Open-source Tapex-large-finetuned-wikisql Model - Empowering Efficient Processing of Table Reasoning Tasks

Tapex Large Finetuned Wikisql

Developed by microsoft

TAPEX is a table pre-training model learned through neural SQL executors, based on the BART architecture, specifically designed for table reasoning tasks.

Large Language Model

Transformers

EnglishOpen Source License:MIT #Table Question Answering #SQL Execution Pretraining #Table Reasoning

Downloads 676

Release Time : 3/8/2022

Model Overview

TAPEX achieves table pre-training by learning neural SQL executors on a synthetic corpus, which is obtained by automatically synthesizing executable SQL queries. The model is based on the BART architecture, featuring a bidirectional encoder and an autoregressive decoder.

Model Features

Table Reasoning Capability

Learned through neural SQL executors, endowing the model with powerful table reasoning capabilities.

Synthetic Corpus Pre-training

Uses automatically synthesized executable SQL queries as the pre-training corpus.

BART Architecture

Combines the advantages of bidirectional encoders and autoregressive decoders.

Model Capabilities

Table Question Answering

SQL Query Execution

Table Data Reasoning

Use Cases

Data Query

Simple Table Question Answering

Answer simple questions about table data

For example, 'In which year did Beijing host the Olympics?', the model correctly answers '2008.0'

Data Statistics

Perform statistical queries on table data

For example, 'How many schools has player No. 3 attended?', the model correctly answers '1.0'

🚀 TAPEX (large-sized model)

TAPEX is a pre - training approach that empowers existing models with table reasoning skills. It was proposed in a research paper and shows good performance in table question - answering tasks.

✨ Features

TAPEX is a conceptually simple and empirically powerful pre - training approach for table reasoning.
It is based on the BART architecture, a transformer encoder - encoder (seq2seq) model.
This specific model is fine - tuned on the WikiSQL dataset.

📦 Installation

No installation steps are provided in the original document, so this section is skipped.

💻 Usage Examples

Basic Usage

Here is how to use this model in transformers:

from transformers import TapexTokenizer, BartForConditionalGeneration
import pandas as pd

tokenizer = TapexTokenizer.from_pretrained("microsoft/tapex-large-finetuned-wikisql")
model = BartForConditionalGeneration.from_pretrained("microsoft/tapex-large-finetuned-wikisql")

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

# tapex accepts uncased input since it is pre-trained on the uncased corpus
query = "In which year did beijing host the Olympic Games?"
encoding = tokenizer(table=table, query=query, return_tensors="pt")

outputs = model.generate(**encoding)

print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
# [' 2008.0']

Advanced Usage

No advanced usage examples are provided in the original document.

📚 Documentation

Model description

TAPEX (Table Pre - training via Execution) is a conceptually simple and empirically powerful pre - training approach to empower existing models with table reasoning skills. TAPEX realizes table pre - training by learning a neural SQL executor over a synthetic corpus, which is obtained by automatically synthesizing executable SQL queries.

TAPEX is based on the BART architecture, the transformer encoder - encoder (seq2seq) model with a bidirectional (BERT - like) encoder and an autoregressive (GPT - like) decoder.

This model is the tapex - base model fine - tuned on the WikiSQL dataset.

Intended Uses

You can use the model for table question answering on relatively simple questions. Some solveable questions are shown below (corresponding tables now shown):

Question	Answer
tell me what the notes are for south australia	no slogan on current series
what position does the player who played for butler cc (ks) play?	guard - forward
how many schools did player number 3 play at?	1.0
how many winning drivers in the kraco twin 125 (r2) race were there?	1.0
for the episode(s) aired in the u.s. on 4 april 2008, what were the names?	"bust a move" part one, "bust a move" part two

How to Eval

Please find the eval script here.

BibTeX entry and citation info

@inproceedings{
    liu2022tapex,
    title={{TAPEX}: Table Pre-training via Learning a Neural {SQL} Executor},
    author={Qian Liu and Bei Chen and Jiaqi Guo and Morteza Ziyadi and Zeqi Lin and Weizhu Chen and Jian - Guang Lou},
    booktitle={International Conference on Learning Representations},
    year={2022},
    url={https://openreview.net/forum?id=O50443AsCP}
}

📄 License

This model is released under the MIT license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご