Concept
180 pack self-tests
Twenty public practice questions per knowledge pack, written from the pack's own text, used to see whether a pack helps.
From the record
- In technical terms
- 9 packs × 20 agent-written items whose fact sits in a known chunk; lets the retrieval chain be traced end to end. Dev data, never claim evidence.
Linked facts · 1
- One matching knowledge pack helps the small model a little on questions written from that pack.One matching knowledge pack helps the small model a little on questions written from that pack.Opus +0.042 (+0.024…+0.062); Haiku +0.032 (+0.016…+0.050)Set180 pack self-testsStatusdev (agent-written from pack text)JudgeClaude Opus (L2) and Claude Haiku (L1)n180Seeds1Date2026-10-13CaveatPack-matched dev questions, not a bounty-bar measurement.Sourceresearch record
reports/REPORT-PACKS-2026-10-13.md:8Open the data file