#tensorflow related work(nlp and image related, text classification, image caption, seq2seq..)
mainly tf work, may do some pytorch related

## incase not find dependence, make sure set PYTHONPATH to include hasky/util so we can find gezi and melt ##./applications/ ###./applications/tf-record/ show how to write and read TFRecord(tensorflow standard dataa format) ###./applications/text-classification reading libsvm format then do text classification ###./applications/text-regression reading libsvm format then do text regression ###./applications/text-binary-classification reading libsvm format then do text binary classification, evaluate by auc ###./applications/sparse-tensor-classification/ (depreciated) this is self contained mlp classification example showing how to read sparse TFRecord and train a mlp classifier

Notice some examples like text-classification will move to deepiu

##./deepiu main application root
###./deepiu/image-caption image-caption related work now support
discriminant method:
bow,
rnn,
cnn(TODO)
generative method:
show_and_tell
show_attend_and_tell(TODO)
Input with both image and text(TODO) ###./deepiu/text-sum app with long text as input(like image title, ct0) and predict shortter summary text(like click query)
supporting method:
seq2seq
seq2seq_attetion
seq2seq_attetion_copy(TODO)
###./deepiu/seq2seq common seq2seq codes used for image-caption, text-sum and other applications ###./deepiu/util common util for deepiu

##./util ###./util/gezi common lib ###./util/melt common tensorflow related lib, you can view it similar as tf.contrib ####./util/melt/flow like tf.supervisor, make train and test flow easier,
main functions include, application only make graph and pass train_ops and evaluate_ops to flow
flow will do model save, log save(auto add all one dimentional shape tensors to tensorboard), show elapsed time ...

publish lib

1. gezi ./util/gezi
2. melt ./util/melt
3. hasky ./util/hasky
4. deepiu ./deepiu