Loading...
#SafeguardingLLMsviaModel‑AgnosticLatentSafetySignalsfromDarkKnowledge论文阅读笔记>论文arXiv链接：[https://arxiv.org/a
Beyond Refusal Patterns 论文总结与翻译
#BeyondRefusalPatterns论文总结与翻译>论文信息：arXiv:2610.07023|BeyondRefusalPatterns:Safe‑RoleInternalizationforRobus
Distributionally Robust Mixture‑of‑Experts Training（DRMoET）论文总结与翻译
#DistributionallyRobustMixture‑of‑ExpertsTraining（DRMoET）论文总结与翻译>论文信息：arXiv:2610.07207，NeurIPS2026，作者：XinTeng,M
CoDe‑LoRA 论文阅读总结
#CoDe‑LoRA论文阅读总结>论文信息：CoDe‑LoRA:MitigatingtheOrthogonalityDilemmainContinualLearningofLLMsviaKnowledgeCo
Dynamic Positional Attention Modulation 论文阅读笔记
#DynamicPositionalAttentionModulation论文阅读笔记>论文信息：arXiv:2610.07848，KDD2026，标题：*DynamicPositionalAttentionModul
Illusory Pattern Perception Drives Spurious Inference in Large Language Models 论文总结与译文
#IllusoryPatternPerceptionDrivesSpuriousInferenceinLargeLanguageModels论文总结与译文>论文信息：arXiv:2610.07791，NeurIPS
Understanding and Enhancing Backdoor Persistency in LLM Agent Post‑Training 论文总结与翻译
#UnderstandingandEnhancingBackdoorPersistencyinLLMAgentPost‑Training论文总结与翻译>论文arXiv：https://arxiv.org/abs/26
MILO: Efficient Many‑shot In‑Context Learning with Block‑wise Low‑rank Compression
>说明：arXiv摘要页面仅公开**Abstract**，未直接展示Introduction、Conclusion完整原文；PDF需要读取完整论文。当前可获取完整的只有Abstract，因此我完成：论文整体总结、创新点、Abstract中
Rationale‑Guided Policy Optimization（RGPO）论文总结与翻译
#Rationale‑GuidedPolicyOptimization（RGPO）论文总结与翻译>论文信息：arXiv:2610.07342｜NeurIPS2026｜[论文仓库](https://github.com/VietH
Component and Dimension Sparsity in Transformer Refusal Mechanisms 论文总结
#ComponentandDimensionSparsityinTransformerRefusalMechanisms论文总结>论文信息：arXiv:2610.06903，接收于COLM2026，作者：Vincen
CodeGraph: Open‑Taxonomy Knowledge Graph for Source Code with Wikidata Grounding
#CodeGraph:Open‑TaxonomyKnowledgeGraphforSourceCodewithWikidataGrounding>论文arXiv链接：[https://arxiv.org/abs/26
LogicTrack: Auditing Reasoning Trajectories of Large Language Models with Formal Logic Solvers 论文总结与
#LogicTrack:AuditingReasoningTrajectoriesofLargeLanguageModelswithFormalLogicSolvers论文总结与翻译>论文arXiv链接：htt
Accelerating Dense LLMs via L0‑regularized Mixture‑of‑Experts 论文总结与翻译
#AcceleratingDenseLLMsviaL0‑regularizedMixture‑of‑Experts论文总结与翻译>论文arXiv链接：https://arxiv.org/abs/2609.21672##
Abstention and Noise Filtering: Two Missing Primitives of Softmax Attention 论文阅读总结
#AbstentionandNoiseFiltering:TwoMissingPrimitivesofSoftmaxAttention论文阅读总结>论文arXiv链接：[https://arxiv.org/abs/
Block‑Sparse Attention with Semantic‑Geometric Decoupled Routing（arXiv:2609.22884）论文总结与翻译
#Block‑SparseAttentionwithSemantic‑GeometricDecoupledRouting（arXiv:2609.22884）论文总结与翻译>论文链接：[https://arxiv.org/ab
GVPO++: Group Variance Policy Optimization for LLM Post‑Training and On‑Policy Distillation
#GVPO++:GroupVariancePolicyOptimizationforLLMPost‑TrainingandOn‑PolicyDistillation>论文链接：https://arxiv.org/a
RBS‑Attention 论文阅读总结
#RBS‑Attention论文阅读总结>论文信息：arXiv:2609.20971，RBS‑Attention:Radius‑BoundedSparsePrefillforLong‑ContextLargeLangu
Random Attention: Rethinking KV Cache Eviction for Efficient Reasoning（arXiv:2609.03430）
#RandomAttention:RethinkingKVCacheEvictionforEfficientReasoning（arXiv:2609.03430）##一、文章主要内容总结###研究背景大模型在长链
SWE‑Gate 论文阅读总结
#SWE‑Gate论文阅读总结>论文信息：arXiv:2609.04167，SWE‑Gate:PassingFunctionalTestsIsNotEnoughforSoftwareEngineeringAgen
What Matters for Aggressive Decoding‑Time KV Eviction? Temporal Aggregation and Ranking Preservation
#WhatMattersforAggressiveDecoding‑TimeKVEviction?TemporalAggregationandRankingPreservation>arXiv:2609.0351
