lab notes / six stations

Lab notes. The maze, walked.

Six chapters, one labyrinth. Each post is a station deeper into the engine — walk them top-down for the full arc, or jump in anywhere. Hardfacts only: every claim traces to a JSON in benchmarks/external/results/ or an audit transcript in benchmarks/audit/.

Operator commentary · 2026— · 6 stations · 0 fabricated numbers

First time? Walk top-down from station 01 — the arc only makes sense in order.

Just want the numbers? Jump to the summit, then the research page.

Methodology-first? Start at the junction, where we audit the benchmarks themselves.

station 06 · the arena2026-07-05flagship · external

The arena we did not build: EverMemBench, official harness.

Our first third-party benchmark — their harness, their judge, their protocol, unmodified. 48.7% category mean against EverMemOS’s published 44.6 on the leaderboard protocol (gpt-4.1-mini answers), and 51.5% with deepseek-v4-flash at a fifth of the answer cost. Plus the honest part: the five engine bugs the arena surfaced — multi-word tsquery crash, idle-connection corpses, event-time-blind imports, a silent supersedes no-op, the REM cold-start flood — every one fixed and shipped in 1.0.0-rc7 before we wrote a word about winning.

48.7 vs 44.6 mean 53.5% pooled · flash 5 bugs fixed & shipped
enter station 06 →

station 05 · the summit2026-05-18flagship

Inside the 100-iteration loop.

The full bench-driven engineering loop: iter00 anchor R@5 = 0.6851 to iter100 champion R@5 = 0.8426, across four architectural eras (retrieval → formation → rerank-feedback → top-K sharpness). The empirical proof that the bottleneck migrated upward in the cognition stack — and it is as honest about the failed levers as the winning ones: naive canonicalization regression, salience 3.0 over-boost, the KU rebake we rolled back.

iter00 → iter100 R@5 0.6851 → 0.8426 R@10 = 0.9000 R@1 = 0.6255 4 eras
enter station 05 →

station 04 · the junction2026-05-18flagship · methodology

Inception Benchmarking: the benchmark that did not exist.

Why we built our own memory benchmark instead of trusting the published ones. The Hindsight int() truncation bug — 9 of 10 ability evaluators silently dropped 0.5 partial credits. The 4/20 rubric-defect rate on BEAM-10M conv-1, cross-confirmed in a blind audit. The 16pp judge spread on identical answers. The 10 models published as “0/N · not viable” that we ran at 90%. If your benchmark cannot survive its own audit, your #1 is decoration, not signal.

9/10 evaluators bugged 4/20 rubric defects 16pp judge spread 0/N → 18/20
enter station 04 →

station 03 · the climb2026-05-18benchmarks · formation

Formation beats retrieval-tuning.

100 iterations on the hard LongMemEval-oracle 500q across four architectural eras. R@5 climbed 0.6851 → 0.8426 (+15.75pp absolute, crossing the 0.84 stretch target), R@10 broke the 0.90 barrier, ssu R@10 hit a perfect 1.0000. The retrieval-side knob surface saturated at 0.7404; memory formation broke through it; the rerank-feedback discovery pushed further still.

R@5 = 0.8426 R@10 = 0.9000 MRR = 0.7124 ssu R@10 = 1.0000
enter station 03 →

station 02 · the corridor2026-05-13infrastructure

The bench corpus now lives in Postgres.

One pg_restore, one sha256, four BEAM scales plus the full LongMemEval triplet. Plus a 282× bulk-write refactor and the two pg_attribute-namespace bugs that were silently leaking embedding dimensions across schemas.

957 MB snapshot ~3 s restore 88 min → 75 s 207 provenance rows
enter station 02 →

station 01 · the entrance2026-05-10audit

Memory benchmarks should measure memory.

The note that started the maze. Why we ran the same 10 models Hindsight published as “not viable”, and what 270 million parameters can do when you stop gating on JSON. Eight rounds of GPT-5.5 review, v2 NO → v8 unconditional yes, no residual caveat.

R@5 = 0.9787 MRR = 0.9114 188/200 = 94.0% 0 errors
enter station 01 →

More stations as the maze grows. Audit transcripts at benchmarks/audit/, result JSONs at benchmarks/external/results/. Full claim-evidence bundle — 1.6 GB, restorable pg_dump plus all iter JSONs, the 8-round audit and the bench-loop logs — on ProtonDrive, SHA-256 263e2494….

Read the notes.
Then run the bench yourself.

Every number on this page is reproducible from the bundle. The pod that produced them installs in one line.

one line · no sudo
curl -fsSL https://api.mazemaker.dev/install.sh | bash