This repository is implementation of Homework 2 for CSIE5431 Applied Deep Learning course in 2023 Fall semester at National Taiwan University.
Step 1: Install the main packages
To set the environment, you can run this command:
pip install -r config/requirements.txt
Step 2: Install the evaluation metric
To install the evaluation metric, you can run this command:
git clone https://github.com/moooooser999/ADL23-HW2.git
cd ADL23-HW2
pip install -e tw_rouge
To download the datasets and model checkpoint, you can run the commad:
bash ./download.sh
To reproduce our best result, you can run the commad:
bash ./run.sh <data file> <output file>
To train the summarization model, you can run the commad:
python train.py
To test the summarization model, you can run the commad:
python test.py
To evaluate the submission file, you can run the command
python ADL23-HW2/eval.py -r public.jsonl -s submission.jsonl
Method | Rouge-1 | Rouge-2 | Rouge-L |
Our | 26.8510 | 10.7346 | 23.9712 |
Baseline | 22.0 | 8.5 | 20.5 |
We implemented the code on an environment running Ubuntu 22.04.1, utilizing a 12th Generation Intel(R) Core(TM) i7-12700 CPU, along with a single NVIDIA GeForce RTX 4090 GPU equipped with 24 GB of dedicated memory.
We thank the Hugging Face repository: https://github.com/huggingface/transformers
@misc{
title = {T5 Chinese News Summarization},
author = {Jia-Wei Liao},
url = {https://github.com/jwliao1209/Chinese-News-Summarization},
year = {2023}
}