Shows
Daily Paper CastGARDO: Reinforcing Diffusion Models without Reward Hacking
🤗 Upvotes: 23 | cs.LG, cs.AI, cs.CV
Authors:
Haoran He, Yuxiao Ye, Jie Liu, Jiajun Liang, Zhiyong Wang, Ziyang Yuan, Xintao Wang, Hangyu Mao, Pengfei Wan, Ling Pan
Title:
GARDO: Reinforcing Diffusion Models without Reward Hacking
Arxiv:
http://arxiv.org/abs/2512.24138v1
Abstract:
Fine-tuning diffusion models via online reinforcement learning (RL) has shown great potential for enhancing text-to-image alignment. However, since precisely specifying a ground-truth objective for visual tasks remains challenging, the models are often optimized using a proxy reward that only partially captures the true goal. Thi...
2026-01-0724 min
EKL Battery Brew极速充电（XFC）- 锂离子电池的技术挑战与突破路径| Battery Brew 7本期主题：极速充电（XFC）电池——15分钟充满80%的科学奥秘节目概述本期节目深入探讨**极速充电（Extreme Fast Charging, XFC）锂离子电池的技术挑战与突破路径。XFC目标是15分钟内充满80%电量，同时保持高能量密度（>240 Wh/kg）和长循环寿命（>2000次），这对电动汽车、无人机和eVTOL（电动垂直起降飞行器）等“低空经济”至关重要。内容基于2025年发表在《Chemical Reviews》上的权威综述： Extremely Fast-Charging Batteries: Principle, Strategies, Detection, and Prediction 作者：Hao Liu, Liyuan Zhao, Yusheng Ye*, Xintao Yang, Yongxin Zhang, Qianya Li, Ruixing Li, Han Liu, Biao Huang, Feng Wu, Renjie Chen*, Li Li*（北京理工大学团队） DOI: 10.1021/acs.chemrev.5c00203（Chemical Reviews, 2025, 125(20), 9553–9678）综述系统剖析XFC的瓶颈、锂离子传输“六步走”机制、多尺度提速策略、失效检测以及AI预测模型，指引从实验室到商业化的系统工程。感谢收听！更多电池前沿论文&解读，探索X账号：@[EKL_Batteries](x.com)
2025-12-2821 minDaily Paper CastSemanticGen: Video Generation in Semantic Space
🤗 Upvotes: 78 | cs.CV
Authors:
Jianhong Bai, Xiaoshi Wu, Xintao Wang, Xiao Fu, Yuanxing Zhang, Qinghe Wang, Xiaoyu Shi, Menghan Xia, Zuozhu Liu, Haoji Hu, Pengfei Wan, Kun Gai
Title:
SemanticGen: Video Generation in Semantic Space
Arxiv:
http://arxiv.org/abs/2512.20619v2
Abstract:
State-of-the-art video generative models typically learn the distribution of video latents in the VAE space and map them to pixels using a VAE decoder. While this approach can generate high-quality videos, it suffers from slow convergence and is computationally expensive when generating long videos. In thi...
2025-12-2522 minDaily Paper CastKling-Omni Technical Report
🤗 Upvotes: 112 | cs.CV
Authors:
Kling Team, Jialu Chen, Yuanzheng Ci, Xiangyu Du, Zipeng Feng, Kun Gai, Sainan Guo, Feng Han, Jingbin He, Kang He, Xiao Hu, Xiaohua Hu, Boyuan Jiang, Fangyuan Kong, Hang Li, Jie Li, Qingyu Li, Shen Li, Xiaohan Li, Yan Li, Jiajun Liang, Borui Liao, Yiqiao Liao, Weihong Lin, Quande Liu, Xiaokun Liu, Yilun Liu, Yuliang Liu, Shun Lu, Hangyu Mao, Yunyao Mao, Haodong Ouyang, Wenyu Qin, Wanqi Shi, Xiaoyu Shi, Lianghao Su, Haozhi Sun, Peiqin Sun, Pengfei Wan, Chao Wang, Chenyu Wang, Meng Wang, Qiulin Wang, Runqi Wang, Xintao Wang, Xuebo Wang, Zek...
2025-12-2024 minDaily Paper CastSVG-T2I: Scaling Up Text-to-Image Latent Diffusion Model Without Variational Autoencoder
🤗 Upvotes: 33 | cs.CV
Authors:
Minglei Shi, Haolin Wang, Borui Zhang, Wenzhao Zheng, Bohan Zeng, Ziyang Yuan, Xiaoshi Wu, Yuanxing Zhang, Huan Yang, Xintao Wang, Pengfei Wan, Kun Gai, Jie Zhou, Jiwen Lu
Title:
SVG-T2I: Scaling Up Text-to-Image Latent Diffusion Model Without Variational Autoencoder
Arxiv:
http://arxiv.org/abs/2512.11749v1
Abstract:
Visual generation grounded in Visual Foundation Model (VFM) representations offers a highly promising unified pathway for integrating visual understanding, perception, and generation. Despite this potential, training large-scale text-to-image diffusion models entirely within the VFM representation space rem...
2025-12-1622 minDaily Paper CastMultiShotMaster: A Controllable Multi-Shot Video Generation Framework
🤗 Upvotes: 50 | cs.CV
Authors:
Qinghe Wang, Xiaoyu Shi, Baolu Li, Weikang Bian, Quande Liu, Huchuan Lu, Xintao Wang, Pengfei Wan, Kun Gai, Xu Jia
Title:
MultiShotMaster: A Controllable Multi-Shot Video Generation Framework
Arxiv:
http://arxiv.org/abs/2512.03041v1
Abstract:
Current video generation techniques excel at single-shot clips but struggle to produce narrative multi-shot videos, which require flexible shot arrangement, coherent narrative, and controllability beyond text prompts. To tackle these challenges, we propose MultiShotMaster, a framework for highly controllable multi-shot video generation. We extend a pretrained single-shot model by...
2025-12-0427 min
加尔讲英语【外刊英语精读】中国露营潮 | camping trendCamping trend goes wild, with room to grow露营风起，发展空间广阔Nature experiences receive a boost with enterprises and sites offering various options商家与场地提供多样化选择，自然体验进一步升级Polychromatic tents of various sizes and shapes resembling a mosaic have added vibrancy to forests, lakesides and beaches across the country's vast landscape since early summer.初夏以来，大小不一、形状各异、犹如马赛克的多色帐篷为全国广阔的森林、湖畔和海滩增添了活力。It took Wu Xintao a while before he found an ideal vacant spot to set up camp at Xiangshanhu Park in Nanjing, East China's Jiangsu province, for his first camping trip in June. The park's grasslands have seen an increasing number of campers spread out equipment and lay back to bask in the sunny breeze while chatting with friends and families.花了好一段时间，吴新涛（音译）才在中国东部省份江苏南京的香山湖公园找到一个理...
2025-11-1922 minDaily Paper CastLatent Diffusion Model without Variational Autoencoder
🤗 Upvotes: 30 | cs.CV, cs.AI
Authors:
Minglei Shi, Haolin Wang, Wenzhao Zheng, Ziyang Yuan, Xiaoshi Wu, Xintao Wang, Pengfei Wan, Jie Zhou, Jiwen Lu
Title:
Latent Diffusion Model without Variational Autoencoder
Arxiv:
http://arxiv.org/abs/2510.15301v2
Abstract:
Recent progress in diffusion-based visual generation has largely relied on latent diffusion models with variational autoencoders (VAEs). While effective for high-fidelity synthesis, this VAE+diffusion paradigm suffers from limited training efficiency, slow inference, and poor transferability to broader vision tasks. These issues stem from a key limitation of VAE lat...
2025-10-2125 minDaily Paper CastUniVideo: Unified Understanding, Generation, and Editing for Videos
🤗 Upvotes: 47 | cs.CV
Authors:
Cong Wei, Quande Liu, Zixuan Ye, Qiulin Wang, Xintao Wang, Pengfei Wan, Kun Gai, Wenhu Chen
Title:
UniVideo: Unified Understanding, Generation, and Editing for Videos
Arxiv:
http://arxiv.org/abs/2510.08377v1
Abstract:
Unified multimodal models have shown promising results in multimodal content generation and editing but remain largely limited to the image domain. In this work, we present UniVideo, a versatile framework that extends unified modeling to the video domain. UniVideo adopts a dual-stream design, combining a Multimodal Large Language Model (MLLM) for ins...
2025-10-1126 minDaily Paper CastVideoCanvas: Unified Video Completion from Arbitrary Spatiotemporal Patches via In-Context Conditioning
🤗 Upvotes: 36 | cs.CV
Authors:
Minghong Cai, Qiulin Wang, Zongli Ye, Wenze Liu, Quande Liu, Weicai Ye, Xintao Wang, Pengfei Wan, Kun Gai, Xiangyu Yue
Title:
VideoCanvas: Unified Video Completion from Arbitrary Spatiotemporal Patches via In-Context Conditioning
Arxiv:
http://arxiv.org/abs/2510.08555v1
Abstract:
We introduce the task of arbitrary spatio-temporal video completion, where a video is generated from arbitrary, user-specified patches placed at any spatial location and timestamp, akin to painting on a video canvas. This flexible formulation naturally unifies many existing controllable video generation tasks--including first-frame ima...
2025-10-1124 minDaily Paper CastHarnessing Uncertainty: Entropy-Modulated Policy Gradients for Long-Horizon LLM Agents
🤗 Upvotes: 34 | cs.LG, cs.CL
Authors:
Jiawei Wang, Jiacai Liu, Yuqian Fu, Yingru Li, Xintao Wang, Yuan Lin, Yu Yue, Lin Zhang, Yang Wang, Ke Wang
Title:
Harnessing Uncertainty: Entropy-Modulated Policy Gradients for Long-Horizon LLM Agents
Arxiv:
http://arxiv.org/abs/2509.09265v1
Abstract:
In long-horizon tasks, recent agents based on Large Language Models (LLMs) face a significant challenge that sparse, outcome-based rewards make it difficult to assign credit to intermediate steps. Previous methods mainly focus on creating dense reward signals to guide learning, either through traditional reinforcement lea...
2025-09-1320 minDaily Paper CastARIA: Training Language Agents with Intention-Driven Reward Aggregation
🤗 Upvotes: 26 | cs.CL
Authors:
Ruihan Yang, Yikai Zhang, Aili Chen, Xintao Wang, Siyu Yuan, Jiangjie Chen, Deqing Yang, Yanghua Xiao
Title:
ARIA: Training Language Agents with Intention-Driven Reward Aggregation
Arxiv:
http://arxiv.org/abs/2506.00539v1
Abstract:
Large language models (LLMs) have enabled agents to perform complex reasoning and decision-making through free-form language interactions. However, in open-ended language action environments (e.g., negotiation or question-asking games), the action space can be formulated as a joint distribution over tokens, resulting in an exponentially large action space. Sampling actions in suc...
2025-06-0423 minDaily Paper CastScaling Image and Video Generation via Test-Time Evolutionary Search
🤗 Upvotes: 33 | cs.CV, cs.AI, cs.LG
Authors:
Haoran He, Jiajun Liang, Xintao Wang, Pengfei Wan, Di Zhang, Kun Gai, Ling Pan
Title:
Scaling Image and Video Generation via Test-Time Evolutionary Search
Arxiv:
http://arxiv.org/abs/2505.17618v1
Abstract:
As the marginal cost of scaling computation (data and parameters) during model pre-training continues to increase substantially, test-time scaling (TTS) has emerged as a promising direction for improving generative model performance by allocating additional computation at inference time. While TTS has demonstrated significant success across multiple language tasks, the...
2025-05-2724 minDaily Paper CastFlow-GRPO: Training Flow Matching Models via Online RL
🤗 Upvotes: 36 | cs.CV, cs.AI
Authors:
Jie Liu, Gongye Liu, Jiajun Liang, Yangguang Li, Jiaheng Liu, Xintao Wang, Pengfei Wan, Di Zhang, Wanli Ouyang
Title:
Flow-GRPO: Training Flow Matching Models via Online RL
Arxiv:
http://arxiv.org/abs/2505.05470v1
Abstract:
We propose Flow-GRPO, the first method integrating online reinforcement learning (RL) into flow matching models. Our approach uses two key strategies: (1) an ODE-to-SDE conversion that transforms a deterministic Ordinary Differential Equation (ODE) into an equivalent Stochastic Differential Equation (SDE) that matches the original model's marginal distribution at all...
2025-05-1023 minDaily Paper CastA Survey of Interactive Generative Video
🤗 Upvotes: 31 | cs.CV
Authors:
Jiwen Yu, Yiran Qin, Haoxuan Che, Quande Liu, Xintao Wang, Pengfei Wan, Di Zhang, Kun Gai, Hao Chen, Xihui Liu
Title:
A Survey of Interactive Generative Video
Arxiv:
http://arxiv.org/abs/2504.21853v1
Abstract:
Interactive Generative Video (IGV) has emerged as a crucial technology in response to the growing demand for high-quality, interactive video content across various domains. In this paper, we define IGV as a technology that combines generative capabilities to produce diverse high-quality video content with interactive features that enable user eng...
2025-05-0322 min
二分电台#26 大串台之一起聊聊新加坡 “捕蛇者说”的主播 laike9m 来到了新加坡！我和捕蛇者说的 xintao、laike9m 以及代码之外的 勾股 在现场录制了一起节目，一起畅聊了有关新加坡和加州的方方面面~
00:12 大串台开场：捕蛇者说 laike9m 和 xintao、代码之外 勾股、二分电台 AB
01:37 laike9m 对新加坡的第一印象 & 和加州的天气对比
06:56 本期真正的主题：对新加坡最喜欢和最不喜欢的地方
07:17 勾股：第一个喜欢的点是“轻松”，少有分心的事情（更专注）
11:42 华人社会的教育“怕输”论
13:2 AB：换个视角看看不同上升渠道的新加坡社会
16:07 勾股：对新加坡教育的双面观察（既卷，也有世界前几的人均教育资源）
20:37 Laike9m: 从同事身上了解到的加州教育情况（小朋友都在学 Python）
22:08 xintao: GovTech 公司 与 政府网站的体验讨论
25:52 Cooling Singapore 项目
28:07 美国的公共服务情况
29:47 新加坡的人口规模与填岛计划
33:12 加州的堂食分享，包括“白人饭”、墨西哥菜、中餐、越南餐等等
39:12 新加坡食材吐槽，鸡肉、猪肉、急冻海鲜、批发与零售的区别
43:24 新加坡的“食阁”分类：小贩中心 Hawker Center、开放咖啡店 Coffee Shop、狭义食阁或有空调的连锁咖啡店例如 Kopitiam 和 FoodRepublic
45:52 时代的光谱：新加坡的一些福建广东食物仿佛看到了几十年前的影子（例如红龟粿）
47:16 新加坡的多民族融合政策与大家长式的管理
51:31 新加坡最大的福利：HDB，准购要求与制度。（顺便分享最贵的 HDB 之一 Pinnacle）
57:18 新加坡的让人不喜欢的地方：第一个点就是贵（通涨、车子、啤酒、旅游、追星）
01:02:58 对比其他移民国家，新加坡拿到绿卡前的几年比较难熬（各方面成本较高）
01:04:12 除了贵，对现在想来新加坡旅游和工作的朋友，我们会劝退吗？
01:06:57 新加坡绿卡申请是个黑盒（内部政策会有很多动态性和不确定性）
01:11:07 新加坡绿卡续签更加严格
01:11:57 美国和新加坡的看病区别
01:19:02 新加坡的休闲娱乐生活：美食、徒步、健身撸铁、游泳、烧烤 party、看世界各地的新电影（基本都有中文字幕）、大量演唱会脱口秀商业比赛（亚洲和欧美来的都有）、周边旅游（海岛、雨林、爬山、日韩台）
01:29:22 在新加坡办各地签证特别方便
01:32:03 人人夸的樟宜机场
01:33:52 xintao：徒步、骑行、马拉松活动的设计
01:36:07 新加坡政府的健康生活倡导：Apps、糖分盐分控制、各类球场和基础设施、户外健身操
01:44:27 新加坡的技术 Meetup 和 Conference
01:49:07 新加坡的互联网发展和近年中国大陆人才输出有关系吗？
01:52:26 总结时间：每个主播的工作生活和本次聊天的感受
参考链接：
PISA 2022 results
Understanding Singapore Math
新加坡副总理尚达曼打脸BBC主持人四十分钟全文
李光耀观天下
新加坡饮料健康等级系统 Measures for Nutri-Grade Beverages ，另外上海也试行饮料营养分级
运动换取代金券的 App Lumihealth
Stranger Soccer
National Parks
Visa Technology Traineeship Program
串台链接：
代码之外
捕蛇者说
音频处理：
西市独柳工作室
相关信息：
公众号：Android高效开发、南瓜饼日常
二分电台官网
关于和版权信息
AB 的联系方式：
关于 AB
2024-04-151h 57
代码之外 Beyond Code大串台之一起聊聊新加坡本期由我们的固定嘉宾勾股和另外两个电台《二分电台》、《捕蛇者说》的四位主播录了一期特别的大串台节目，从衣食住行等各个方面聊了聊新加坡。其中 AB、xintao、勾股三位生活在新加坡深耕多年，另外一位 Laike9m 居住在美国加州湾区，所以部分内容也和美国加州做了一些对比。00:25 自我介绍03:00 天气开场06:57 新加坡简单轻松的感觉11:36 新加坡和加州基础教育21:56 新加坡和加州政府网站和公共服务32:59 新加坡和加州的饮食47:04 新加坡的家长式政府：HDB、民族大融合57:06 新加坡的生活成本1:06:00 新加坡的工作机会1:11:45 美国和加州的看病区别1:18:51 新加坡的休闲娱乐生活1:35:55 新加坡对健康的重视1:44:15 新加坡的技术氛围1:52:14 总结时间相关链接：- PISA 2022 results https://www.oecd.org/publication/pisa-2022-results/- Singapore Math https://en.wikipedia.org/wiki/Singapore_math- 时任新加坡副总理的尚达曼接受圣加仑研讨会 (St. Gallen Symposium) 的访谈 https://www.youtube.com/watch?v=hpwPciW74b8- 中文字幕版 https://www.bilibili.com/video/BV1ft411Z7wC/- 李光耀观天下 https://book.douban.com/subject/26413154/- Measures for Nutri-Grade Beverages https://hpb.gov.sg/healthy-living/food-beverage/nutri-grade- Lumihealth https://www.lumihealth.sg/- Stranger Soccer https://www.strangersoccer.com/- National Parks https://www.nparks.gov.sg/串台链接：- 二分电台 https://binary.2bab.me/- 捕蛇者说 https://pythonhunter.org/音频处理：- 西市独柳工作室 https://xishiduliu.com/---代码之外 Beyond Code 是一档由 GeekPlux 和 Randy 共同主持的程序员闲聊播客节目。节目地址：https://bento.me/beyondcode 我们的节目同时会发布视频版，在 YouTube 和 Bilibili 搜索「代码之外」都能找到我们。欢迎在 https://to.beyondcodefm.com/ 向我们来信，我们会在下
2024-04-151h 57
Oriente PressParole d'ordine: sicurezza e ricchezza condivisahttps://ogzero.org/tag/cina/Bri, Idailu, è ormai qualcosa di acquisito che procede per inerzia o comunque proprio il mondo meno interconnesso non ne prevede più tanto i presupposti; ma comunque è una proposta di Xi e quindi in qualche modo bisogna continuare a investire nella Nuova via della seta, come dimostrano gli accordi di Cosco di questi giorni per quote di controllo del porto di Amburgo. Sgomberiamo subito il campo degli argomenti che mandano in solluchero il mainstream poco avvezzo alla reale società cinese in cui troviamo invece immerso Gabriele Battaglia (@Chen_the_Tramp), che ci parla dal suo covo...
2022-10-2845 min
捕蛇者说Ep 35. 和 Gray 聊聊那些年遇到的神奇 Bug 如果喜欢我们的节目，欢迎通过爱发电打赏支持：https://afdian.net/@pythonhunter
最近我们和 Gray 聊了聊 Debug，会分成上下两期，这一期我们主要聊了一些遇到的神奇 bug，以及解决的思路，下一期，我们会聊 debug 的一些工具。
嘉宾
Gray
主播
Manjusaka
laike9m
laixintao
时间线
03:02 Gray 遇到的 HTTP 下载文件的 bug
10:59 Xintao 遇到的 HTTP 跳转 HTTPS 问题
15:37 Manjusaka 遇到的 Python Asyncio 的问题
24:08 Laike9m: 不要浪费太多时间在一个 bug 上，即时寻求帮助
25:20 Xintao 语雀编辑器的一个神奇的 bug
33:40 监控的 P99 毛刺现象
35:30 uwsgi 中 hping3 信号丢失的一个bug
42:30 推荐 git 的 bisect 工具
44:52 laike9m 遇到的 latency 问题
52:43 有关 Python 的 malloc 的一个问题
57:24 Golang 1.12 内存泄漏？
60:00 Python 中 re.complie cache 的行为问题
链接
git bisect
Debug 一个在 uWSGI 下使用 subprocess 卡住的问题 | 卡瓦邦噶！
爱发电上赞助
2022-01-131h 04
捕蛇者说Ep 32. 和李辉聊聊自由职业（上） 如果喜欢我们的节目，欢迎通过爱发电打赏支持：https://afdian.net/@pythonhunter
嘉宾
李辉
主播
Adam Wen
小白
laixintao
laike9m
时间线
00:01:21 自由职业经历
00:04:54 找工作的失败经历
00:08:01 编程视频课程
00:10:13 看视频学编程
00:11:31 做外包的奇葩遭遇
00:17:26 小白的职业规划
00:19:45 今天星期几 & 晚上几点睡 & 闹钟哪家强？
00:25:47 毕业后先不要立刻开始工作？
00:28:10 推荐环节
00:28:32 李辉的推荐
00:30:56 xintao 没有什么要推荐
00:31:32 Adam 的推荐
00:34:03 「阿里五型人格」（阿里巴巴的小白兔、野狗、大牛、老牛、老白兔，分别指代什么？？）
00:34:50 小白的推荐
00:36:35 laike9m 的推荐
00:37:21 嘉宾自带环节：你未来 3～5 年的阶段目标
00:38:01 李辉的阶段目标
00:40:15 laike9m 的阶段目标
00:42:16 小白的阶段目标
00:45:19 xintao 的阶段目标
00:46:17 知识管理工具
00:49:19 Adam 的阶段目标
00:52:27 结语
链接
00:28:32 娱乐至死
00:31:32 奈飞文化手册
00:31:58 No Rules Rules
00:34:50 Python 神经网络编程
00:36:35 Async Python is Not Faster
00:36:56 Ignore All Web Performance Benchmarks, Including This One
爱发电上赞助
2021-09-1153 min
2021-03-091h 15
捕蛇者说Ep 26. 和 xintao 聊聊新加坡的工作与生活 如果喜欢我们的节目，欢迎通过爱发电打赏支持：https://afdian.net/@pythonhunter
主播
Manjusaka
laike9m
laixintao
时间轴
00:02:00 为什么 xintao 会离开阿里？
00:22:43 办理新加坡签证
00:28:30 新加坡的生活成本和税收
00:29:57 在新加坡租房
00:43:20 新加坡的日常生活
00:58:17 应对诈骗
01:03:13 xintao 在 Shopee 的工作，Shopee 的公司文化
01:06:06 如何进入 Shopee 工作？
01:11:05 Manjusaka 的招人广告
链接
What is Site Reliability Engineering (SRE)?
Google December 2020 services outage
智能运维系列（一）| AIOps 的崛起与实践
关于《Fluent Python》中文版中“期物”这个翻译的讨论
组屋
我在新加坡一个月的生活费明细 - by laixintao
Join Shopee & Work with Me! - xintao 的内推链接
PyCon US 2021
爱发电上赞助
2021-03-071h 15