Alignment, interpretability, red-teaming, bias and privacy: the research on what these systems do when they misbehave.
arXiv:2607.28639v3 Announce Type: replace-cross Abstract: Small instruction-tuned language models reach phones, cars, and offline assistants through knowledge distillation, and whether a distilled mo...
arXiv:2608.19760v3 Announce Type: replace-cross Abstract: Step-level credit evaluation can conflate numerical association with ordering and branch recovery with position-selection value. We test thes...
arXiv:2610.31718v1 Announce Type: new Abstract: Financial transaction networks are naturally represented as multigraphs, where repeated transactions between the same pair of accounts are modeled as d...
arXiv:2610.32184v1 Announce Type: new Abstract: Machine learning on resting-state functional magnetic resonance imaging (rs-fMRI) brain networks is promising for diagnosing Autism Spectrum Disorder (...
arXiv:2610.32362v1 Announce Type: new Abstract: Machine learning (ML) models used in clinical workflows often perform systematically differently across patient groups, raising concerns about equity a...
arXiv:2610.32772v1 Announce Type: new Abstract: GRPO shares a response-level advantage across tokens, leaving unresolved how strongly each token should contribute to learning. Updates induced by the...
arXiv:2610.31959v1 Announce Type: cross Abstract: Short-term building load forecasting remains challenging because load patterns can change rapidly with weather and operating conditions. Data-driven...
arXiv:2610.33258v1 Announce Type: new Abstract: Content moderation compresses a graded, contested human judgment into a binary act (leave up / take down) over billions of items, at a scale where runn...
arXiv:2610.34653v1 Announce Type: new Abstract: Latency-sensitive spoken agents must know not only which tool to call, but also when they have heard enough to act. Waiting for the turn to complete de...
arXiv:2610.35254v1 Announce Type: new Abstract: The number of paper submissions to major AI conferences has increased dramatically in recent years, substantially increasing the peer-review workload o...
arXiv:2601.10307v2 Announce Type: replace Abstract: Enhancing the moral alignment of Large Language Models (LLMs) is a critical challenge in AI safety. Current alignment techniques often act as super...
arXiv:2607.18066v2 Announce Type: replace Abstract: The value alignment of large language models (LLMs) is crucial for ensuring responses align with human intention and value preferences. However, mo...
arXiv:2610.15910v2 Announce Type: replace Abstract: Deep Research agents rely on long-horizon reasoning and iterative tool use to solve complex information-seeking tasks. However, adaptive decision-m...
arXiv:2610.33019v1 Announce Type: new Abstract: As multimodal agents engage in sustained interactions, long-term memory serves two distinct but complementary functions: fine-grained evidence recall a...
arXiv:2610.34190v1 Announce Type: new Abstract: Open-vocabulary instance segmentation (OVIS) provides a promising solution for marine perception by recognizing and segmenting categories beyond predef...
arXiv:2610.34926v1 Announce Type: new Abstract: An action seen once and the same action seen three times receive the same action-recognition label, so that supervision does not require a model to cou...
arXiv:2610.35729v1 Announce Type: cross Abstract: Imagery captured by unmanned aerial vehicle (UAV) systems increasingly raises privacy concerns, as UAV aerial surveillance routinely captures sensiti...
The paper presents a child-centered harm taxonomy for AI systems and analyzes 1,331 publicly reported child-related AI incidents. It finds that harms are concentrated in representational, dignity-related, and psychological categories, often linked to deployment-context, governance, and misuse failures rather than model defects. The study highlights that model-level evaluations alone cannot anticipate the harm pathways most frequently documented for children, and recommends broader mitigations such as age assurance, consent verification, safety-by-design UX constraints, incident disclosure norms, and targeted auditing regimes.
The paper investigates how aligning rounding locations between inference and training can stabilize reinforcement learning for large language models using BF16 precision. By retaining selected intermediates in FP32 until their BF16 outputs in three key regions—RoPE, SwiGLU, and residual RMSNorm—the authors prevent training collapse across four Qwen-family models. The approach keeps the inference engine, matrix‑multiplication and attention kernels, and learning objective unchanged.
Large Language Models: A Plain-Language Overview for Engineers explains that LLMs are neural networks trained on vast text collections to predict subsequent words, and outlines key concepts such as Transformers, pre‑training, fine‑tuning, reinforcement learning from human feedback, retrieval‑augmented generation, chain‑of‑thought prompting, and tool‑using agents. It also discusses efficiency techniques like FlashAttention and QLoRA, provides a life‑cycle figure and comparative tables, and offers a checklist for evaluating LLM claims, including checks for data contamination, model‑based grading, long‑context handling, and hallucination. The article aims to give engineers a clear picture of how training changes and prompts affect generative AI tools like ChatGPT.