Protocol overview
In a PoT scheme, a service provider needs to show a network of coordinator nodes that they have properly trained a model as requested by a client. This process also involves a group of validators who check the work. The scheme helps in picking out the best-trained model.
The PoT protocol includes four main steps:
Claim: In this step, the service provider says they have trained the model using the client's instructions and data. They keep a record of their work. This claim is their way of asking for a reward for their work and is necessary for the next steps of the process. They might need help from other services for storing the model or setting it up.
Validate: Miners in the network check the service provider's work. They send out a report stating how well the model works and who made it. This step is crucial for recognizing the service provider's efforts.
Verify: It works by checking the validators. Any member of the network can do this to make sure the validators did their job right. If someone finds a mistake and reports it, they could get a reward. This ensures everyone is honest and accurate.
Finalize: The final step is where the decision-makers, known as coordinators, review all the information and decide who gets the rewards. The best model gets the most reward, but the validators also get some. This encourages everyone to do their best and be honest in their validation.
Overall, the PoT scheme is a way to ensure that when a service provider trains a model, their work is thoroughly checked and validated. The scheme promotes quality work, honesty, and active participation by rewarding the best model and the validators who maintain the process's integrity.
Last updated
