用分布视角重新拆解知识蒸馏，带你理解模型压缩背后的数学本质，适合深度学习进阶读者。
arXiv:2608.15215v1 Announce Type: cross Abstract: Token-level knowledge distillation (KD) matches two conditional distributions per position, yet the …
理论剖析Transformer表达能力的边界，揭示大模型底层机制，值得算法研究者细读。
arXiv:2608.12671v1 Announce Type: new Abstract: Multi-layer transformers form the critical component of essentially all large language models (LLMs) i…
用平均场理论拆解思维链推理的动态机制，为理解大模型推理过程提供全新数学视角
arXiv:2608.05152v1 Announce Type: cross Abstract: Large language models (LLMs) with chain-of-thought reasoning have been widely applied in recent year…
大模型内部状态新假说，24页图解重塑认知，理论控必读
arXiv:2607.19360v2 Announce Type: replace Abstract: Large language models (LLMs) adapt rapidly through fine-tuning and in-context learning, yet it rem…
从理论层面重新审视LLM对齐的奖励分布假设，揭示重尾效应对对齐边界的影响，值得关注。
arXiv:2604.10727v2 Announce Type: replace-cross Abstract: Classical information-theoretic learning bounds typically rely on KL mutual information and …
把LLM输出看作动力系统轨迹，这项研究首次给出可区分性的理论保证，值得深读。
arXiv:2607.28667v1 Announce Type: cross Abstract: Recent work has shown that classifying large language models (LLMs)' responses can be distinguished …
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…
颠覆认知：GNN本质是高级启发式算法而非特征学习器，理论推导+实验验证给出新视角
arXiv:2601.13465v4 Announce Type: replace Abstract: Graph neural networks are usually treated as auxiliaries for combinatorial optimization: they imit…
强化学习理论新突破：线性Bellman完备MDP在确定性转移下实现端到端高效求解，值得算法研究者细读。
arXiv:2603.23461v2 Announce Type: replace Abstract: We study reinforcement learning (RL) with linear function approximation in Markov Decision Process…
首次用闭式模型刻画GRPO训练动态，为强化学习调参提供可预测的理论框架。
arXiv:2606.30789v1 Announce Type: new Abstract: Group Relative Policy Optimization (GRPO) has become a standard tool for improving the reasoning abili…
深度揭示稀疏性如何破解大模型层数诅咒，图证详实，值得一读。
arXiv:2603.15389v2 Announce Type: replace Abstract: Recent work has demonstrated the curse of depth in large language models (LLMs), where later layer…
首次从理论层面证明同质深度网络在持续学习中的收敛性质，为灾难性遗忘提供数学解释
arXiv:2606.30559v1 Announce Type: new Abstract: We characterize weakly regularized continual classification in homogeneous models as sequential projec…
理论揭示：判别模型的条件分布等价性如何约束内部表示的唯一性，为理解模型表征相似性提供新视角
arXiv:2602.15438v3 Announce Type: replace-cross Abstract: For a broad family of discriminative models that includes autoregressive language models, id…
大模型并非真贝叶斯，而是在期望上趋同，这一视角颠覆传统认知，值得细读。
arXiv:2507.11768v3 Announce Type: replace-cross Abstract: Bayesian accounts of in-context learning face a direct objection: exact posterior predictive…
流式逆求解器到底在逼近什么？这篇论文用后验传输视角给出新答案。
arXiv:2606.24516v1 Announce Type: new Abstract: A growing family of training-free solvers -- FlowDPS, FLOWER, PnP-Flow and their diffusion ancestors (…
探讨大语言模型与人类表征模式的差异，从认知科学视角剖析AI理解能力
arXiv:2606.21616v1 Announce Type: new Abstract: Much work on the cognitive foundations of AI has focussed on comparisons between the ways in which Lar…
在线镜像下降的近似误差竟有隐藏代价，理论推导揭示收敛性新边界，优化算法研究者必读。
arXiv:2511.22283v2 Announce Type: replace Abstract: Online mirror descent (OMD) is a fundamental algorithmic paradigm that underlies many algorithms i…
33页论文揭示ReLU神经网络输出范围如何受层结构和参数约束，理论进展值得关注
arXiv:2508.03867v2 Announce Type: replace-cross Abstract: We introduce a class of algebraic varieties naturally associated with ReLU neural networks, …
Q-learning新理论突破，细粒度依赖遗憾界优化算法性能。
arXiv:2510.06647v2 Announce Type: replace-cross Abstract: We study fine-grained gap-dependent regret bounds for model-free reinforcement learning in e…
探讨LLM作为语言学模态模型的新视角，论文从理论层面解析大语言模型的语言学应用价值
arXiv:2606.10467v1 Announce Type: new Abstract: The rapid advancement of large language models (LLMs) has intensified debates about their significance…