Iterative-Document-Representation-Learning-Towards-Summarization-with-Polishing

This is the official codes for the paper: Iterative Document Representation Learning Towards Summarization with Polishing

Requirements

---

• Python 2.7
• Tensorflow 1.3.0

CNN/Daily Mail dataset

---

Preprocessed data can be found here.

Use story2TF.py to transform data into .bin format and put in tfrecord directory.

How to train

---

python single.py --train_dir=train_dir --train_path=path_to_training_data --validation_path=path_to_validation_data

How to test

---

python test.py --train_dir=path_to_train_dir --test_path=test_dir