针对LLM代理的伪装MCP攻击，提出改进偏好对齐的拒绝训练方法，附RAG免训练对齐方案。
Article URL: https://github.com/johnhalloran321/mcp_safety_training Comments URL: https://news.ycombinator.com/item?id=48834106 Points: 1 # Comments: …
被ICASSP 2026接收的PINN训练新方法，提出无模块化冲突避免策略，提升泛化能力。
arXiv:2606.20156v1 Announce Type: new Abstract: Physics-informed neural networks (PINNs) have become a powerful framework for solving PDEs by embeddin…
提出一种无需额外训练的扩散模型，仅需单张图像即可生成高质量样本，CVPR 2026论文。
arXiv:2606.04299v1 Announce Type: new Abstract: We consider the problem of generating images whose internal structure -- defined by the distribution o…
一步扩散模型实现免训练图像反演，速度与质量兼得。
arXiv:2606.01380v1 Announce Type: new Abstract: In this work, we introduce a novel training-free inversion (TFinv) framework for one-step diffusion mo…
无需训练即可避免路由坍缩，专家特征向量为MoE提供高效非学习路由方案。
arXiv:2605.30992v1 Announce Type: new Abstract: Sparse Mixture of Experts (SMoE) architectures improve the training efficiency of Large Language Model…
提出一种无需训练的向量量化新方法，利用高斯VAE实现高效表示学习，为量化领域开辟新路径。
arXiv:2512.06609v3 Announce Type: replace Abstract: Vector-quantized variational autoencoders (VQ-VAEs) are discrete autoencoders that compress images…
用扩散模型实现免训练贝叶斯滤波，解决高维粒子滤波扩展性难题
arXiv:2605.20028v1 Announce Type: new Abstract: Bayesian filtering is a well-known problem that aims to estimate plausible states of a dynamical syste…
无需额外训练，通过进化合并大模型权重即可提升推理性能——前沿的免训练规模扩展方法。
arXiv:2605.14386v1 Announce Type: cross Abstract: We present Darwin Family, a framework for training-free evolutionary merging of large language model…