfjL8va?dl=0) as reference for a project we are working on. But, unfortunately, running the go_defend.py file with the dataset we collected from FakeNewsNet will have some errors. Specifically, the error is caused when model.fit is called and it is in the sentence lstm part. Please find exact error below:

Traceback (most recent call last):
File "go_defend.py", line 85, in
saved_model_filename=SAVED_MODEL_FILENAME)
File "/home/shreeshaa/d/dEFEND-5/fake news detection/dEFEND/defend.py", line 389, in train
callbacks=callbacks)
File "/home/shreeshaa/anaconda3/envs/d/lib/python3.7/site-packages/keras/engine/training.py", line 1039, in fit
validation_steps=validation_steps)
File "/home/shreeshaa/anaconda3/envs/d/lib/python3.7/site-packages/keras/engine/training_arrays.py", line 199, in fit_loop
outs = f(ins_batch)
File "/home/shreeshaa/anaconda3/envs/d/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py", line 2715, in call
return self._call(inputs)
File "/home/shreeshaa/anaconda3/envs/d/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py", line 2675, in _call
fetched = self._callable_fn(*array_vals)
File "/home/shreeshaa/anaconda3/envs/d/lib/python3.7/site-packages/tensorflow/python/client/session.py", line 1439, in call
run_metadata_ptr)
File "/home/shreeshaa/anaconda3/envs/d/lib/python3.7/site-packages/tensorflow/python/framework/errors_impl.py", line 528, in exit
c_api.TF_GetCode(self.status.status))
tensorflow.python.framework.errors_impl.InvalidArgumentError: Inputs to operation sentence_lstm/while_1/Select_1 of type Select must have the same size and shape. Input 0: [20,12000] != input 1: [20,100]
[[{{node sentence_lstm/while_1/Select_1}}]]

Any advice would be extremely helpful. Thanks!

[DeepAccount]

I am also stuck with same problem. After changing dependencies according to https://github.com/cuilimeng/dEFEND-web/blob/master/requirements.txt. I get a different error
You are trying to load a weight file containing 5 layers into a model with 4 layers

Error is coming in load module step
self.model = load_model(./static/saved_models/politifact_Defend_model.h5)

[CODE-FOR]

Have you fixed it? I occur the same problem.

[namphn]

I am also stuck with same problem. After changing dependencies according to https://github.com/cuilimeng/dEFEND-web/blob/master/requirements.txt. I get a different error
You are trying to load a weight file containing 5 layers into a model with 4 layers

Error is coming in load module step
self.model = load_model(./static/saved_models/politifact_Defend_model.h5)

Have you fixed it? I occur the same problem

[CODE-FOR]

I am also stuck with same problem. After changing dependencies according to https://github.com/cuilimeng/dEFEND-web/blob/master/requirements.txt. I get a different error
You are trying to load a weight file containing 5 layers into a model with 4 layers
Error is coming in load module step
self.model = load_model(./static/saved_models/politifact_Defend_model.h5)

Have you fixed it? I occur the same problem

I kind of fixed it by changing code in _build_model

        sentence_input = Input(shape=(self.MAX_SENTENCE_LENGTH,), dtype='int32')
        embedded_sequences = embedding_layer(sentence_input)
        # change
        embedded_sequences = Lambda(lambda x: x, output_shape=lambda s: s)(embedded_sequences)
        l_lstm = Bidirectional(GRU(100, return_sequences=True), name='word_lstm')(Reshape((120, 100))(embedded_sequences))
        # change
        l_att = AttLayer(name='word_attention')(l_lstm)
        sentEncoder = Model(sentence_input, l_att)

        comment_input = Input(shape=(self.MAX_COMS_LENGTH,), dtype='int32')
        com_embedded_sequences = com_embedding_layer(comment_input)
        # change
        com_embedded_sequences = Lambda(lambda x: x, output_shape=lambda s: s)(com_embedded_sequences)
        c_lstm = Bidirectional(GRU(100, return_sequences=True), name='comment_lstm')(Reshape((100, 120))(com_embedded_sequences))
        # change
        c_att = AttLayer(name='comment_word_attention')(c_lstm)
        comEncoder = Model(comment_input, c_att, name='comment_word_level_encoder')

Maybe just reshape is enough, but after reshaping. I occur some strange bug.
And by adding a lambda layer can fix it.
The model seems to run successfully.

You can try this.

[namphn]

c_lstm = Bidirectional(GRU(100, return_sequences=True), name='comment_lstm')(Reshape((100, 120))(com_embedded_sequences))

Oh my god!!!
You saved my life =))))) thanks so munch!!!!!!

[peteoleary]

The above code fixed the problem for me but now I am seeing:

`Exception has occurred: RuntimeError
Unable to create link (name already exists)
File "/Users/pete_o/Documents/Dev/InfoAuthN/news-dj-service/our3.py", line 400, in train
callbacks=callbacks)
File "/Users/pete_o/Documents/Dev/InfoAuthN/news-dj-service/go_defend.py", line 96, in
embeddings_path='saved_models', saved_model_dir='saved_models', saved_model_filename=SAVED_MODEL_FILENAME )

The exception is thrown from:

self.model.fit([encoded_train_c, encoded_train_x], y=train_y, validation_data=([encoded_val_c, encoded_val_x], val_y), batch_size=batch_size, epochs=epochs, verbose=1, callbacks=callbacks)
`

[peteoleary]

Opened new issue #14

Let me know if there is another Github repo or forum where I could post this.

[gmyfreedom]

c_lstm = Bidirectional(GRU(100, return_sequences=True), name='comment_lstm')(Reshape((100, 120))(com_embedded_sequences))

Oh my god!!! You saved my life =))))) thanks so munch!!!!!!

c_lstm = Bidirectional(GRU(100, return_sequences=True), name='comment_lstm')(Reshape((100, 120))(com_embedded_sequences))

Oh my god!!! You saved my life =))))) thanks so munch!!!!!!

Hi，have you run the project successfully? I have some problem in getting twitter API. I want to know if this project can run successfully.

[gmyfreedom]

Hi，have you run the project successfully? I have some problem in getting twitter API. I want to know if this project can run successfully.