Training MAI
At the current moment, the training and fine-tuning of MAI, our official AI chatbot, is exclusively handled by Mirror. We utilize a strategic compilation of diverse data sources to create robust language models capable of understanding the nuances of blockchain technology and the rapidly evolving DeFi landscape. This data pipeline includes:
Live Ethereum network activity, including transaction data, smart contract deployments, and on-chain analytics
Defi (decentralized finance) research documents, whitepapers, and technical specifications from leading projects and protocols
Target videos from educational resources, project updates, and community discussions related to blockchain and cryptocurrency
Target social media activity across platforms like Twitter, Reddit, and crypto-focused forums, capturing sentiment, news, and trends
User conversations and queries, providing insights into common pain points and information needs
The combination of on-chain data, authoritative research materials, multimedia resources, community discussions, and real-world queries enables our models to develop a comprehensive understanding of the subject matter, allowing MAI to provide accurate and contextually relevant responses to users.
However, in the future, we plan to leverage the power of the Mainet ecosystem to crowdsource and incentivize contributions to MAI's training process. Users will be able to participate in creating and validating data, as well as offering computational resources for model training, all while earning $MAI tokens for their contributions.
Through the Mainet platform, users can:
Contribute Data: Run aNodes to gather, curate, and validate data relevant to MAI's training, earning $MAI tokens for their data services.
Offer Computational Resources: Operate tNodes and list their hardware capabilities and hourly rates on the marketplace, earning $MAI tokens for providing computational resources used in training MAI.
This open and collaborative approach not only democratizes the development of cutting-edge AI models like MAI but also creates a self-sustaining ecosystem where individuals and organizations can actively participate and benefit from the process.
Last updated