feat: evaluation framework with 63 anchor specs and pilot results#343

Merged

rdmueller merged 9 commits into

LLM-Coding:mainfrom

raifdmueller:feat/eval-l1-questions

Mar 24, 2026

Commits on Mar 24, 2026

feat: generate L1 recognition questions for 60 Tier 3 anchors (#331)

raifdmueller
and
claude
committed
feat: full L1 evaluation run with 63 specs across 3 models

raifdmueller
and
claude
committed
feat: add HTML evaluation report generator

raifdmueller
and
claude
committed
fix: use latest results, remove stale runs, regenerate report

raifdmueller
and
claude
committed
feat: generate L2 Application questions for all 59 anchors (#332)

raifdmueller
and
claude
committed
feat: add progress counter [n/total] to evaluation output

raifdmueller
and
claude
committed
fix: calibrate Feynman Technique paraphrase

raifdmueller
and
claude
committed
feat: full L1+L2 evaluation run — 193 questions × 3 models

raifdmueller
and
claude
committed
fix: SemVer spec syntax error, HTML escaping in report

raifdmueller
and
claude
committed