polyphony · claude-opus-5-5
✓ pass≠ not comparableMetrics
| prose.score | 0.935 |
| prose.words | 3,632 |
Labels
| cadence | on-demand |
| experiment | w16-r2-1 |
| harness_sha | 47632720c43762568ae9087c9bc6abedfe3aee3f |
| judge_provider | groq+anthropic |
| judge_self_graded | True |
| repeat | 1 |
| trigger | workflow_dispatch |
Compared with the previous run
Baseline 2026-10-24 06:14 UTC · 100% → this run 100% · code diff ↗
- + eval:prose
- − eval:attribution
- − eval:continuity
- − eval:extraction
- − eval:ingestion
- − eval:retrieval
Cases (1)
words=3632 path=planned rubrics={'R1': 0.9825, 'R2': None, 'R3': 0.9825, 'R4': 0.9575, 'R5': 0.9125} seams=0/0scene 1 R1: All required canon characters appear correctly and behave as described, with no extraneous or mischaracterized figures. | scene 1 R3: All planned beats appear in correct order with their specified changes and no unplanned major events. | scene 1 R4: The scene maintains consistent first‑person‑limited past tense POV, matches the gothic yet courteous tone, and presents varied, concrete prose without labels or repetitions. | scene 1 R5: The prose aligns with canon and planned beats, containing no contradictions and fulfilling the continuity criteria fully. | scene 1 R2: not applicable: under two cast speakers with voice samples have attributed lines | scene 2 R1: All required canon characters appear correctly and behave as described, with no extraneous or mischaracterized figures. | scene 2 R3: All five planned beats appear sequentially, each change occurs as described, with no extra or out‑of‑order events. | scene 2 R4: The scene maintains consistent POV and tense, matches the requested tone, and reads as polished narrative prose without structural errors. | scene 2 R5: The prose aligns with canon and prior text, introduces no contradictions, and fulfills the planned beats without error. | scene 2 R2: not applicable: under two cast speakers with voice samples have attributed lines | scene 3 R1: All principal characters required appear correctly with accurate names and behavior, meeting the rubric fully. | scene 3 R3: The scene realizes every planned beat in the correct order with all specified changes and no extra unplanned events. | scene 3 R4: The scene maintains consistent POV and past tense, matches the tone, avoids labels, and reads as publishable prose with only minor generic phrasing. | scene 3 R5: The scene follows the beats and prior text closely, with only minor slips: the window looks onto a mist-filled gorge yet the glass lands on courtyard stones, and Aldous the solicitor thinks of himself as a 'tired clerk'. | scene 3 R2: not applicable: under two cast speakers with voice samples have attributed lines | scene 4 R1: All three beat-cast principals (Aldous Kerr, Count Vasska, and Nora Vance, who is present in the locket photograph and spoken name) appear exactly as planned with exact names, and the only other named figure, Mr. Hawkins, is a minor reference with no scene role, so the cast is faithful; the Count is | scene 4 R3: All five beats are realised in order with their changes landing (signing, locket falling open, Nora's name and town given with the map traced, cold locket returned and the Count's gaze shifting, Aldous alone and uneasy); the only small deviations are minor additions such as 'Carfax' being named and | scene 4 R4: The close-third past-tense POV from Aldous is consistent, the Gothic tone of courtesy over menace fits the prior text, and the prose is concrete and varied with no labels or summary; minor slips include a few generic phrases, 'God help him', and the 'for a week' timing. | scene 4 R5: The scene follows the prior text and all five planned beats (razor cut, the empty glass, hairy palm, cold fingers, crucifix, Nora's name and town) without contradicting anything; the only risk is a few invented details (Mr. Hawkins, Exeter, Carfax, Thornmouth, 'a week') that the canon neither suppor | scene 4 R2: not applicable: under two cast speakers with voice samples have attributed lines