数据来源：HuggingFace Papers Latest Papers1. Compile by Training: Turning Natural-Language Specifications into Local Neural FunctionsAbstract: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 ...

数据来源：HuggingFace Papers Latest Papers1. Unlocking Lossless Speedups in LLMs via Discrete DiffusionAbstract:Large Language Models (LLMs) owe much of their success to next-token prediction (NTP), but their autoregressive (AR) structure requires slow, sequential token generation. To overcome this bottleneck, we introduce diffusion-augmented LLMs, a new class of models that defines an AR model distribution while using diffusion to draw multiple tokens in parallel from that distribution. We decoupl ...

数据来源：HuggingFace Papers Latest Papers1. Bilevel Coordinated Reflection: A Game-Theoretic Approach to Multi-Agent LLM SystemsAbstract:Multi-agent LLM systems commonly use an orchestrator to decompose a task for a team of workers and then improve through textual reflection. Despite strong empirical results, these systems lack a unified account of coordination, memory improvement, and the role of external verification. We model orchestrator-worker interaction as a bilevel coordination game: under ...

数据来源：HuggingFace Papers Latest Papers1. Show-Harness: Just a VLM Agent Can Play RobotsAbstract:Foundation vision-language models (VLMs) exhibit broad intelligence about the world, yet translating this intelligence into robot control remains challenging. We present Show-Harness, an Embodied Harness that enables VLMs to “play” robots through a compact semantic interface linking intent to action. Show-Harness exposes discrete semantic action units that VLMs can naturally reason over, while embodi ...

数据来源：ArXiv Domain LLM Domain Papers1. Modality Discrepancy Transformer for Ambivalence and Hesitancy RecognitionAbstract:Ambivalence and hesitancy (A/H) are affective states in which individuals express contradictory signals across facial, vocal, and linguistic channels. Automatically recognising A/H in clinical videos requires detecting cross-modal disagreement — the signal that standard fusion methods suppress. Based on the conflict-aware multimodal fusion framework of Bekhouche et al., we p ...

数据来源：HuggingFace Papers Latest Papers1. NCP-ArchPreview Technical Report: Moving towards Latent Space Language Models through Next Concept PredictionAbstract:We introduce NCP-ArchPreview, a latent-space language model that pushes autoregressive pretraining beyond standard next-token prediction (NTP). Alongside NTP, the model learns through Next Concept Prediction (NCP) to predict discrete concepts that span multiple tokens, introducing an explicit and more challenging concept-level objective w ...

数据来源：HuggingFace Papers Latest Papers1. NCP-ArchPreview Technical Report: Moving towards Latent Space Language Models through Next Concept PredictionAbstract:We introduce NCP-ArchPreview, a latent-space language model that pushes autoregressive pretraining beyond standard next-token prediction (NTP). Alongside NTP, the model learns through Next Concept Prediction (NCP) to predict discrete concepts that span multiple tokens, introducing an explicit and more challenging concept-level objective w ...

数据来源：HuggingFace Papers Latest Papers1. Omni-Streaming ThinkingAbstract:Streaming omni-modal models must decide what and when to answer from the video chunks and synchronized audio observed so far. Visual cues often support an interpretation before an utterance or sound event is complete. If that interpretation enters memory as a fact, later reasoning can keep relaying it even after audio contradicts it. We call this failure premature cross-modal commitment. We propose Omni-Streaming Thinking ...

数据来源：HuggingFace Papers Latest Papers1. Continual Learning Mechanisms Compose for Long-Horizon MemorizationAbstract:Language models may need to internalize information that arrives over time and retain it through many subsequent updates. To study this challenge, we introduce long-horizon memorization, a setting in which a model learns 100 query-answer tasks through continual supervised fine-tuning without retaining earlier training examples or receiving task identifiers at inference. Sequentia ...

数据来源：HuggingFace Papers Latest Papers1. DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache CompressionAbstract:The widespread adoption of long-horizon agents has made model workloads increasingly input-heavy. Although prior work has substantially reduced the cost of long-context computation, prefill remains computationally expensive, and large KV caches continue to strain HBM and SSD capacity and data-transfer bandwidth. Together, these compute, storage, and bandwidth demands constitute the p ...

微博 排名 话题 1 台风白海豚已达17级 2 超强台风白海豚来了 3 强军制胜不负荣光 4 周星驰质疑董宇辉看大话西游500遍 5 辽宁凌晨炸街人员全被带走调查 6 房主任被停演3个月 7 歌手排名 8 外军想回家解放军回复fight 9 国盛证券被证监会立案 10 南美足联回应国际足联 11 曝高以翔生前女友官宣生子 12 连云港消防回应火灾母女坠楼 13 医保个人账户里的钱怎么用才不浪费 14 李光洙起哄金宇彬申敏儿 15 辅助生殖漏洞让不忠者有可乘之机 16 房主任致歉 17 张雅琪挑战玩手机 18 低价大流量卡将集体退场 19 三大运营商将停止第三方互联网渠道办卡 20 三大运营商发布公告 21 公瑾爆蛋创始人发声 22 上海一顾客因大盘鸡里有土豆报警 23 孕妇室内骑平衡车与人相撞致流产 24 Jennie的纹身 25 原来到一定年龄就想生小孩了 26 建议中年人要长期坚持力量训练 27 苏醒唱歌时何炅表情 28 美军一架F35战机在加州基地坠毁 2 ...

微博 排名 话题 1 公职人员停车受阻暴力损坏地锁 2 郑钦文回应无缘多伦多站正赛 3 中国新型战机超视距猎杀 4 直观感受到什么是家道中落 5 女儿说房主任愧疚式教育 6 不吃压力并展示了基建 7 曾舜晞说孟子义拍戏动过真感情 8 雷军说晚安被指营销 9 婚外胚胎案当事人女儿刚考上一本 10 郑钦文无缘多伦多站正赛 11 部分民办本科出现招生难 12 谢贤前女友CoCo直播回应与张柏芝关系 13 董璇也怀疑周也是自己女儿 14 发现赚钱方式对人影响很大 15 王一博曾是YG第一 16 蜘蛛侠 屏摄 17 徐洁云回应雷军说晚安被指营销 18 存储芯片利好来了 19 马斯克关注DeepSeek账号 20 肖战怎么瘦了这么多 21 已婚男子与军嫂同居被当场抓获 22 邹市明怀念从前的冉莹颖 23 240名警力出动抓获飙车炸街103人 24 黄子韬给罗正刷礼物 25 婚外胚胎案女子对原配叫嚣我没错 26 郑钦文进攻体系全面失灵 27 蔡徐坤默片 28 王玉雯20个蒜扒了一 ...

微博 排名 话题 1 赛里木湖景区7人被行拘 2 赛里木湖按人收费改为按车收费 3 6图看中欧班列开行规模稳步向上 4 迪丽热巴裙子里面是牛仔裤 5 赛里木湖景区情况通报 6 显卡跳涨 7 河南三支一扶案件相关负责人停职 8 蔡徐坤说终于撕掉爱豆标签 9 手机不好卖 芯片出货量暴跌 10 孙颖莎因伤退赛 11 网友吐槽天才女友女主四败爱因斯坦 12 释小龙时隔30年再次探班周星驰 13 宜宾地震 14 可以抵抗空虚的低成本爱好 15 认知差 人生真正的分水岭 16 八仙 17 北京地铁14号线 18 河南将严惩三支一扶作弊人员 19 PCB大牛股业绩暴涨1562% 20 C罗乔治娜婚礼更多细节曝光 21 赛里木湖多名负责人被免职处分 22 左奇函 陈奕恒 23 刚出生宝宝不哭反而开心大笑 24 弟媳照顾去世大哥5个孩子5年 25 迪丽热巴和御木本总裁的合照 26 Lucas19岁了 27 人扮美人鱼把鱼看懵了 28 窦文涛问鲁豫是否抑郁 29 王垲智齐昕苒感情时 ...

微博 排名 话题 1 笔试第一称被第二名花钱劝弃考 2 泰航拒绝20多名中国乘客登机 3 6图看上半年经济发展动能向新 4 天才女友语文成绩148 5 抱冬瓜睡觉的第一批受害者出现了 6 原配称不会再就婚姻家事细节公开回应 7 台风白海豚 8 鸿蒙智行回应竹知了事件 9 王一博只有一部待播作品了 10 商竣程2比1卢布列夫 11 港媒报道李荣浩中回力镖 12 短剧 降薪 13 王腾的小米SU7Ultra泊车辅助遇bug 14 王一博工作室真的很爱王一博 15 虞书欣赛场贝微微既视感 16 怪不得有人看病能在诊室待很久 17 张凌赫小学证件照 18 医生漏诊获刑 19 重案六组消失的警号 20 心理学上有个词叫螃蟹效应 21 35岁女子1天5瓶可乐喝成60岁脸 22 死刑改死缓被害人家属十年后才知晓 23 浙江宣传谈竹知了事件 24 张凌赫素人时期依旧很权威 25 张帅晋级多伦多赛32强 26 湛江教育局回应考生被传话劝退 27 国内机票燃油费再下调 28 终于知道为 ...

微博 排名 话题 1 中传取消艺考意味着什么 2 姆巴佩官宣恋情 3 预拨3.3亿元支持8省市抢险救灾 4 梅姨真实姓名首曝光 5 李亚鹏向地铁吐血女孩捐99999元 6 老祖宗怎么可以聪明成这样 7 河南带薪休假新政为何引全网关注 8 这一秒过火没破万 9 时代少年团退场魔术失败了 10 余承东曾称扛不住了也可能涨价 11 王源易烊千玺换蛋糕不给王俊凯说 12 TFBOYS换置顶 13 被梅姨拐卖者称已被养父母删除 14 台风白海豚大概率登陆浙江 15 胚胎案妻子称早已没什么可失去 16 黎彼得去世 17 最近你买黄金了没 18 美国禁止进口中国机器人 19 刘宇宁知道自己是大男主的反应 20 中国人心目中天庭的样子在外网爆火 21 TFBOYS这组头像用了九年了 22 侯卓成 王垲智 23 日本女网红自杀过程被完整直播 24 养大后才发现买了头牛魔王 25 孩子吵闹2小时乘客叹气反被家长怼 26 福州煎饼叔叔去世 27 侯卓成回应聊天记录 28 中国名将樊振东的 ...