Skip to content

Latest commit

 

History

History
24 lines (16 loc) · 1.12 KB

Readme.md

File metadata and controls

24 lines (16 loc) · 1.12 KB

Ranking Sentences for Extractive Summarisation using Reinforcement Learning

This is a re-implementation of the paper Ranking Sentences for Extractive Summarisation using Reinforcement Learning in PyTorch. Find the original code here.

Running the summarizer:

NOTE: to run on CPU, set --gpu_no argument to -1

  • Download model weights from here and save it.
  • Download Glove cache from here and save it.
  • Run: 
    cd notebooks

ipython run.py -- --input_folder=<path/to/document_folder> --output_folder=<path/to/summary_folder> --weights=<path/to/saved/weights/> --glove=<path/to/glove/cache> --gpu_no=0 --summary_len=3

Training

  • Download dataset (neuralsum.zip) from here.
  • Run preprocess.sh <path/to/neuralsum.zip> to process data.
    • Script for training