skip to content

dashboard miners my fleet live training research source x

DISTRIBUTED POST-TRAININGBITTENSOR SUBNET 81

Products Dashboard Lab Mining Research

Sign in Open workspace ↗

Menu +

Products ↗Dashboard ↗Lab ↗Mining ↗Research ↗Open workspace ↗Sign in ↗

All tasks SFT generation RL generation

RELIQUARYDistributed post-training. Open evidence.

Documentation Lab Research Workspace ↗

Reinforcement-learning training on a decentralized network, on Bittensor Subnet 81. Rollout generation is distributed across independent operators; the objective is to lower the cost of an RL run.

/dashboard · live
/research
/status
source · reliquary
x · @reliquary_ai

/dashboard · leaderboard
/ledger · rollout records
/explorer · proof windows
/mesh · validator design

/forge · checkpoint evidence
/forge/cycles · live attestations
/training · the training loop
/paper · protocol spec

mit · reliquadotai · 2026

RL / CODE · MATH · LOGIC

Generation.

Live validator activity and sealed-window evidence. Each source loads independently.

Three environments.

Results from the last scored window in R2. These are separate from the live window above.

seats per lane

reading the archive

Training & checkpoints.

training · published evidence

read from reliquary/training and reliquary/checkpoints

reading the forge