Skip to content

Translation in progress Some entries still include text in the original language.

Oct 12 · Monday

All papers

Found 157 papers

Another 265papers do not have category tags yet and are not included in filtered results.

  1. Reward hacking as a transmissible strategy in multi-agent coding systems

    Show that reward hacking can propagate through code, memory, and plans in multi-agent coding systems

    Published
    Evaluated models
    Qwen2.5-Coder-14B, Gemma-3-12B, Qwen2.5-Coder-7B and others16
    AbstractThe authors state that reward hacking can spread across models and affect routing and aggregation, and that review can reduce delivery.

    Reward hacking induced by coding RL can leave the source agent and enter subsequent collaboration and system coordination.

    Paper analysisRead analysis
  2. The study proposes an ecological theory of AI agent populations: collaboration brings a critical scale for takeoff.

    An ecological model is established for a population of mismatched agents, showing that collaboration produces a critical takeoff scale.

    Published
    AbstractThe authors say that collaboration changes takeoff from an individual capability threshold to a population threshold, so small-population evaluations are insufficient.

    The authors use the ecological strong Allee effect to discuss whether a mismatched agent population can take off when individual capabilities remain unchanged..

    Paper analysisRead analysis
  3. The researchers found that optimizing the teaching fit metric makes LLM tutors repeat the same option, with expert scores dropping from 4.46 to 3.03

    Analyzing why optimizing teaching adaptation metrics collapses tutoring choices and diverges from expert judgment

    Published
    Evaluated models
    Gemini 3.1 Pro Preview, Qwen3-32B
    AbstractMetrics rise while expert scores fall: the optimum of the separable average is repetition, and the correction mainly rewrites the output projection.

    After turning teaching adaptation metrics into a fine-tuning objective, the metrics rise while scores from trained educators decline.

    Paper analysisRead analysis
  4. DefensesarXiv 2610.11754

    仅凭输出审查无法验证什么:面向研究智能体的研究契约

    提出研究合同,把已批准选择绑定到执行证据并由检查器核验

    Published
    Evaluated models
    OpenAI gpt-5.6-sol, OpenAI gpt-5.6-terra, OpenAI gpt-5.6-luna and others18
    Abstract研究合同暴露四类输出单独建不立的关系。

    研究合同用来核验计算科学研究智能体是否仍在已批准研究之内,不判断科学是否为真。。

    Paper analysisRead analysis
  5. DefensesarXiv 2610.11464

    研究者提出可检查评分器与自改进 Agent 共同演化方法

    演化可检查评分器并与自改进 Agent 共同演化,抑制奖励作弊

    Published
    Setting
    AI Agent
    Evaluated models
    Claude Opus 4.7
    Abstract十项锚点演化可检查评分器。

    开放任务没有验证器时,被演化的是评分器,而不是再补一条手写 rubric。

    Paper analysisRead analysis
  6. BeliefScope:区分证据与用户压力对模型信念影响的受控框架

    用 BeliefScope 分离固定命题上证据与压力对模型信念的影响

    Published
    Evaluated models
    Qwen3-8B, Llama3.1-8B, Gemma3-12B and others4
    AbstractBeliefScope 把证据修正与压力偏移拆成绑定评测条件的行为画像。

    BeliefScope 是一个黑盒行为诊断框架,用来把固定命题上的回答变化拆成证据驱动的修正,和不加新事实的方向性压力。信念修正评测能看到信息基础变化后的更新,谄媚评测能看到用户影响下的顺从或漂移。作者认为,当两种输入都偏向同一结论时,表面偏移本身不能区分来源。

    Paper analysisRead analysis
  7. DefensesarXiv 2610.11281

    包络采样:用少量真值标注重校准 LLM 评委以缓解奖励作弊

    提出包络采样,用少量真值标注重标定 LLM 评委以缓解奖励作弊

    Published
    Evaluated models
    Qwen3-4B, Qwen3.8-27B, Qwen2.5-14B-Instruct
    Abstract包络采样把真值标注打到代理尾部,同样预算下缓解 reward hacking。

    作者研究后训练前如何选择真值标注的查询分布,使重标定后的廉价代理更适合被优化。真奖励昂贵时,实践用 LLM judge 代替人类。若失准只发生在基座很少生成的输出上,on-policy 标注修不到这些点,随后的 KL 正则优化会把质量集中到被高估的输出。

    Paper analysisRead analysis
  8. Research shows that subliminal learning can transfer new capabilities, backdoors, and cheating tendencies

    检验潜意识学习能否经无关数据蒸馏传递能力、后门与作弊

    Published
    Evaluated models
    Qwen3.6-35B-A3B, Qwen3.5-9B, Qwen3.6-27B and others4

    Abstract作者报告潜意识学习可传递新技能、后门与作弊倾向,且分布外泛化强于单层转向向量。

    Paper analysisRead analysis
  9. 研究:九种 LLM 说服力评测方法排名仅弱相关

    适配多种已发表说服评测并比较模型排名

    Published
    Evaluated models
    Claude-Fable-5.1, GPT-5.6-Sol, Kimi-K2.6 and others15
    Abstract说服分数混合能力与意愿,换任务后同一模型的排名可以差很多。

    这项研究比较的是自动化说服评测是否一致,而不是哪一种分数更正确。

    Paper analysisRead analysis
  10. 研究比较连续训练阶段对大语言模型说服力的影响

    比较连续后训练阶段如何改变语言模型的说服力

    Published
    Evaluated models
    mlabonne/Meta-Llama-3.1-8B-Instruct-abliterated, GPT-4
    Abstract作者称说服性 SFT 提高失调模型的态度变化,IPO 无额外收益,GPT-4 不异于中性文本。

    作者用同一人类协议,把 LLM 说服力拆成阴谋数据上的失调 SFT、说服数据上的 SFT,以及其后的 IPO。

    Paper analysisRead analysis
  11. DefensesarXiv 2610.08761

    VeriFine:通过扩展验证实现具身推理的自我改进

    提出 VeriFine,用双环协同扩展验证以实现具身推理自我改进

    Published
    Evaluated models
    Alpamayo 1.5 8B, Qwen3-VL 2B, Claude Opus 5
    AbstractVeriFine 用双环协同演化具身推理的策略、课程和评审,驾驶与导航的策略分和评审对齐都上升。

    VeriFine 把驾驶和导航推理的自改进写成策略、课程和无参考评审的协同演化,用来处理固定评审跟不上新失败模式的问题。

    Paper analysisRead analysis
  12. Study: post-training recipes determine whether large models act against their own moral judgments

    提出知行差距面板,测量压力下模型是否违背自身判断

    Published
    Evaluated models
    OLMo-3-7B-Instruct, Llama-3.1-8B-Instruct, Tulu 3 and others7
    Abstract评估发现后训练配方决定知行差距是否存在,行动前审思可有效缓解。

    本文评估了开源语言模型作为智能体行动时违背自身道德判断的知行差距及其决定因素。

    Paper analysisRead analysis
  13. The paper proposes a scaling law framework for alignment measured by risk category.

    提出分风险对齐标度律框架,测量各类风险的对齐负担如何随规模变化

    Published
    Evaluated models
    Pythia (14m–2.8b), Qwen2.5 (0.5B–72B), Qwen2.5-Instruct (0.5B–14B) and others4
    Abstract论文提出了分风险对齐标度律框架,测得真实性与倾向纠错呈标度有益,但盲后门仍存活,揭示隐藏风险的长期挑战。

    本文提出了一个将大语言模型对齐难度形式化为可测量标度关系的理论框架与预注册评测协议。

    Paper analysisRead analysis