ExcelFormer

KO | EN

gitlite — search

#javascript #python #hacktoberfest #react #ai #typescript #llm #go #golang #android #machine-learning #rust #deep-learning #linux

(★ 68)

The pioneering neural network surpassing extremely-tuned XGboost and Catboost on varied tabular datasets.

ExcelFormer 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# ExcelFormer

This repository will include the original implementation and experiment codes of [**ExcelFormer**](https://arxiv.org/abs/2301.02819). TabFormer is a pioneering neural network can surpass extensively-tuned XGboost, Catboost, and most tuned previous deep learning approaches on most of tabular data prediction tasks, in the supervised learning manner. It can be a go-to choice on tabualr dataset prediction competitions (e.g., Kaggle).

Even without hyper-parameter tuning, TabFormer performs comparable to tuned models. After hyper-parameter tuning, TabFormer typically outperforms them.

The implementation of TabFormer in the original paper is `bin/excel_former.py`.

## How to test your model

You can test your models by adding them to `bin` directory and `bin/__init__.py`. Keep the same API we used in other models, and write your own evaluation script (`run_default_config_excel.py` as a reference).

## Datasets:

The datasets (96 small tabular datasets + 21 large tabular datasets) are available at: https://huggingface.co/datasets/jyansir/excelformer.

## Future work

We will organize our previous works on **tabular prediction** into [Tabular AI Research](https://github.com/pytabular-ai) group for industrial use (e.g. further architecture optimization or acceleration / compilation). If you want to include our model as a baseline in your paper, please use the version in this repository rather than the industrial one in the group repository.

ExcelFormer

ExcelFormer 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

ExcelFormer

ExcelFormer 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README