Curating the frontier of Artificial Intelligence

QuoteBench: How Matched Scores Can Hide Command-Path Failures

LLM 编码代理通过接口发出 Bash 命令，这些命令可以序列化, 包装, 并重新解析模型输出。仅匹配的执行分数并不能区分...

LLM coding agents issue Bash commands through interfaces that may serialize, wrap, and reparse model output. Matched execution scores alone cannot dis...

OmniScientist: An Omni-Modal Omni-Discipline AI Scientist

基础模型的最新进展使 AI 科学家能够自动化日益完整的研究工作流程,，从假设生成到分析……

Recent advances in foundation models have enabled AI scientists to automate increasingly complete research workflows, from hypothesis generation and c...

Chinese firm tops Micron and Kioxia in shipments of NAND memory chips

北京 — 据 Counterpoint Research 称，长江存储科技公司正在关键芯片领域, 中迅速获得市场份额。该公司,还...

BEIJING — Yangtze Memory Technologies is rapidly gaining market share in a critical chip sector, according to Counterpoint Research. The company, also...

From Apple to Ford: How Chinese tech is becoming harder for global companies to ignore

华盛顿一直在加紧努力遏制北京'的技术野心,，但从人工智能到电动汽车电池,中国...

Washington has been stepping up efforts to curb Beijing's technological ambitions, but from artificial intelligence to electric vehicle batteries, Chi...

Who Thinks Best Depends on How Long You Let Them: Budget-Dependent Rankings in LLM Evaluation

大型语言模型的标准评估假设在推理条件下模型排名稳定。我们通过改变...来挑战这个假设。

Standard evaluation of large language models assumes stable model rankings across inference conditions. We challenge this assumption by varying the to...

How to Spend Your Oracle Budget: Practical Guidance for Protein Structure Prediction Models

蛋白质结构预测的基础模型在某些目标上仍然不可靠。外部预言机可以标记并纠正这些故障,，但生物......

Foundation models for protein structure prediction remain unreliable on certain targets. External oracles can flag and correct these failures, but bio...

An Agentic Workflow for Legacy HPC Modernization: Converting the Two-Electron-Integral Core of GAMESS

对遗留 Fortran 进行现代化改造是一个很大的问题:，转换是单独的例行程序,，但代码库可能非常庞大,，并且跨越了大部分...

Modernizing legacy Fortran is a problem of volume: the transformations are individually routine, but the codebases can be enormous, and across much of...

VAKRA: Evaluating Multi-Hop Reasoning Across APIs and Retrieval Under Tool-Use Policies

部署在企业环境中的代理必须跨结构化 API 和文档集合进行推理,，但现有基准评估这些功能...

Agents deployed in enterprise settings must reason across structured APIs and document collections, yet existing benchmarks evaluate these capabilitie...

Constructing Dynamic Master Logic Models as Knowledge Graphs for Complex System Diagnostics Using Retrieval-Augmented Large Language Models

动态主逻辑(DML)通过将功能目标链接到底层结构来提供表示系统行为的分层框架...

Dynamic Master Logic (DML) provides a hierarchical framework for representing system behavior by linking functional objectives to underlying structura...

EVs dominate China’s car market: 5 takeaways from the country's latest auto sales data

北京— 在中国'竞争激烈的汽车市场,中，幸存者已经出现，该市场日益由电动汽车主导。这里有br...

BEIJING — Survivors have emerged in China's fiercely competitive car market, which is increasingly dominated by electric-powered vehicles. Here are br...