Implementation of 'Pretraining-Based Natural Language Generation for Text Summarization'
- python 2.7
- PyTorch: 1.0.1.post2
pip install -r requirements.txtto install required packages
- Download chunk CNN/DailyMail data from: https://github.com/JafferWilson/Process-Data-of-CNN-DailyMail
python news_data_reader.pyto create pickle file that will be used in my data-loader
Running the model
For me, the model was too big for my GPU, so I used smaller parameters as following for debugging purpose.
CUDA_VISIBLE_DEVICES=3 python main.py --cuda --batch_size=2 --hop 4 --hidden_dim 100
Note to reviewer:
- Although I implemented the core-part (2-step summary generation using BERT), I didn't have enough time to implement RL section.
- The 2nd decoder process is very time-consuming (since it needs to create BERT context vector for each timestamp).