H2025.10.01 | 自对弈零标注训练;MCP代理深度评测

2025.10.01 | 自对弈零标注训练;MCP代理深度评测

HuggingFace 每日AI论文速递·2025 年 10 月 2 日·11 分钟
在小宇宙播放
创作者写的节目说明

本期的 15 篇论文如下:

[] 🎮 Vision-Zero: Scalable VLM Self-Improvement via Strategic Gamified Self-Play(Vision-Zero:基于策略化博弈自对弈的可扩展视觉语言模型自我提升)

[] 🔥 MCPMark: A Benchmark for Stress-Testing Realistic and Comprehensive MCP Use(MCPMark:面向真实且全面的MCP应用场景的压力测试基准)

[] 🐣 The Dragon Hatchling: The Missing Link between the Transformer and Models of the Brain(幼龙破壳: Transformer 与大脑模型之间缺失的环节)

[] 🤥 TruthRL: Incentivizing Truthful LLMs via Reinforcement Learning(TruthRL:通过强化学习激励大模型说真话)

[] 🌊 OceanGym: A Benchmark Environment for Underwater Embodied Agents(OceanGym:面向水下具身智能体的综合基准环境)

[] ⚡ DC-VideoGen: Efficient Video Generation with Deep Compression Video Autoencoder(DC-VideoGen:基于深度压缩视频自编码器的高效视频生成)

[] 🔍 Who's Your Judge? On the Detectability of LLM-Generated Judgments(谁是你的评审?大模型生成评审意见的检测性研究)

[] ✂ Winning the Pruning Gamble: A Unified Approach to Joint Sample and Token Pruning for Efficient Supervised Fine-Tuning(赢得剪枝豪赌:统一样本-令牌剪枝的高效监督微调新方法)

[] 👁 Learning to See Before Seeing: Demystifying LLM Visual Priors from Language Pre-training(未见先识:从语言预训练解密大模型视觉先验)

[] 🧠 Thinking Sparks!: Emergent Attention Heads in Reasoning Models During Post Training(思维火花!后训练阶段推理模型中涌现的专用注意力头)

[] 🧪 VitaBench: Benchmarking LLM Agents with Versatile Interactive Tasks in Real-world Applications(VitaBench:面向真实场景多功能交互任务的LLM智能体评测基准)

[] ⚡ dParallel: Learnable Parallel Decoding for dLLMs(dParallel:面向扩散大语言模型的可学习并行解码)

[] 🎯 IMG: Calibrating Diffusion Models via Implicit Multimodal Guidance(IMG:通过隐式多模态引导校准扩散模型)

[] 🎬 MotionRAG: Motion Retrieval-Augmented Image-to-Video Generation(MotionRAG:基于运动检索增强的图像到视频生成)

[] 🐬 Efficient Audio-Visual Speech Separation with Discrete Lip Semantics and Multi-Scale Global-Local Attention(基于离散唇部语义与多尺度全局-局部注意力的高效视听语音分离)

【关注我们】

您还可以在以下平台找到我们,获得播客内容以外更多信息

小红书: AI速递


在小宇宙查看该单集文稿

说明里标出的时间点,点一下跳过去

HuggingFace 每日AI论文速递 · 更多单集

看全部 739 期 →

HuggingFace 每日AI论文速递

H2025.10.01 | 自对弈零标注训练;MCP代理深度评测

HuggingFace 每日AI论文速递 · 2025 年 10 月 2 日 · 11 分钟

2025.10.01 | 自对弈零标注训练;MCP代理深度评测

在小宇宙播放
创作者写的节目说明

本期的 15 篇论文如下:

[] 🎮 Vision-Zero: Scalable VLM Self-Improvement via Strategic Gamified Self-Play(Vision-Zero:基于策略化博弈自对弈的可扩展视觉语言模型自我提升)

[] 🔥 MCPMark: A Benchmark for Stress-Testing Realistic and Comprehensive MCP Use(MCPMark:面向真实且全面的MCP应用场景的压力测试基准)

[] 🐣 The Dragon Hatchling: The Missing Link between the Transformer and Models of the Brain(幼龙破壳: Transformer 与大脑模型之间缺失的环节)

[] 🤥 TruthRL: Incentivizing Truthful LLMs via Reinforcement Learning(TruthRL:通过强化学习激励大模型说真话)

[] 🌊 OceanGym: A Benchmark Environment for Underwater Embodied Agents(OceanGym:面向水下具身智能体的综合基准环境)

[] ⚡ DC-VideoGen: Efficient Video Generation with Deep Compression Video Autoencoder(DC-VideoGen:基于深度压缩视频自编码器的高效视频生成)

[] 🔍 Who's Your Judge? On the Detectability of LLM-Generated Judgments(谁是你的评审?大模型生成评审意见的检测性研究)

[] ✂ Winning the Pruning Gamble: A Unified Approach to Joint Sample and Token Pruning for Efficient Supervised Fine-Tuning(赢得剪枝豪赌:统一样本-令牌剪枝的高效监督微调新方法)

[] 👁 Learning to See Before Seeing: Demystifying LLM Visual Priors from Language Pre-training(未见先识:从语言预训练解密大模型视觉先验)

[] 🧠 Thinking Sparks!: Emergent Attention Heads in Reasoning Models During Post Training(思维火花!后训练阶段推理模型中涌现的专用注意力头)

[] 🧪 VitaBench: Benchmarking LLM Agents with Versatile Interactive Tasks in Real-world Applications(VitaBench:面向真实场景多功能交互任务的LLM智能体评测基准)

[] ⚡ dParallel: Learnable Parallel Decoding for dLLMs(dParallel:面向扩散大语言模型的可学习并行解码)

[] 🎯 IMG: Calibrating Diffusion Models via Implicit Multimodal Guidance(IMG:通过隐式多模态引导校准扩散模型)

[] 🎬 MotionRAG: Motion Retrieval-Augmented Image-to-Video Generation(MotionRAG:基于运动检索增强的图像到视频生成)

[] 🐬 Efficient Audio-Visual Speech Separation with Discrete Lip Semantics and Multi-Scale Global-Local Attention(基于离散唇部语义与多尺度全局-局部注意力的高效视听语音分离)

【关注我们】

您还可以在以下平台找到我们,获得播客内容以外更多信息

小红书: AI速递


在小宇宙查看该单集文稿

说明里标出的时间点,点一下跳过去

HHuggingFace 每日AI论文速递

HuggingFace 每日AI论文速递

面向中文AI从业者与爱好者的每日AI论文速递,工作日高密度速览前沿论文,月末以月度热门论文特辑补充,形成每日速递加月末特…

已收录 739 期 · 最近更新 11 月 10 日

在哪里听

https://feed.xyzfm.space/7jtt3j3lbbra复制

内容版权归创作者,声湃按创作者开放的 RSS 收录。了解收录与 RSS →

更多单集 全部 739 期 →