The list as it stood that day: the papers researchers were upvoting on Hugging Face's daily papers. Papers stay on arXiv.
We therefore introduce The Trinity, a recursive self-improvement framework coupling world generation with Harness evolution through three roles: Creator, Explorer, and Evolver.
Weidong Tang, Ruiheng Wang, Jingxuan Zhang and 17 moreWe observe that critic values vary primarily across turns, motivating optimization anchored at turns with finer credit allocation among tokens.
Tao Ren, Hui Yang, Jia Deng and 16 moreIn this paper, we propose Denoising Token Credit Assignment (DTCA), a framework for token-level credit assignment in reinforcement learning for diffusion language models.
Hongze Chen, Miao Pan, Lizhou Liang and 7 moreWe introduce VibeVideoBench, a benchmark for evaluating agents on code-driven video generation.
Muxi Diao, Wenbin Xu, Lele Yang and 21 moreThrough a large-scale evaluation across multiple benchmarks and model backbones, we find many dedicated memory systems underperform naive RAG in many cases, indicating a negative net value.
Xuan He, Sining Zhoubian, Ruozhen Yang and 8 moreWe introduce PDRVLA, whose lightweight Physical Dynamics Module extracts physical tokens from vision--language features to condition action generation without online video generation.
Hanqing Wang, Zhaotian Cai, Ruichuan An and 12 moreStreaming audio-visual target speaker extraction (AVTSE) is essential for latency-sensitive applications.
Wendi Sang, Kai Li, Yifan Li and 2 moreWe introduce Test-Time Memory Evolution (TTME), a plug-and-play test-time adaptation method, as an alternative to TTT.
Yepeng Liu, Shengmai Chen, Xuandong Zhao and 4 moreWe present Hy-Nomas, a video world model with an 8B diffusion backbone that supports free-form text instructions and continuous camera control across multiple interaction turns.
Qiaoling Chen, Jiaxiang Cheng, Xueqing Deng and 35 moreUsing multimodal understanding as a proxy for semantic fidelity, we revisit this assumption and find quantization error unreliable, as discrete representations with substantially larger errors can achieve better understanding.
Jialv Zou, Xiaotong Li, Zhengrong Yue and 6 moreIn long-term interactions with large language model (LLM) agents, users reveal habits through their choices and actions.
Junming Zhang, Rubin Wei, Xu Cheng and 7 moreWe introduce ThinkWorld, a VLA learning framework that grounds action reasoning in world dynamics and strengthens it through reinforcement learning with world imagination.
Chaehyun Kim, Chi-Pin Huang, Jisoo Kim and 4 moreIn this work, we explore Code as Video Caption, a paradigm where executable programs serve as a complementary representation of visual content through graphical primitives, layout rules, and functions of time.
Jiafu Tang, Yifan Xu, Eric Tang and 11 moreWe introduce OvisOCR3, a 0.8B model for a broad range of OCR-related tasks. OvisOCR3 unifies document parsing, general OCR, key information extraction (KIE), document figure captioning, and cross-page table reconstruction in a single model.
Yu Xia, Yinglun Li, Yuhui Chen and 12 moreIn this work, we study the value of truncation in looped transformer training in a systematic way: we jointly vary model size, training tokens, and forward and backward loop counts across nearly 600 training runs and fit a unified scaling law that captures the effects of looping…
Xuyuan Xiong, Siyu Chen, Zihan Shao and 5 moreOur key insight is that these tasks share an underlying capability, which we call connecting evidence: identifying relevant visual facts, binding them to the correct entities, recovering their relations, and composing them into an answer.
Lai Wei, Yatai Li, Liangbo He and 10 moreGenerating videos of a specific subject from a few reference images remains challenging when the camera reveals surfaces that are not visible in the input.
Junyeong Ahn, Geonwoo Kim, Kinam Kim and 4 moreWe present EmbeddingGemma 2, an open multimodal embedding model which natively represents text, images, audio, video, and combinations thereof in a unified space.
Henrique Schechter Vera, Sahil Dua, Jovan Andonov and 70 moreWe show that channel selection needs no calibration data. First, a calibrated threshold does not fix how much a selector reads, because the number of channels passing it varies across inputs; Prox with thresholds calibrated for an 18% read budget actually reads only 12.9% on…
Geunhyoung Kim, Sung Ju HwangWe present MatAnyone-E, an efficient extension of MatAnyone 2 that focuses computation where it is most needed.
Shuyun Wang, Shangchen Zhou, Peiqing Yang and 3 more