eigen

For VB
Join VBs DAO guys!
Our process

Inspiration 👾

Entropy Labs has developed a concept called EIGEN, which is a platform that hosts DAOs (Decentralized Autonomous Organizations) focusing on training AI models. Our team was inspired by the concept of DATA DAOs and wanted to explore how Web3 could improve data availability and decentralization compared to Web2 platforms like Kaggle and User testing. In our brainstorming sessions, we identified various use cases for data, such as real-time user feedback for product testing, research data on topics like air quality, and data for training AI models. However, we realized that privacy concerns might hinder the willingness of users to share certain types of data, leading you to drop some ideas.

What it does 👽

Eigen is a platform that fundamentally transforms the process of fine-tuning AI models by making them more accessible, faster, and easier. It achieves this by creating a decentralized, community-driven ecosystem where data contribution is incentivized, thereby encouraging a wide range of individuals to participate in the AI model training process.

Traditionally, fine-tuning AI models has been a time-consuming and complex process, often requiring significant resources and expertise in data collection and processing. This has been a major barrier to entry for many individuals and organizations looking to leverage AI technologies.

Eigen addresses this challenge head-on. By incentivizing data contribution, it encourages a diverse range of individuals to contribute the data needed to fine-tune AI models. This not only accelerates the data collection process but also ensures a diverse and robust dataset, which is crucial for training effective AI models.

Moreover, Eigen's innovative approach ensures that contributors are rewarded for their efforts. Every time the AI models trained with the contributed data are used, the contributors receive a royalty split. This creates a sustainable and equitable reward mechanism that recognizes and values the contributions of each individual.

The flow of the platform includes the creation of a DAO by the owner, volunteers contributing data, the interaction between volunteers/DAO members and the DAO, the owner reviewing data to train models, storage providers claiming contributions to store data, and end-users using the output.

How we built it 👁

We used the design thinking process (Discover > Define > Make ) x n

The following technologies were used to develop Eigen:

Next.js for the frontend
Solidity for smart contract development
Chainlink VRF for democratically and fairly choosing moderators in the DataDAO
Chainlink automation for fine tuning the model when the number of data sources criteria is met.
Polygon blockchain for deploying smart contracts

Challenges we ran into 🤖

Experience POV : Craft a flow that would explain the story was tough. The challenge was to make it look fun and not intimidating and give a community vibe to it. Keep the interactions familiar but also exciting. In all, not make it look like work but fun.

Technical POV : Integrating a DAO and then using chainlink functions to fine-tune the model based on the data ingested was hard. Making sure that flagged content is not passed on to the fine-tuner was also challenging.

Accomplishments that we're proud of 🙌

We're proud of successfully developing an innovative platform that democratizes the fine-tuning of AI models by incentivizing community participation. The implementation of a sustainable and equitable reward mechanism that recognizes individual contributions is another significant achievement.

What we learned ✌️

We learned the immense potential of community engagement in data collection and AI model training. The project also provided valuable insights into the synergy between blockchain and AI, the importance of fair reward systems, and the transformative potential of Decentralized Autonomous Organizations (DAOs).