Chain Lens Method Detailed: Contrastive Decoding, Attention Masking, State Re-injection
voooooogel · x · 2026-11-12
voooooogel shares the chain lens method and pseudocode: transport the hidden state to the final-layer basis with the J-lens, then roll out tokens with the MTP head. Three key refinements: contrastive decoding against a prior averaged from many hidden states; attention masking so the head attends only to its own token chain; and hidden-state re-injection, without which completions degrade to generic values like round numbers.
Related event: Chain Lens: Training-Free Multi-Token Readout from Model Activations(6 posts)→
More from Research
- Intology Founder at YC Paper Club: Drug Failures Stem From Picking Wrong Targets, Not Bad Molecules — yusufroohani · 2026-11-12
- DeepMind's FlipTest benchmark: best video world models score below random guessing — AlexGDimakis · 2026-11-12
- AI Predicts Tooth Decay Risk from Brazilian Survey Data for Public Health Planning — Brighter-Side-News · 2026-11-12
- Meta FAIR open-sources e-means: 10–45x faster k-means with lower loss than Lloyd's algorithm — generativist · 2026-11-12
- Google and Artificial Analysis launch MKA benchmark to verify agent deliverables — OnlineInference · 2026-11-12
- AI models can prove anything, making 'mathematically trivial' ICLR reviews obsolete — nanjiang_cs · 2026-11-12