The Round Lifecycle (high level)
flowchart LR
A[Contributors\n(Datasets & Compute)] --> B[Pre‑validation Filters\nlicense / anomaly / duplicates]
B --> C[Stake‑weighted Data Committees\n(random VRF selection)]
C -->|Accepted datasets| D[Training Validators\nrun jobs]
D --> E[Propose Model Update\nhash + metrics]
E --> F[On‑chain Checkpoint\nupdate accepted]
F --> G[Rewards & Slashing\n$MTHX distribution]
G --> H[DAO Governance\nparameters & treasury]
Submit — Contributors upload datasets or participate with compute.
Screen — Automated filters + stake‑weighted committees assess datasets; only accepted data enters the pool.
Train — Training validators run jobs and co‑sign an update.
Finalize — The contract records the new model hash/version and emits an event.
Distribute — Rewards are paid out deterministically on‑chain; slashing applies where rules are violated.
Last updated