How YOLO learns
The self-learning ledger
A record of changes, not results. YOLO’s analytical brains adjust their own parameters through evidence-gated mechanisms — re-qualification, calibration refits, staged promotions, a human-gated weekend lab. This page is the audit trail of those changes: what changed, when, and the evidence reference behind each entry. It documents the process; it is not financial advice and it is not a promise that any change improves outcomes.
Every entry below was compiled from a durable record another mechanism already wrote — the ledger itself never causes a change, and adopting a new detector always remains a human act behind a pre-registered gate. The companion record of how well the AI expert’s stated confidence has matched reality lives at the public reliability page; the map of the brains themselves is in How YOLO thinks.
Recording has just begun. The ledger compiles once a day (05:10 UTC) from the brains’ own learning records; entries appear here as those mechanisms commit changes. An empty page today is an honest zero, not a missing feature.