E006 — Diversity Search
Diversity Search records verified real-time milestones with zero model calls, original run evidence and exact replay.
Read the updateExplore the laboratory
Experiments
Six experiments separate infrastructure, perception, controllers, Qwen selection and automated policy search.
Each experiment records its branch, evidence, findings and useful limitations. The initial release is a frozen checkpoint supplied by the laboratory.
Validated checkpoint: 2026-10-05. Data last synchronized: .
Latest record verified from reviewed published run evidence. Registry counts retain the last full dataset checkpoint; record publication does not refresh the complete dataset. This is a snapshot, not a live counter.
Registry count checkpoint: 2026-10-05T08:28:16.420013+00:00. The latest reviewed record and its evidence update separately from full dataset totals.
E005 — Policy Search milestone
Automated policy search reached a real-time milestone with an exact deterministic replay and no model calls during the run.
Read the updateE004 — Qwen selection & timing
A higher-level Qwen policy selector was tested against deterministic controllers and Search.
Read the updateE003 — Controllers & Search
Fast deterministic controllers and automated policy exploration form a separate branch from model-controlled gameplay.
Read the updateE002 — Perception benchmark
On 200 visible-player cases, Qwen 9B with 4× nearest-neighbour scaling and three-frame history measured 90% Exact and 1.5% Opposite errors.
Read the updateE001 — Infrastructure & first gameplay
The first automated cohorts established observation, real game inputs, telemetry and deterministic replay.
Read the update