← polyphony · eval · frankenstein · ci
polyphony · claude-opus-5-5
✓ passMetrics
| attribution.accuracy | 0.267 (was 0.400) |
| attribution.chance | 0.333 |
| attribution.score | 0.267 (was 0.400) |
| continuity.detection_recall | 1 (was 0.500) |
| continuity.false_positive_rate | 0 |
| continuity.score | 1 (was 0.500) |
| extraction.f1 | 0.636 (was 0.737) |
| extraction.precision | 0.467 (was 0.583) |
| extraction.recall | 1 |
| extraction.score | 0.636 (was 0.737) |
| ingestion.score | 0.643 (was 0.667) |
| retrieval.mrr | 0.833 |
| retrieval.precision_at_k | 0.815 |
| retrieval.score | 0.815 |
Labels
| cadence | on-demand |
| experiment | w15-post-2 |
| harness_sha | ececd1595c9bfb5b766355795d60833b965ec6a6 |
| judge_provider | groq |
| judge_self_graded | False |
| repeat | 1 |
| trigger | workflow_dispatch |
Compared with the previous run
Baseline 2026-10-24 06:05 UTC · 100% → this run 100%
± score moved · 3
- eval:attribution: 0.40 → 0.27
- eval:extraction: 0.74 → 0.64
- eval:continuity: 0.50 → 1.00
Cases (5)
gated 4/4 passed
eval:continuityscore 1.00✓ pass
score=1precision=0.667n_findings=3n_injected=2quote_verbatim=1detection_recall=1n_control_findings=0false_positive_rate=0
Output
detection_recall=1.0 fpr=0.0
eval:extractionscore 0.64✓ pass
f1=0.636score=0.636recall=1windows=2precision=0.467f1_selected=0.714recall_selected=0.714precision_selected=0.714principal_fields_rate=1principal_mentions_min=4
Output
F1=0.6364 P=0.4667 R=1.0 predicted=['Soraya', 'Fenwick', 'the creature', 'his father', 'Alina', 'Emeric Vandel', 'Isolde Sarrow', 'Hollis Crane', 'Cosima', 'Rowan Merrow', 'Uncle Thomas', "Crane's father", "Vandel's father", 'the lieutenant', 'the master']
eval:ingestionscore 0.64✓ pass
score=0.643grounded=9avg_chunks=8.43n_characters=14
Output
grounded=9/14 avg_chunks=8.43
eval:retrievalscore 0.81✓ pass
mrr=0.833score=0.815n_queries=9precision_at_k=0.815
Output
precision@k=0.8148 mrr=0.8333
reported 1 reported
eval:attribution0.27 · reported
n=30empty=0score=0.267chance=0.333correct=8accuracy=0.267mean_gap=0.050generated=30
Output
accuracy=0.2667 chance=0.3333 n=30