Cannot replicate

[rajivpoddar]

I trained the model as specified in the readme but cannot replicate the results. The following is what I get.

Input: you must have girlfriend Output: than than than than than than than than than than

Is this because of the training/dataset?

[essamgoda]

hi @braindead i have problem in training it show error

batched_corrective_tokens = tf.pack(
AttributeError: 'module' object has no attribute 'pack'

in file

text_corrector_models.py line 91

tensorflow V 1.0.0

[xushenkun]

@braindead I also get this kind of results. Have you solved it? @essamgoda You should use stack instead of pack.

[essamgoda]

another error show

@xushenkun

TypeError: Input 'y' of 'Mul' Op has type float32 that does not match type int32 of argument 'x'.

[derofim]

@essamgoda Maybe change labels and inputs, as in in API https://www.tensorflow.org/versions/r1.1/api_docs/python/tf/nn/sampled_softmax_loss def sampled_loss(labels, inputs): labels = tf.reshape(labels, [-1, 1]) return tf.nn.sampled_softmax_loss(w_t, b, labels, inputs, num_samples, self.target_vocab_size)

another error "Ran out of memory"

 File "correct_text.py", line 434, in main
    train(data_reader, FLAGS.train_path, FLAGS.val_path, FLAGS.model_path)
  File "correct_text.py", line 181, in train
    target_weights, bucket_id, False)
  File "text_corrector_models.py", line 297, in step
    outputs = session.run(output_feed, input_feed)

Command used: python correct_text.py --train_path "./data/movie_dialog_train.txt" --val_path "./data/movie_dialog_val.txt" --config DefaultMovieDialogConfig --data_reader_type MovieDialogReader --model_path "./movie_dialog_model"

Used data and source code: https://www.dropbox.com/s/oo7e9vg5hcrnjke/deep-text-corrector-master.rar?dl=0

[essamgoda]

@derofim thanks alot it's worked

your code worked (take a long time)

try to reduce number of layer and size if your resource is limited i use gpu ( Nvidia 940 mx 4 gb)

[essamgoda]

@braindead @xushenkun the same error did you solve it ?

[liuyichaosoftware]

@essamgoda hi, how much steps have you trained? when I run 20000 times in half an hour, it showed....

Input: you must have girlfriend Output: girlfriend girlfriend girlfriend girlfriend girlfriend girlfriend girlfriend girlfriend girlfriend girlfriend

[essamgoda]

@liuyichaosoftware I also run 20000 times

[bittlingmayer]

Same here.

[HarshdeepGupta]

@braindead @essamgoda I am having the same issues with the model. Did you eventually solve it? Or is it unsolvable?

[qinbill]

Problem fixed. Please see my fork.

[arianpasquali]

@braindead @essamgoda @HarshdeepGupta I have the same kind of output here:

na na na na na na na na na na
what what what what what what what what what what
gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim gim
... ... ... ... ... ... ... ... ... ...
up up up up up up up up up up
ya ya ya ya ya ya ya ya ya ya ya ya ya ya ya

I didnt change parameters. I also run 20000 times. Did anyone eventually manage to get expected results?

[arianpasquali]

@qinbill what exactly did you fixed? did you get expected results?

[shibing624]

how to get expected results, and which parameters should change? output: what what what what what what what what what

[Baidi96]

Problem fixed. Please see my fork.

Hi, which problem did you fix? problem 'cannot replicate the results'? I tried your fork with my pertained model but still get the same kind of results.