Attention control works on single words too: +124% focus, -21% ignore
Sauers_ · x · 2026-11-19
Extending to word-level control, Sauers shows that telling Qwen to "focus your attention on word [W4]" boosts attention to that word by +124%, while "completely ignore" reduces it by -21% (both p<10⁻⁵), with per-layer decomposition again peaking around layer 28 — showing attention control works beyond paragraph granularity.
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