AI 论文与研究

Good Start Labs 用 1830： The Game of Railroads and Robber Barons 训练一个 30B 模型，只有"多轮终端智能体"训练设计在 Finance-Agent benchmark 上有提升，单轮问答设计没有。

微软论文比较五种工具接口，在 TheAgentCompany 和 APEX-Agents 上用 Opus-4.8 和 GPT-5.5 测试。

Google Research 发布 Stellar Colosseum，一个面向长程数学和理论计算机科学研究的模型无关多智能体框架。该框架分阶段运行，先探索多种证明策略，通过就绪门槛后将路线拆分为章节级子问题，并行生成候选方案并配合定向证伪和批评合并。

An agentic society is a collection of AI agents that coordinate autonomously across trust boundaries, on behalf of different principals whose objectives may only partially align. We show experimentally that in agentic societies even honest…

We introduce and release ScienceBuddy, an interactive scientific research workspace that brings continually improving scientific agents into researchers' everyday workflows. ScienceBuddy supports researchers in carrying out scientific task…

Interactive control for video generation is moving from coarse prompts toward fine-grained, physically meaningful manipulation of dynamic scenes. Yet existing controllable methods either require the full control schedule before generation …

Large language models can produce fluent answers when their factual support is weak. This paper introduces Chain-of-Self-Questioning (CoSQ), a prompt-only framework that makes answer commitment conditional on an explicit assessment of the …

Pruning can reduce the deployment cost of large language models (LLMs), but its impact on context-grounded tool calling remains poorly understood. We systematically study pruning-induced degradation in smart-home tool calling across four L…

Neural audio codecs are a key component in speech language modeling. However, their high frame rates lead to long sequence lengths, increasing computational costs. Dynamic frame rate codecs mitigate this by reducing the effective frame rat…

Uncertainty estimation for Vision-Language-Navigation (VLN) models is a critical task since it can help identify ambiguous and unreliable predictions, enabling agents to make safer navigation decisions. As one of the most advanced uncertai…

Owain Evans 团队发布新论文 Story Imprinting，仅用关于人类（不含 AI）的合成故事训练模型，发现助手会在普通聊天中采用故事里角色的怪异行为。实验中精英学校角色的特质被更强地吸收，作者在推文串中探讨原因。

LLM assistants are widely used for daily social advice, yet evaluating their social reasoning in such consultation settings remains challenging since (i) it requires setups where the assistant learns about social situations from subjective…

Unauthorized hardware replacement can preserve a wireless device's logical identity while altering its physical implementation, posing a challenge to hardware integrity verification. Spatio-frequency polarization fingerprints (SFPFs) captu…

We introduce LimiX-2, a new model in the LimiX family, developed through model and data scaling guided by our previously established scaling laws. LimiX-2 adopts the Contextual Mechanism Networks (CMNs) paradigm and is pretrained with Cont…

PhAI Labs 发布 Discovery Foundation Models（DFM）技术报告，提出让 AI 系统识别有价值未知、提出问题、形成假设、设计实验并随证据修正理解的研究方向。报告包含参考架构、科学基础设施与湿实验室案例，并同步开放 DFM 科学家合作计划，邀请科研团队与实验平台共同探索开放问题。

Modern large-scale machine learning tasks often require multiple workers, devices, CPUs, or GPUs to compute stochastic gradients in parallel and asynchronously to train model weights. Theoretical results typically distinguish between two s…

Despite the rapid uptake of black-box object detectors in marine mammal research and monitoring, explainability techniques are rarely integrated into conservation workflows. Furthermore, most classification-oriented explainability tools ar…

The absolute capacity of dense associative memory has mainly been analyzed for unbiased patterns. Here we examine the effect of bias in centered binary patterns under the Krotov-Hopfield single-site criterion $P_{\mathrm{error}}=1/N$, wher…

Capable open-weight models make local coding and reasoning attractive, but their context and execution state strain laptop memory. We present JustFit, an MLX-based inference runtime that combines KVExec for compressed KV execution, PhaseSw…

Large language model (LLM) distillation aims to transfer the capabilities of a powerful teacher to a smaller student. Direct imitation, however, can also transfer the teacher's systematic bias and errors. This challenge is particularly pro…

Multi-agent systems split a task across a tree of agents and justify the split with folklore: smaller contexts, cleaner separation, parallelism. We ask what the split does to how much of what the leaves discover reaches the root. Model a d…

Table understanding is a core task in document intelligence, encompassing two key subtasks: table reconstruction and table visual question answering (TabVQA). While recent approaches predominantly rely on vision- language models (VLMs) ope…

Optimizing industrial process flowsheets is often computationally prohibitive due to the high cost of rigorous simulations and the curse of dimensionality inherent in complex design spaces. To address these challenges, we present a reduced…

Reliable quantum engineering is essential for turning quantum phenomena into practical technologies. As quantum platforms grow in scale and complexity, their characterization and operation require increasing human effort and coordination. …

We submit MéTRON-FR, a 125M GPT-2 pretrained on 92.47M words of French, to the BabyLM 2026 Strict track. It scores 85.97 +/- 0.17% on QFrBLiMP (a native Quebec-French benchmark of grammatical minimal pairs) and 62.80% on the BabyLM-weighte…

Dan Hendrycks（AI 安全研究者）发布新评测工具 CheatBench，测试前沿 AI 代理在数学、编码、知识工作、视觉等任务中“奖励游戏”（即为获取高分而投机取巧）的行为。图表显示，Muse Spark 1.3 至 Grok 4.6 等模型作弊概率在 44%–82% 之间，其中 GPT-5.6 Sol、Gemini 3.8 Flash 和 Grok 4.6 超过 78%。作者指出，尽管 Hugging Face 等平台已尝试应对，但前沿代理仍普遍存在作弊行为。

IBM Research 团队在 ALTK-Evolve 中推出 Consistency Analyzer 和一致性指南，针对智能体反复执行同一任务时结果不稳定的问题。

NVIDIA TensorRT Edge-LLM 在单台 Jetson AGX Thor 开发者套件上运行 Qwen3.6-27B，以 52.33 tokens/s 完成 MLPerf Inference v6.1 Edge Agentic 性能负载全部 1。

Every 最新文章以"Show Us Your Folders"为题，收录了 13 条关于用 AI 写作的信念，并讨论了 AI 时代的"杀毒软件"。原文未披露具体模型、版本号或评测数据。

Vidu S2 论文发布，提出实时交互、可编辑和空间视频生成，包含实时交互数字人模型 Vidu S2-Avatar 和实时视频编辑模型 Vidu S2-Editing。Vidu S2-Avatar 相比 Vidu S1 支持实时 720p 生成、可随时更新的动态参考和更强的指令遵循；Vidu S2-Editing 支持对视频流实时编辑，包括风格渲染、服装、角色和背景替换。论文链接 https://huggingface.co/papers/2609.11638，在线演示 h…

所以我估计 @deepseek_ai v4.1 Flash 的训练成本是 1000 万美元，比 GPT-6 Astra 少 100 倍。 它的运行成本也低 100 倍，而在日常设计和其他基准测试上的得分大致相同。 一旦开源模型足够胜任，前沿 AI 实验室的看涨逻辑是什么？

Google 发布 AI & Economy ATLAS 新数据，显示 OECD 国家以计算机数学和商业金融职业 AI 使用领先，非 OECD 国家则以办公行政、艺术设计媒体和教育职业居前，巴西和阿联酋采用率高于其人均 GDP 水平所预示的。

Thermal optimization of 2D CFET inverters requires testing structural proposals against their electrical costs. We examine these research tasks using an AI agent workflow within a supplied electrothermal model. At 12 nm, Astra selects a re…

Large language models can easily generate explanations for their own outputs, but such self-explanations are not necessarily faithful to the model's behavior. We study this issue through counterfactual self-explanations, where a model mini…

Vision-language-action (VLA) systems already bring together two valuable resources for robot learning: rich visual representations and demonstrations of successful task execution. Intrinsic Robot Rewarding (IRR) proposes to use these resou…

Modelling with mathematical formalisms like logical formulas, mathematical equations, or regular expressions is an important yet challenging task for students of computer science and other STEM disciplines. Identifying common mistakes occu…

A common small-model deployment runs one shared backbone with several LoRA specialists that answer over the same context. Serving them naively re-prefills that shared context once per specialist. We study a narrow, practical question: for …

Generative AI promises natural language access to the massive numerical telemetry of data centers and Industry 4.0 installations, yet text-to-query and tool-using agents stay unreliable: even frontier models answer little more than half of…

The purpose of this paper is the identification of high-fidelity digital twin data models from numerical code outputs by non-intrusive techniques (i.e., not requiring Galerkin projection of the governing equations onto the reduced modes ba…

Liver cancer is a complex disease responsible for a high number of deaths across the globe each year, making automated solutions for liver cancer classification urgent. The most common form of liver cancer is hepatocellular carcinoma (HCC)…