LLM Digest

Story

arxiv_llm_reliability · Oct 22, 2026 · paper

Source brief

Spectral Smoothing and Spiking Perturbations Reconfigure Capability Profiles and Selective Task Performance in Large Language Models

arxiv.orgOct 22, 2026
original source linked

In brief

Singular-value spectra encode important structure in transformer feed-forward projections, yet their causal role in downstream behavior remains underexplored. We study post-hoc spectral smoothing and spiking intervent...

Feed lens

eval

Read the original at arxiv.org →Open in live feed Read that day’s brief

Spectral Smoothing and Spiking Perturbations Reconfigure Capability Profiles and Selective Task Performance in Large Language Models

Earlier in this thread 4 items

Self-Assessment Reinforcement Learning for Search-State Awareness in Retrieval-Augmented Large Language Models

Hallucinations of Large Language Models in Cross-Language Code Translation: Phenomena and Mitigation

Tool Recommendation for Multi-Step Agents: Finding Every Tool a Task Needs

SkillBrowser: Selective Skill Exposure for Scalable Skill-Using Agents