|
|
--- |
|
|
title: README |
|
|
emoji: 🏃 |
|
|
colorFrom: gray |
|
|
colorTo: purple |
|
|
sdk: static |
|
|
pinned: false |
|
|
thumbnail: >- |
|
|
https://cdn-uploads.huggingface.co/production/uploads/68ec78cca886edada26b90b0/Gxhtz7kQOxTC7j_xX7QJ3.png |
|
|
short_description: '**Madlab** is a self‑hosted tool for fine‑tuning language mo' |
|
|
--- |
|
|
|
|
|
# MadlabOSS |
|
|
|
|
|
|
|
|
 |
|
|
|
|
|
**Madlab** is a self‑hosted tool for fine‑tuning language models on your own hardware. |
|
|
It automates the tedious parts of training so you can focus on experimentation and iteration. |
|
|
|
|
|
## What Madlab does |
|
|
- Formats datasets automatically |
|
|
- Synthetic dataset generator |
|
|
- Generates training configurations |
|
|
- Runs supervised fine‑tuning with PyTorch/Transformers |
|
|
- Converts models to GGUF |
|
|
- Evaluates outputs with built‑in metrics |
|
|
- Works with many Hugging Face models |
|
|
|
|
|
## Tech Stack |
|
|
- **Frontend:** React |
|
|
- **Backend:** Node.js |
|
|
- **Training:** Python (PyTorch + Transformers) |
|
|
|
|
|
## Philosophy |
|
|
Madlab aims to make local fine‑tuning: |
|
|
- simple |
|
|
- reproducible |
|
|
- transparent |
|
|
- hardware‑friendly |
|
|
|
|
|
## On this Organization |
|
|
This Hugging Face organization hosts: |
|
|
- example models trained with Madlab |
|
|
- reference outputs |
|
|
- compatibility tests |
|
|
|
|
|
Madlab is open‑source and built for the community. |
