https://github.com/sb-ai-lab/LADS

Last synced: 3 months ago
JSON representation

Host: GitHub
URL: https://github.com/sb-ai-lab/LADS
Owner: sb-ai-lab
License: bsd-3-clause
Created: 2025-06-02T13:58:59.000Z (12 months ago)
Default Branch: main
Last Pushed: 2025-09-16T07:41:32.000Z (9 months ago)
Last Synced: 2026-02-10T22:55:28.055Z (4 months ago)
Language: Python
Size: 1.79 MB
Stars: 16
Watchers: 0
Forks: 2
Open Issues: 1
Metadata Files:
- Readme: README.md
- License: LICENSE

Awesome Lists containing this project

awesome-agentic-machine-learning - LADS (LightAutoDS) - AutoML agentic system combining LLM code generation with AutoGluon, LightAutoML, and FEDOT. | ![GitHub stars](https://img.shields.io/github/stars/sb-ai-lab/LADS?style=flat-square) | (AutoML Agents)

README

LightAutoDS-Tab

![architecture](./image/architecture.png)

**LightAutoDS-Tab**, a multi-AutoML agentic system for tasks with tabular data, which combines an LLM-based code generation with several AutoML tools.

## ✨ Demo
[Watch the Video](https://www.youtube.com/watch?v=5e8eADd_HWE)

## 🧑‍💻 User interface

The interface includes two main panels:

1. The **right panel** provides detailed *technical insights* into each step of the ML pipeline construction, offering transparency for expert users.
1. The **left panel** presents a simplified, *non-technical summary* of the process, making it easy for non-experts to follow along and understand the results

## 🚀 Quick Start

**Step 1: Clone the repository**

```shell
git clone https://github.com/sb-ai-lab/LADS.git
cd LADS
```

**Step 2: Create conda environment**

```shell
conda env create -f environment.yml
conda activate LightAutoDS
```

**Step 3. Set up environment variables**

You need to create a `.env` file in the root directory of the project.

```shell
cp .env_example .env
```

You will need to fill in the required API keys and other environment variables in the `.env` file.

You can also change some parameters in [`config.yml`](./config.yml).

**Step 4: Run the application**

```shell
streamlit run app.py
```

Your application will be hosted on [http://localhost:8501](http://localhost:8501) by default.

## 📊 Result
We evaluated our framework on eight Kaggle ML datasets and compared it with two state-of-the-art open-source solutions: AutoKaggle and AIDE.

To ensure consistency across competitions, we use the Normalized Performance Score (NPS). This score standardizes the results, with a higher value indicating better performance.