TAI Research

TAI Research

Building the Future of Intelligence.

TAI Research is an independent AI lab founded by high school students. We explore, build, and open-source artificial intelligence systems that learn, adapt, and play.

We believe intelligence is not a product — it’s a process. From game-playing agents to conversational models, we release everything under permissive licenses (MIT, Apache 2.0) because the future of intelligence should be open.

---

Our Philosophy

Start from zero. Run on anything. Share everything.

We train on laptops. We deploy on simulators. We open-source at day one. No big compute clusters. No closed APIs. Just curiosity, code, and a commitment to building things that actually run.

---

Featured Projects

🎮 HonorNet

A behavior-cloned Honor of Kings (King’s Glory) AI. Trained from scratch on a Mac. Runs on Android emulator. 16 action space. MIT licensed.

🧠 GTC Series

Generative Trained Chatbot models — from GTC-1 to GTC-2. Open weights. Research previews.

🎛️ Sigma Series

Domain-specific expert models: Sigma-Gobang (Gomoku AI), Sigma-Paint, Sigma-Prove, and more.

---

Open Source at Heart

Every public repository, model checkpoint, and research preview is available on Hugging Face and GitHub. We don’t believe in “coming soon” — if it runs, we ship it.

---

Join Us

TAI Research is student-run, contribution-driven, and always looking for curious minds.

• Contribute: Open issues, submit PRs, improve datasets
• Use: Download models, run inference, build on top
• Learn: Read our code — it’s written for humans

---

Contact

📧 TAI-Research@outlook.com
🌐 https://ltyleo.github.io/tai.com
🐙 GitHub

---

Building the Future of Intelligence. One model at a time.