arXiv:2506. 07406v3 Announce Type: replace-cross Abstract: Understanding the internal representations of large language models (LLMs) is a central challenge in interpretability research.
arXiv:2609.37731v1 Announce Type: new Abstract: Activation space and parameter space provide complementary views of model computation. Activations represent information, while weights read, transform...
arXiv:2602. 06065v3 Announce Type: replace-cross Abstract: Understanding how the structure of language can be learned from sentences alone is a central question in both cognitive science and machine learning.
arXiv:2607. 29378v1 Announce Type: cross Abstract: Large language models (LLMs) generate text by auto-regressively sampling the next token.
arXiv:2601. 22594v2 Announce Type: replace-cross Abstract: The high-level concepts that a neural network uses to perform computation need not be aligned to individual neurons (Smolensky, 1986).
arXiv:2608.21664v1 Announce Type: new Abstract: Safe deployment of increasingly capable models will likely come to rely on latent-space monitoring as a complement to behavioral evaluations, especiall...
arXiv:2608.29034v1 Announce Type: cross Abstract: A wide range of methods have been proposed for interpreting language models, delivering important insights into their inner workings. However, differ...
arXiv:2607. 08399v1 Announce Type: cross Abstract: Large language models process prompts by propagating activations through dozens of layers before generating a response.
arXiv:2608. 15448v1 Announce Type: cross Abstract: Large language models increasingly rely on sampling as a driver of their own improvement, making the fidelity of their learned distributions more critical than ever.
arXiv:2510. 07884v2 Announce Type: replace-cross Abstract: Weak-to-strong generalization provides a promising paradigm for scaling large language models (LLMs) by training stronger models on samples from aligned weaker ones, without requiring human feedback or explicit reward modeling.
arXiv:2609.17376v1 Announce Type: new Abstract: Large language models (LLMs) trained on next-token prediction exhibit remarkable in-context learning (ICL) abilities, yet the representations that supp...
arXiv:2606. 26620v1 Announce Type: cross Abstract: Sparse autoencoders (SAEs) have emerged as a powerful tool for decomposing superposed language model representations into sparse and interpretable features.
Large language models (LLMs) trained on next-token prediction exhibit remarkable in-context learning (ICL) abilities, yet the representations that support ICL remain poorly understood. We consider suc...