Manifold curvature and intrinsic dimension predict layerwise SAE width exponents and asymptotic floors across Gemma models, with cross-model transfer of the geometric regression, establishing a transferable geometric law instead of a universal scaling law.
Connor Kissane, Robert Krzyzanowski, Arthur Conmy, and Neel Nanda
7 Pith papers cite this work. Polarity classification is still indexing.
7
Pith papers citing it
citation-role summary
background 1
dataset 1
method 1
citation-polarity summary
years
2026 7verdicts
UNVERDICTED 7representative citing papers
LOCA identifies an average of six minimal interpretable changes in intermediate representations that causally induce refusal on otherwise successful jailbreaks for Gemma and Llama models.
LLMs show a grounding gap with humans on abstract concepts, with property-generation correlations at most r=0.37 versus human-to-human r>0.9, though larger models align better on explicit rating tasks and internal SAE features capture some grounding dimensions.
Activation steering is cast as constrained optimization that minimizes collateral damage by weighting perturbations according to the empirical second-moment matrix of activations instead of assuming isotropy.
Blocking a fixed set of latent features during fine-tuning reduces emergent misalignment by up to 95% across six domains with no loss in target task performance.
ConceptTracer supplies an interactive interface and saliency/selectivity metrics to locate concept-responsive neurons in neural representations, shown on TabPFN.
Qwen-Scope provides open-source sparse autoencoders for Qwen models that function as practical interfaces for steering, evaluating, data workflows, and optimizing large language models.
citing papers explorer
-
Qwen-Scope: Turning Sparse Features into Development Tools for Large Language Models
Qwen-Scope provides open-source sparse autoencoders for Qwen models that function as practical interfaces for steering, evaluating, data workflows, and optimizing large language models.