◇ 今日论文 · Co-RL: Unsupervised Reasoning Emerges from Diverse Cohort in Multi-agent RL

一句话人话版

这是一篇今日 AI 论文的线索卡：讲什么、谁写的、在哪里读全文。

编辑判断

适合当作今日前沿线索追踪；要复现实验请读原文。

基于字段派生 · 置信度：中（功能未逐项实测）

打开官网 / 来源 ↗

人读卡片 · 五问

① 这是什么

这是一篇今日 AI 论文的线索卡：讲什么、谁写的、在哪里读全文。（原文：今日论文《今日论文 · Co-RL: Unsupervised Reasoning Emerges from Diver…）

② 值不值得用

进入 HF 每日精选说明有社区关注度；是否与你的问题相关需要读原文判断，本条目不评价研究质量。

③ 怎么开始

先读摘要，需要再读全文（https://huggingface.co/papers/2608.17253）；对照论文 ID（2608.17253）可找实现与讨论。

展开：④ 关键证据 / ⑤ 注意事项与坑

④ 关键证据

论文 ID 2608.17253
作者 Yunhao Yang、Yuexin Bian、Yunjie Tian、Di Fu、Tianjin Huang、Yuanyuan Shi

⑤ 注意事项与坑

摘要来自作者原文，未做同行评议级核验；引用请以正式发表版本为准。

一句话使用前提

适合：追踪今日 AI 论文前沿的读者。

不适合：需要完整复现实验细节的场景（请读原文）。

状态：机检通过（结构与溯源校验）；功能未逐项实测（未校验）。

最小验证（3 步）

读摘要 — Co-RL: Unsupervised Reasoning Emerges from Diverse Cohort in Multi-agent RL
核作者/日期 — Yunhao Yang, Yuexin Bian, Yunjie Tian · 2026-08-19
跑 CLI — npx aiskillready search 17253

快速开始

可以这样用

npx aiskillready search 17253   # 本站 CLI：真实可跑（aiskillready@0.2.0）

说明：`npx aiskillready` 是本站已发布 CLI（npm aiskillready@0.2.0），命令真实可跑；第三方命令按官方文档给出、未逐一实测，本平台不编造结论。

术语速览：SKILL.md / veridrop / schema 是什么？（点开）

SKILL.md — 技能包的说明文件：告诉 Agent 这个技能怎么装、怎么用。
veridrop — 第三方 API 中转观测源：本站「观测」类条目的数据来源。
schema — 数据结构的约定版本：机器读 JSON 的字段规范。
Skill 客户端 — 支持「技能包」标准、能安装运行 Agent Skills 的 App / 工具。
Skill 注册表 — 收录「谁家发布了哪些技能」的公开目录（如 skills.sh）。

信任层：机检 / 成本 / 置信度 / schema（展开查看）

条目 ID：ASR-PAPER-20260820-2608-17253 ｜类型：论文｜ schema v0.5

✅ 机检通过 💰 升维成本 $0.0000 🧭 置信度中 📌 schema 0.5 🛂 审核 agent reviewed 🧪 功能未验证（Top20 实测计划）

相关推荐（应该知道的）

◇ 今日论文 · OmniScientist: An Omni-Modal Omni-Discipline AI Scientist

Recent advances in foundation models have enabled AI scientists to automate increasingly complete research workflows, from hypothesis generation and code execution to manuscript preparation. Yet workflow coverage alone does not provide access to the full evidence on which scientific discovery depends. Existing systems typically reason over text, code, labels, or precomputed summaries, leaving scie

相似

◇ 今日论文 · Scaling Creative Writing Beyond Story-Centric Data with Attribute-Guided Genre Expansion

High-quality creative writing data for large language models (LLMs) remains dominated by story-centric data, limiting models' ability to follow the structural and functional conventions of diverse creative formats. We propose an attribute-guided genre expansion framework for scaling creative writing data beyond story generation. By separating thematic breadth from genre-form control, our framework

相似

◇ 今日论文 · Training Leaves Traces: Centered Residual Signatures for Language Model Lineage Verification

Open-weight language models are fine-tuned, quantized, pruned, and merged, yet their provenance is often undocumented. We study data-free white-box lineage verification: can weights alone reveal whether two compatible model checkpoints share ancestry? Residual training produces a shared identity-aligned component in branch products, so this structure alone cannot establish ancestry. We remove it

反馈

这条不对 / 我执行报错了？带 entry_id 一键提 Issue，回流后修订版本。

🐞 纠错 / 报错 → 提 Issue

AI 区块 · 机器可读（同一事实源）

AI 区块 · 机器可读 JSON（点击展开）

{
  "schema_version": "0.5",
  "entry_id": "ASR-PAPER-20260820-2608-17253",
  "record_type": "paper",
  "title": "今日论文 · Co-RL: Unsupervised Reasoning Emerges from Diverse Cohort in Multi-agent RL",
  "observed_at": "2026-08-20",
  "source": {
    "name": "huggingface.co/api/daily_papers",
    "url": "https://huggingface.co/api/daily_papers?limit=10",
    "snapshot_hash": "d637c477ff6da03cf6c527f6d81da12836edef97191c108db740c2967537d66e",
    "fetched_at": "2026-08-20T04:51:45+00:00"
  },
  "confidence": "medium",
  "human_stream": {
    "summary": "Reinforcement learning (RL) has emerged as a powerful approach for improving reasoning in language and vision-language models, yet its strongest successes still depend heavily on ground-truth supervision (e.g., verifiable reward). Such annotations are costly to obtain and become increasingly scarce as reasoning capabilities advance beyond what humans can reliably evaluate. Self-rewarding RL reduce",
    "key_points": [
      "论文 ID 2608.17253",
      "作者 Yunhao Yang、Yuexin Bian、Yunjie Tian、Di Fu、Tianjin Huang、Yuanyuan Shi",
      "发布 2026-08-19T00:00:00.000Z"
    ],
    "note": "来源为 HuggingFace daily papers 公开 API；完整阅读请走原文链接。",
    "what_it_is": "今日论文《今日论文 · Co-RL: Unsupervised Reasoning Emerges from Diverse Cohort in Multi-agent RL》（Yunhao Yang、Yuexin Bian、Yunjie Tian、Di Fu、Tianjin Huang、Yuanyuan Shi），来自 HF Daily Papers 公开源。",
    "worth_it": "进入 HF 每日精选说明有社区关注度；是否与你的问题相关需要读原文判断，本条目不评价研究质量。",
    "how_to_start": "先读摘要，需要再读全文（https://huggingface.co/papers/2608.17253）；对照论文 ID（2608.17253）可找实现与讨论。",
    "evidence": [
      "论文 ID 2608.17253",
      "作者 Yunhao Yang、Yuexin Bian、Yunjie Tian、Di Fu、Tianjin Huang、Yuanyuan Shi"
    ],
    "cautions": "摘要来自作者原文，未做同行评议级核验；引用请以正式发表版本为准。"
  },
  "ai_stream": {
    "structured": {
      "paper_id": "2608.17253",
      "title": "Co-RL: Unsupervised Reasoning Emerges from Diverse Cohort in Multi-agent RL",
      "authors": [
        "Yunhao Yang",
        "Yuexin Bian",
        "Yunjie Tian",
        "Di Fu",
        "Tianjin Huang",
        "Yuanyuan Shi",
        "Ziang Xiao",
        "Nuno Vasconcelos",
        "Yijiang Li"
      ],
      "published_at": "2026-08-19T00:00:00.000Z",
      "url": "https://huggingface.co/papers/2608.17253",
      "source_label": "hf-daily-papers"
    },
    "raw": [
      {
        "paper_id": "2608.17253",
        "title": "Co-RL: Unsupervised Reasoning Emerges from Diverse Cohort in Multi-agent RL",
        "authors": [
          "Yunhao Yang",
          "Yuexin Bian",
          "Yunjie Tian",
          "Di Fu",
          "Tianjin Huang",
          "Yuanyuan Shi",
          "Ziang Xiao",
          "Nuno Vasconcelos",
          "Yijiang Li"
        ],
        "summary": "Reinforcement learning (RL) has emerged as a powerful approach for improving reasoning in language and vision-language models, yet its strongest successes still depend heavily on ground-truth supervision (e.g., verifiable reward). Such annotations are costly to obtain and become increasingly scarce as reasoning capabilities advance beyond what humans can reliably evaluate. Self-rewarding RL reduces this dependence by enabling models to derive reward signals from their own completions. However, training solely on self-generated feedback can reinforce existing biases and suboptimal behaviors, re",
        "published_at": "2026-08-19T00:00:00.000Z",
        "submitted_on_daily_at": null,
        "url": "https://huggingface.co/papers/2608.17253"
      }
    ]
  },
  "token_cost": {
    "total": 0.0,
    "currency": "USD",
    "breakdown": {
      "crawl": 0.0,
      "clean": 0.0,
      "elevate": 0.0,
      "verify": 0.0
    }
  },
  "machine_verified": true,
  "review": {
    "status": "agent_reviewed",
    "reviewer": "pipeline-validate",
    "reviewed_at": "2026-08-20T08:33:14+00:00",
    "comments": "schema 0 error + 溯源一致 + 成本达标"
  },
  "provenance": {
    "extracted_by": "codex",
    "extracted_at": "2026-08-20T04:56:41+00:00",
    "pipeline": "collect_hf_papers.py + build_entries.py v0.2（规则管线）",
    "access_urls": [
      "https://huggingface.co/api/daily_papers?limit=10"
    ],
    "card_generated_at": "2026-08-20T08:33:14+00:00",
    "card_pipeline": "enrich_human_stream.py v0.1（规则模板，零 Token）"
  }
}