English(EN)The Joint Effect of Quantization and Sampling Temperature on LLM Safety Alignment: A Factorial Analysis
新研究应对大语言模型对齐、安全和优化挑战
作者PulseAugur 编辑部·[13 个来源]·
研究人员正在探索改进大语言模型（LLM）对齐和可靠性的新方法。一项研究发现字节对编码（BPE）分词中存在一个漏洞，该漏洞可能被利用来绕过安全机制，导致多个模型系列产生有害输出。另一篇论文提出了一个名为HAL的框架，通过优化明确的、可解释的对话特征来诱导大语言模型产生类似人类的对话行为。此外，一个名为Object Aligner的新库提供了一种可配置的方法来评估JSON模式相似度，这对于大语言模型提示优化和工具使用非常有用。最后，对大语言模型量化的研究揭示了对齐-多样性权衡，并提出了一个任务感知敏感性分析框架（TASA）来提高模型性能和鲁棒性，而另一项研究则侧重于自改进在线大语言模型对齐算法的理论收敛特性。
AI
arXiv:2607.01239v1 Announce Type: cross Abstract: Character-level perturbations bypass safety alignment in modern LLMs despite leaving prompts human-readable. We identify and test a central structural mechanism: BPE tokenization fragments safety-critical words into sub-word piece…
arXiv cs.AI
TIER_1English(EN)·Tiancheng Hu, Benjamin Minixhofer, Nigel Collier·
arXiv:2510.17426v3 Announce Type: replace-cross Abstract: The "alignment tax" of post-training is typically framed as a drop in task accuracy. We show it also involves a severe loss of calibration, making models overconfident, less reliable, and model outputs less diverse. We sho…
arXiv:2601.02813v3 Announce Type: replace Abstract: Aligning language models to qualitative behavioral traits, such as human-likeness, remains difficult because they are hard to define, measure, and optimize. As a result, improvements in human-like behavior are largely driven by …
arXiv:2607.01972v1 Announce Type: cross Abstract: Large language models (LLMs) are often asked to produce JSON conforming to a fixed schema, powering information extraction, tool calling, agentic planning, and knowledge-graph construction. Measuring how closely an output matches …
Large language models (LLMs) are often asked to produce JSON conforming to a fixed schema, powering information extraction, tool calling, agentic planning, and knowledge-graph construction. Measuring how closely an output matches a gold reference is essential yet surprisingly har…
Large language models (LLMs) are often asked to produce JSON conforming to a fixed schema, powering information extraction, tool calling, agentic planning, and knowledge-graph construction. Measuring how closely an output matches a gold reference is essential yet surprisingly har…
arXiv cs.LG
TIER_1English(EN)·Fei Wang, Chao Xue, Taoran Liu, Li Shen, Ye Liu, ChangXing Ding·
arXiv:2607.00908v1 Announce Type: new Abstract: Mixed-precision quantization (MPQ) has become a key technique for deploying large language models under stringent memory and compute constraints. We first identify a phenomenon that we term the Perplexity Illusion: layers ranked as …
Mixed-precision quantization (MPQ) has become a key technique for deploying large language models under stringent memory and compute constraints. We first identify a phenomenon that we term the Perplexity Illusion: layers ranked as important by perplexity-based sensitivity show l…
arXiv:2606.31524v1 Announce Type: cross Abstract: The Self-Improving Alignment (SAIL) algorithm addresses distribution shift by reducing a bilevel formulation of the problem to an efficient, single-level method. Empirically, SAIL has demonstrated strong performance on this task. …
The Self-Improving Alignment (SAIL) algorithm addresses distribution shift by reducing a bilevel formulation of the problem to an efficient, single-level method. Empirically, SAIL has demonstrated strong performance on this task. However, a formal analysis of its convergence prop…
arXiv:2606.29581v1 Announce Type: cross Abstract: Modern LLM deployments routinely compress models and raise sampling temperature to reduce cost, latency, or repetition, yet safety evaluations usually treat these choices as fixed implementation details. This leaves a practical un…
arXiv:2510.06096v3 Announce Type: replace-cross Abstract: The objectives that Large Language Models (LLMs) implicitly optimize remain dangerously opaque, making trustworthy alignment and auditing a grand challenge. While Inverse Reinforcement Learning (IRL) can infer reward funct…
arXiv cs.AI
TIER_1English(EN)·Yanbo Wang, Minzheng Wang, Jian Liang, Lu Wang, Yongcan Yu, Ran He·
arXiv:2602.13562v2 Announce Type: replace-cross Abstract: While reasoning models have achieved remarkable success in complex reasoning tasks, their increasing power necessitates stringent safety measures. For safety alignment, the core challenge lies in the inherent trade-off bet…
