Daily Papers · 2026-09-07
📄 论文日报 | 来源：Hugging Face Daily Papers
🗓 Episode：【周末特辑】9月第1周最火AI论文 | 仓库蒸馏技能，提升研究智能体；模拟学生反馈，优化AI导师。
1. 从仓库到技能：将GitHub代码库蒸馏为AI4AI技能
- 英文标题：Repo-To-Skill: Distilling GitHub Repositories Into AI4AI Skills
- arXiv：2609.02749
💡 一句话解读：这篇研究要解决AI做科研时缺乏“操作知识”的问题，即知道方法但不会实际运行。他们提出DisCo系统，把GitHub上大量代码库提炼成可复用的“技能”，让AI代理在任务中直接调用，避免每次重新摸索。实验显示，加了技能后AI在多个基准上的表现显著提升。
自主代理开始端到端地执行机器学习（ML）研究。这些代理将模型骨干与规划、执行、记忆和验证的框架相结合，但该架构仍将领域特定的操作知识（operational knowledge）排除在代理之外。这种知识并非不存在，而是以面向人类读者的形式出现在代码库和论文中，且规模过大，无法在任务期间加载。一旦将其提炼为紧凑、经过验证的技能，这些知识便可在任务间复用，而无需每次运行重新发现。我们提出了DisCo，一个技能驱动的科研代理，它创建技能并在研究中使用技能。其提炼以两种互补形式进行：任务无关型，将领域广泛使用的代码库浓缩为可复用技能；任务导向型，生成具体任务所需的技能。前者应用于开放生态系统，产生了AREX技能库，包含从1000个广泛使用的ML代码库中提炼的5000多个已验证技能，组织成20个领域和178个能力族。在固定GPT-5.5骨干、研究框架和下游执行预算的情况下，配备技能的科研代理在MLE-bench上得分提高134.3%，在PaperBench上提高34.4%，在FrontierCS上提高9.2%，在PassNet上提高14.0%，相比无技能的同一代理。这些收益来自在该固定设置下添加提炼的操作上下文。
📄 English Abstract
Abstract: …call this missing layer operational knowledge, the know-how that separates knowing a method from making it work. That knowledge is not absent from the field. It appears in repositories and papers, but in forms written for human readers and too large to load during a task. Once… ▽ More Autonomous agents are beginning to carry out machine-learning (ML) research end to end. These agents combine a model backbone with a harness for planning, execution, memory, and verification, but this architecture still leaves domain-specific know-how outside the agent. We call this missing layer operational knowledge, the know-how that separates knowing a method from making it work. That knowledge is not absent from the field. It appears in repositories and papers, but in forms written for human readers and too large to load during a task. Once distilled into compact, verified skills, this knowledge can be reused across tasks rather than rediscovered during each run. We present DisCo, a skill-powered research agent that creates skills and uses them during research. Its distillation runs in two complementary forms: task-agnostic, condensing the field's widely used repositories into reusable skills, and task-oriented, producing the skills a concrete task calls for. The former, applied across the open ecosystem, yields the AREX-Skill Library, with 5,000+ verified skills distilled from 1,000 widely used ML repositories and organized into 20 areas and 178 capability families. With the GPT-5.5 backbone, research harness, and downstream execution budget held fixed, the skill-equipped research agent scores 134.3% higher on MLE-bench, 34.4% higher on PaperBench, 9.2% higher on FrontierCS, and 14.0% higher on PassNet than the same agent without skills. These gains come from adding distilled operating context under that fixed setup. △ Less
2. StudentSim：训练基于大语言模型的学生模拟器
- 英文标题：StudentSim: Training LLM-based Student Simulators
- arXiv：2609.01591
💡 一句话解读：AI 辅导系统要因材施教，但收集真实学生反馈太慢太贵。现有模拟器要么学不会解释纠错，要么模仿不准。这篇论文提出 StudentSim 训练框架，先用大量数据训练通用模型，再针对每个学生微调，造出既像学生又能接受指导的模拟器，并配套评估标准。实验证明它比 GPT-5.4 更准，还能用来训练更好的 AI 老师。
AI 辅导系统在适应每个学生的优势、弱点和偏好指导时最为有用，但关于哪种指导对哪种学生有效的证据稀疏、获取缓慢且成本高昂。学生模拟器可以作为代理提供这种信号，但现有方法存在局限：状态跟踪模型拟合学生行为但难以处理解释或纠正，而大语言模型角色扮演能流畅遵循指导但无法可靠匹配被模仿学生的能力。我们提出 StudentSim，一个训练框架，通过池化训练后跟每个学生的专门化，将稀疏的每学生数据转化为个性化模拟器。生成的模拟器既能反映学生自身的回答，也能在教师指导下更新这些回答。我们还引入 StudentSimEval，一个标准化协议，涵盖国际象棋、第二语言英语写作和数学三个领域的 60 名学生，使用公开的学习者数据集，其中包含用于研究的去标识化记录。StudentSimEval 测量行为保真度（F）——模拟器匹配学生回答的程度，以及指导响应性（R）——模拟器在教师指导下更新的容易程度，所有方法都在相同记录上进行拟合和评估。在所有三个领域中，StudentSim 在两个指标上都优于 GPT-5.4。在国际象棋中，StudentSim 达到 F=0.51 和 R=0.91，而 GPT-5.4 为 0.23 和 0.72，Maia2 为 0.45 和 0.27。作为概念验证，使用 StudentSim 作为教师强化学习的奖励模型，产生了一个国际象棋教师，专家人类评价其比无强化学习基线和针对 GPT-5.4 模拟器奖励训练的教师更准确、指导更好、更个性化。代码可在 https://github.com/microsoft/StudentSim 获取。
📄 English Abstract
Abstract: AI tutors are most useful when they adapt to each student's strengths, weaknesses, and preferred guidance, but evidence about which guidance works for which… ▽ More AI tutors are most useful when they adapt to each student's strengths, weaknesses, and preferred guidance, but evidence about which guidance works for which student is sparse, slow, and costly to collect from real learners. Student simulators can provide this signal as a proxy, yet existing approaches are limited: state-tracking models fit student behavior but struggle to process explanations or corrections, while LLM role-play follows guidance fluently but does not reliably match the competence of the student being imitated. We present StudentSim, a training framework that turns sparse per-student data into individualized simulators through pooled training followed by per-student specialization. The resulting simulators both mirror a student's own responses and update them under tutor guidance. We also introduce StudentSimEval, a standardized protocol covering 60 students across chess, second-language English writing, and mathematics, using public learner datasets with de-identified records shared for research. StudentSimEval measures behavioral fidelity (F), or how well a simulator matches a student's responses, and guidance responsiveness (R), or how readily it updates under tutor guidance, with all methods fit and evaluated on the same records. Across all three domains, StudentSim outperforms GPT-5.4 on both metrics. In chess, StudentSim reaches F=0.51 and R=0.91, compared with 0.23 and 0.72 for GPT-5.4 and 0.45 and 0.27 for Maia2. As a proof of concept, using StudentSim as a reward model for tutor reinforcement learning produces a chess tutor that expert humans rate as more accurate, better-guided, and more personalized than a no-RL baseline and a tutor trained against a GPT-5.4 simulator reward. Code is available at https://github.com/microsoft/StudentSim. △ Less
3. Qwen-Drive-1.0：迈向自动驾驶视觉语言基础模型的第一步
- 英文标题：Qwen-Drive-1.0: An Initial Step towards a Vision-Language Foundation Model for Autonomous Driving
- arXiv：2609.00111
💡 一句话解读：这篇研究要解决自动驾驶中如何让AI同时理解路况、回答问题和规划行驶路线的问题。他们提出了一个叫Qwen-Drive-1.0的系统，它像给AI装上一双‘鸟眼’，能看清周围3D环境，并在此基础上做出驾驶决策。通过分阶段训练，既学会了开车，又保留了通用对话能力。
我们提出了Qwen-Drive-1.0，这是迈向自动驾驶视觉语言基础模型的第一步。Qwen-Drive-1.0保留了预训练视觉语言模型（VLM）的架构，并在一个统一框架内整合了3D感知、视觉问答和运动规划。一个外部鸟瞰图（BEV）感知头联合执行3D目标检测、语义占用预测和BEV地图分割。它作为共享表示中可访问的3D信息的探针，并为3D场景结构提供了显式、可检查的接口。一个规划专家（Planning Expert）基于共享的VLM表示来生成未来的自我轨迹。一种分阶段的训练方案将驾驶监督与通用视觉语言数据相结合，以获取驾驶特定能力，同时有助于保留广泛的视觉理解和指令跟随能力。实验表明，在保持通用视觉语言能力的同时，实现了强大的3D感知和驾驶场景理解。在开环、伪闭环和闭环设置中的全面评估进一步显示了极具竞争力的运动规划性能。
📄 English Abstract
Abstract: We present Qwen-… ▽ More We present Qwen-Drive-1.0, an initial step towards a vision-language foundation model for autonomous driving. Qwen-Drive-1.0 retains the architecture of the pretrained vision-language model (VLM) and integrates 3D perception, visual question answering, and motion planning within a unified framework. An external bird's-eye-view (BEV) perception head jointly performs 3D object detection, semantic occupancy prediction, and BEV map segmentation. It serves as a probe of the 3D information accessible from the shared representations and provides an explicit, inspectable interface to 3D scene structure. A Planning Expert conditions on shared VLM representations to generate future ego trajectories. A staged training recipe combines driving supervision with general-purpose vision-language data to acquire driving-specific competence while helping preserve broad visual understanding and instruction-following capabilities. Experiments demonstrate strong 3D perception and driving scene understanding while largely preserving general vision-language capability. Comprehensive evaluations across open-loop, pseudo-closed-loop, and closed-loop settings further show highly competitive motion-planning performance. △ Less
4. 通过训练进行编译：将自然语言规范转化为本地神经函数
- 英文标题：Compile by Training: Turning Natural-Language Specifications into Local Neural Functions
- arXiv：2609.04199
💡 一句话解读：很多常见的文本处理任务，用规则很难实现，但每次调用大型远程模型又慢又贵。这篇研究提出一种“编译”方法，把自然语言描述转换成一个小型、可复用的本地函数。它先用大模型生成示例来训练一个小模型，之后这个小模型就能独立运行，无需再依赖大模型。
许多重复出现的文本功能易于描述但难以用规则实现，而每次输入都调用大型远程模型会带来重复成本、延迟和对提供商的依赖。我们提出了“通过训练进行编译”（compile by training），它将自然语言规范转化为可复用的神经函数。在编译时，教师模型生成任务特定的示例，用于训练一个紧凑解释器的小型适配器。生成的函数无需教师即可运行，并且可以像普通软件一样存储、版本控制和组合。在FuzzyBench-Hard上，一个快速编译器（Program-as-Weights）未能产生任何精确匹配的子集上，通过训练进行编译达到了83.6%的语义准确率。这种更高的准确率伴随着更高的编译时间成本：大约一分钟，而快速编译器只需几秒。我们将该编译器部署在公共交互式服务中，并展示了编译函数在多站点网站助手、语言控制的3D虚拟角色以及双向英语-克劳迪什翻译器中的应用。
📄 English Abstract
Abstract: Many recurring text functions are easy to describe but difficult to implement with rules, while calling a large remote model for every input introduces repeated cost, latency, and dependency on a provider. We present… ▽ More Many recurring text functions are easy to describe but difficult to implement with rules, while calling a large remote model for every input introduces repeated cost, latency, and dependency on a provider. We present compile by training, which turns a natural-language specification into a reusable neural function. At compile time, teacher models generate task-specific examples that are used to train a small adapter for a compact interpreter. The resulting function runs without the teachers and can be stored, versioned, and composed like ordinary software. On FuzzyBench-Hard, a subset on which the Program-as-Weights fast compiler produced no exact matches, compile by training reaches 83.6% semantic accuracy. This higher accuracy comes with a higher compile-time cost: roughly a minute rather than seconds for the fast compiler. We deploy the compiler in a public interactive service and demonstrate compiled functions in a multi-site website helper, a language-controlled 3D avatar, and a bidirectional English-Claudish translator. △ Less
5. HarnessDev：大语言模型能否创建并演化自己的智能体执行框架？
- 英文标题：HarnessDev: Can LLMs Create and Evolve Their Own Agent Harness?
- arXiv：2609.01437
💡 一句话解读：这篇研究想解决一个问题：大语言模型能不能自己搭建和优化自己的“执行框架”（就像给机器人造身体和工具）？他们设计了一个叫HarnessDev的测试，分两步：先让模型从零开始造一个能用的系统，再让它根据运行结果自己改进。结果发现，模型造的系统在写作和机器学习实验上能赶上甚至超过人类设计的，但在代码和搜索上还差得远，而且改进效果不稳定。
随着智能体从研究原型转向部署工具，其能力越来越依赖于模型外部的执行基础设施，即所谓的智能体执行框架（agent harness）。在保持模型权重不变的情况下改变这一框架可以显著改变任务性能。当前的智能体评估通常在选定的框架下报告下游性能，而模型自身开发框架的能力相对未被探索。我们引入了HarnessDev，一个将评估单元从任务输出转向可运行基础设施的基准。HarnessDev涵盖两个阶段：在创建（Creation）阶段，智能体从最小种子和少量案例出发，构建一个完整的执行系统；在演化（Evolution）阶段，它从自己创建的框架出发，利用下游执行反馈迭代修改，以提高基准性能。我们随后评估每个构建框架的能力（在保留基准上的任务成功率）和效率（执行令牌成本）。报告的创建结果涵盖六个创建者大语言模型、四个领域和五个下游基准，共2,207个独特下游实例，且隐藏评估任务未参与开发。我们发现，生成的框架在代码以及搜索和研究方面仍明显落后于成熟的人工工程参考，但在写作和机器学习实验方面达到或超过所选参考，且执行成本差异很大。演化带来了一些性能提升，但不稳定，且仅部分转移到保留任务上。使用固定运行时模型的实验进一步表明，收益强烈依赖于执行框架的模型，表明跨模型的迁移有限。
📄 English Abstract
Abstract: As agents move from research prototypes to deployed tools, their capability increasingly depends on model-external execution infrastructure, commonly termed the… ▽ More As agents move from research prototypes to deployed tools, their capability increasingly depends on model-external execution infrastructure, commonly termed the agent harness. Changing this harness while holding model weights fixed can substantially alter task performance. Current agent evaluations typically report downstream performance under a chosen harness, leaving a model's ability to develop the harness itself comparatively underexplored. We introduce HarnessDev, a benchmark that shifts the unit of evaluation from task outputs to runnable infrastructure. HarnessDev covers two stages. In Creation, the agent starts from a minimal seed and a small number of cases, then builds a complete execution system. In Evolution, it starts from its own created harness and iteratively revises it using downstream execution feedback, with the goal of improving benchmark performance. We then evaluate each constructed harness on capability (task success on held-out benchmarks) and efficiency (execution-token cost). The reported Creation results cover six creator LLMs, four domains, and five downstream benchmarks totaling 2,207 unique downstream instances, with hidden evaluation tasks withheld from development. We find that generated harnesses remain substantially behind mature human-engineered references on code and on search and research, while matching or exceeding the selected references on writing and machine-learning experimentation, with large variation in execution cost. Evolution produces some performance gains, but they are unstable and transfer only partially to held-out tasks. Experiments with a fixed runtime model further show that the gains depend strongly on the model executing the harness, indicating limited transfer across models. △ Less