Repository navigation
docs(moral): full M-J test report (md+html) + README/docs/blind-judge updates - #20
Merged
Merged
Conversation
… updates Another project consumes this repo's docs from git, so this brings the moral subsystem's documentation in line with the merged, calibrated state. - NEW eval/MORAL_REPORT.md + eval/MORAL_REPORT.html: the full moral-tension test report. Moral behaviour is judged on a fuzzy 1-5 quality scale (believability + curve_plausibility), NOT PASS/FAIL. Covers the Gate-C corpus (126 vignettes, weak situations lifted to 4.0/4.0), litmus (4.83/5), the half-life calibration (serious-guilt ~72h via the Secret moral_weight drip, suspicion 24h), and the deterministic guarantees (pre-filter 18/81, 399 passed, base 1400 corpus byte-identical across the whole M-MEM+M-J merge). - FINAL_report.md + .html (base blind-judge): note that the PASS numbers still hold after the moral merge (opt-in, base corpus byte-identical) + cross-link to the moral report. - README: Reports line + a "Moral-tension overlay -- merged, calibrated" status entry; M-MEM marked merged. - docs/moral_tension_PLAN.md, moral_calibration_plan.md, moral_tension_impl_spec.md: STATUS banners (IMPLEMENTED + CALIBRATED / CLOSED) pointing at the report. GitHub wiki updated separately (new Moral Tension page + Home/Believability-Testing links). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_016rWVvvhWijSuLMEhzjiUhu
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Brings the moral subsystem's documentation in line with the merged, calibrated state. Another project consumes these docs from git, so this is a deliberate, accurate doc pass.
New: full moral test report
eval/MORAL_REPORT.md+eval/MORAL_REPORT.html— moral behaviour is judged on a fuzzy 1–5 quality scale (believability + curve_plausibility), not PASS/FAIL. Covers the Gate-C corpus (126 vignettes; weak situations betrayal/suspicion lifted 3.05/3.29 → 4.0/4.0), litmus (4.83/5), half-life calibration (serious-guilt ~72h via the Secretmoral_weightdrip; suspicion 24h), and the deterministic guarantees (pre-filter 18/81, 399 passed, base 1400 corpus byte-identical across the whole M-MEM+M-J merge).Updated
FINAL_report.md+.html): the base PASS numbers still hold after the moral merge (opt-in, base corpus byte-identical) + cross-link to the moral report.moral_tension_PLAN,moral_calibration_plan,moral_tension_impl_spec): STATUS banners pointing at the report.Docs/report-only — no engine or test changes.
🤖 Generated with Claude Code