Alignment, interpretability, red-teaming, bias and privacy: the research on what these systems do when they misbehave.
arXiv:2610.31944v1 Announce Type: new Abstract: Emergency medical dispatch increasingly relies on AI-enabled decision support systems when making ambulance dispatch decisions. In this process, accura...
arXiv:2610.32175v1 Announce Type: new Abstract: Differentially Private Stochastic Gradient Descent (DPSGD) has become the de facto standard for training deep neural networks with rigorous privacy gua...
arXiv:2610.32465v1 Announce Type: new Abstract: Adversarial sharpness penalty per iteration of a base optimizer, known as sharpness-aware minimization (SAM), is an accepted technique to improve gener...
arXiv:2610.32630v1 Announce Type: new Abstract: VICReg prevents collapse with a hinge on the standard deviation of each embedding dimension, and its authors argue from the gradient that a hinge on th...
arXiv:2610.32215v2 Announce Type: cross Abstract: The growing computational power of modern GPU data centers has let a few industrial giants train and deploy large language models (LLMs) on massive a...
arXiv:2605.11735v2 Announce Type: replace Abstract: Accurate spatio-temporal cellular traffic analysis is fundamental to the reliable operation of modern mobile networks, as it underpins load forecas...
arXiv:2610.11146v3 Announce Type: replace Abstract: Training-only teacher access lets a compact student deploy with local features and exported weights. This established distillation interface does n...
arXiv:2610.33697v1 Announce Type: new Abstract: Large Language Models (LLMs) are increasingly used for medical self-diagnosis. However, the way demographic cues shape outputs in these interactions re...
arXiv:2610.33749v1 Announce Type: new Abstract: As machine translation systems approach human-level quality, traditional evaluation methodologies struggle to detect subtle translation errors. We crit...
arXiv:2610.33813v1 Announce Type: new Abstract: Fine-tuning a language model for classification involves many design decisions. Comparisons of low-rank adaptation (LoRA) with full fine-tuning are com...
arXiv:2610.33899v1 Announce Type: new Abstract: LLMs are increasingly being used by users from diverse demographics with pluralistic values and distinct backgrounds. Maintaining safety while respecti...
arXiv:2610.34281v1 Announce Type: new Abstract: Large Language Models (LLMs) have revolutionized Artificial Intelligence (AI) by their extraordinary capabilities, like contextual understanding, few-s...
arXiv:2610.34572v1 Announce Type: new Abstract: The largest open German text datasets are built from broad web crawls, which provide scale but limited coverage of specific domains. We introduce Refin...
arXiv:2610.34809v1 Announce Type: new Abstract: Language models now answer questions alongside other claims: a note in the system prompt, a message from another agent, their own earlier answer. We as...
arXiv:2610.35054v1 Announce Type: new Abstract: Transportation agencies maintain extensive crash databases to identify safety priorities, evaluate countermeasures, and guide resource allocation. Maki...
arXiv:2610.35142v1 Announce Type: new Abstract: Large language models (LLMs) often produce plausible but incorrect answers, motivating reinforcement learning methods that reward correct answers, pena...
arXiv:2610.35261v1 Announce Type: new Abstract: Idioms have long been a challenge for machine translation, and while large language models (LLMs) have proven increasingly adept at handling them, whet...
arXiv:2610.35269v1 Announce Type: new Abstract: Lean verifies that a formal proof establishes its stated proposition under declared assumptions, but does not check whether that proposition faithfully...
arXiv:2610.34756v1 Announce Type: cross Abstract: Low-Rank Adaptation (LoRA) has emerged as the dominant paradigm for parameter-efficient fine-tuning of large language models. Standard all-layer LoRA...
arXiv:2602.10657v2 Announce Type: replace Abstract: How much of zero-shot benchmark accuracy is captured by word-level alignment between a pre-training corpus and the benchmark target? We study this...