ACADEMY
MODULE 4
LESSON 3
Training Tasks, Validation, and Neural Net Growth
Let’s go through a training cycle:
Task assignment
The network pushes a training segment — a slice of a model or dataset — to miners.
Local computation
Miners process it using CPUs or GPUs. Each miner is effectively running part of a federated training system.
Submission
The results are sent to one or more Computors for validation.
Validation and tick inclusion
If the work is accurate, efficient, and useful, the Computor includes it in a tick.
Reinforcement
The system reinforces valid outputs and feeds them forward into future tasks.