HN Trends Tracker

HN Trends Tracker

backend ok

Repost history

Canonical URL: https://joinhandshake.com/research/ai/deepswe-reward-hacking

Current story

#
LLM makes decisions to raise its training scores, and ignore user directives(joinhandshake.com)
HN #49,886,2062026-09-29 00:07 UTC
3 pointsby guardiangod28 days ago1 comments

Prior reposts (1 of 3 total)

#
Speculative Reward Hacking in Coding Agents(joinhandshake.com)
HN #49,851,2662026-09-25 23:13 UTC
3 pointsby _jonas31 days ago2 comments