← back to paper

Review history

arxiv: 2508.09019 · 2 revisions

Activation Steering for Bias Mitigation: An Interpretable Approach to Safer LLMs

1. 2026-08-15 UNVERDICTED LOW v1.4.0-alltime-deepseek-medium novelty 4.0
   80261 ms 10686 in 8469 out 2026-08-15T17:29:55.291877+00:00
2. 2026-08-05 CONDITIONAL LOW v1.4.0-alltime-deepseek-medium novelty 3.0
   132527 ms 10660 in 12841 out 2026-08-05T21:15:53.489899+00:00