For balanced Gaussian class projections, OOD AUROC is a linear function of MCS to the reference probe because both are sigmoid-shaped functions of the probe SNR on test data.
arXiv preprint arXiv:2509.23024 , year=
8 Pith papers cite this work. Polarity classification is still indexing.
8
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
years
2026 8roles
background 1polarities
support 1representative citing papers
Language models produce overcomplete reasoning traces where on average 46% of steps can be removed while preserving the answer in 86% of cases, with necessity concentrated in the top three steps.
Dynamic Rollout Editing reduces overthinking in RL-trained LLMs by editing post-answer continuations in successful rollouts and preferring the edited versions within GRPO groups.
Introduces LOES, a constructive spectral method to select task-discriminative subspaces from intermediate layer embeddings, and GeoReg for enforcing simplicial class geometry during fine-tuning, with reported gains increasing with model depth across modalities.
Representation geometry in language models aligns with the unembedding readout subspace in a scale-dependent manner, preserved throughout training in large models but progressively lost in late layers of small models despite continued loss improvement.
LRD framework with Frenet, NRS, and GFMI metrics shows layer-wise structure in 31 models provides usable signal for model selection and pruning on MTEB tasks.
Spectral analysis of activations and gradients provides new diagnostics that link batch size to representation geometry, early covariance tails to token efficiency, and spectral shifts to learning dynamics in decoder-only LLMs, backed by a mechanistic model.
citing papers explorer
-
Comparing Linear Probes with Mahalanobis Cosine Similarity
For balanced Gaussian class projections, OOD AUROC is a linear function of MCS to the reference probe because both are sigmoid-shaped functions of the probe SNR on test data.
-
Uncovering the Latent Potential of Deep Intermediate Representations
Introduces LOES, a constructive spectral method to select task-discriminative subspaces from intermediate layer embeddings, and GeoReg for enforcing simplicial class geometry during fine-tuning, with reported gains increasing with model depth across modalities.
-
Scale Determines Whether Language Models Organize Representation Geometry for Prediction
Representation geometry in language models aligns with the unembedding readout subspace in a scale-dependent manner, preserved throughout training in large models but progressively lost in late layers of small models despite continued loss improvement.
-
Layer-wise Representation Dynamics: An Empirical Investigation Across Embedders and Base LLMs
LRD framework with Frenet, NRS, and GFMI metrics shows layer-wise structure in 31 models provides usable signal for model selection and pruning on MTEB tasks.
- From Dispersion to Attraction: Spectral Dynamics of Hallucination Across Whisper Model Scales