eng-0143
window recorded
citable URL: https://halobench.com/records/eng-0143/ — this address never moves; the anchor /records/#eng-0143 keeps resolving
608.6 Whmean 182.04 W − 10.1 W idle floor (box-idle-all-empty) = 171.94 W attributable
window 2026-08-17T00:18:28Z → 2026-08-17T03:39:04Z · provenance recorded
method wall-meter · aihydra energy sensor via Home Assistant (counter-difference on sensor.hardware_ai_hydra_energy, raw ~10s-resolution history samples, linearly interpolated to the run-meta window) · no contention
per task 304.3021 Wh · 9.2203 p at 30.3 p/kWh
covers run-0360
aihydra qwen36-27b-mtp-fullbench tau2 arm2 (continuation, task-ids 21-25) window (00:18:28Z-03:39:04Z, 12036s / 3.34h). wh_per_task is Wh-per-CORRECT-ANSWER for THIS ARM ONLY: 608.6042 / 2 = 304.3021 Wh, 9.2203 pence at 30.3 p/kWh — a small-N figure (only 2 of 5 passed, including task 23's ~90-minute cumulative wall time) that is NOT representative alone. See clm-0076 and eng-0141 for the COMBINED 26-task figure (2055.19 Wh / 16 correct = 128.45 Wh per correct answer), the one that belongs on the model page. Counter interpolated from raw ~10s-resolution HA history samples pulled fresh this session: start 23.8800284894 kWh, end 24.4886327086 kWh.
Wh, never joules — read as a cumulative-counter difference across the window; delta over the idle baseline is the only figure that means anything
Cited by — computed at build time, never stored
claims clm-0076