Chain lens: training-free multi-token activation readout beats J-lens 0.42 vs 0.32
voooooogel · x · 2026-11-12
vgel and Raymond Douglas release a draft paper introducing the chain lens, which extends single-token lenses like the J-lens into multi-token readouts via contrastive decoding on an MTP head — no training needed. It scores 0.42 vs 0.32 for J-lens across all 28 WorkspaceBench families, beats a hindsight composite of the best untrained lenses (0.49 vs 0.41), and even surpasses a trained oracle lens on arithmetic chain intermediates (0.42 vs 0.25), though readouts remain fragmentary.
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