1
OpenAI 首席科学家回应“AI 推理不透明”争议：复杂度未跳跃增长
IT之家 9 月 3 日消息，OpenAI 首席科学家雅库布・帕乔基（Jakub Pachocki）昨日（9 月 2 日）在 X 平台发布推文， 回应 AI 模型 Astra“不可监控 / 推理不透明”的争议。 帕乔基强调，包括 Astra 在内的前沿模型，计算图深度与 GPT-4 相差不超过两倍，…
2
The Value of a Prompt: An LLM-Relative Kolmogorov-Complexity Approach
用列文-柯尔莫哥洛夫复杂度度量提示词价值，给“好提示”一个可计算的理论标尺。
arXiv:2608.16438v1 Announce Type: new Abstract: In a world where valuable artifacts are increasingly created, completed, or processed by LLMs, the cen…
3
From Caveman to Expert Analyst: Energy Consumption of Variable LLM Tasks
从原始人式推理到专家级分析，看不同LLM任务能耗差异，为绿色AI提供量化视角。
arXiv:2608.12350v1 Announce Type: cross Abstract: The energy demand growth and environmental impacts of artificial intelligence (AI) have generated su…
4
NASA 曾为国际空间站笔记本统一充电接口，减少航天任务复杂度
IT之家 8 月 8 日消息，NASA 工程师乔治 · 劳什（George Roush）8 月 2 日披露了一项 NASA 在早期针对国际空间站（ISS）笔记本电脑的改装细节。 为确保所有送往空间站的笔记本电脑使用统一的电源接口，NASA 将设备原有的调制解调器端口移除，替换为统一的“炮型插头”（c…
5
GPTQ-2D: Cubic-Time Two-Sided Adaptive Rounding
提出GPTQ-2D算法，实现三次时间复杂度的双向自适应舍入，有效提升大模型量化精度。
arXiv:2607.27042v1 Announce Type: cross Abstract: Adaptive rounding methods such as GPTQ, or equivalently Babai's nearest plane algorithm, round a rea…
6
IKEA Complexity Index：为 2 万件宜家家具计算“安装难度分”
查询宜家2万件家具的安装难度分，根据零件数、步骤和用户反馈计算，帮你避开组装噩梦
买宜家家具的时候，我们通常会关注价格、尺寸和外观。但还有一个很少有人提前考虑的问题：它到底有多难装？ 有些家具打开包装后，半小时就能完成；有些柜子则需要面对几十块板材、上百颗螺丝，以及一张让人反复确认也无法看懂的说明书。 IKEA Complexity Index（宜家复杂性指数） 有一位开发者做了
7
前 R 星制作人揭秘《GTA6》为何首发不上 PC：资源调配与优化复杂度是主因
IT之家 7 月 13 日消息，前 Rockstar 制作人约翰 · 里奇奥（IT之家注：John Riccitiello）昨天出席 Kiwi Talkz 播客节目，谈到 R 星旗下作品为何总是首发上主机、过一两年后才上 PC。 这名参与过《侠盗猎车手 V》《荒野大镖客：救赎》的制作人表示，R 星游…
8
Prompting Complexity: Shortest Prompts for Texts and Behaviors in LLMs
定义LLM提示复杂度的新概念，探究指令调优模型最短合理提示的边界。
arXiv:2607.06145v1 Announce Type: new Abstract: In this paper, we define the quantity of prompting complexity: for a fixed instruction-tuned language …
9
Graph Neural Networks are Heuristics
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…
10
Kwai Summary Attention Technical Report
快手新注意力机制直击长上下文痛点，线性复杂度替代二次方，为下一代LLM推理提速。
arXiv:2604.24432v2 Announce Type: replace-cross Abstract: Long-context ability, has become one of the most important iteration direction of next-gener…
11
Decoupling Code Complexity from Newcomer Participation: A Causal Study of AI Coding Agent Adoption in OSS
用因果推断揭示AI编程助手如何重塑开源社区，新手参与门槛是否真正降低？实证研究给出答案。
arXiv:2607.01810v1 Announce Type: cross Abstract: Open-source projects depend on a steady inflow of newcomers. A growing concern is that AI coding age…
12
How AI settled the complexity of the oldest SGD algorithm
AI最终破解了最古老随机梯度下降算法的复杂度难题，数学理论迎来新突破。
arXiv:2606.29593v1 Announce Type: new Abstract: In 1937, Stefan Kaczmarz proposed a simple algorithm for solving systems of linear equations. This alg…
13
TriViewBench: Controlled Complexity Scaling for Multi-View Structural Reasoning in MLLMs
多模态大模型评估新基准，用可控复杂度细测多视角结构推理。
arXiv:2606.26029v1 Announce Type: cross Abstract: Multimodal Large Language Models (MLLMs) demonstrate strong performance on standard visual question …
14
Hierarchical Attention via Domain Decomposition
高效近似注意力机制新突破：用域分解实现层次化注意力，降低计算复杂度。
arXiv:2606.18525v1 Announce Type: new Abstract: We propose a hierarchical attention mechanism based on two-level overlapping Schwarz domain decomposit…
15
How fast can you find a good hypothesis?
如何高效搜索好的假设？这篇论文从理论上探讨假设空间搜索的时间下界与算法效率，对机器学习模型选择有启发。
arXiv:2509.03734v3 Announce Type: replace-cross Abstract: In the hypothesis selection problem, we are given sample and query access to finite set of c…
16
DecoSearch: Complexity-Aware Routing and Plan-Level Repair for Text-to-SQL
复杂度感知路由+计划级修复，这篇论文为Text-to-SQL推理提供了更鲁棒的方案。
arXiv:2606.17821v1 Announce Type: new Abstract: Large Language Models (LLMs) have demonstrated remarkable capabilities in translating natural language…
17
SICI: A Semantic-Pragmatic Complexity Index Reveals Regime Shifts in LLM Stance Detection
研究提出SICI指数，揭示了大模型立场检测中语义-语用复杂度的机制转变，技术深度与创新兼具。
arXiv:2606.13189v1 Announce Type: new Abstract: Prompt-based LLMs are increasingly used for stance detection, but harder examples are not always repai…
18
T1-Bench: Benchmarking Multi-Scenario Agents in Real-World Domains
全新多场景代理基准T1-Bench，全面评测LLM在真实领域中的复杂交互能力
arXiv:2606.11070v1 Announce Type: cross Abstract: Recent advances in reasoning and tool-calling capabilities of large language models (LLMs) have enab…
19
Singular Learning Theory: AI learns like ice melts
将AI学习比作冰融化，用参数效应重新理解模型复杂度，颠覆传统视角。
Article URL: https://tcz.hu/blog/2026/02/26/singular-learning-theory/ Comments URL: https://news.ycombinator.com/item?id=48476343 Points: 1 # Comments…
20
Algorithmic and Minimax Complexities in Kernel Bandits
探究核赌博机问题的算法与极小极大复杂度，理论机器学习新进展
arXiv:2606.11171v1 Announce Type: new Abstract: Gaussian-process upper confidence bound (GP-UCB) and decision-estimation-coefficient (DEC) methods may…