For model trainers
As model trainers, your role is pivotal in the secure and efficient training of AI models. Here are your responsibilities:
Homomorphic Encryption for Privacy: You can encrypt the parameters homomorphically and send the ciphertext along with the public key to data owners. This allows data owners to train within their trusted environments without exposing sensitive data.
Secure Gradient Aggregation: After training, one of the data owners can act as an aggregator to combine the gradients. The aggregated gradient is then sent back to you, the trainer, maintaining the confidentiality of the data throughout the process.
Model Packaging as NFT ($Model): Once you've trained a model, you can package it into an NFT ($Model) for use by others. Users who wish to utilize your model will pay a certain number of tokens as compensation.
Token-Based Ecosystem for Improvement: The tokens you earn can be reinvested into further training, allowing you to refine and enhance model performance, which in turn delivers a better experience to users.
Your expertise not only contributes to the advancement of AI models but also nurtures a token-based ecosystem that rewards continuous improvement and collaboration.
Last updated