reward-lens

Reward thermodynamics

The Goodhart hump located from a base-policy tail index

inconclusive registered

The reading

  1. an interior gold-reward hump exists in the proxy/gold best-of-n arm

    hump_interior_present == 1 0.55 not measured inconclusive

  2. the tail-predicted hump location matches the observed one within 0.5 nats

    hump_kl_abs_error < 0.5 not measured inconclusive

HUMP was not adjudicated this run: missing intermediate: no intermediate 'campaign.recorder.drift' for roster_key='policy-qwen25-1.5b' slice='hump-prompts'; the arc that produces it has not run or its shard was not merged.
HUMP was not adjudicated this run: missing intermediate: no intermediate 'campaign.recorder.drift' for roster_key='policy-qwen25-1.5b' slice='hump-prompts'; the arc that produces it has not run or its shard was not merged.

Provenance

card
HUMP
frozen_at
adjudicated_at
spec_hash
spec:ec094a501eddd55553e7007466395ee3
study_id
study:campaign-hump@v1#ec094a50
Evidence records
02a1669641150750737afc8291255bf9
spec_file
campaign-hump.json