Tracer
← polyphony · eval · dracula · ci

polyphony · gemini-2.5-flash

✗ fail↓ regression 100% → 50%
Pass rate
50%
Cases
4 (2 failed)
Mode
eval
Suite
dracula
Env
ci
Started
2026-07-15 07:32 UTC
Duration
—
Cost
—
Tokens in/out
— / —
Commit
—
Branch
—
Source
—

Labels

legacy_envci-dracula

Compared with the previous run

Baseline 2026-07-14 20:01 UTC · 100% → this run 50%

cases added / removed
  • + eval:attribution
  • + eval:continuity
  • + eval:prose

Cases (4, 2 failed, failures first)

eval:attributionscore 0.17✗ fail
Output
score=0.1666 ±0.1666 (n=2)  accuracy=0.0 chance=0.3333 n=3
eval:prosescore 0.45✗ fail
Output
score=0.45 ±0.45 (n=2)  words=0
⚖︎ Judge rationale

empty scene (terminal status=failed: generation failed, blocked, or quota-exhausted)

eval:continuityscore 1.00✓ pass
Output
score=1.0 ±0.0 (n=2)  detection_recall=1.0 fpr=1.0
eval:retrievalscore 0.82✓ pass
Output
score=0.8182 ±0.0 (n=2)  precision@k=0.8182 mrr=0.9545