2026.10.12 | 扩散模型RL微调高效稳定;多模态模型空间工作区

2026.10.12 | 扩散模型RL微调高效稳定;多模态模型空间工作区

12分钟 ·
播放数59
·
评论数0

【赞助商】
OpenClaw快报
每天五分钟,听听 OpenClaw 快报,带你了解最新动态和业内讨论
传送门 www.xiaoyuzhoufm.com

【目录】
本期的 15 篇论文如下:

00:28 ⚡ DiffusionART: Efficient, Stable and Scalable Reinforcement of Diffusion Models(DiffusionART:高效、稳定且可扩展的扩散模型强化学习)
01:13 🧭 S-Space: Exploring Spatial Workspace in Multimodal Language Models(S-Space:探索多模态语言模型中的空间工作区)
01:58 🤖 Reflex: Visually Grounded Reactive Humanoid Control(Reflex:视觉接地的反应式人形机器人控制)
02:48 🖥 SoftwareGym: Scaling Computer-Use Agent Training with Diverse Synthesized Software Environments(SoftwareGym:利用多样化合成软件环境扩展计算机使用智能体训练)
03:30 🤖 EmbodiedCLBench: Evaluating Continual Learning for Self-Evolving Embodied Agents(EmbodiedCLBench:评估自进化具身智能体的持续学习)
04:18 🎧 Hear the World in Stereo: Learning Dynamic Spatial Correspondence for Immersive Joint Video-Audio Generation(用立体声聆听世界:学习动态空间对应以实现沉浸式音视频联合生成)
04:58 🧶 VisionWeave: Weaving Elastic Visual Representations as a Native Capability of MLLMs(VisionWeave:将弹性视觉表征编织为多模态大语言模型的原生能力)
05:41 🤖 ProgramGym: Scaling Training Environments for Long-Horizon Program Reproduction Agents(ProgramGym:扩展长时程程序复现智能体的训练环境)
06:29 🖥 Work2Skill: Capturing Human GUI Experience for Reusable Agent Skill Induction(Work2Skill:捕获人类 GUI 经验以实现可复用智能体技能归纳)
07:05 🎬 SepGen: Multi-Stem Audio-Video Separation and Generation in a Single Model(SepGen:在单一模型中实现多音轨音视频分离与生成)
07:48 🔄 Continually Post-training Small Models via Synthetic Self-distillation(通过合成自蒸馏持续后训练小型模型)
08:33 🎮 Can AI Agents Learn Their Way to the Top? Evaluating Heuristic Learning in a Long-Running Game Agent Competition(AI 智能体能否通过学习登顶?评估长期运行的游戏智能体竞赛中的启发式学习)
09:19 ⚡ SoloQ: Calibration-Free Quantization for Diffusion Language Models(SoloQ:面向扩散语言模型的无校准量化)
10:05 🔍 PolyOCR-Venus: Unified OCR Foundation Models for Text-Centric Visual Intelligence(PolyOCR-Venus:面向以文本为中心的视觉智能的统一OCR基础模型)
10:45 🤖 RSI-Anything: Unlocking Recursive Self-Improvement with World Models(RSI-Anything:用世界模型解锁递归自我改进)

【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