Aurelius subnet 37
Aurelius presents Subnet 37 as a network for improving AI moral reasoning. Miners submit structured ethical-dilemma scenarios, and validators score them through a staged pipeline that can run accepted scenarios through generative-agent simulations. The resulting transcripts and evaluations are intended to support better alignment datasets and moral-reasoning benchmark performance.
About Aurelius subnet 37
Aurelius is an AI-alignment subnet that generates and evaluates moral-reasoning scenarios for training and benchmarking.
AI systems often struggle with nuanced moral reasoning, and alignment datasets need richer, scenario-based evidence rather than only static question-answer examples.
AI alignment researchers, model evaluators, safety teams, and miners interested in scenario design and simulation-based evaluation.
Team and ownership
Austin McCaffrey is the founder associated with Aurelius, which represents new leadership for SN37 after a prior Macrocosmos-led finetuning operation.
- Austin McCaffrey, Founder: SN37 moved from a Macrocosmos-led finetuning operation to an AI alignment project under his management.
Key pages:
learn about Bittensor,
Bittensor glossary,
Bittensor FAQ,
Bittensor subnet screener,
portfolio tracker,
subnet baskets,
Neuralteq validator,
and fees.