Affine presents netuid 120 as a reason-mining subnet with champion and challenger dynamics. The README describes miners submitting models that are evaluated across tasks such as program abduction or coding, with validators comparing challengers against the current champion. Affine is best framed as a model-improvement and reasoning-evaluation subnet.
About Affine subnet 120
Affine is a Bittensor reasoning subnet where miners compete to improve models on defined evaluation environments.
Open reasoning models need adversarial, repeatable evaluation loops that reward incremental improvements rather than one-off benchmarks.
Reasoning model developers, RL researchers, Bittensor miners, validators, and users comparing model-training subnets.
Team and ownership
Affine publishes an official dashboard, Discord, X account, and GitHub repository, but reliable named founder, executive, and core-team profiles are not available.