Few-Shot Table-to-Text Generation with Prototype Memory

Authors: Yixuan Su, Zaiqiao Meng, Simon Baker, and Nigel Collier

Code for EMNLP 2021 paper Few-Shot Table-to-Text Generation with Prototype Memory

1. Download Data and Pre-trained Models:

(1) Download Data link

unzip the data.zip and replace it with the empty data folder

(2) Pre-trained Checkpoints link

unzip the checkpoints.zip and replace it with empty checkpoints folder

2. Prototype Selector

(1) Enviornment Installation:

pip install -r prototype_selector_requirements.txt

(2) Training of Few-Shot-k setting for humans dataset:

cd ./prototype_selector/sh_folder/training/human/
chmod +x ./human-few-shot-k.sh
./human-few-shot-k.sh

(3) Inference of Few-Shot-k setting for humans dataset:

cd ./prototype_selector/sh_folder/inference/human/
chmod +x ./inference_human-few-shot-k.sh
./inference_human-few-shot-k.sh

3. Generator

(1) Enviornment Installation:

pip install -r generator_requirements.txt

(2) Training of Few-Shot-k setting for humans dataset:

cd ./generator/training/human/
chmod +x ./human-few-shot-k.sh
./human-few-shot-k.sh

(3) Inference of Few-Shot-k setting for humans dataset:

cd ./generator/inference/human/
chmod +x ./human-few-shot-k-inference.sh
./human-few-shot-k-inference.sh

4. Citation

If you find our paper and code useful, please kindly cite our paper:

@inproceedings{su-etal-2021-shot-table,
    title = "Few-Shot Table-to-Text Generation with Prototype Memory",
    author = "Su, Yixuan  and
      Meng, Zaiqiao  and
      Baker, Simon  and
      Collier, Nigel",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2021",
    month = nov,
    year = "2021",
    address = "Punta Cana, Dominican Republic",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.findings-emnlp.77",
    pages = "910--917",
    abstract = "Neural table-to-text generation models have achieved remarkable progress on an array of tasks. However, due to the data-hungry nature of neural models, their performances strongly rely on large-scale training examples, limiting their applicability in real-world applications. To address this, we propose a new framework: Prototype-to-Generate (P2G), for table-to-text generation under the few-shot scenario. The proposed framework utilizes the retrieved prototypes, which are jointly selected by an IR system and a novel prototype selector to help the model bridging the structural gap between tables and texts. Experimental results on three benchmark datasets with three state-of-the-art models demonstrate that the proposed framework significantly improves the model performance across various evaluation metrics.",
}