Chain Lens Confabulates Less Than Oracle Lens on Poetry Task, Authors Puzzled
voooooogel · x · 2026-11-12
voooooogel reports the chain lens reads workspace content well and confabulates less than the oracle lens on their poetry task — oddly diverging from WorkspaceBench results. On WorkspaceBench, logit + chain lens without transport gives 2x the improvement of J-lens alone, and it's much cheaper to run than activation oracles, though hard-to-parse outputs can cause hallucinations.
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