metadata

language: en
tags:
  - table-fact-checking
datasets:
  - tabfact

ReasTAP

ReasTAP is a table reasoning model proposed in the EMNLP 2022 paper ReasTAP: Injecting Table Reasoning Skills During Pre-training via Synthetic Reasoning Examples. The original Github repository is https://github.com/Yale-LILY/ReasTAP.

Description

Yale-LILY/reastap-large-finetuned-tabfact is initialized with Yale-LILY/reastap-large and finetuned on TabFact.

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import pandas as pd

tokenizer = AutoTokenizer.from_pretrained("Yale-LILY/reastap-large-finetuned-tabfact")
model = AutoModelForSequenceClassification.from_pretrained("Yale-LILY/reastap-large-finetuned-tabfact")

data = {
    "year": [1896, 1900, 1904, 2004, 2008, 2012],
    "city": ["athens", "paris", "st. louis", "athens", "beijing", "london"]
}
table = pd.DataFrame.from_dict(data)

query = "beijing hosts the olympic games in 2012"
encoding = tokenizer(table=table, query=query, return_tensors="pt")

outputs = model(**encoding)
output_id = int(outputs.logits[0].argmax(dim=0))
print(model.config.id2label[output_id])

Reference

@inproceedings{zhao-etal-2022-reastap,
    title = "{R}eas{TAP}: Injecting Table Reasoning Skills During Pre-training via Synthetic Reasoning Examples",
    author = "Zhao, Yilun  and
      Nan, Linyong  and
      Qi, Zhenting  and
      Zhang, Rui  and
      Radev, Dragomir",
    booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing",
    month = dec,
    year = "2022",
    address = "Abu Dhabi, United Arab Emirates",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.emnlp-main.615",
    pages = "9006--9018",
    abstract = "Reasoning over tabular data requires both table structure understanding and a broad set of table reasoning skills. Current models with table-specific architectures and pre-training methods perform well on understanding table structures, but they still struggle with tasks that require various table reasoning skills. In this work, we develop ReasTAP to show that high-level table reasoning skills can be injected into models during pre-training without a complex table-specific architecture design. We define 7 table reasoning skills, such as numerical operation, temporal comparison, and conjunction. Each reasoning skill is associated with one example generator, which synthesizes questions over semi-structured tables according to the sampled templates. We model the table pre-training task as a sequence generation task and pre-train ReasTAP to generate precise answers of the synthetic examples. ReasTAP is evaluated on four benchmarks covering three downstream tasks including 1) WikiSQL-Weak and WikiTQ for Table Question Answering, 2) TabFact for Table Fact Verification, and 3) LogicNLG for Faithful Table-to-Text Generation. Experimental results demonstrate that ReasTAP achieves new state-of-the-art results on all of them and delivers a significant improvement under low-resource setting. Our code is publicly available at https://github.com/Yale-LILY/ReasTAP.",
}