---
title: README
emoji: 🐢
colorFrom: purple
colorTo: pink
sdk: static
pinned: false
---

## Panda Villa Tech Limited

<p align="center">
    <img src="https://raw.githubusercontent.com/PandaVT/DataTager/main/assert/PandaVilla_logo.jpg" width="650" style="margin-bottom: 0.2;"/>
<p>
<h5 align="center"> Grow Together ⭐  </h5>
<h4 align="center"> [<a href="https://github.com/PandaVT/DataTager">GitHub</a> | <a href="https://datatager.com/">DataTager</a>]</h4>  
  
**Long-term Focus:**

- Our company is dedicated to long-term specialization in **synthetic data**, **metaphysics**, and **psychology LLM**, exploring how these fields can intersect with AI.

**Product:** DataTager

**Website:** [DataTager.com](https://DataTager.com/)

- **Description:** DataTager is a tool designed to evaluate and generate the training data needed for large language models. We believe it's more important for individuals and enterprises to fine-tune large models easily and create models tailored to their specific business needs, rather than just choosing models with the highest benchmarks.

**Philosophy:**

- We published a paper titled "AnyTaskTune," advocating that **Task Fine-Tuning** based on real-world scenarios is crucial. This approach is more significant than using universally high-scoring models.

**Resources:**

- We have open-sourced various subtask datasets across multiple domains to support the community. These resources are available on our website for anyone interested in specific task fine-tuning.

Explore more on how to fine-tune your tasks efficiently with our resources at [DataTager.com](https://DataTager.com/).