egan

Karma: 171

Measuring Spurious Correlations with Feature Strength

egan11 Aug 2026 18:05 UTC

36 points

1 comment10 min readLW link

Reward Laundering: LLMs Can Gain Unintended Behaviors by Deciding When to Earn Their Rewards

egan, abhayesian and Jozdien

31 Jul 2026 15:48 UTC

76 points

3 comments4 min readLW link

Hint-based CoT faithfulness evals still mostly work on Claude

egan30 Jul 2026 20:44 UTC

24 points

0 comments6 min readLW link

Research Sabotage in ML Codebases

egan, Vivek Hebbar and Julian Stastny

30 Apr 2026 0:26 UTC

63 points

4 comments6 min readLW link

How will we do SFT on models with opaque reasoning?

Alek Westover, Vivek Hebbar and egan

21 Feb 2026 0:00 UTC

32 points

17 comments7 min readLW link

Four Downsides of Training Policies Online

Alek Westover and egan

4 Jan 2026 3:17 UTC

30 points

4 comments3 min readLW link