Blog
Articles, research notes and updates from the Reliquary team, newest first. Papers and the released model are collected on the research page.
A Market Mechanism for Rollout Selection in Group-Relative RL
Independent operators choose the prompts, commit their own compute before the outcome is known, and are paid only for groups a validator recomputes and finds to carry reward variance.
ReadPaying for what we can actually consume
An update to our incentive mechanism.
ReadReliquary 2B: training done
The results, and what they taught us about the training regime.
ReadEight rollouts are not eight votes
How Reliquary finds the learning frontier today—and the evidence-gated path from equal-difficulty rollouts to a verified behavioral-diversity ledger.
ReadReliquary: a market for the learning frontier
Subnet 81 — decentralized RL training where miners compete to find the prompts the model learns from, accelerating the training process.
Read