Timing matters: attention injection during reading boosts Qwen's report by up to +580%
Sauers_ · x · 2026-11-19
A key variable in Sauers's experiments: injecting attention while Qwen 3 32B reads and answers the "where is your attention" question raises the probability of naming the boosted target by +97% (paragraphs) to +580% (words), whereas injection during answer generation yields -18% (paragraphs) to +3% (words). Timing strongly modulates the introspection effect, though answer-time injection isn't strictly required.
More from Research
- FoodTruck Bench, an LLM agent business-sim benchmark, lands on Papers with Code — NielsRogge · 2026-11-19
- MIT reads and steers LLMs' internal physics intuition via hidden-state shifts in Gemma — ProfBuehlerMIT · 2026-11-19
- harness-hub open-sources RL-trained LoRA adapters tuned per agent harness, recipe coming Friday — yb2698 · 2026-11-19
- RL-Trained Hermes LoRA on Qwen3.8-27B Adds +13.7 Pts on SWE-Bench Pro While Using Fewer Tokens — yb2698 · 2026-11-19
- Freenet releases new video on network routing with old-school ML — Sanity · 2026-11-19
- Empire AI creates 120+ early-career research positions in New York, Cornell deadline Dec 15 — dianarycai · 2026-11-19