Introduction 🚀

Introduction to the Training Gym

What is the Training Gym?

The Training Gym is a platform where users can participate in AI-related “races” (multi-task challenges) for fun and rewards. Participants can choose from:

  • Free Races: Require holding a certain amount of $VIRAL but do not require any token to be staked or spent.

  • Staked Races: Require staking $VIRAL, and reward pools come from the participants’ stakes, plus any forfeited tokens from those who fail.

By mixing free and staked races, we aim to onboard newcomers quickly while rewarding advanced users with higher stakes and higher potential gains.

Want to know when new free races open up? Join our Telegram for real-time notifications and community discussions.

Key Features

  • Community-Submitted Tasks: Users can submit their own tasks, contributing to a wide range of diverse data and challenge types—spanning everything from simple code exercises to advanced AI tasks.

  • Automated Trap Tasks: Certain tasks have evaluators that detect malicious or incompetent submissions.

  • Token Holding & Staking: For free races, just hold $VIRAL for a specific period to qualify. For staked races, you can risk your $VIRAL for bigger potential payouts.

  • Clear Incentives: Successful participants in staked races can earn more than they risk, thanks to the redistribution of forfeited stakes from those who fail trap tasks.

  • Build your own agent: One of the most exciting features of the Training Gym is the ability to export your performance data to build your own AI agent. By participating in races, you generate high-quality training trajectories that can be:

    • Exported in OpenAI JSONL, ChatML, or raw video + JSON formats.

    • Used to fine-tune your own Large Action Model (LAM).

    • Applied directly to your own projects, whether for gaming, office automation, or custom workflows.

    • Coming soon: Trade your data on a agentic skill market

The Training Gym not only rewards participants but also enables them to take the next step in developing powerful, personalized AI agents.

Last updated