随机权重平均遇上数据增强，简单组合就能显著提升模型泛化，炼丹必备技巧。
arXiv:2608.14373v1 Announce Type: new Abstract: The symmetries of a learning task have become an important factor in designing modern deep learning so…
自我改进的LLM智能体会把成功轨迹固化为可复用技能，一次不安全的成功可能就此潜伏成系统风险。
arXiv:2608.12851v1 Announce Type: new Abstract: Self-improving LLM agents convert successful trajectories into persistent cross-task state. An unsafe …
联邦蒸馏遇上数据分布偏移？域感知代理方案给出稳健新解，跨域协作更可靠。
arXiv:2608.08525v1 Announce Type: new Abstract: Federated Learning is a distributed machine learning paradigm that trains a global model by aggregatin…
大模型并非只能逐步推理，新观点称LLM也能像人类一样“跳跃”思考，值得一读。
Article URL: https://yongzx.github.io/blog/2026/08/08/llm-can-jump Comments URL: https://news.ycombinator.com/item?id=49232254 Points: 4 # Comments: 0
提出泛化感知的结构化剪枝方法，在压缩大模型规模的同时保持泛化能力，兼顾效率与性能。
arXiv:2603.13418v2 Announce Type: replace-cross Abstract: Structured pruning is widely applied to compress large language models (LLMs), but its perfo…
从语义相似度视角衡量LLM生成新颖性，突破词汇重叠局限，更精准评估泛化能力。
arXiv:2510.27313v3 Announce Type: replace-cross Abstract: Generation novelty is a key indicator of an LLM's ability to generalize, yet measuring it ag…
多语言LLM中指令层次遵从性因语言而异，揭示语言对模型安全对齐的影响。
arXiv:2607.23545v1 Announce Type: new Abstract: Instruction hierarchy (IH) requires models to prioritize instructions by source, ensuring that higher-…
通过层级技能编译（Hierarchical Skill Compilation），让LLM代理在开放场景中高效组合与复用技能，突破任务上限
arXiv:2508.14751v2 Announce Type: replace Abstract: We study goal-conditioned reinforcement learning in partially observable environments with sparse …
利用基础模型的强泛化能力，无需源数据即可高效适应目标域，摆脱传统调阈值和聚类等繁琐操作。
arXiv:2607.17653v1 Announce Type: cross Abstract: Source-free universal domain adaptation (SF-UniDA) adapts a pre-trained source model to an unlabeled…
提出一种简单域泛化方法，显著增强现代视觉语言模型下像素级图像篡改检测的鲁棒性。
arXiv:2607.18230v1 Announce Type: cross Abstract: Modern vision-language models (VLMs) have significantly improved image generation and editing capabi…
微调LLM看似无害的数据集可能暗含意识形态泛化，这项研究揭示了隐藏风险。
arXiv:2607.14888v1 Announce Type: new Abstract: Finetuning language models on small, curated datasets is standard practice for adapting them to specif…
利用未标注数据提升神经群体解码的泛化能力，打通稀疏标注与海量数据的桥梁。
arXiv:2607.14086v1 Announce Type: new Abstract: Robust and accurate neural decoders are integral to neurotechnologies such as brain-computer interface…
大模型内部发现通用倒数机制，助你在多种任务中精准控制输出长度
arXiv:2607.12279v1 Announce Type: cross Abstract: Writing a sentence of exactly twelve words; ending a DNA sequence at the right codon; formatting an …
LLM时代语音合成与转换加剧深度伪造风险，这项新基准测试揭露了现有欺骗检测器的泛化短板。
arXiv:2607.11706v1 Announce Type: cross Abstract: Modern LLM-driven text-to-speech (TTS) and voice conversion (VC) systems produce synthetic speech th…
一篇探讨对比学习框架下弱到强泛化的新论文，理论分析和实验验证结合，为AI大模型泛化研究提供新视角。
arXiv:2510.07884v2 Announce Type: replace-cross Abstract: Weak-to-strong generalization provides a promising paradigm for scaling large language model…
一作质问持续学习何时真正需要「学习」，揭示任务无关场景下模型无需更新即可泛化，引发对学习本质的再思考。
arXiv:2607.07847v1 Announce Type: new Abstract: As large language models (LLMs) become increasingly capable, the next question is how can we enable mo…
从模型多样性切入，LEMUR 2 为AI泛化能力提供新思路，值得算法研究者细读。
arXiv:2607.06839v1 Announce Type: new Abstract: Existing NAS benchmarks (e.g., NAS-Bench, NATS-Bench) cover only narrow, task-specific regions of the …
低成本智能体在ARC-AGI基准上实现惊人推理性能，兼顾效率与泛化。
arXiv:2607.06764v1 Announce Type: new Abstract: Recent progress on ARC-AGI-1 from disclosed architectures has come broadly from two regimes: heavy tes…
蚂蚁灵波开源新一代具身基座模型，支持20+构型泛化，多自由度操控，跨域任务领先，开发者可快速上手。
IT之家 7 月 8 日消息，蚂蚁集团旗下蚂蚁灵波科技今日正式官宣新一代具身基座模型 LingBot-VLA 2.0 全面开源。官方称，作为今年 1 月开源的 LingBot-VLA 1.0 的全面升级，LingBot-VLA 2.0 在构型泛化、自由度支持和落地效率等方面实现了显著提升。 IT之家…
扩散模型技术跨界3D交互，让机器人理解开放世界物体操作方式，知识泛化能力大升级。
arXiv:2508.01651v2 Announce Type: replace Abstract: 3D affordance grounding aims to understand how diverse objects can be manipulated, making it a cor…