1
卡普空开源 RE:Dox 结构化数据引擎，JSON 反序列化最高提速 2.84 倍
IT之家 10 月 5 日消息，卡普空宣布，将旗下最新的数据处理及序列化基础库“RE:Dox”以开源软件（OSS）形式公开。该项目是卡普空次世代游戏引擎计划 REX（RE neXt ENGINE）的基础技术之一，开发者可以自由使用。 卡普空早在 2023 年 10 月便公布了 REX 引擎计划，目标…
2
清华稳准智能联合发布LimiX-2，结构化数据基础模型登顶国际评测榜单
9月16日，稳准智能联合清华大学发布新一代数据大模型 LimiX-2，模型参数规模提升至400M。
3
Enhancing LLMs in Predictive Political QA with Semi-Structured Data
半结构化数据让大模型预测政治走向更靠谱？看这项研究如何提升问答准确性。
arXiv:2608.21218v1 Announce Type: new Abstract: Predictive political question answering (QA), such as predicting how a political actor will vote, goes…
4
Show HN: Structural code grep across public GitHub repositories
用AST模式语法跨GitHub海量仓库做结构化代码搜索，找模式快人一步。
Codemod grep searches public code by syntax and structure, not just text or regex. Use ast-grep patterns to find code by shape across repositories. Co…
5
美国专家示警：学生依赖“AI 代写”会削弱思考能力
用大纲式笔记梳理思考脉络，让写作变成真正的思维训练，告别AI代写依赖。
IT之家 8 月 23 日消息，美国学生使用 AI 完成作业、甚至代写整篇论文的现象已经十分普遍，也有不少学校允许学生在一定范围内借助 AI 工具。 据《纽约时报》当地时间 17 日报道，越来越多专家担心，问题可能不只是学生会不会写文章，而是长期依赖 AI 可能 削弱他们本身的思考能力 。 认知心理…
6
Reinforcing Consistency in Video MLLMs with Structured Rewards
视频多模态模型常“看走眼”？这项研究用结构化奖励强化推理一致性，为视频理解立新标尺。
arXiv:2604.01460v2 Announce Type: replace Abstract: Multimodal large language models (MLLMs) have achieved remarkable progress in video understanding.…
7
From Sequence to Structure: Relational Uncertainty Propagation for LLM Agents
不再只关注序列预测，而是引入关系不确定性传播，让LLM智能体在结构化决策中更可靠，值得深入研读。
arXiv:2608.16002v1 Announce Type: cross Abstract: Reliable uncertainty quantification (UQ) is essential for deploying large language model (LLM) agent…
8
Training Fair Tabular Foundation Models
表格数据基础模型如何兼顾公平性？ICML spotlight论文提出全新训练思路，AI公平性研究者必读。
arXiv:2608.14211v1 Announce Type: cross Abstract: Tabular Foundation Models (TFMs) have emerged as leading methods for tabular predictive tasks, lever…
9
Unifying Depth and Width Pruning for LLMs via Binary Knapsack Optimization
把LLM剪枝难题化为背包问题，SNIPER统一深度+宽度剪枝，精准控预算！
arXiv:2608.12953v1 Announce Type: cross Abstract: Structured pruning is a promising approach for compressing large language models (LLMs), yet existin…
10
Ask HN: How do you develop more deterministic LLM pipelines?
面对LLM输出不确定？看看开发者如何构建更可靠的流水线，减少人工核验成本。
It feels like with LLMs I develop a prompt and then get some kind of output that's not very well structured and requires some kind of human oversight …
11
High-Layer Attention Pruning with Rescaling
大模型剪枝新招：高层注意力剪枝加缩放，推理提速不损精度，论文原文速看。
arXiv:2507.01900v3 Announce Type: replace-cross Abstract: Pruning is a highly effective approach for compressing large language models (LLMs), signifi…
12
Mitigating Over-Personalization in LLMs via Structured Memory
用结构化记忆给LLM“降噪”，破解过度个性化难题，AI调校新思路。
arXiv:2608.08300v1 Announce Type: new Abstract: Conversational assistants increasingly rely on persistent long-term memory to personalize responses ac…
13
Why Pasting Figma Screenshots into Claude Fails (And What to Use Instead)
Figma截图喂给Claude会丢失哪些关键信息？用结构化上下文替代像素猜测。
Here's the workflow that's become default in every design-to-code team right now: export a frame from Figma, paste the PNG into Claude or Cursor, type…
14
SVRepair: Structured Visual Reasoning for Automated Program Repair
把代码修复变成“看图找茬”，用结构化视觉推理定位缺陷，AI编程新思路。
arXiv:2602.06090v2 Announce Type: replace-cross Abstract: Large language models (LLMs) have recently been applied to Automated Program Repair (APR), y…
15
GPrune-LLM: Generalization-Aware Structured Pruning for Large Language Models
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…
16
RAP: KV-Cache Compression via RoPE-Aligned Pruning
直击长上下文推理痛点，用RoPE对齐剪枝实现KV-Cache压缩，兼顾效率与精度。
arXiv:2602.02599v4 Announce Type: replace-cross Abstract: Long-context inference in large language models (LLMs) is bottlenecked by the memory and com…
17
SG-Layout: Structured Scene Graph-Guided Layout Generation with LLMs
利用场景图引导大模型生成空间布局，破解几何关系建模难题，AI布局新思路。
arXiv:2608.01106v1 Announce Type: cross Abstract: Understanding and generating spatially coherent layouts from natural language remains a fundamental …
18
Keep It Real: A Schema-Level Deferral Channel for Concise LLM Dialogue [pdf]
大模型对话总是冗长？Schema级延迟通道带来更简洁的LLM交互新思路。
Article URL: https://raw.githubusercontent.com/noamt/kir/cfdc97de467de8d84d60f073773c0dcfd9975849/paper/arxiv/main.pdf Comments URL: https://news.ycom…
19
Baikal: Structured Search for Deep Research over Data Lakes
面向数据湖的结构化搜索新框架，为深度研究提供高效精准的信息定位方案
arXiv:2607.27726v1 Announce Type: cross Abstract: Deep research over data lakes requires an LLM agent to investigate evidence across thousands of hete…
20
Show HN: Emboss – a Python PDF engine that keeps documents as structured data
一个用语义模型保持文档结构化、自动生成无障碍PDF的Python引擎，EU可访问法案下必备。
Article URL: https://github.com/GGChamp85/Emboss Comments URL: https://news.ycombinator.com/item?id=49106579 Points: 5 # Comments: 1