◇ Paper · MOSS-VL Technical Report

{
  "schema_version": "0.5",
  "entry_id": "ASR-PAPER-20260818-2608-15045",
  "record_type": "paper",
  "title": "今日论文 · MOSS-VL Technical Report",
  "observed_at": "2026-08-18",
  "source": {
    "name": "huggingface.co/api/daily_papers",
    "url": "https://huggingface.co/api/daily_papers?limit=10",
    "snapshot_hash": "652c1f13fb67dd93f7d60fecc11565d77173419a1e4693958922f4b73738fb42",
    "fetched_at": "2026-08-18T10:51:18+00:00"
  },
  "confidence": "medium",
  "human_stream": {
    "summary": "We present MOSS-VL, an open vision-language model family that treats real-time interaction -- perceiving while it speaks -- as a first-class capability. It is co-designed across the stack: the language decoder attends to vision only through gated cross-attention, so the model can naturally see incoming frames while generating; a synthesized interaction corpus supervises when to speak, when to stay",
    "key_points": [
      "论文 ID 2608.15045",
      "作者 Pengyu Wang、Chenkun Tan、Shaojun Zhou、Qirui Zhou、Yanxin Chen、Xingyang He",
      "发布 2026-08-15T00:00:00.000Z"
    ],
    "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 · MOSS-VL Technical Report》（Pengyu Wang、Chenkun Tan、Shaojun Zhou、Qirui Zhou、Yanxin Chen、Xingyang He），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/2608.15045）；对照论文 ID（2608.15045）可找实现与讨论。",
    "evidence": [
      "论文 ID 2608.15045",
      "作者 Pengyu Wang、Chenkun Tan、Shaojun Zhou、Qirui Zhou、Yanxin Chen、Xingyang He"
    ],
    "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": "2608.15045",
      "title": "MOSS-VL Technical Report",
      "authors": [
        "Pengyu Wang",
        "Chenkun Tan",
        "Shaojun Zhou",
        "Qirui Zhou",
        "Yanxin Chen",
        "Xingyang He",
        "Huazheng Zeng",
        "Jijun Cheng",
        "Chenghao Wang",
        "Xiaomeng Qian",
        "Pengfei Wang",
        "Zhan Huang",
        "Shanqing Gao",
        "Wei Huang",
        "Longjun Cao",
        "Wu Ran",
        "Jie Liu",
        "Changtai Zhu",
        "Hongkai Wang",
        "Yixian Tian",
        "Chenghao Liu",
        "Zhen Ye",
        "Xinghao Wang",
        "Botian Jiang",
        "Guoguo Feng",
        "Zhaoye Fei",
        "Ruixiao Li",
        "Mingshu Chen",
        "Yang Gao",
        "Qinyuan Cheng",
        "Shimin Li",
        "Xipeng Qiu"
      ],
      "published_at": "2026-08-15T00:00:00.000Z",
      "url": "https://huggingface.co/papers/2608.15045",
      "source_label": "hf-daily-papers"
    },
    "raw": [
      {
        "paper_id": "2608.15045",
        "title": "MOSS-VL Technical Report",
        "authors": [
          "Pengyu Wang",
          "Chenkun Tan",
          "Shaojun Zhou",
          "Qirui Zhou",
          "Yanxin Chen",
          "Xingyang He",
          "Huazheng Zeng",
          "Jijun Cheng",
          "Chenghao Wang",
          "Xiaomeng Qian",
          "Pengfei Wang",
          "Zhan Huang",
          "Shanqing Gao",
          "Wei Huang",
          "Longjun Cao",
          "Wu Ran",
          "Jie Liu",
          "Changtai Zhu",
          "Hongkai Wang",
          "Yixian Tian",
          "Chenghao Liu",
          "Zhen Ye",
          "Xinghao Wang",
          "Botian Jiang",
          "Guoguo Feng",
          "Zhaoye Fei",
          "Ruixiao Li",
          "Mingshu Chen",
          "Yang Gao",
          "Qinyuan Cheng",
          "Shimin Li",
          "Xipeng Qiu"
        ],
        "summary": "We present MOSS-VL, an open vision-language model family that treats real-time interaction -- perceiving while it speaks -- as a first-class capability. It is co-designed across the stack: the language decoder attends to vision only through gated cross-attention, so the model can naturally see incoming frames while generating; a synthesized interaction corpus supervises when to speak, when to stay silent, and when to revise; and a staged curriculum concentrates all real-time-specific training in one light final stage over a strong offline foundation. Offline, MOSS-VL-Instruct is competitive at",
        "published_at": "2026-08-15T00:00:00.000Z",
        "submitted_on_daily_at": null,
        "url": "https://huggingface.co/papers/2608.15045"
      }
    ]
  },
  "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": "2026-08-18T10:54:20+00:00",
    "comments": "机检通过（schema 0 error + 溯源一致 + 成本达标）"
  },
  "provenance": {
    "extracted_by": "codex",
    "extracted_at": "2026-08-18T10:53:05+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": "2026-08-18T10:53:05+00:00",
    "card_pipeline": "enrich_human_stream.py v0.1（规则模板，零 Token）"
  }
}