kunthawat/huggingface-daily-paper-feed
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

bot: update RSS feed

Browse Source
This commit is contained in:
github-actions[bot]
2025-12-21 00:02:39 +00:00
parent b583c42a93
commit 044bb0de8f
1 changed files with 37 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files

60
feed.xml
View File

@@ -7,18 +7,18 @@
<docs>http://www.rssboard.org/rss-specification</docs> <docs>http://www.rssboard.org/rss-specification</docs>
<generator>python-feedgen</generator> <generator>python-feedgen</generator>
<language>en</language> <language>en</language>
<lastBuildDate>Sat, 20 Dec 2025 00:02:40 +0000</lastBuildDate> <lastBuildDate>Sun, 21 Dec 2025 00:02:39 +0000</lastBuildDate>
<item> <item>
<title>LLaDA2.0: Scaling Up Diffusion Language Models to 100B</title> <title>LLaDA2.0: Scaling Up Diffusion Language Models to 100B</title>
<link>https://arxiv.org/abs/2512.15745</link> <link>https://arxiv.org/abs/2512.15745</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.15745.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Tiwei Bie, Maosong Cao, Kun Chen, Lun Du, Mingliang Gong, Zhuochen Gong, Yanmei Gu, Jiaqi Hu, Zenan Huang, Zhenzhong Lan, Chengxi Li, Chongxuan Li, Jianguo Li, Zehuan Li, Huabin Liu, Ling Liu, Guoshan Lu, Xiaocheng Lu, Yuxin Ma, Jianfeng Tan, Lanning Wei, Ji-Rong Wen, Yipeng Xing, Xiaolu Zhang, Junbo Zhao, Da Zheng, Jun Zhou, Junlin Zhou, Zhanchao Zhou, Liwang Zhu, Yihong Zhuang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 52&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; This paper presents LLaDA2.0 -- a tuple of discrete diffusion large language models (dLLM) scaling up to 100B total parameters through systematic conversion from auto-regressive (AR) models -- establishing a new paradigm for frontier-scale deployment. Instead of costly training from scratch, LLaDA2.0 upholds knowledge inheritance, progressive adaption and efficiency-aware design principle, and seamless converts a pre-trained AR model into dLLM with a novel 3-phase block-level WSD based training scheme: progressive increasing block-size in block diffusion (warm-up), large-scale full-sequence diffusion (stable) and reverting back to compact-size block diffusion (decay). Along with post-training alignment with SFT and DPO, we obtain LLaDA2.0-mini (16B) and LLaDA2.0-flash (100B), two instruction-tuned Mixture-of-Experts (MoE) variants optimized for practical deployment. By preserving the advantages of parallel decoding, these models deliver superior performance and efficiency at the frontier scale. Both models were open-sourced.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.15745.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Tiwei Bie, Maosong Cao, Kun Chen, Lun Du, Mingliang Gong, Zhuochen Gong, Yanmei Gu, Jiaqi Hu, Zenan Huang, Zhenzhong Lan, Chengxi Li, Chongxuan Li, Jianguo Li, Zehuan Li, Huabin Liu, Ling Liu, Guoshan Lu, Xiaocheng Lu, Yuxin Ma, Jianfeng Tan, Lanning Wei, Ji-Rong Wen, Yipeng Xing, Xiaolu Zhang, Junbo Zhao, Da Zheng, Jun Zhou, Junlin Zhou, Zhanchao Zhou, Liwang Zhu, Yihong Zhuang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 56&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; This paper presents LLaDA2.0 -- a tuple of discrete diffusion large language models (dLLM) scaling up to 100B total parameters through systematic conversion from auto-regressive (AR) models -- establishing a new paradigm for frontier-scale deployment. Instead of costly training from scratch, LLaDA2.0 upholds knowledge inheritance, progressive adaption and efficiency-aware design principle, and seamless converts a pre-trained AR model into dLLM with a novel 3-phase block-level WSD based training scheme: progressive increasing block-size in block diffusion (warm-up), large-scale full-sequence diffusion (stable) and reverting back to compact-size block diffusion (decay). Along with post-training alignment with SFT and DPO, we obtain LLaDA2.0-mini (16B) and LLaDA2.0-flash (100B), two instruction-tuned Mixture-of-Experts (MoE) variants optimized for practical deployment. By preserving the advantages of parallel decoding, these models deliver superior performance and efficiency at the frontier scale. Both models were open-sourced.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.15745</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.15745</guid>
<pubDate>Wed, 10 Dec 2025 09:26:18 +0000</pubDate> <pubDate>Wed, 10 Dec 2025 09:26:18 +0000</pubDate>
</item> </item>
<item> <item>
<title>Bidirectional Normalizing Flow: From Data to Noise and Back</title> <title>Bidirectional Normalizing Flow: From Data to Noise and Back</title>
<link>https://arxiv.org/abs/2512.10953</link> <link>https://arxiv.org/abs/2512.10953</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.10953.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Yiyang Lu, Qiao Sun, Xianbang Wang, Zhicheng Jiang, Hanhong Zhao, Kaiming He&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 1&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Normalizing Flows (NFs) have been established as a principled framework for generative modeling. Standard NFs consist of a forward process and a reverse process: the forward process maps data to noise, while the reverse process generates samples by inverting it. Typical NF forward transformations are constrained by explicit invertibility, ensuring that the reverse process can serve as their exact analytic inverse. Recent developments in TARFlow and its variants have revitalized NF methods by combining Transformers and autoregressive flows, but have also exposed causal decoding as a major bottleneck. In this work, we introduce Bidirectional Normalizing Flow (BiFlow), a framework that removes the need for an exact analytic inverse. BiFlow learns a reverse model that approximates the underlying noise-to-data inverse mapping, enabling more flexible loss functions and architectures. Experiments on ImageNet demonstrate that BiFlow, compared to its causal decoding counterpart, improves generation quality while accelerating sampling by up to two orders of magnitude. BiFlow yields state-of-the-art results among NF-based methods and competitive performance among single-evaluation ("1-NFE") methods. Following recent encouraging progress on NFs, we hope our work will draw further attention to this classical paradigm.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.10953.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Yiyang Lu, Qiao Sun, Xianbang Wang, Zhicheng Jiang, Hanhong Zhao, Kaiming He&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 2&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Normalizing Flows (NFs) have been established as a principled framework for generative modeling. Standard NFs consist of a forward process and a reverse process: the forward process maps data to noise, while the reverse process generates samples by inverting it. Typical NF forward transformations are constrained by explicit invertibility, ensuring that the reverse process can serve as their exact analytic inverse. Recent developments in TARFlow and its variants have revitalized NF methods by combining Transformers and autoregressive flows, but have also exposed causal decoding as a major bottleneck. In this work, we introduce Bidirectional Normalizing Flow (BiFlow), a framework that removes the need for an exact analytic inverse. BiFlow learns a reverse model that approximates the underlying noise-to-data inverse mapping, enabling more flexible loss functions and architectures. Experiments on ImageNet demonstrate that BiFlow, compared to its causal decoding counterpart, improves generation quality while accelerating sampling by up to two orders of magnitude. BiFlow yields state-of-the-art results among NF-based methods and competitive performance among single-evaluation ("1-NFE") methods. Following recent encouraging progress on NFs, we hope our work will draw further attention to this classical paradigm.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.10953</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.10953</guid>
<pubDate>Thu, 11 Dec 2025 18:59:55 +0000</pubDate> <pubDate>Thu, 11 Dec 2025 18:59:55 +0000</pubDate>
</item> </item>
@@ -36,10 +36,24 @@
<guid isPermaLink="false">https://arxiv.org/abs/2512.12576</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.12576</guid>
<pubDate>Sun, 14 Dec 2025 07:03:51 +0000</pubDate> <pubDate>Sun, 14 Dec 2025 07:03:51 +0000</pubDate>
</item> </item>
<item>
<title>Reasoning Within the Mind: Dynamic Multimodal Interleaving in Latent Space</title>
<link>https://arxiv.org/abs/2512.12623</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.12623.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Chengzhi Liu, Yuzhe Yang, Yue Fan, Qingyue Wei, Sheng Liu, Xin Eric Wang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 1&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent advancements in Multimodal Large Language Models (MLLMs) have significantly enhanced cross-modal understanding and reasoning by incorporating Chain-of-Thought (CoT) reasoning in the semantic space. Building upon this, recent studies extend the CoT mechanism to the visual modality, enabling models to integrate visual information during reasoning through external tools or explicit image generation. However, these methods remain dependent on explicit step-by-step reasoning, unstable perception-reasoning interaction and notable computational overhead. Inspired by human cognition, we posit that thinking unfolds not linearly but through the dynamic interleaving of reasoning and perception within the mind. Motivated by this perspective, we propose DMLR, a test-time Dynamic Multimodal Latent Reasoning framework that employs confidence-guided latent policy gradient optimization to refine latent think tokens for in-depth reasoning. Furthermore, a Dynamic Visual Injection Strategy is introduced, which retrieves the most relevant visual features at each latent think token and updates the set of best visual patches. The updated patches are then injected into latent think token to achieve dynamic visual-textual interleaving. Experiments across seven multimodal reasoning benchmarks and various model architectures demonstrate that DMLR significantly improves reasoning and perception performance while maintaining high inference efficiency.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.12623</guid>
<pubDate>Sun, 14 Dec 2025 10:07:45 +0000</pubDate>
</item>
<item>
<title>Improving Recursive Transformers with Mixture of LoRAs</title>
<link>https://arxiv.org/abs/2512.12880</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.12880.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Mohammadmahdi Nouriborji, Morteza Rohanian, Omid Rohanian&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 1&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Parameter sharing in recursive transformers reduces model size but collapses layer-wise expressivity. We propose Mixture of LoRAs (MoL), a lightweight conditional-computation mechanism that inserts Low-Rank Adaptation (LoRA) experts inside a shared feed-forward network (FFN). MoL enables token-conditional weight-space modulation of the shared FFN without untying backbone parameters, unlike prior approaches that add fixed or externally attached adapters. We pretrain a modernised recursive architecture, ModernALBERT, integrating rotary embeddings, GeGLU, FlashAttention, and a distillation-based initialisation. Across GLUE, SQuAD-v2, and BEIR, ModernALBERT (50M--120M) achieves state-of-the-art performance among compact models and surpasses larger fully parameterised baselines. We also propose an expert-merging procedure that compresses MoL into a single adapter at inference while preserving accuracy, enabling efficient deployment. Our results show that conditional weight-space modulation effectively restores the expressivity lost under aggressive parameter sharing in recursive transformers.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.12880</guid>
<pubDate>Sun, 14 Dec 2025 23:39:30 +0000</pubDate>
</item>
<item> <item>
<title>Seedance 1.5 pro: A Native Audio-Visual Joint Generation Foundation Model</title> <title>Seedance 1.5 pro: A Native Audio-Visual Joint Generation Foundation Model</title>
<link>https://arxiv.org/abs/2512.13507</link> <link>https://arxiv.org/abs/2512.13507</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.13507.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Heyi Chen, Siyan Chen, Xin Chen, Yanfei Chen, Ying Chen, Zhuo Chen, Feng Cheng, Tianheng Cheng, Xinqi Cheng, Xuyan Chi, Jian Cong, Jing Cui, Qinpeng Cui, Qide Dong, Junliang Fan, Jing Fang, Zetao Fang, Chengjian Feng, Han Feng, Mingyuan Gao, Yu Gao, Dong Guo, Qiushan Guo, Boyang Hao, Qingkai Hao, Bibo He, Qian He, Tuyen Hoang, Ruoqing Hu, Xi Hu, Weilin Huang, Zhaoyang Huang, Zhongyi Huang, Donglei Ji, Siqi Jiang, Wei Jiang, Yunpu Jiang, Zhuo Jiang, Ashley Kim, Jianan Kong, Zhichao Lai, Shanshan Lao, Yichong Leng, Ai Li, Feiya Li, Gen Li, Huixia Li, JiaShi Li, Liang Li, Ming Li, Shanshan Li, Tao Li, Xian Li, Xiaojie Li, Xiaoyang Li, Xingxing Li, Yameng Li, Yifu Li, Yiying Li, Chao Liang, Han Liang, Jianzhong Liang, Ying Liang, Zhiqiang Liang, Wang Liao, Yalin Liao, Heng Lin, Kengyu Lin, Shanchuan Lin, Xi Lin, Zhijie Lin, Feng Ling, Fangfang Liu, Gaohong Liu, Jiawei Liu, Jie Liu, Jihao Liu, Shouda Liu, Shu Liu, Sichao Liu, Songwei Liu, Xin Liu, Xue Liu, Yibo Liu, Zikun Liu, Zuxi Liu, Junlin Lyu, Lecheng Lyu, Qian Lyu, Han Mu, Xiaonan Nie, Jingzhe Ning, Xitong Pan, Yanghua Peng, Lianke Qin, Xueqiong Qu, Yuxi Ren, Kai Shen, Guang Shi, Lei Shi, Yan Song, Yinglong Song, Fan Sun, Li Sun, Renfei Sun, Yan Sun, Zeyu Sun, Wenjing Tang, Yaxue Tang, Zirui Tao, Feng Wang, Furui Wang, Jinran Wang, Junkai Wang, Ke Wang, Kexin Wang, Qingyi Wang, Rui Wang, Sen Wang, Shuai Wang, Tingru Wang, Weichen Wang, Xin Wang, Yanhui Wang, Yue Wang, Yuping Wang, Yuxuan Wang, Ziyu Wang, Guoqiang Wei, Wanru Wei, Di Wu, Guohong Wu, Hanjie Wu, Jian Wu, Jie Wu, Ruolan Wu, Xinglong Wu, Yonghui Wu, Ruiqi Xia, Liang Xiang, Fei Xiao, XueFeng Xiao, Pan Xie, Shuangyi Xie, Shuang Xu, Jinlan Xue, Shen Yan, Bangbang Yang, Ceyuan Yang, Jiaqi Yang, Runkai Yang, Tao Yang, Yang Yang, Yihang Yang, ZhiXian Yang, Ziyan Yang, Songting Yao, Yifan Yao, Zilyu Ye, Bowen Yu, Jian Yu, Chujie Yuan, Linxiao Yuan, Sichun Zeng, Weihong Zeng, Xuejiao Zeng, Yan Zeng, Chuntao Zhang, Heng Zhang, Jingjie Zhang, Kuo Zhang, Liang Zhang, Liying Zhang, Manlin Zhang, Ting Zhang, Weida Zhang, Xiaohe Zhang, Xinyan Zhang, Yan Zhang, Yuan Zhang, Zixiang Zhang, Fengxuan Zhao, Huating Zhao, Yang Zhao, Hao Zheng, Jianbin Zheng, Xiaozheng Zheng, Yangyang Zheng, Yijie Zheng, Jiexin Zhou, Jiahui Zhu, Kuan Zhu, Shenhan Zhu, Wenjia Zhu, Benhui Zou, Feilong Zuo&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 30&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent strides in video generation have paved the way for unified audio-visual generation. In this work, we present Seedance 1.5 pro, a foundational model engineered specifically for native, joint audio-video generation. Leveraging a dual-branch Diffusion Transformer architecture, the model integrates a cross-modal joint module with a specialized multi-stage data pipeline, achieving exceptional audio-visual synchronization and superior generation quality. To ensure practical utility, we implement meticulous post-training optimizations, including Supervised Fine-Tuning (SFT) on high-quality datasets and Reinforcement Learning from Human Feedback (RLHF) with multi-dimensional reward models. Furthermore, we introduce an acceleration framework that boosts inference speed by over 10X. Seedance 1.5 pro distinguishes itself through precise multilingual and dialect lip-syncing, dynamic cinematic camera control, and enhanced narrative coherence, positioning it as a robust engine for professional-grade content creation. Seedance 1.5 pro is now accessible on Volcano Engine at https://console.volcengine.com/ark/region:ark+cn-beijing/experience/vision?type=GenVideo.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.13507.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Heyi Chen, Siyan Chen, Xin Chen, Yanfei Chen, Ying Chen, Zhuo Chen, Feng Cheng, Tianheng Cheng, Xinqi Cheng, Xuyan Chi, Jian Cong, Jing Cui, Qinpeng Cui, Qide Dong, Junliang Fan, Jing Fang, Zetao Fang, Chengjian Feng, Han Feng, Mingyuan Gao, Yu Gao, Dong Guo, Qiushan Guo, Boyang Hao, Qingkai Hao, Bibo He, Qian He, Tuyen Hoang, Ruoqing Hu, Xi Hu, Weilin Huang, Zhaoyang Huang, Zhongyi Huang, Donglei Ji, Siqi Jiang, Wei Jiang, Yunpu Jiang, Zhuo Jiang, Ashley Kim, Jianan Kong, Zhichao Lai, Shanshan Lao, Yichong Leng, Ai Li, Feiya Li, Gen Li, Huixia Li, JiaShi Li, Liang Li, Ming Li, Shanshan Li, Tao Li, Xian Li, Xiaojie Li, Xiaoyang Li, Xingxing Li, Yameng Li, Yifu Li, Yiying Li, Chao Liang, Han Liang, Jianzhong Liang, Ying Liang, Zhiqiang Liang, Wang Liao, Yalin Liao, Heng Lin, Kengyu Lin, Shanchuan Lin, Xi Lin, Zhijie Lin, Feng Ling, Fangfang Liu, Gaohong Liu, Jiawei Liu, Jie Liu, Jihao Liu, Shouda Liu, Shu Liu, Sichao Liu, Songwei Liu, Xin Liu, Xue Liu, Yibo Liu, Zikun Liu, Zuxi Liu, Junlin Lyu, Lecheng Lyu, Qian Lyu, Han Mu, Xiaonan Nie, Jingzhe Ning, Xitong Pan, Yanghua Peng, Lianke Qin, Xueqiong Qu, Yuxi Ren, Kai Shen, Guang Shi, Lei Shi, Yan Song, Yinglong Song, Fan Sun, Li Sun, Renfei Sun, Yan Sun, Zeyu Sun, Wenjing Tang, Yaxue Tang, Zirui Tao, Feng Wang, Furui Wang, Jinran Wang, Junkai Wang, Ke Wang, Kexin Wang, Qingyi Wang, Rui Wang, Sen Wang, Shuai Wang, Tingru Wang, Weichen Wang, Xin Wang, Yanhui Wang, Yue Wang, Yuping Wang, Yuxuan Wang, Ziyu Wang, Guoqiang Wei, Wanru Wei, Di Wu, Guohong Wu, Hanjie Wu, Jian Wu, Jie Wu, Ruolan Wu, Xinglong Wu, Yonghui Wu, Ruiqi Xia, Liang Xiang, Fei Xiao, XueFeng Xiao, Pan Xie, Shuangyi Xie, Shuang Xu, Jinlan Xue, Shen Yan, Bangbang Yang, Ceyuan Yang, Jiaqi Yang, Runkai Yang, Tao Yang, Yang Yang, Yihang Yang, ZhiXian Yang, Ziyan Yang, Songting Yao, Yifan Yao, Zilyu Ye, Bowen Yu, Jian Yu, Chujie Yuan, Linxiao Yuan, Sichun Zeng, Weihong Zeng, Xuejiao Zeng, Yan Zeng, Chuntao Zhang, Heng Zhang, Jingjie Zhang, Kuo Zhang, Liang Zhang, Liying Zhang, Manlin Zhang, Ting Zhang, Weida Zhang, Xiaohe Zhang, Xinyan Zhang, Yan Zhang, Yuan Zhang, Zixiang Zhang, Fengxuan Zhao, Huating Zhao, Yang Zhao, Hao Zheng, Jianbin Zheng, Xiaozheng Zheng, Yangyang Zheng, Yijie Zheng, Jiexin Zhou, Jiahui Zhu, Kuan Zhu, Shenhan Zhu, Wenjia Zhu, Benhui Zou, Feilong Zuo&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 32&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent strides in video generation have paved the way for unified audio-visual generation. In this work, we present Seedance 1.5 pro, a foundational model engineered specifically for native, joint audio-video generation. Leveraging a dual-branch Diffusion Transformer architecture, the model integrates a cross-modal joint module with a specialized multi-stage data pipeline, achieving exceptional audio-visual synchronization and superior generation quality. To ensure practical utility, we implement meticulous post-training optimizations, including Supervised Fine-Tuning (SFT) on high-quality datasets and Reinforcement Learning from Human Feedback (RLHF) with multi-dimensional reward models. Furthermore, we introduce an acceleration framework that boosts inference speed by over 10X. Seedance 1.5 pro distinguishes itself through precise multilingual and dialect lip-syncing, dynamic cinematic camera control, and enhanced narrative coherence, positioning it as a robust engine for professional-grade content creation. Seedance 1.5 pro is now accessible on Volcano Engine at https://console.volcengine.com/ark/region:ark+cn-beijing/experience/vision?type=GenVideo.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.13507</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.13507</guid>
<pubDate>Mon, 15 Dec 2025 16:36:52 +0000</pubDate> <pubDate>Mon, 15 Dec 2025 16:36:52 +0000</pubDate>
</item> </item>
@@ -60,7 +74,7 @@
<item> <item>
<title>Nemotron-Math: Efficient Long-Context Distillation of Mathematical Reasoning from Multi-Mode Supervision</title> <title>Nemotron-Math: Efficient Long-Context Distillation of Mathematical Reasoning from Multi-Mode Supervision</title>
<link>https://arxiv.org/abs/2512.15489</link> <link>https://arxiv.org/abs/2512.15489</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.15489.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Wei Du, Shubham Toshniwal, Branislav Kisacanin, Sadegh Mahdavi, Ivan Moshkov, George Armstrong, Stephen Ge, Edgar Minasyan, Feng Chen, Igor Gitman&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 0&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; High-quality mathematical reasoning supervision requires diverse reasoning styles, long-form traces, and effective tool integration, capabilities that existing datasets provide only in limited form. Leveraging the multi-mode generation ability of gpt-oss-120b, we introduce Nemotron-Math, a large-scale mathematical reasoning dataset containing 7.5M solution traces across high, medium, and low reasoning modes, each available both with and without Python tool-integrated reasoning (TIR). The dataset integrates 85K curated AoPS problems with 262K community-sourced StackExchange-Math problems, combining structured competition tasks with diverse real-world mathematical queries. We conduct controlled evaluations to assess the dataset quality. Nemotron-Math consistently outperforms the original OpenMathReasoning on matched AoPS problems. Incorporating StackExchange-Math substantially improves robustness and generalization, especially on HLE-Math, while preserving accuracy on math competition benchmarks. To support efficient long-context training, we develop a sequential bucketed strategy that accelerates 128K context-length fine-tuning by 2--3times without significant accuracy loss. Overall, Nemotron-Math enables state-of-the-art performance, including 100\% maj@16 accuracy on AIME 2024 and 2025 with Python TIR.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.15489.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Wei Du, Shubham Toshniwal, Branislav Kisacanin, Sadegh Mahdavi, Ivan Moshkov, George Armstrong, Stephen Ge, Edgar Minasyan, Feng Chen, Igor Gitman&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 1&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; High-quality mathematical reasoning supervision requires diverse reasoning styles, long-form traces, and effective tool integration, capabilities that existing datasets provide only in limited form. Leveraging the multi-mode generation ability of gpt-oss-120b, we introduce Nemotron-Math, a large-scale mathematical reasoning dataset containing 7.5M solution traces across high, medium, and low reasoning modes, each available both with and without Python tool-integrated reasoning (TIR). The dataset integrates 85K curated AoPS problems with 262K community-sourced StackExchange-Math problems, combining structured competition tasks with diverse real-world mathematical queries. We conduct controlled evaluations to assess the dataset quality. Nemotron-Math consistently outperforms the original OpenMathReasoning on matched AoPS problems. Incorporating StackExchange-Math substantially improves robustness and generalization, especially on HLE-Math, while preserving accuracy on math competition benchmarks. To support efficient long-context training, we develop a sequential bucketed strategy that accelerates 128K context-length fine-tuning by 2--3times without significant accuracy loss. Overall, Nemotron-Math enables state-of-the-art performance, including 100\% maj@16 accuracy on AIME 2024 and 2025 with Python TIR.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.15489</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.15489</guid>
<pubDate>Wed, 17 Dec 2025 14:37:41 +0000</pubDate> <pubDate>Wed, 17 Dec 2025 14:37:41 +0000</pubDate>
</item> </item>
@@ -88,7 +102,7 @@
<item> <item>
<title>Adaptation of Agentic AI</title> <title>Adaptation of Agentic AI</title>
<link>https://arxiv.org/abs/2512.16301</link> <link>https://arxiv.org/abs/2512.16301</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16301.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Pengcheng Jiang, Jiacheng Lin, Zhiyi Shi, Zifeng Wang, Luxi He, Yichen Wu, Ming Zhong, Peiyang Song, Qizheng Zhang, Heng Wang, Xueqiang Xu, Hanwen Xu, Pengrui Han, Dylan Zhang, Jiashuo Sun, Chaoqi Yang, Kun Qian, Tian Wang, Changran Hu, Manling Li, Quanzheng Li, Hao Peng, Sheng Wang, Jingbo Shang, Chao Zhang, Jiaxuan You, Liyuan Liu, Pan Lu, Yu Zhang, Heng Ji, Yejin Choi, Dawn Song, Jimeng Sun, Jiawei Han&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 43&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Cutting-edge agentic AI systems are built on foundation models that can be adapted to plan, reason, and interact with external tools to perform increasingly complex and specialized tasks. As these systems grow in capability and scope, adaptation becomes a central mechanism for improving performance, reliability, and generalization. In this paper, we unify the rapidly expanding research landscape into a systematic framework that spans both agent adaptations and tool adaptations. We further decompose these into tool-execution-signaled and agent-output-signaled forms of agent adaptation, as well as agent-agnostic and agent-supervised forms of tool adaptation. We demonstrate that this framework helps clarify the design space of adaptation strategies in agentic AI, makes their trade-offs explicit, and provides practical guidance for selecting or switching among strategies during system design. We then review the representative approaches in each category, analyze their strengths and limitations, and highlight key open challenges and future opportunities. Overall, this paper aims to offer a conceptual foundation and practical roadmap for researchers and practitioners seeking to build more capable, efficient, and reliable agentic AI systems.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16301.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Pengcheng Jiang, Jiacheng Lin, Zhiyi Shi, Zifeng Wang, Luxi He, Yichen Wu, Ming Zhong, Peiyang Song, Qizheng Zhang, Heng Wang, Xueqiang Xu, Hanwen Xu, Pengrui Han, Dylan Zhang, Jiashuo Sun, Chaoqi Yang, Kun Qian, Tian Wang, Changran Hu, Manling Li, Quanzheng Li, Hao Peng, Sheng Wang, Jingbo Shang, Chao Zhang, Jiaxuan You, Liyuan Liu, Pan Lu, Yu Zhang, Heng Ji, Yejin Choi, Dawn Song, Jimeng Sun, Jiawei Han&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 71&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Cutting-edge agentic AI systems are built on foundation models that can be adapted to plan, reason, and interact with external tools to perform increasingly complex and specialized tasks. As these systems grow in capability and scope, adaptation becomes a central mechanism for improving performance, reliability, and generalization. In this paper, we unify the rapidly expanding research landscape into a systematic framework that spans both agent adaptations and tool adaptations. We further decompose these into tool-execution-signaled and agent-output-signaled forms of agent adaptation, as well as agent-agnostic and agent-supervised forms of tool adaptation. We demonstrate that this framework helps clarify the design space of adaptation strategies in agentic AI, makes their trade-offs explicit, and provides practical guidance for selecting or switching among strategies during system design. We then review the representative approaches in each category, analyze their strengths and limitations, and highlight key open challenges and future opportunities. Overall, this paper aims to offer a conceptual foundation and practical roadmap for researchers and practitioners seeking to build more capable, efficient, and reliable agentic AI systems.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16301</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16301</guid>
<pubDate>Thu, 18 Dec 2025 08:38:51 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 08:38:51 +0000</pubDate>
</item> </item>
@@ -109,42 +123,42 @@
<item> <item>
<title>N3D-VLM: Native 3D Grounding Enables Accurate Spatial Reasoning in Vision-Language Models</title> <title>N3D-VLM: Native 3D Grounding Enables Accurate Spatial Reasoning in Vision-Language Models</title>
<link>https://arxiv.org/abs/2512.16561</link> <link>https://arxiv.org/abs/2512.16561</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16561.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Yuxin Wang, Lei Ke, Boqiang Zhang, Tianyuan Qu, Hanxun Yu, Zhenpeng Huang, Meng Yu, Dan Xu, Dong Yu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 16&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; While current multimodal models can answer questions based on 2D images, they lack intrinsic 3D object perception, limiting their ability to comprehend spatial relationships and depth cues in 3D scenes. In this work, we propose N3D-VLM, a novel unified framework that seamlessly integrates native 3D object perception with 3D-aware visual reasoning, enabling both precise 3D grounding and interpretable spatial understanding. Unlike conventional end-to-end models that directly predict answers from RGB/RGB-D inputs, our approach equips the model with native 3D object perception capabilities, enabling it to directly localize objects in 3D space based on textual descriptions. Building upon accurate 3D object localization, the model further performs explicit reasoning in 3D, achieving more interpretable and structured spatial understanding. To support robust training for these capabilities, we develop a scalable data construction pipeline that leverages depth estimation to lift large-scale 2D annotations into 3D space, significantly increasing the diversity and coverage for 3D object grounding data, yielding over six times larger than the largest existing single-image 3D detection dataset. Moreover, the pipeline generates spatial question-answering datasets that target chain-of-thought (CoT) reasoning in 3D, facilitating joint training for both 3D object localization and 3D spatial reasoning. Experimental results demonstrate that our unified framework not only achieves state-of-the-art performance on 3D grounding tasks, but also consistently surpasses existing methods in 3D spatial reasoning in vision-language model.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16561.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Yuxin Wang, Lei Ke, Boqiang Zhang, Tianyuan Qu, Hanxun Yu, Zhenpeng Huang, Meng Yu, Dan Xu, Dong Yu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 17&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; While current multimodal models can answer questions based on 2D images, they lack intrinsic 3D object perception, limiting their ability to comprehend spatial relationships and depth cues in 3D scenes. In this work, we propose N3D-VLM, a novel unified framework that seamlessly integrates native 3D object perception with 3D-aware visual reasoning, enabling both precise 3D grounding and interpretable spatial understanding. Unlike conventional end-to-end models that directly predict answers from RGB/RGB-D inputs, our approach equips the model with native 3D object perception capabilities, enabling it to directly localize objects in 3D space based on textual descriptions. Building upon accurate 3D object localization, the model further performs explicit reasoning in 3D, achieving more interpretable and structured spatial understanding. To support robust training for these capabilities, we develop a scalable data construction pipeline that leverages depth estimation to lift large-scale 2D annotations into 3D space, significantly increasing the diversity and coverage for 3D object grounding data, yielding over six times larger than the largest existing single-image 3D detection dataset. Moreover, the pipeline generates spatial question-answering datasets that target chain-of-thought (CoT) reasoning in 3D, facilitating joint training for both 3D object localization and 3D spatial reasoning. Experimental results demonstrate that our unified framework not only achieves state-of-the-art performance on 3D grounding tasks, but also consistently surpasses existing methods in 3D spatial reasoning in vision-language model.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16561</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16561</guid>
<pubDate>Thu, 18 Dec 2025 14:03:44 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 14:03:44 +0000</pubDate>
</item> </item>
<item> <item>
<title>Trainable Log-linear Sparse Attention for Efficient Diffusion Transformers</title> <title>Trainable Log-linear Sparse Attention for Efficient Diffusion Transformers</title>
<link>https://arxiv.org/abs/2512.16615</link> <link>https://arxiv.org/abs/2512.16615</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16615.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Yifan Zhou, Zeqi Xiao, Tianyi Wei, Shuai Yang, Xingang Pan&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 2&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Diffusion Transformers (DiTs) set the state of the art in visual generation, yet their quadratic self-attention cost fundamentally limits scaling to long token sequences. Recent Top-K sparse attention approaches reduce the computation of DiTs by compressing tokens into block-wise representation and selecting a small set of relevant key blocks, but still suffer from (i) quadratic selection cost on compressed tokens and (ii) increasing K required to maintain model quality as sequences grow. We identify that their inefficiency is due to the single-level design, as a single coarse level is insufficient to represent the global structure. In this paper, we introduce Log-linear Sparse Attention (LLSA), a trainable sparse attention mechanism for extremely long token sequences that reduces both selection and attention costs from quadratic to log-linear complexity by utilizing a hierarchical structure. LLSA performs hierarchical Top-K selection, progressively adopting sparse Top-K selection with the indices found at the previous level, and introduces a Hierarchical KV Enrichment mechanism that preserves global context while using fewer tokens of different granularity during attention computation. To support efficient training, we develop a high-performance GPU implementation that uses only sparse indices for both the forward and backward passes, eliminating the need for dense attention masks. We evaluate LLSA on high-resolution pixel-space image generation without using patchification and VAE encoding. LLSA accelerates attention inference by 28.27x and DiT training by 6.09x on 256x256 pixel token sequences, while maintaining generation quality. The results demonstrate that LLSA offers a promising direction for training long-sequence DiTs efficiently. Code is available at: https://github.com/SingleZombie/LLSA&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16615.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Yifan Zhou, Zeqi Xiao, Tianyi Wei, Shuai Yang, Xingang Pan&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 3&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Diffusion Transformers (DiTs) set the state of the art in visual generation, yet their quadratic self-attention cost fundamentally limits scaling to long token sequences. Recent Top-K sparse attention approaches reduce the computation of DiTs by compressing tokens into block-wise representation and selecting a small set of relevant key blocks, but still suffer from (i) quadratic selection cost on compressed tokens and (ii) increasing K required to maintain model quality as sequences grow. We identify that their inefficiency is due to the single-level design, as a single coarse level is insufficient to represent the global structure. In this paper, we introduce Log-linear Sparse Attention (LLSA), a trainable sparse attention mechanism for extremely long token sequences that reduces both selection and attention costs from quadratic to log-linear complexity by utilizing a hierarchical structure. LLSA performs hierarchical Top-K selection, progressively adopting sparse Top-K selection with the indices found at the previous level, and introduces a Hierarchical KV Enrichment mechanism that preserves global context while using fewer tokens of different granularity during attention computation. To support efficient training, we develop a high-performance GPU implementation that uses only sparse indices for both the forward and backward passes, eliminating the need for dense attention masks. We evaluate LLSA on high-resolution pixel-space image generation without using patchification and VAE encoding. LLSA accelerates attention inference by 28.27x and DiT training by 6.09x on 256x256 pixel token sequences, while maintaining generation quality. The results demonstrate that LLSA offers a promising direction for training long-sequence DiTs efficiently. Code is available at: https://github.com/SingleZombie/LLSA&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16615</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16615</guid>
<pubDate>Thu, 18 Dec 2025 14:53:12 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 14:53:12 +0000</pubDate>
</item> </item>
<item> <item>
<title>DeContext as Defense: Safe Image Editing in Diffusion Transformers</title> <title>DeContext as Defense: Safe Image Editing in Diffusion Transformers</title>
<link>https://arxiv.org/abs/2512.16625</link> <link>https://arxiv.org/abs/2512.16625</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16625.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Linghui Shen, Mingyue Cui, Xingyi Yang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 21&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; In-context diffusion models allow users to modify images with remarkable ease and realism. However, the same power raises serious privacy concerns: personal images can be easily manipulated for identity impersonation, misinformation, or other malicious uses, all without the owner's consent. While prior work has explored input perturbations to protect against misuse in personalized text-to-image generation, the robustness of modern, large-scale in-context DiT-based models remains largely unexamined. In this paper, we propose DeContext, a new method to safeguard input images from unauthorized in-context editing. Our key insight is that contextual information from the source image propagates to the output primarily through multimodal attention layers. By injecting small, targeted perturbations that weaken these cross-attention pathways, DeContext breaks this flow, effectively decouples the link between input and output. This simple defense is both efficient and robust. We further show that early denoising steps and specific transformer blocks dominate context propagation, which allows us to concentrate perturbations where they matter most. Experiments on Flux Kontext and Step1X-Edit show that DeContext consistently blocks unwanted image edits while preserving visual quality. These results highlight the effectiveness of attention-based perturbations as a powerful defense against image manipulation.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16625.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Linghui Shen, Mingyue Cui, Xingyi Yang&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 23&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; In-context diffusion models allow users to modify images with remarkable ease and realism. However, the same power raises serious privacy concerns: personal images can be easily manipulated for identity impersonation, misinformation, or other malicious uses, all without the owner's consent. While prior work has explored input perturbations to protect against misuse in personalized text-to-image generation, the robustness of modern, large-scale in-context DiT-based models remains largely unexamined. In this paper, we propose DeContext, a new method to safeguard input images from unauthorized in-context editing. Our key insight is that contextual information from the source image propagates to the output primarily through multimodal attention layers. By injecting small, targeted perturbations that weaken these cross-attention pathways, DeContext breaks this flow, effectively decouples the link between input and output. This simple defense is both efficient and robust. We further show that early denoising steps and specific transformer blocks dominate context propagation, which allows us to concentrate perturbations where they matter most. Experiments on Flux Kontext and Step1X-Edit show that DeContext consistently blocks unwanted image edits while preserving visual quality. These results highlight the effectiveness of attention-based perturbations as a powerful defense against image manipulation.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16625</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16625</guid>
<pubDate>Thu, 18 Dec 2025 15:01:44 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 15:01:44 +0000</pubDate>
</item> </item>
<item> <item>
<title>REGLUE Your Latents with Global and Local Semantics for Entangled Diffusion</title> <title>REGLUE Your Latents with Global and Local Semantics for Entangled Diffusion</title>
<link>https://arxiv.org/abs/2512.16636</link> <link>https://arxiv.org/abs/2512.16636</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16636.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Giorgos Petsangourakis, Christos Sgouropoulos, Bill Psomas, Theodoros Giannakopoulos, Giorgos Sfikas, Ioannis Kakogeorgiou&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 19&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Latent diffusion models (LDMs) achieve state-of-the-art image synthesis, yet their reconstruction-style denoising objective provides only indirect semantic supervision: high-level semantics emerge slowly, requiring longer training and limiting sample quality. Recent works inject semantics from Vision Foundation Models (VFMs) either externally via representation alignment or internally by jointly modeling only a narrow slice of VFM features inside the diffusion process, under-utilizing the rich, nonlinear, multi-layer spatial semantics available. We introduce REGLUE (Representation Entanglement with Global-Local Unified Encoding), a unified latent diffusion framework that jointly models (i) VAE image latents, (ii) compact local (patch-level) VFM semantics, and (iii) a global (image-level) [CLS] token within a single SiT backbone. A lightweight convolutional semantic compressor nonlinearly aggregates multi-layer VFM features into a low-dimensional, spatially structured representation, which is entangled with the VAE latents in the diffusion process. An external alignment loss further regularizes internal representations toward frozen VFM targets. On ImageNet 256x256, REGLUE consistently improves FID and accelerates convergence over SiT-B/2 and SiT-XL/2 baselines, as well as over REPA, ReDi, and REG. Extensive experiments show that (a) spatial VFM semantics are crucial, (b) non-linear compression is key to unlocking their full benefit, and (c) global tokens and external alignment act as complementary, lightweight enhancements within our global-local-latent joint modeling framework. The code is available at https://github.com/giorgospets/reglue .&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16636.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Giorgos Petsangourakis, Christos Sgouropoulos, Bill Psomas, Theodoros Giannakopoulos, Giorgos Sfikas, Ioannis Kakogeorgiou&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 21&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Latent diffusion models (LDMs) achieve state-of-the-art image synthesis, yet their reconstruction-style denoising objective provides only indirect semantic supervision: high-level semantics emerge slowly, requiring longer training and limiting sample quality. Recent works inject semantics from Vision Foundation Models (VFMs) either externally via representation alignment or internally by jointly modeling only a narrow slice of VFM features inside the diffusion process, under-utilizing the rich, nonlinear, multi-layer spatial semantics available. We introduce REGLUE (Representation Entanglement with Global-Local Unified Encoding), a unified latent diffusion framework that jointly models (i) VAE image latents, (ii) compact local (patch-level) VFM semantics, and (iii) a global (image-level) [CLS] token within a single SiT backbone. A lightweight convolutional semantic compressor nonlinearly aggregates multi-layer VFM features into a low-dimensional, spatially structured representation, which is entangled with the VAE latents in the diffusion process. An external alignment loss further regularizes internal representations toward frozen VFM targets. On ImageNet 256x256, REGLUE consistently improves FID and accelerates convergence over SiT-B/2 and SiT-XL/2 baselines, as well as over REPA, ReDi, and REG. Extensive experiments show that (a) spatial VFM semantics are crucial, (b) non-linear compression is key to unlocking their full benefit, and (c) global tokens and external alignment act as complementary, lightweight enhancements within our global-local-latent joint modeling framework. The code is available at https://github.com/giorgospets/reglue .&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16636</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16636</guid>
<pubDate>Thu, 18 Dec 2025 15:10:42 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 15:10:42 +0000</pubDate>
</item> </item>
<item> <item>
<title>JustRL: Scaling a 1.5B LLM with a Simple RL Recipe</title> <title>JustRL: Scaling a 1.5B LLM with a Simple RL Recipe</title>
<link>https://arxiv.org/abs/2512.16649</link> <link>https://arxiv.org/abs/2512.16649</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16649.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Bingxiang He, Zekai Qu, Zeyuan Liu, Yinghao Chen, Yuxin Zuo, Cheng Qian, Kaiyan Zhang, Weize Chen, Chaojun Xiao, Ganqu Cui, Ning Ding, Zhiyuan Liu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 12&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent advances in reinforcement learning for large language models have converged on increasing complexity: multi-stage training pipelines, dynamic hyperparameter schedules, and curriculum learning strategies. This raises a fundamental question: Is this complexity necessary? We present JustRL, a minimal approach using single-stage training with fixed hyperparameters that achieves state-of-the-art performance on two 1.5B reasoning models (54.9\% and 64.3\% average accuracy across nine mathematical benchmarks) while using 2times less compute than sophisticated approaches. The same hyperparameters transfer across both models without tuning, and training exhibits smooth, monotonic improvement over 4,000+ steps without the collapses or plateaus that typically motivate interventions. Critically, ablations reveal that adding ``standard tricks'' like explicit length penalties and robust verifiers may degrade performance by collapsing exploration. These results suggest that the field may be adding complexity to solve problems that disappear with a stable, scaled-up baseline. We release our models and code to establish a simple, validated baseline for the community.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16649.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Bingxiang He, Zekai Qu, Zeyuan Liu, Yinghao Chen, Yuxin Zuo, Cheng Qian, Kaiyan Zhang, Weize Chen, Chaojun Xiao, Ganqu Cui, Ning Ding, Zhiyuan Liu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 14&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent advances in reinforcement learning for large language models have converged on increasing complexity: multi-stage training pipelines, dynamic hyperparameter schedules, and curriculum learning strategies. This raises a fundamental question: Is this complexity necessary? We present JustRL, a minimal approach using single-stage training with fixed hyperparameters that achieves state-of-the-art performance on two 1.5B reasoning models (54.9\% and 64.3\% average accuracy across nine mathematical benchmarks) while using 2times less compute than sophisticated approaches. The same hyperparameters transfer across both models without tuning, and training exhibits smooth, monotonic improvement over 4,000+ steps without the collapses or plateaus that typically motivate interventions. Critically, ablations reveal that adding ``standard tricks'' like explicit length penalties and robust verifiers may degrade performance by collapsing exploration. These results suggest that the field may be adding complexity to solve problems that disappear with a stable, scaled-up baseline. We release our models and code to establish a simple, validated baseline for the community.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16649</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16649</guid>
<pubDate>Thu, 18 Dec 2025 15:21:25 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 15:21:25 +0000</pubDate>
</item> </item>
<item> <item>
<title>FrameDiffuser: G-Buffer-Conditioned Diffusion for Neural Forward Frame Rendering</title> <title>FrameDiffuser: G-Buffer-Conditioned Diffusion for Neural Forward Frame Rendering</title>
<link>https://arxiv.org/abs/2512.16670</link> <link>https://arxiv.org/abs/2512.16670</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16670.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Ole Beisswenger, Jan-Niklas Dihlmann, Hendrik P. A. Lensch&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 2&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Neural rendering for interactive applications requires translating geometric and material properties (G-buffer) to photorealistic images with realistic lighting on a frame-by-frame basis. While recent diffusion-based approaches show promise for G-buffer-conditioned image synthesis, they face critical limitations: single-image models like RGBX generate frames independently without temporal consistency, while video models like DiffusionRenderer are too computationally expensive for most consumer gaming sets ups and require complete sequences upfront, making them unsuitable for interactive applications where future frames depend on user input. We introduce FrameDiffuser, an autoregressive neural rendering framework that generates temporally consistent, photorealistic frames by conditioning on G-buffer data and the models own previous output. After an initial frame, FrameDiffuser operates purely on incoming G-buffer data, comprising geometry, materials, and surface properties, while using its previously generated frame for temporal guidance, maintaining stable, temporal consistent generation over hundreds to thousands of frames. Our dual-conditioning architecture combines ControlNet for structural guidance with ControlLoRA for temporal coherence. A three-stage training strategy enables stable autoregressive generation. We specialize our model to individual environments, prioritizing consistency and inference speed over broad generalization, demonstrating that environment-specific training achieves superior photorealistic quality with accurate lighting, shadows, and reflections compared to generalized approaches.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16670.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Ole Beisswenger, Jan-Niklas Dihlmann, Hendrik P. A. Lensch&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 3&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Neural rendering for interactive applications requires translating geometric and material properties (G-buffer) to photorealistic images with realistic lighting on a frame-by-frame basis. While recent diffusion-based approaches show promise for G-buffer-conditioned image synthesis, they face critical limitations: single-image models like RGBX generate frames independently without temporal consistency, while video models like DiffusionRenderer are too computationally expensive for most consumer gaming sets ups and require complete sequences upfront, making them unsuitable for interactive applications where future frames depend on user input. We introduce FrameDiffuser, an autoregressive neural rendering framework that generates temporally consistent, photorealistic frames by conditioning on G-buffer data and the models own previous output. After an initial frame, FrameDiffuser operates purely on incoming G-buffer data, comprising geometry, materials, and surface properties, while using its previously generated frame for temporal guidance, maintaining stable, temporal consistent generation over hundreds to thousands of frames. Our dual-conditioning architecture combines ControlNet for structural guidance with ControlLoRA for temporal coherence. A three-stage training strategy enables stable autoregressive generation. We specialize our model to individual environments, prioritizing consistency and inference speed over broad generalization, demonstrating that environment-specific training achieves superior photorealistic quality with accurate lighting, shadows, and reflections compared to generalized approaches.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16670</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16670</guid>
<pubDate>Thu, 18 Dec 2025 15:41:08 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 15:41:08 +0000</pubDate>
</item> </item>
@@ -158,14 +172,14 @@
<item> <item>
<title>Kling-Omni Technical Report</title> <title>Kling-Omni Technical Report</title>
<link>https://arxiv.org/abs/2512.16776</link> <link>https://arxiv.org/abs/2512.16776</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16776.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; 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, Zekun Wang, Min Wei, Tiancheng Wen, Guohao Wu, Xiaoshi Wu, Zhenhua Wu, Da Xie, Yingtong Xiong, Yulong Xu, Sile Yang, Zikang Yang, Weicai Ye, Ziyang Yuan, Shenglong Zhang, Shuaiyu Zhang, Yuanxing Zhang, Yufan Zhang, Wenzheng Zhao, Ruiliang Zhou, Yan Zhou, Guosheng Zhu, Yongjie Zhu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 110&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; We present Kling-Omni, a generalist generative framework designed to synthesize high-fidelity videos directly from multimodal visual language inputs. Adopting an end-to-end perspective, Kling-Omni bridges the functional separation among diverse video generation, editing, and intelligent reasoning tasks, integrating them into a holistic system. Unlike disjointed pipeline approaches, Kling-Omni supports a diverse range of user inputs, including text instructions, reference images, and video contexts, processing them into a unified multimodal representation to deliver cinematic-quality and highly-intelligent video content creation. To support these capabilities, we constructed a comprehensive data system that serves as the foundation for multimodal video creation. The framework is further empowered by efficient large-scale pre-training strategies and infrastructure optimizations for inference. Comprehensive evaluations reveal that Kling-Omni demonstrates exceptional capabilities in in-context generation, reasoning-based editing, and multimodal instruction following. Moving beyond a content creation tool, we believe Kling-Omni is a pivotal advancement toward multimodal world simulators capable of perceiving, reasoning, generating and interacting with the dynamic and complex worlds.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16776.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; 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, Zekun Wang, Min Wei, Tiancheng Wen, Guohao Wu, Xiaoshi Wu, Zhenhua Wu, Da Xie, Yingtong Xiong, Yulong Xu, Sile Yang, Zikang Yang, Weicai Ye, Ziyang Yuan, Shenglong Zhang, Shuaiyu Zhang, Yuanxing Zhang, Yufan Zhang, Wenzheng Zhao, Ruiliang Zhou, Yan Zhou, Guosheng Zhu, Yongjie Zhu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 125&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; We present Kling-Omni, a generalist generative framework designed to synthesize high-fidelity videos directly from multimodal visual language inputs. Adopting an end-to-end perspective, Kling-Omni bridges the functional separation among diverse video generation, editing, and intelligent reasoning tasks, integrating them into a holistic system. Unlike disjointed pipeline approaches, Kling-Omni supports a diverse range of user inputs, including text instructions, reference images, and video contexts, processing them into a unified multimodal representation to deliver cinematic-quality and highly-intelligent video content creation. To support these capabilities, we constructed a comprehensive data system that serves as the foundation for multimodal video creation. The framework is further empowered by efficient large-scale pre-training strategies and infrastructure optimizations for inference. Comprehensive evaluations reveal that Kling-Omni demonstrates exceptional capabilities in in-context generation, reasoning-based editing, and multimodal instruction following. Moving beyond a content creation tool, we believe Kling-Omni is a pivotal advancement toward multimodal world simulators capable of perceiving, reasoning, generating and interacting with the dynamic and complex worlds.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16776</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16776</guid>
<pubDate>Thu, 18 Dec 2025 17:08:12 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 17:08:12 +0000</pubDate>
</item> </item>
<item> <item>
<title>RePlan: Reasoning-guided Region Planning for Complex Instruction-based Image Editing</title> <title>RePlan: Reasoning-guided Region Planning for Complex Instruction-based Image Editing</title>
<link>https://arxiv.org/abs/2512.16864</link> <link>https://arxiv.org/abs/2512.16864</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16864.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Tianyuan Qu, Lei Ke, Xiaohang Zhan, Longxiang Tang, Yuqi Liu, Bohao Peng, Bei Yu, Dong Yu, Jiaya Jia&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 8&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Instruction-based image editing enables natural-language control over visual modifications, yet existing models falter under Instruction-Visual Complexity (IV-Complexity), where intricate instructions meet cluttered or ambiguous scenes. We introduce RePlan (Region-aligned Planning), a plan-then-execute framework that couples a vision-language planner with a diffusion editor. The planner decomposes instructions via step-by-step reasoning and explicitly grounds them to target regions; the editor then applies changes using a training-free attention-region injection mechanism, enabling precise, parallel multi-region edits without iterative inpainting. To strengthen planning, we apply GRPO-based reinforcement learning using 1K instruction-only examples, yielding substantial gains in reasoning fidelity and format reliability. We further present IV-Edit, a benchmark focused on fine-grained grounding and knowledge-intensive edits. Across IV-Complex settings, RePlan consistently outperforms strong baselines trained on far larger datasets, improving regional precision and overall fidelity. Our project page: https://replan-iv-edit.github.io&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16864.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Tianyuan Qu, Lei Ke, Xiaohang Zhan, Longxiang Tang, Yuqi Liu, Bohao Peng, Bei Yu, Dong Yu, Jiaya Jia&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 9&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Instruction-based image editing enables natural-language control over visual modifications, yet existing models falter under Instruction-Visual Complexity (IV-Complexity), where intricate instructions meet cluttered or ambiguous scenes. We introduce RePlan (Region-aligned Planning), a plan-then-execute framework that couples a vision-language planner with a diffusion editor. The planner decomposes instructions via step-by-step reasoning and explicitly grounds them to target regions; the editor then applies changes using a training-free attention-region injection mechanism, enabling precise, parallel multi-region edits without iterative inpainting. To strengthen planning, we apply GRPO-based reinforcement learning using 1K instruction-only examples, yielding substantial gains in reasoning fidelity and format reliability. We further present IV-Edit, a benchmark focused on fine-grained grounding and knowledge-intensive edits. Across IV-Complex settings, RePlan consistently outperforms strong baselines trained on far larger datasets, improving regional precision and overall fidelity. Our project page: https://replan-iv-edit.github.io&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16864</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16864</guid>
<pubDate>Thu, 18 Dec 2025 18:34:23 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 18:34:23 +0000</pubDate>
</item> </item>
@@ -179,14 +193,14 @@
<item> <item>
<title>FlashPortrait: 6x Faster Infinite Portrait Animation with Adaptive Latent Prediction</title> <title>FlashPortrait: 6x Faster Infinite Portrait Animation with Adaptive Latent Prediction</title>
<link>https://arxiv.org/abs/2512.16900</link> <link>https://arxiv.org/abs/2512.16900</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16900.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Shuyuan Tu, Yueming Pan, Yinming Huang, Xintong Han, Zhen Xing, Qi Dai, Kai Qiu, Chong Luo, Zuxuan Wu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 8&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Current diffusion-based acceleration methods for long-portrait animation struggle to ensure identity (ID) consistency. This paper presents FlashPortrait, an end-to-end video diffusion transformer capable of synthesizing ID-preserving, infinite-length videos while achieving up to 6x acceleration in inference speed. In particular, FlashPortrait begins by computing the identity-agnostic facial expression features with an off-the-shelf extractor. It then introduces a Normalized Facial Expression Block to align facial features with diffusion latents by normalizing them with their respective means and variances, thereby improving identity stability in facial modeling. During inference, FlashPortrait adopts a dynamic sliding-window scheme with weighted blending in overlapping areas, ensuring smooth transitions and ID consistency in long animations. In each context window, based on the latent variation rate at particular timesteps and the derivative magnitude ratio among diffusion layers, FlashPortrait utilizes higher-order latent derivatives at the current timestep to directly predict latents at future timesteps, thereby skipping several denoising steps and achieving 6x speed acceleration. Experiments on benchmarks show the effectiveness of FlashPortrait both qualitatively and quantitatively.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16900.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Shuyuan Tu, Yueming Pan, Yinming Huang, Xintong Han, Zhen Xing, Qi Dai, Kai Qiu, Chong Luo, Zuxuan Wu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 9&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Current diffusion-based acceleration methods for long-portrait animation struggle to ensure identity (ID) consistency. This paper presents FlashPortrait, an end-to-end video diffusion transformer capable of synthesizing ID-preserving, infinite-length videos while achieving up to 6x acceleration in inference speed. In particular, FlashPortrait begins by computing the identity-agnostic facial expression features with an off-the-shelf extractor. It then introduces a Normalized Facial Expression Block to align facial features with diffusion latents by normalizing them with their respective means and variances, thereby improving identity stability in facial modeling. During inference, FlashPortrait adopts a dynamic sliding-window scheme with weighted blending in overlapping areas, ensuring smooth transitions and ID consistency in long animations. In each context window, based on the latent variation rate at particular timesteps and the derivative magnitude ratio among diffusion layers, FlashPortrait utilizes higher-order latent derivatives at the current timestep to directly predict latents at future timesteps, thereby skipping several denoising steps and achieving 6x speed acceleration. Experiments on benchmarks show the effectiveness of FlashPortrait both qualitatively and quantitatively.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16900</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16900</guid>
<pubDate>Thu, 18 Dec 2025 18:56:05 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 18:56:05 +0000</pubDate>
</item> </item>
<item> <item>
<title>Alchemist: Unlocking Efficiency in Text-to-Image Model Training via Meta-Gradient Data Selection</title> <title>Alchemist: Unlocking Efficiency in Text-to-Image Model Training via Meta-Gradient Data Selection</title>
<link>https://arxiv.org/abs/2512.16905</link> <link>https://arxiv.org/abs/2512.16905</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16905.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Kaixin Ding, Yang Zhou, Xi Chen, Miao Yang, Jiarong Ou, Rui Chen, Xin Tao, Hengshuang Zhao&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 18&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent advances in Text-to-Image (T2I) generative models, such as Imagen, Stable Diffusion, and FLUX, have led to remarkable improvements in visual quality. However, their performance is fundamentally limited by the quality of training data. Web-crawled and synthetic image datasets often contain low-quality or redundant samples, which lead to degraded visual fidelity, unstable training, and inefficient computation. Hence, effective data selection is crucial for improving data efficiency. Existing approaches rely on costly manual curation or heuristic scoring based on single-dimensional features in Text-to-Image data filtering. Although meta-learning based method has been explored in LLM, there is no adaptation for image modalities. To this end, we propose **Alchemist**, a meta-gradient-based framework to select a suitable subset from large-scale text-image data pairs. Our approach automatically learns to assess the influence of each sample by iteratively optimizing the model from a data-centric perspective. Alchemist consists of two key stages: data rating and data pruning. We train a lightweight rater to estimate each sample's influence based on gradient information, enhanced with multi-granularity perception. We then use the Shift-Gsampling strategy to select informative subsets for efficient model training. Alchemist is the first automatic, scalable, meta-gradient-based data selection framework for Text-to-Image model training. Experiments on both synthetic and web-crawled datasets demonstrate that Alchemist consistently improves visual quality and downstream performance. Training on an Alchemist-selected 50% of the data can outperform training on the full dataset.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16905.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Kaixin Ding, Yang Zhou, Xi Chen, Miao Yang, Jiarong Ou, Rui Chen, Xin Tao, Hengshuang Zhao&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 21&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Recent advances in Text-to-Image (T2I) generative models, such as Imagen, Stable Diffusion, and FLUX, have led to remarkable improvements in visual quality. However, their performance is fundamentally limited by the quality of training data. Web-crawled and synthetic image datasets often contain low-quality or redundant samples, which lead to degraded visual fidelity, unstable training, and inefficient computation. Hence, effective data selection is crucial for improving data efficiency. Existing approaches rely on costly manual curation or heuristic scoring based on single-dimensional features in Text-to-Image data filtering. Although meta-learning based method has been explored in LLM, there is no adaptation for image modalities. To this end, we propose **Alchemist**, a meta-gradient-based framework to select a suitable subset from large-scale text-image data pairs. Our approach automatically learns to assess the influence of each sample by iteratively optimizing the model from a data-centric perspective. Alchemist consists of two key stages: data rating and data pruning. We train a lightweight rater to estimate each sample's influence based on gradient information, enhanced with multi-granularity perception. We then use the Shift-Gsampling strategy to select informative subsets for efficient model training. Alchemist is the first automatic, scalable, meta-gradient-based data selection framework for Text-to-Image model training. Experiments on both synthetic and web-crawled datasets demonstrate that Alchemist consistently improves visual quality and downstream performance. Training on an Alchemist-selected 50% of the data can outperform training on the full dataset.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16905</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16905</guid>
<pubDate>Thu, 18 Dec 2025 18:57:58 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 18:57:58 +0000</pubDate>
</item> </item>
@@ -200,21 +214,21 @@
<item> <item>
<title>Exploration v.s. Exploitation: Rethinking RLVR through Clipping, Entropy, and Spurious Reward</title> <title>Exploration v.s. Exploitation: Rethinking RLVR through Clipping, Entropy, and Spurious Reward</title>
<link>https://arxiv.org/abs/2512.16912</link> <link>https://arxiv.org/abs/2512.16912</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16912.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Peter Chen, Xiaopeng Li, Ziniu Li, Wotao Yin, Xi Chen, Tianyi Lin&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 8&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; This paper examines the exploration-exploitation trade-off in reinforcement learning with verifiable rewards (RLVR), a framework for improving the reasoning of Large Language Models (LLMs). Recent studies suggest that RLVR can elicit strong mathematical reasoning in LLMs through two seemingly paradoxical mechanisms: spurious rewards, which suppress exploitation by rewarding outcomes unrelated to the ground truth, and entropy minimization, which suppresses exploration by pushing the model toward more confident and deterministic outputs, highlighting a puzzling dynamic: both discouraging exploitation and discouraging exploration improve reasoning performance, yet the underlying principles that reconcile these effects remain poorly understood. We focus on two fundamental questions: (i) how policy entropy relates to performance, and (ii) whether spurious rewards yield gains, potentially through the interplay of clipping bias and model contamination. Our results show that clipping bias under spurious rewards reduces policy entropy, leading to more confident and deterministic outputs, while entropy minimization alone is insufficient for improvement. We further propose a reward-misalignment model explaining why spurious rewards can enhance performance beyond contaminated settings. Our findings clarify the mechanisms behind spurious-reward benefits and provide principles for more effective RLVR training.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16912.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Peter Chen, Xiaopeng Li, Ziniu Li, Wotao Yin, Xi Chen, Tianyi Lin&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 9&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; This paper examines the exploration-exploitation trade-off in reinforcement learning with verifiable rewards (RLVR), a framework for improving the reasoning of Large Language Models (LLMs). Recent studies suggest that RLVR can elicit strong mathematical reasoning in LLMs through two seemingly paradoxical mechanisms: spurious rewards, which suppress exploitation by rewarding outcomes unrelated to the ground truth, and entropy minimization, which suppresses exploration by pushing the model toward more confident and deterministic outputs, highlighting a puzzling dynamic: both discouraging exploitation and discouraging exploration improve reasoning performance, yet the underlying principles that reconcile these effects remain poorly understood. We focus on two fundamental questions: (i) how policy entropy relates to performance, and (ii) whether spurious rewards yield gains, potentially through the interplay of clipping bias and model contamination. Our results show that clipping bias under spurious rewards reduces policy entropy, leading to more confident and deterministic outputs, while entropy minimization alone is insufficient for improvement. We further propose a reward-misalignment model explaining why spurious rewards can enhance performance beyond contaminated settings. Our findings clarify the mechanisms behind spurious-reward benefits and provide principles for more effective RLVR training.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16912</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16912</guid>
<pubDate>Thu, 18 Dec 2025 18:59:27 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 18:59:27 +0000</pubDate>
</item> </item>
<item> <item>
<title>Depth Any Panoramas: A Foundation Model for Panoramic Depth Estimation</title> <title>Depth Any Panoramas: A Foundation Model for Panoramic Depth Estimation</title>
<link>https://arxiv.org/abs/2512.16913</link> <link>https://arxiv.org/abs/2512.16913</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16913.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Xin Lin, Meixi Song, Dizhe Zhang, Wenxuan Lu, Haodong Li, Bo Du, Ming-Hsuan Yang, Truong Nguyen, Lu Qi&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 27&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; In this work, we present a panoramic metric depth foundation model that generalizes across diverse scene distances. We explore a data-in-the-loop paradigm from the view of both data construction and framework design. We collect a large-scale dataset by combining public datasets, high-quality synthetic data from our UE5 simulator and text-to-image models, and real panoramic images from the web. To reduce domain gaps between indoor/outdoor and synthetic/real data, we introduce a three-stage pseudo-label curation pipeline to generate reliable ground truth for unlabeled images. For the model, we adopt DINOv3-Large as the backbone for its strong pre-trained generalization, and introduce a plug-and-play range mask head, sharpness-centric optimization, and geometry-centric optimization to improve robustness to varying distances and enforce geometric consistency across views. Experiments on multiple benchmarks (e.g., Stanford2D3D, Matterport3D, and Deep360) demonstrate strong performance and zero-shot generalization, with particularly robust and stable metric predictions in diverse real-world scenes. The project page can be found at: https://insta360-research-team.github.io/DAP_website/ {https://insta360-research-team.github.io/DAP\_website/}&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16913.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Xin Lin, Meixi Song, Dizhe Zhang, Wenxuan Lu, Haodong Li, Bo Du, Ming-Hsuan Yang, Truong Nguyen, Lu Qi&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 29&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; In this work, we present a panoramic metric depth foundation model that generalizes across diverse scene distances. We explore a data-in-the-loop paradigm from the view of both data construction and framework design. We collect a large-scale dataset by combining public datasets, high-quality synthetic data from our UE5 simulator and text-to-image models, and real panoramic images from the web. To reduce domain gaps between indoor/outdoor and synthetic/real data, we introduce a three-stage pseudo-label curation pipeline to generate reliable ground truth for unlabeled images. For the model, we adopt DINOv3-Large as the backbone for its strong pre-trained generalization, and introduce a plug-and-play range mask head, sharpness-centric optimization, and geometry-centric optimization to improve robustness to varying distances and enforce geometric consistency across views. Experiments on multiple benchmarks (e.g., Stanford2D3D, Matterport3D, and Deep360) demonstrate strong performance and zero-shot generalization, with particularly robust and stable metric predictions in diverse real-world scenes. The project page can be found at: https://insta360-research-team.github.io/DAP_website/ {https://insta360-research-team.github.io/DAP\_website/}&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16913</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16913</guid>
<pubDate>Thu, 18 Dec 2025 18:59:29 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 18:59:29 +0000</pubDate>
</item> </item>
<item> <item>
<title>StereoPilot: Learning Unified and Efficient Stereo Conversion via Generative Priors</title> <title>StereoPilot: Learning Unified and Efficient Stereo Conversion via Generative Priors</title>
<link>https://arxiv.org/abs/2512.16915</link> <link>https://arxiv.org/abs/2512.16915</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16915.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Guibao Shen, Yihua Du, Wenhang Ge, Jing He, Chirui Chang, Donghao Zhou, Zhen Yang, Luozhou Wang, Xin Tao, Ying-Cong Chen&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 32&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; The rapid growth of stereoscopic displays, including VR headsets and 3D cinemas, has led to increasing demand for high-quality stereo video content. However, producing 3D videos remains costly and complex, while automatic Monocular-to-Stereo conversion is hindered by the limitations of the multi-stage ``Depth-Warp-Inpaint'' (DWI) pipeline. This paradigm suffers from error propagation, depth ambiguity, and format inconsistency between parallel and converged stereo configurations. To address these challenges, we introduce UniStereo, the first large-scale unified dataset for stereo video conversion, covering both stereo formats to enable fair benchmarking and robust model training. Building upon this dataset, we propose StereoPilot, an efficient feed-forward model that directly synthesizes the target view without relying on explicit depth maps or iterative diffusion sampling. Equipped with a learnable domain switcher and a cycle consistency loss, StereoPilot adapts seamlessly to different stereo formats and achieves improved consistency. Extensive experiments demonstrate that StereoPilot significantly outperforms state-of-the-art methods in both visual fidelity and computational efficiency. Project page: https://hit-perfect.github.io/StereoPilot/.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16915.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Guibao Shen, Yihua Du, Wenhang Ge, Jing He, Chirui Chang, Donghao Zhou, Zhen Yang, Luozhou Wang, Xin Tao, Ying-Cong Chen&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 35&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; The rapid growth of stereoscopic displays, including VR headsets and 3D cinemas, has led to increasing demand for high-quality stereo video content. However, producing 3D videos remains costly and complex, while automatic Monocular-to-Stereo conversion is hindered by the limitations of the multi-stage ``Depth-Warp-Inpaint'' (DWI) pipeline. This paradigm suffers from error propagation, depth ambiguity, and format inconsistency between parallel and converged stereo configurations. To address these challenges, we introduce UniStereo, the first large-scale unified dataset for stereo video conversion, covering both stereo formats to enable fair benchmarking and robust model training. Building upon this dataset, we propose StereoPilot, an efficient feed-forward model that directly synthesizes the target view without relying on explicit depth maps or iterative diffusion sampling. Equipped with a learnable domain switcher and a cycle consistency loss, StereoPilot adapts seamlessly to different stereo formats and achieves improved consistency. Extensive experiments demonstrate that StereoPilot significantly outperforms state-of-the-art methods in both visual fidelity and computational efficiency. Project page: https://hit-perfect.github.io/StereoPilot/.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16915</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16915</guid>
<pubDate>Thu, 18 Dec 2025 18:59:50 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 18:59:50 +0000</pubDate>
</item> </item>
@@ -235,28 +249,28 @@
<item> <item>
<title>EasyV2V: A High-quality Instruction-based Video Editing Framework</title> <title>EasyV2V: A High-quality Instruction-based Video Editing Framework</title>
<link>https://arxiv.org/abs/2512.16920</link> <link>https://arxiv.org/abs/2512.16920</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16920.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Jinjie Mai, Chaoyang Wang, Guocheng Gordon Qian, Willi Menapace, Sergey Tulyakov, Bernard Ghanem, Peter Wonka, Ashkan Mirzaei&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 8&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; While image editing has advanced rapidly, video editing remains less explored, facing challenges in consistency, control, and generalization. We study the design space of data, architecture, and control, and introduce EasyV2V, a simple and effective framework for instruction-based video editing. On the data side, we compose existing experts with fast inverses to build diverse video pairs, lift image edit pairs into videos via single-frame supervision and pseudo pairs with shared affine motion, mine dense-captioned clips for video pairs, and add transition supervision to teach how edits unfold. On the model side, we observe that pretrained text-to-video models possess editing capability, motivating a simplified design. Simple sequence concatenation for conditioning with light LoRA fine-tuning suffices to train a strong model. For control, we unify spatiotemporal control via a single mask mechanism and support optional reference images. Overall, EasyV2V works with flexible inputs, e.g., video+text, video+mask+text, video+mask+reference+text, and achieves state-of-the-art video editing results, surpassing concurrent and commercial systems. Project page: https://snap-research.github.io/easyv2v/&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16920.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Jinjie Mai, Chaoyang Wang, Guocheng Gordon Qian, Willi Menapace, Sergey Tulyakov, Bernard Ghanem, Peter Wonka, Ashkan Mirzaei&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 12&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; While image editing has advanced rapidly, video editing remains less explored, facing challenges in consistency, control, and generalization. We study the design space of data, architecture, and control, and introduce EasyV2V, a simple and effective framework for instruction-based video editing. On the data side, we compose existing experts with fast inverses to build diverse video pairs, lift image edit pairs into videos via single-frame supervision and pseudo pairs with shared affine motion, mine dense-captioned clips for video pairs, and add transition supervision to teach how edits unfold. On the model side, we observe that pretrained text-to-video models possess editing capability, motivating a simplified design. Simple sequence concatenation for conditioning with light LoRA fine-tuning suffices to train a strong model. For control, we unify spatiotemporal control via a single mask mechanism and support optional reference images. Overall, EasyV2V works with flexible inputs, e.g., video+text, video+mask+text, video+mask+reference+text, and achieves state-of-the-art video editing results, surpassing concurrent and commercial systems. Project page: https://snap-research.github.io/easyv2v/&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16920</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16920</guid>
<pubDate>Thu, 18 Dec 2025 18:59:57 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 18:59:57 +0000</pubDate>
</item> </item>
<item> <item>
<title>Next-Embedding Prediction Makes Strong Vision Learners</title> <title>Next-Embedding Prediction Makes Strong Vision Learners</title>
<link>https://arxiv.org/abs/2512.16922</link> <link>https://arxiv.org/abs/2512.16922</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16922.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Sihan Xu, Ziqiao Ma, Wenhao Chai, Xuweiyi Chen, Weiyang Jin, Joyce Chai, Saining Xie, Stella X. Yu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 47&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Inspired by the success of generative pretraining in natural language, we ask whether the same principles can yield strong self-supervised visual learners. Instead of training models to output features for downstream use, we train them to generate embeddings to perform predictive tasks directly. This work explores such a shift from learning representations to learning models. Specifically, models learn to predict future patch embeddings conditioned on past ones, using causal masking and stop gradient, which we refer to as Next-Embedding Predictive Autoregression (NEPA). We demonstrate that a simple Transformer pretrained on ImageNet-1k with next embedding prediction as its sole learning objective is effective - no pixel reconstruction, discrete tokens, contrastive loss, or task-specific heads. This formulation retains architectural simplicity and scalability, without requiring additional design complexity. NEPA achieves strong results across tasks, attaining 83.8% and 85.3% top-1 accuracy on ImageNet-1K with ViT-B and ViT-L backbones after fine-tuning, and transferring effectively to semantic segmentation on ADE20K. We believe generative pretraining from embeddings provides a simple, scalable, and potentially modality-agnostic alternative to visual self-supervised learning.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16922.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Sihan Xu, Ziqiao Ma, Wenhao Chai, Xuweiyi Chen, Weiyang Jin, Joyce Chai, Saining Xie, Stella X. Yu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 54&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Inspired by the success of generative pretraining in natural language, we ask whether the same principles can yield strong self-supervised visual learners. Instead of training models to output features for downstream use, we train them to generate embeddings to perform predictive tasks directly. This work explores such a shift from learning representations to learning models. Specifically, models learn to predict future patch embeddings conditioned on past ones, using causal masking and stop gradient, which we refer to as Next-Embedding Predictive Autoregression (NEPA). We demonstrate that a simple Transformer pretrained on ImageNet-1k with next embedding prediction as its sole learning objective is effective - no pixel reconstruction, discrete tokens, contrastive loss, or task-specific heads. This formulation retains architectural simplicity and scalability, without requiring additional design complexity. NEPA achieves strong results across tasks, attaining 83.8% and 85.3% top-1 accuracy on ImageNet-1K with ViT-B and ViT-L backbones after fine-tuning, and transferring effectively to semantic segmentation on ADE20K. We believe generative pretraining from embeddings provides a simple, scalable, and potentially modality-agnostic alternative to visual self-supervised learning.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16922</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16922</guid>
<pubDate>Thu, 18 Dec 2025 18:59:58 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 18:59:58 +0000</pubDate>
</item> </item>
<item> <item>
<title>The World is Your Canvas: Painting Promptable Events with Reference Images, Trajectories, and Text</title> <title>The World is Your Canvas: Painting Promptable Events with Reference Images, Trajectories, and Text</title>
<link>https://arxiv.org/abs/2512.16924</link> <link>https://arxiv.org/abs/2512.16924</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16924.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Hanlin Wang, Hao Ouyang, Qiuyu Wang, Yue Yu, Yihao Meng, Wen Wang, Ka Leong Cheng, Shuailei Ma, Qingyan Bai, Yixuan Li, Cheng Chen, Yanhong Zeng, Xing Zhu, Yujun Shen, Qifeng Chen&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 16&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; We present WorldCanvas, a framework for promptable world events that enables rich, user-directed simulation by combining text, trajectories, and reference images. Unlike text-only approaches and existing trajectory-controlled image-to-video methods, our multimodal approach combines trajectories -- encoding motion, timing, and visibility -- with natural language for semantic intent and reference images for visual grounding of object identity, enabling the generation of coherent, controllable events that include multi-agent interactions, object entry/exit, reference-guided appearance and counterintuitive events. The resulting videos demonstrate not only temporal coherence but also emergent consistency, preserving object identity and scene despite temporary disappearance. By supporting expressive world events generation, WorldCanvas advances world models from passive predictors to interactive, user-shaped simulators. Our project page is available at: https://worldcanvas.github.io/.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16924.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Hanlin Wang, Hao Ouyang, Qiuyu Wang, Yue Yu, Yihao Meng, Wen Wang, Ka Leong Cheng, Shuailei Ma, Qingyan Bai, Yixuan Li, Cheng Chen, Yanhong Zeng, Xing Zhu, Yujun Shen, Qifeng Chen&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 19&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; We present WorldCanvas, a framework for promptable world events that enables rich, user-directed simulation by combining text, trajectories, and reference images. Unlike text-only approaches and existing trajectory-controlled image-to-video methods, our multimodal approach combines trajectories -- encoding motion, timing, and visibility -- with natural language for semantic intent and reference images for visual grounding of object identity, enabling the generation of coherent, controllable events that include multi-agent interactions, object entry/exit, reference-guided appearance and counterintuitive events. The resulting videos demonstrate not only temporal coherence but also emergent consistency, preserving object identity and scene despite temporary disappearance. By supporting expressive world events generation, WorldCanvas advances world models from passive predictors to interactive, user-shaped simulators. Our project page is available at: https://worldcanvas.github.io/.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16924</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16924</guid>
<pubDate>Thu, 18 Dec 2025 18:59:59 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 18:59:59 +0000</pubDate>
</item> </item>
<item> <item>
<title>Generative Refocusing: Flexible Defocus Control from a Single Image</title> <title>Generative Refocusing: Flexible Defocus Control from a Single Image</title>
<link>https://arxiv.org/abs/2512.16923</link> <link>https://arxiv.org/abs/2512.16923</link>
<description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16923.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Chun-Wei Tuan Mu, Jia-Bin Huang, Yu-Lun Liu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 25&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Depth-of-field control is essential in photography, but getting the perfect focus often takes several tries or special equipment. Single-image refocusing is still difficult. It involves recovering sharp content and creating realistic bokeh. Current methods have significant drawbacks. They need all-in-focus inputs, depend on synthetic data from simulators, and have limited control over aperture. We introduce Generative Refocusing, a two-step process that uses DeblurNet to recover all-in-focus images from various inputs and BokehNet for creating controllable bokeh. Our main innovation is semi-supervised training. This method combines synthetic paired data with unpaired real bokeh images, using EXIF metadata to capture real optical characteristics beyond what simulators can provide. Our experiments show we achieve top performance in defocus deblurring, bokeh synthesis, and refocusing benchmarks. Additionally, our Generative Refocusing allows text-guided adjustments and custom aperture shapes.&lt;/p&gt;</description> <description>&lt;p&gt;&lt;img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2512.16923.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /&gt;&lt;/p&gt;&lt;p&gt;&lt;b&gt;Authors:&lt;/b&gt; Chun-Wei Tuan Mu, Jia-Bin Huang, Yu-Lun Liu&lt;/p&gt;&lt;p&gt;&lt;b&gt;Upvotes:&lt;/b&gt; 28&lt;/p&gt;&lt;p&gt;&lt;b&gt;Summary:&lt;/b&gt; Depth-of-field control is essential in photography, but getting the perfect focus often takes several tries or special equipment. Single-image refocusing is still difficult. It involves recovering sharp content and creating realistic bokeh. Current methods have significant drawbacks. They need all-in-focus inputs, depend on synthetic data from simulators, and have limited control over aperture. We introduce Generative Refocusing, a two-step process that uses DeblurNet to recover all-in-focus images from various inputs and BokehNet for creating controllable bokeh. Our main innovation is semi-supervised training. This method combines synthetic paired data with unpaired real bokeh images, using EXIF metadata to capture real optical characteristics beyond what simulators can provide. Our experiments show we achieve top performance in defocus deblurring, bokeh synthesis, and refocusing benchmarks. Additionally, our Generative Refocusing allows text-guided adjustments and custom aperture shapes.&lt;/p&gt;</description>
<guid isPermaLink="false">https://arxiv.org/abs/2512.16923</guid> <guid isPermaLink="false">https://arxiv.org/abs/2512.16923</guid>
<pubDate>Thu, 18 Dec 2025 18:59:59 +0000</pubDate> <pubDate>Thu, 18 Dec 2025 18:59:59 +0000</pubDate>
</item> </item>