Home › Evidence › Records › run-0274
⚠ This notebook has stopped taking notes: newest record is 23 days old (eng-0280), against an expected cadence of 14 days.

run-0274

capability
citable URL: https://halobench.com/records/run-0274/ — this address never moves; the anchor /records/#run-0274 keeps resolving
date 2026-08-15 · window 2026-08-15T10:48:12Z → 2026-08-15T11:17:09Z
suite nemotron3-super-q4km-screen · config cfg-0072
depth 32,768 tokens · scoring deterministic

Metrics

metricvalue
tasks_passed4
tasks_total5
mean_score0.8

turns: total 130 · median 28 · p90 44 · max 44 — machine-written trailer, parsed at build time

Guard chain

not a performance run — no guard required

Energy

metered window: eng-0084

Per-task trace — the full record, no truncation

taskrewardturnsseconds
0112190
1028249
2132512
3114229
4144554

reward null = infrastructure error, excluded from every paired computation

Cited by — computed at build time, never stored

candidate gate history nemotron3-super