Pointer-Generator for AMI Meeting Dataset

This code is based on OpenNMT.

Requirements

• We use Conda python 3.7 and strongly recommend that you create a new environment.
  • conda create -n pgn python=3.7.
• Run the following command.
  • pip install -r requirements.txt.

Data

You can get data here. Put them under the dir data/*.

Reproduce Results

You can follow the following steps to reproduce the best results in our paper.

download checkpoints

Download checkpoints here. Put the checkpoint under the dir ckpt/ami.pt.

translate

• sh ./scripts/infer.sh

test rouge score

• Change pyrouge.Rouge155() to your local path.
  • Output format >> ROUGE(1/2/L): xx.xx-xx.xx-xx.xx
• python test_rouge.py -c summaries/ami.txt

ROUGE score

You will get following ROUGE scores.

	ROUGE-1	ROUGE-2	ROUGE-L
AMI	50.91	17.75	24.59

From Scratch

Preprocess

Run the following commands:

• sh ./scripts/preprocess.sh
• sh ./scripts/embedding.sh

Train

Run the following command:

• sh ./scripts/train.sh

Translate

Run the following command:

• sh ./scripts/infer.sh
  • set up model_name param first.