Prime Intellect(@PrimeIntellect):The Goodfire team used Prime Intellect to train activation probes to detect reward hacking. With them, they are able to catch reward hacking in various models. Their probes are performing similarly or better than frontier LLM-as-judge setups, while being more efficient.

Prime Intellect

@PrimeIntellect

Open Superintelligence Stack

加入 June 2020

44 正在关注 84.7K 粉丝

Prime Intellect@PrimeIntellect

2026.09.17 18:44

The Goodfire team used Prime Intellect to train activation probes to detect reward hacking. With them, they are able to catch reward hacking in various models. Their probes are performing similarly or better than frontier LLM-as-judge setups, while being more efficient.

显示更多