A demo code for the weibo-comments creation.
The modules are implemented based on the DeepLearning 0.1 documentation (http://deeplearning.net/tutorial/).
The code is written in Theano.
To use it you will also need: coco-caption envaluate package, cPickle,scikit-learn, matplotlib,tqdm,jieba,gensim and PIL.
Before running the code make sure that you have set floatX to float32 in Theano settings.
Dataset is not available for now because of the authority issue.
Make a subfolder named lib, and put activations.py, inits.py , rng.py, theano_utils.py and updates.py in it.
Train: weibo_train_v3.py
Envalue:weibo_v3_enval_with_beamsearch.py
If you found this code or our paper useful, please consider citing the following paper:
@inproceedings{IJCAI20weibo,
author = {Shijie Yang, Liang Li, Shuhui Wang, Weigang Zhang, Qingming Huang, Qi Tian},
title = {A Structured Latent Variable Recurrent Network with Stochastic Attention for GeneratingWeibo Comments},
booktitle = {IJCAI},
year = {2020}
}