Phase 1 — Mainnet Launch
Goal
Launch a minimal but credible alignment subnet that produces structured supervised fine-tuning datasets.
What is happening
The alignment data engine is switched on. Miners submit prompts, validators score completions using open-source moderation tools, and the first structured datasets begin flowing on-chain.
Why it matters
This establishes proof-of-concept that Bittensor can generate alignment datasets at scale through a distributed network of miners and validators.
Unlocks:
Content Moderation Alignment Datasets
These datasets can be used by open-source model developers and alignment researchers for supervised fine-tuning or as baselines for early post-training experiments.
Value Accrual
Subnet Ecosystem
Aurelius begins producing its foundational asset: alignment data.
Early datasets carry immediate value for alignment researchers.
The subnet gains initial visibility, attracting early participants and sparking the first commercial conversations.