{{ c.k }}
{{ c.v }}
{{ c.d }}
Current scoring: Per-subject set F1: twice the number of shared features divided by the total number of predicted and reference features. Two empty sets score 1.
Current subjects and volunteer comparison methods
All 100 current campaign subjects are shown. GPT-6 Astra received the campaign tutorial with low reasoning and the optional field guide off. These are direct API responses from one run on 2026-10-01.
29 subjects came from volunteer-feature-enriched samples. Volunteer comparisons describe this selected cohort and do not estimate performance on the full survey.
Subject-balanced set-F1 against frozen annotations. Selection-biased descriptive comparison: 29 subjects came from volunteer-feature-enriched samples. Not an independent evaluation baseline.
Subject {{ s.id }}
Benchmark target
{{ s.truth }}
{{ modelSub }} answer
{{ s.answer }}
Score: {{ s.scoreText }}
Average volunteer
{{ s.volPct }}
{{ s.volSub }}
Volunteer consensus
{{ s.majorityText }}
Individual volunteer feature sets
{{ g.label }}
{{ row.answer }}
{{ row.pct }}
How this subject was scored
{{ c.name }}
{{ c.scorePct }}
{{ fact.name }}: {{ fact.detail }}
Structured model response
{{ s.raw }}
No subjects match these filters.