diff --git a/feed_monthly.xml b/feed_monthly.xml index e371a56..332a55a 100644 --- a/feed_monthly.xml +++ b/feed_monthly.xml @@ -7,741 +7,741 @@ http://www.rssboard.org/rss-specification python-feedgen en - Sun, 01 Jun 2025 00:14:50 +0000 + Tue, 01 Jul 2025 00:13:58 +0000 - PixelHacker: Image Inpainting with Structural and Semantic Consistency - https://arxiv.org/abs/2504.20438 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2504.20438.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ziyang Xu, Kangsheng Duan, Xiaolei Shen, Zhifeng Ding, Wenyu Liu, Xiaohu Ruan, Xiaoxin Chen, Xinggang Wang</p><p><b>Upvotes:</b> 42</p><p><b>Summary:</b> Image inpainting is a fundamental research area between image editing and image generation. Recent state-of-the-art (SOTA) methods have explored novel attention mechanisms, lightweight architectures, and context-aware modeling, demonstrating impressive performance. However, they often struggle with complex structure (e.g., texture, shape, spatial relations) and semantics (e.g., color consistency, object restoration, and logical correctness), leading to artifacts and inappropriate generation. To address this challenge, we design a simple yet effective inpainting paradigm called latent categories guidance, and further propose a diffusion-based model named PixelHacker. Specifically, we first construct a large dataset containing 14 million image-mask pairs by annotating foreground and background (potential 116 and 21 categories, respectively). Then, we encode potential foreground and background representations separately through two fixed-size embeddings, and intermittently inject these features into the denoising process via linear attention. Finally, by pre-training on our dataset and fine-tuning on open-source benchmarks, we obtain PixelHacker. Extensive experiments show that PixelHacker comprehensively outperforms the SOTA on a wide range of datasets (Places2, CelebA-HQ, and FFHQ) and exhibits remarkable consistency in both structure and semantics. Project page at https://hustvl.github.io/PixelHacker.</p> - https://arxiv.org/abs/2504.20438 - Tue, 29 Apr 2025 05:28:36 +0000 + Large Language Models for Data Synthesis + https://arxiv.org/abs/2505.14752 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.14752.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yihong Tang, Menglin Kong, Lijun Sun</p><p><b>Upvotes:</b> 49</p><p><b>Summary:</b> Generating synthetic data that faithfully captures the statistical structure of real-world distributions is a fundamental challenge in data modeling. Classical approaches often depend on strong parametric assumptions or manual structural design and struggle in high-dimensional or heterogeneous domains. Recent progress in Large Language Models (LLMs) reveals their potential as flexible, high-dimensional priors over real-world distributions. However, when applied to data synthesis, standard LLM-based sampling is inefficient, constrained by fixed context limits, and fails to ensure statistical alignment. Given this, we introduce LLMSynthor, a general framework for data synthesis that transforms LLMs into structure-aware simulators guided by distributional feedback. LLMSynthor treats the LLM as a nonparametric copula simulator for modeling high-order dependencies and introduces LLM Proposal Sampling to generate grounded proposal distributions that improve sampling efficiency without requiring rejection. By minimizing discrepancies in the summary statistics space, the iterative synthesis loop aligns real and synthetic data while gradually uncovering and refining the latent generative structure. We evaluate LLMSynthor in both controlled and real-world settings using heterogeneous datasets in privacy-sensitive domains (e.g., e-commerce, population, and mobility) that encompass both structured and unstructured formats. The synthetic data produced by LLMSynthor shows high statistical fidelity, practical utility, and cross-data adaptability, positioning it as a valuable tool across economics, social science, urban studies, and beyond.</p> + https://arxiv.org/abs/2505.14752 + Tue, 20 May 2025 13:35:38 +0000 - Grokking in the Wild: Data Augmentation for Real-World Multi-Hop Reasoning with Transformers - https://arxiv.org/abs/2504.20752 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2504.20752.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Roman Abramov, Felix Steinbauer, Gjergji Kasneci</p><p><b>Upvotes:</b> 90</p><p><b>Summary:</b> Transformers have achieved great success in numerous NLP tasks but continue to exhibit notable gaps in multi-step factual reasoning, especially when real-world knowledge is sparse. Recent advances in grokking have demonstrated that neural networks can transition from memorizing to perfectly generalizing once they detect underlying logical patterns - yet these studies have primarily used small, synthetic tasks. In this paper, for the first time, we extend grokking to real-world factual data and address the challenge of dataset sparsity by augmenting existing knowledge graphs with carefully designed synthetic data to raise the ratio phi_r of inferred facts to atomic facts above the threshold required for grokking. Surprisingly, we find that even factually incorrect synthetic data can strengthen emergent reasoning circuits rather than degrade accuracy, as it forces the model to rely on relational structure rather than memorization. When evaluated on multi-hop reasoning benchmarks, our approach achieves up to 95-100% accuracy on 2WikiMultiHopQA - substantially improving over strong baselines and matching or exceeding current state-of-the-art results. We further provide an in-depth analysis of how increasing phi_r drives the formation of generalizing circuits inside Transformers. Our findings suggest that grokking-based data augmentation can unlock implicit multi-hop reasoning capabilities, opening the door to more robust and interpretable factual reasoning in large-scale language models.</p> - https://arxiv.org/abs/2504.20752 - Tue, 29 Apr 2025 13:33:29 +0000 + CASS: Nvidia to AMD Transpilation with Data, Models, and Benchmark + https://arxiv.org/abs/2505.16968 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.16968.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ahmed Heakl, Sarim Hashmi, Gustavo Bertolo Stahl, Seung Hun Eddie Han, Salman Khan, Abdulrahman Mahmoud</p><p><b>Upvotes:</b> 41</p><p><b>Summary:</b> We introduce CASS, the first large-scale dataset and model suite for cross-architecture GPU code transpilation, targeting both source-level (CUDA leftrightarrow HIP) and assembly-level (Nvidia SASS leftrightarrow AMD RDNA3) translation. The dataset comprises 70k verified code pairs across host and device, addressing a critical gap in low-level GPU code portability. Leveraging this resource, we train the CASS family of domain-specific language models, achieving 95% source translation accuracy and 37.5% assembly translation accuracy, substantially outperforming commercial baselines such as GPT-4o, Claude, and Hipify. Our generated code matches native performance in over 85% of test cases, preserving runtime and memory behavior. To support rigorous evaluation, we introduce CASS-Bench, a curated benchmark spanning 16 GPU domains with ground-truth execution. All data, models, and evaluation tools are released as open source to foster progress in GPU compiler tooling, binary compatibility, and LLM-guided hardware translation. Dataset and benchmark are on https://huggingface.co/datasets/MBZUAI/cass{blue{HuggingFace}}, with code at https://github.com/GustavoStahl/CASS{blue{GitHub}}.</p> + https://arxiv.org/abs/2505.16968 + Thu, 22 May 2025 17:48:53 +0000 - Phi-4-Mini-Reasoning: Exploring the Limits of Small Reasoning Language Models in Math - https://arxiv.org/abs/2504.21233 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2504.21233.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Haoran Xu, Baolin Peng, Hany Awadalla, Dongdong Chen, Yen-Chun Chen, Mei Gao, Young Jin Kim, Yunsheng Li, Liliang Ren, Yelong Shen, Shuohang Wang, Weijian Xu, Jianfeng Gao, Weizhu Chen</p><p><b>Upvotes:</b> 43</p><p><b>Summary:</b> Chain-of-Thought (CoT) significantly enhances formal reasoning capabilities in Large Language Models (LLMs) by training them to explicitly generate intermediate reasoning steps. While LLMs readily benefit from such techniques, improving reasoning in Small Language Models (SLMs) remains challenging due to their limited model capacity. Recent work by Deepseek-R1 demonstrates that distillation from LLM-generated synthetic data can substantially improve the reasoning ability of SLM. However, the detailed modeling recipe is not disclosed. In this work, we present a systematic training recipe for SLMs that consists of four steps: (1) large-scale mid-training on diverse distilled long-CoT data, (2) supervised fine-tuning on high-quality long-CoT data, (3) Rollout DPO leveraging a carefully curated preference dataset, and (4) Reinforcement Learning (RL) with Verifiable Reward. We apply our method on Phi-4-Mini, a compact 3.8B-parameter model. The resulting Phi-4-Mini-Reasoning model exceeds, on math reasoning tasks, much larger reasoning models, e.g., outperforming DeepSeek-R1-Distill-Qwen-7B by 3.2 points and DeepSeek-R1-Distill-Llama-8B by 7.7 points on Math-500. Our results validate that a carefully designed training recipe, with large-scale high-quality CoT data, is effective to unlock strong reasoning capabilities even in resource-constrained small models.</p> - https://arxiv.org/abs/2504.21233 - Wed, 30 Apr 2025 00:04:35 +0000 + Don't Look Only Once: Towards Multimodal Interactive Reasoning with Selective Visual Revisitation + https://arxiv.org/abs/2505.18842 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.18842.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jiwan Chung, Junhyeok Kim, Siyeol Kim, Jaeyoung Lee, Min Soo Kim, Youngjae Yu</p><p><b>Upvotes:</b> 37</p><p><b>Summary:</b> We present v1, a lightweight extension to Multimodal Large Language Models (MLLMs) that enables selective visual revisitation during inference. While current MLLMs typically consume visual input only once and reason purely over internal memory, v1 introduces a simple point-and-copy mechanism that allows the model to dynamically retrieve relevant image regions throughout the reasoning process. This mechanism augments existing architectures with minimal modifications, enabling contextual access to visual tokens based on the model's evolving hypotheses. To train this capability, we construct v1g, a dataset of 300K multimodal reasoning traces with interleaved visual grounding annotations. Experiments on three multimodal mathematical reasoning benchmarks -- MathVista, MathVision, and MathVerse -- demonstrate that v1 consistently improves performance over comparable baselines, particularly on tasks requiring fine-grained visual reference and multi-step reasoning. Our results suggest that dynamic visual access is a promising direction for enhancing grounded multimodal reasoning. Code, models, and data will be released to support future research.</p> + https://arxiv.org/abs/2505.18842 + Sat, 24 May 2025 19:30:47 +0000 - Phi-4-reasoning Technical Report - https://arxiv.org/abs/2504.21318 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2504.21318.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Marah Abdin, Sahaj Agarwal, Ahmed Awadallah, Vidhisha Balachandran, Harkirat Behl, Lingjiao Chen, Gustavo de Rosa, Suriya Gunasekar, Mojan Javaheripi, Neel Joshi, Piero Kauffmann, Yash Lara, Caio César Teodoro Mendes, Arindam Mitra, Besmira Nushi, Dimitris Papailiopoulos, Olli Saarikivi, Shital Shah, Vaishnavi Shrivastava, Vibhav Vineet, Yue Wu, Safoora Yousefi, Guoqing Zheng</p><p><b>Upvotes:</b> 46</p><p><b>Summary:</b> We introduce Phi-4-reasoning, a 14-billion parameter reasoning model that achieves strong performance on complex reasoning tasks. Trained via supervised fine-tuning of Phi-4 on carefully curated set of "teachable" prompts-selected for the right level of complexity and diversity-and reasoning demonstrations generated using o3-mini, Phi-4-reasoning generates detailed reasoning chains that effectively leverage inference-time compute. We further develop Phi-4-reasoning-plus, a variant enhanced through a short phase of outcome-based reinforcement learning that offers higher performance by generating longer reasoning traces. Across a wide range of reasoning tasks, both models outperform significantly larger open-weight models such as DeepSeek-R1-Distill-Llama-70B model and approach the performance levels of full DeepSeek-R1 model. Our comprehensive evaluations span benchmarks in math and scientific reasoning, coding, algorithmic problem solving, planning, and spatial understanding. Interestingly, we observe a non-trivial transfer of improvements to general-purpose benchmarks as well. In this report, we provide insights into our training data, our training methodologies, and our evaluations. We show that the benefit of careful data curation for supervised fine-tuning (SFT) extends to reasoning language models, and can be further amplified by reinforcement learning (RL). Finally, our evaluation points to opportunities for improving how we assess the performance and robustness of reasoning models.</p> - https://arxiv.org/abs/2504.21318 - Wed, 30 Apr 2025 05:05:09 +0000 + Will It Still Be True Tomorrow? Multilingual Evergreen Question Classification to Improve Trustworthy QA + https://arxiv.org/abs/2505.21115 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.21115.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Sergey Pletenev, Maria Marina, Nikolay Ivanov, Daria Galimzianova, Nikita Krayko, Mikhail Salnikov, Vasily Konovalov, Alexander Panchenko, Viktor Moskvoretskii</p><p><b>Upvotes:</b> 133</p><p><b>Summary:</b> Large Language Models (LLMs) often hallucinate in question answering (QA) tasks. A key yet underexplored factor contributing to this is the temporality of questions -- whether they are evergreen (answers remain stable over time) or mutable (answers change). In this work, we introduce EverGreenQA, the first multilingual QA dataset with evergreen labels, supporting both evaluation and training. Using EverGreenQA, we benchmark 12 modern LLMs to assess whether they encode question temporality explicitly (via verbalized judgments) or implicitly (via uncertainty signals). We also train EG-E5, a lightweight multilingual classifier that achieves SoTA performance on this task. Finally, we demonstrate the practical utility of evergreen classification across three applications: improving self-knowledge estimation, filtering QA datasets, and explaining GPT-4o retrieval behavior.</p> + https://arxiv.org/abs/2505.21115 + Tue, 27 May 2025 12:35:13 +0000 - Sadeed: Advancing Arabic Diacritization Through Small Language Model - https://arxiv.org/abs/2504.21635 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2504.21635.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zeina Aldallal, Sara Chrouf, Khalil Hennara, Mohamed Motaism Hamed, Muhammad Hreden, Safwan AlModhayan</p><p><b>Upvotes:</b> 58</p><p><b>Summary:</b> Arabic text diacritization remains a persistent challenge in natural language processing due to the language's morphological richness. In this paper, we introduce Sadeed, a novel approach based on a fine-tuned decoder-only language model adapted from Kuwain 1.5B Hennara et al. [2025], a compact model originally trained on diverse Arabic corpora. Sadeed is fine-tuned on carefully curated, high-quality diacritized datasets, constructed through a rigorous data-cleaning and normalization pipeline. Despite utilizing modest computational resources, Sadeed achieves competitive results compared to proprietary large language models and outperforms traditional models trained on similar domains. Additionally, we highlight key limitations in current benchmarking practices for Arabic diacritization. To address these issues, we introduce SadeedDiac-25, a new benchmark designed to enable fairer and more comprehensive evaluation across diverse text genres and complexity levels. Together, Sadeed and SadeedDiac-25 provide a robust foundation for advancing Arabic NLP applications, including machine translation, text-to-speech, and language learning tools.</p> - https://arxiv.org/abs/2504.21635 - Wed, 30 Apr 2025 13:37:24 +0000 + HardTests: Synthesizing High-Quality Test Cases for LLM Coding + https://arxiv.org/abs/2505.24098 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.24098.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhongmou He, Yee Man Choi, Kexun Zhang, Jiabao Ji, Junting Zhou, Dejia Xu, Ivan Bercovich, Aidan Zhang, Lei Li</p><p><b>Upvotes:</b> 44</p><p><b>Summary:</b> Verifiers play a crucial role in large language model (LLM) reasoning, needed by post-training techniques such as reinforcement learning. However, reliable verifiers are hard to get for difficult coding problems, because a well-disguised wrong solution may only be detected by carefully human-written edge cases that are difficult to synthesize. To address this issue, we propose HARDTESTGEN, a pipeline for high-quality test synthesis using LLMs. With this pipeline, we curate a comprehensive competitive programming dataset HARDTESTS with 47k problems and synthetic high-quality tests. Compared with existing tests, HARDTESTGEN tests demonstrate precision that is 11.3 percentage points higher and recall that is 17.5 percentage points higher when evaluating LLM-generated code. For harder problems, the improvement in precision can be as large as 40 points. HARDTESTS also proves to be more effective for model training, measured by downstream code generation performance. We will open-source our dataset and synthesis pipeline at https://leililab.github.io/HardTests/.</p> + https://arxiv.org/abs/2505.24098 + Fri, 30 May 2025 01:00:34 +0000 - WebThinker: Empowering Large Reasoning Models with Deep Research Capability - https://arxiv.org/abs/2504.21776 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2504.21776.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Xiaoxi Li, Jiajie Jin, Guanting Dong, Hongjin Qian, Yutao Zhu, Yongkang Wu, Ji-Rong Wen, Zhicheng Dou</p><p><b>Upvotes:</b> 54</p><p><b>Summary:</b> Large reasoning models (LRMs), such as OpenAI-o1 and DeepSeek-R1, demonstrate impressive long-horizon reasoning capabilities. However, their reliance on static internal knowledge limits their performance on complex, knowledge-intensive tasks and hinders their ability to produce comprehensive research reports requiring synthesis of diverse web information. To address this, we propose WebThinker, a deep research agent that empowers LRMs to autonomously search the web, navigate web pages, and draft research reports during the reasoning process. WebThinker integrates a Deep Web Explorer module, enabling LRMs to dynamically search, navigate, and extract information from the web when encountering knowledge gaps. It also employs an Autonomous Think-Search-and-Draft strategy, allowing the model to seamlessly interleave reasoning, information gathering, and report writing in real time. To further enhance research tool utilization, we introduce an RL-based training strategy via iterative online Direct Preference Optimization (DPO). Extensive experiments on complex reasoning benchmarks (GPQA, GAIA, WebWalkerQA, HLE) and scientific report generation tasks (Glaive) demonstrate that WebThinker significantly outperforms existing methods and strong proprietary systems. Our approach enhances LRM reliability and applicability in complex scenarios, paving the way for more capable and versatile deep research systems. The code is available at https://github.com/RUC-NLPIR/WebThinker.</p> - https://arxiv.org/abs/2504.21776 - Wed, 30 Apr 2025 16:25:25 +0000 + CSVQA: A Chinese Multimodal Benchmark for Evaluating STEM Reasoning Capabilities of VLMs + https://arxiv.org/abs/2505.24120 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.24120.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ai Jian, Weijie Qiu, Xiaokun Wang, Peiyu Wang, Yunzhuo Hao, Jiangbo Pei, Yichen Wei, Yi Peng, Xuchen Song</p><p><b>Upvotes:</b> 49</p><p><b>Summary:</b> Vision-Language Models (VLMs) have demonstrated remarkable progress in multimodal understanding, yet their capabilities for scientific reasoning remains inadequately assessed. Current multimodal benchmarks predominantly evaluate generic image comprehension or text-driven reasoning, lacking authentic scientific contexts that require domain-specific knowledge integration with visual evidence analysis. To fill this gap, we present CSVQA, a diagnostic multimodal benchmark specifically designed for evaluating scientific reasoning through domain-grounded visual question answering.Our benchmark features 1,378 carefully constructed question-answer pairs spanning diverse STEM disciplines, each demanding domain knowledge, integration of visual evidence, and higher-order reasoning. Compared to prior multimodal benchmarks, CSVQA places greater emphasis on real-world scientific content and complex reasoning.We additionally propose a rigorous evaluation protocol to systematically assess whether model predictions are substantiated by valid intermediate reasoning steps based on curated explanations. Our comprehensive evaluation of 15 VLMs on this benchmark reveals notable performance disparities, as even the top-ranked proprietary model attains only 49.6\% accuracy.This empirical evidence underscores the pressing need for advancing scientific reasoning capabilities in VLMs. Our CSVQA is released at https://huggingface.co/datasets/Skywork/CSVQA.</p> + https://arxiv.org/abs/2505.24120 + Fri, 30 May 2025 01:34:25 +0000 - A Survey of Interactive Generative Video - https://arxiv.org/abs/2504.21853 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2504.21853.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jiwen Yu, Yiran Qin, Haoxuan Che, Quande Liu, Xintao Wang, Pengfei Wan, Di Zhang, Kun Gai, Hao Chen, Xihui Liu</p><p><b>Upvotes:</b> 46</p><p><b>Summary:</b> Interactive Generative Video (IGV) has emerged as a crucial technology in response to the growing demand for high-quality, interactive video content across various domains. In this paper, we define IGV as a technology that combines generative capabilities to produce diverse high-quality video content with interactive features that enable user engagement through control signals and responsive feedback. We survey the current landscape of IGV applications, focusing on three major domains: 1) gaming, where IGV enables infinite exploration in virtual worlds; 2) embodied AI, where IGV serves as a physics-aware environment synthesizer for training agents in multimodal interaction with dynamically evolving scenes; and 3) autonomous driving, where IGV provides closed-loop simulation capabilities for safety-critical testing and validation. To guide future development, we propose a comprehensive framework that decomposes an ideal IGV system into five essential modules: Generation, Control, Memory, Dynamics, and Intelligence. Furthermore, we systematically analyze the technical challenges and future directions in realizing each component for an ideal IGV system, such as achieving real-time generation, enabling open-domain control, maintaining long-term coherence, simulating accurate physics, and integrating causal reasoning. We believe that this systematic analysis will facilitate future research and development in the field of IGV, ultimately advancing the technology toward more sophisticated and practical applications.</p> - https://arxiv.org/abs/2504.21853 - Wed, 30 Apr 2025 17:59:02 +0000 + FinMME: Benchmark Dataset for Financial Multi-Modal Reasoning Evaluation + https://arxiv.org/abs/2505.24714 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.24714.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Junyu Luo, Zhizhuo Kou, Liming Yang, Xiao Luo, Jinsheng Huang, Zhiping Xiao, Jingshu Peng, Chengzhong Liu, Jiaming Ji, Xuanzhe Liu, Sirui Han, Ming Zhang, Yike Guo</p><p><b>Upvotes:</b> 36</p><p><b>Summary:</b> Multimodal Large Language Models (MLLMs) have experienced rapid development in recent years. However, in the financial domain, there is a notable lack of effective and specialized multimodal evaluation datasets. To advance the development of MLLMs in the finance domain, we introduce FinMME, encompassing more than 11,000 high-quality financial research samples across 18 financial domains and 6 asset classes, featuring 10 major chart types and 21 subtypes. We ensure data quality through 20 annotators and carefully designed validation mechanisms. Additionally, we develop FinScore, an evaluation system incorporating hallucination penalties and multi-dimensional capability assessment to provide an unbiased evaluation. Extensive experimental results demonstrate that even state-of-the-art models like GPT-4o exhibit unsatisfactory performance on FinMME, highlighting its challenging nature. The benchmark exhibits high robustness with prediction variations under different prompts remaining below 1%, demonstrating superior reliability compared to existing datasets. Our dataset and evaluation protocol are available at https://huggingface.co/datasets/luojunyu/FinMME and https://github.com/luo-junyu/FinMME.</p> + https://arxiv.org/abs/2505.24714 + Fri, 30 May 2025 15:36:19 +0000 - 100 Days After DeepSeek-R1: A Survey on Replication Studies and More Directions for Reasoning Language Models - https://arxiv.org/abs/2505.00551 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.00551.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Chong Zhang, Yue Deng, Xiang Lin, Bin Wang, Dianwen Ng, Hai Ye, Xingxuan Li, Yao Xiao, Zhanfeng Mo, Qi Zhang, Lidong Bing</p><p><b>Upvotes:</b> 36</p><p><b>Summary:</b> The recent development of reasoning language models (RLMs) represents a novel evolution in large language models. In particular, the recent release of DeepSeek-R1 has generated widespread social impact and sparked enthusiasm in the research community for exploring the explicit reasoning paradigm of language models. However, the implementation details of the released models have not been fully open-sourced by DeepSeek, including DeepSeek-R1-Zero, DeepSeek-R1, and the distilled small models. As a result, many replication studies have emerged aiming to reproduce the strong performance achieved by DeepSeek-R1, reaching comparable performance through similar training procedures and fully open-source data resources. These works have investigated feasible strategies for supervised fine-tuning (SFT) and reinforcement learning from verifiable rewards (RLVR), focusing on data preparation and method design, yielding various valuable insights. In this report, we provide a summary of recent replication studies to inspire future research. We primarily focus on SFT and RLVR as two main directions, introducing the details for data construction, method design and training procedure of current replication studies. Moreover, we conclude key findings from the implementation details and experimental results reported by these studies, anticipating to inspire future research. We also discuss additional techniques of enhancing RLMs, highlighting the potential of expanding the application scope of these models, and discussing the challenges in development. By this survey, we aim to help researchers and developers of RLMs stay updated with the latest advancements, and seek to inspire new ideas to further enhance RLMs.</p> - https://arxiv.org/abs/2505.00551 - Thu, 01 May 2025 14:28:35 +0000 + Reflect, Retry, Reward: Self-Improving LLMs via Reinforcement Learning + https://arxiv.org/abs/2505.24726 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.24726.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Shelly Bensal, Umar Jamil, Christopher Bryant, Melisa Russak, Kiran Kamble, Dmytro Mozolevskyi, Muayad Ali, Waseem AlShikh</p><p><b>Upvotes:</b> 244</p><p><b>Summary:</b> We explore a method for improving the performance of large language models through self-reflection and reinforcement learning. By incentivizing the model to generate better self-reflections when it answers incorrectly, we demonstrate that a model's ability to solve complex, verifiable tasks can be enhanced even when generating synthetic data is infeasible and only binary feedback is available. Our framework operates in two stages: first, upon failing a given task, the model generates a self-reflective commentary analyzing its previous attempt; second, the model is given another attempt at the task with the self-reflection in context. If the subsequent attempt succeeds, the tokens generated during the self-reflection phase are rewarded. Our experimental results show substantial performance gains across a variety of model architectures, as high as 34.7% improvement at math equation writing and 18.1% improvement at function calling. Notably, smaller fine-tuned models (1.5 billion to 7 billion parameters) outperform models in the same family that are 10 times larger. Our novel paradigm is thus an exciting pathway to more useful and reliable language models that can self-improve on challenging tasks with limited external feedback.</p> + https://arxiv.org/abs/2505.24726 + Fri, 30 May 2025 15:49:42 +0000 - DeepCritic: Deliberate Critique with Large Language Models - https://arxiv.org/abs/2505.00662 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.00662.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Wenkai Yang, Jingwen Chen, Yankai Lin, Ji-Rong Wen</p><p><b>Upvotes:</b> 52</p><p><b>Summary:</b> As Large Language Models (LLMs) are rapidly evolving, providing accurate feedback and scalable oversight on their outputs becomes an urgent and critical problem. Leveraging LLMs as critique models to achieve automated supervision is a promising solution. In this work, we focus on studying and enhancing the math critique ability of LLMs. Current LLM critics provide critiques that are too shallow and superficial on each step, leading to low judgment accuracy and struggling to offer sufficient feedback for the LLM generator to correct mistakes. To tackle this issue, we propose a novel and effective two-stage framework to develop LLM critics that are capable of deliberately critiquing on each reasoning step of math solutions. In the first stage, we utilize Qwen2.5-72B-Instruct to generate 4.5K long-form critiques as seed data for supervised fine-tuning. Each seed critique consists of deliberate step-wise critiques that includes multi-perspective verifications as well as in-depth critiques of initial critiques for each reasoning step. Then, we perform reinforcement learning on the fine-tuned model with either existing human-labeled data from PRM800K or our automatically annotated data obtained via Monte Carlo sampling-based correctness estimation, to further incentivize its critique ability. Our developed critique model built on Qwen2.5-7B-Instruct not only significantly outperforms existing LLM critics (including the same-sized DeepSeek-R1-distill models and GPT-4o) on various error identification benchmarks, but also more effectively helps the LLM generator refine erroneous steps through more detailed feedback.</p> - https://arxiv.org/abs/2505.00662 - Thu, 01 May 2025 17:03:17 +0000 + REASONING GYM: Reasoning Environments for Reinforcement Learning with Verifiable Rewards + https://arxiv.org/abs/2505.24760 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.24760.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zafir Stojanovski, Oliver Stanley, Joe Sharratt, Richard Jones, Abdulhakeem Adefioye, Jean Kaddour, Andreas Köpf</p><p><b>Upvotes:</b> 64</p><p><b>Summary:</b> We introduce Reasoning Gym (RG), a library of reasoning environments for reinforcement learning with verifiable rewards. It provides over 100 data generators and verifiers spanning multiple domains including algebra, arithmetic, computation, cognition, geometry, graph theory, logic, and various common games. Its key innovation is the ability to generate virtually infinite training data with adjustable complexity, unlike most previous reasoning datasets, which are typically fixed. This procedural generation approach allows for continuous evaluation across varying difficulty levels. Our experimental results demonstrate the efficacy of RG in both evaluating and reinforcement learning of reasoning models.</p> + https://arxiv.org/abs/2505.24760 + Fri, 30 May 2025 16:20:18 +0000 - T2I-R1: Reinforcing Image Generation with Collaborative Semantic-level and Token-level CoT - https://arxiv.org/abs/2505.00703 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.00703.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Dongzhi Jiang, Ziyu Guo, Renrui Zhang, Zhuofan Zong, Hao Li, Le Zhuo, Shilin Yan, Pheng-Ann Heng, Hongsheng Li</p><p><b>Upvotes:</b> 41</p><p><b>Summary:</b> Recent advancements in large language models have demonstrated how chain-of-thought (CoT) and reinforcement learning (RL) can improve performance. However, applying such reasoning strategies to the visual generation domain remains largely unexplored. In this paper, we present T2I-R1, a novel reasoning-enhanced text-to-image generation model, powered by RL with a bi-level CoT reasoning process. Specifically, we identify two levels of CoT that can be utilized to enhance different stages of generation: (1) the semantic-level CoT for high-level planning of the prompt and (2) the token-level CoT for low-level pixel processing during patch-by-patch generation. To better coordinate these two levels of CoT, we introduce BiCoT-GRPO with an ensemble of generation rewards, which seamlessly optimizes both generation CoTs within the same training step. By applying our reasoning strategies to the baseline model, Janus-Pro, we achieve superior performance with 13% improvement on T2I-CompBench and 19% improvement on the WISE benchmark, even surpassing the state-of-the-art model FLUX.1. Code is available at: https://github.com/CaraJ7/T2I-R1</p> - https://arxiv.org/abs/2505.00703 - Thu, 01 May 2025 17:59:46 +0000 + AlphaOne: Reasoning Models Thinking Slow and Fast at Test Time + https://arxiv.org/abs/2505.24863 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.24863.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Junyu Zhang, Runpei Dong, Han Wang, Xuying Ning, Haoran Geng, Peihao Li, Xialin He, Yutong Bai, Jitendra Malik, Saurabh Gupta, Huan Zhang</p><p><b>Upvotes:</b> 95</p><p><b>Summary:</b> This paper presents AlphaOne (alpha1), a universal framework for modulating reasoning progress in large reasoning models (LRMs) at test time. alpha1 first introduces alpha moment, which represents the scaled thinking phase with a universal parameter alpha. Within this scaled pre-alpha moment phase, it dynamically schedules slow thinking transitions by modeling the insertion of reasoning transition tokens as a Bernoulli stochastic process. After the alpha moment, alpha1 deterministically terminates slow thinking with the end-of-thinking token, thereby fostering fast reasoning and efficient answer generation. This approach unifies and generalizes existing monotonic scaling methods by enabling flexible and dense slow-to-fast reasoning modulation. Extensive empirical studies on various challenging benchmarks across mathematical, coding, and scientific domains demonstrate alpha1's superior reasoning capability and efficiency. Project page: https://alphaone-project.github.io/</p> + https://arxiv.org/abs/2505.24863 + Fri, 30 May 2025 17:58:36 +0000 - RM-R1: Reward Modeling as Reasoning - https://arxiv.org/abs/2505.02387 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.02387.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Xiusi Chen, Gaotang Li, Ziqi Wang, Bowen Jin, Cheng Qian, Yu Wang, Hongru Wang, Yu Zhang, Denghui Zhang, Tong Zhang, Hanghang Tong, Heng Ji</p><p><b>Upvotes:</b> 73</p><p><b>Summary:</b> Reward modeling is essential for aligning large language models (LLMs) with human preferences, especially through reinforcement learning from human feedback (RLHF). To provide accurate reward signals, a reward model (RM) should stimulate deep thinking and conduct interpretable reasoning before assigning a score or a judgment. However, existing RMs either produce opaque scalar scores or directly generate the prediction of a preferred answer, making them struggle to integrate natural language critiques, thus lacking interpretability. Inspired by recent advances of long chain-of-thought (CoT) on reasoning-intensive tasks, we hypothesize and validate that integrating reasoning capabilities into reward modeling significantly enhances RM's interpretability and performance. In this work, we introduce a new class of generative reward models -- Reasoning Reward Models (ReasRMs) -- which formulate reward modeling as a reasoning task. We propose a reasoning-oriented training pipeline and train a family of ReasRMs, RM-R1. The training consists of two key stages: (1) distillation of high-quality reasoning chains and (2) reinforcement learning with verifiable rewards. RM-R1 improves LLM rollouts by self-generating reasoning traces or chat-specific rubrics and evaluating candidate responses against them. Empirically, our models achieve state-of-the-art or near state-of-the-art performance of generative RMs across multiple comprehensive reward model benchmarks, outperforming much larger open-weight models (e.g., Llama3.1-405B) and proprietary ones (e.g., GPT-4o) by up to 13.8%. Beyond final performance, we perform thorough empirical analysis to understand the key ingredients of successful ReasRM training. To facilitate future research, we release six ReasRM models along with code and data at https://github.com/RM-R1-UIUC/RM-R1.</p> - https://arxiv.org/abs/2505.02387 - Mon, 05 May 2025 06:11:12 +0000 + ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language Models + https://arxiv.org/abs/2505.24864 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.24864.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Mingjie Liu, Shizhe Diao, Ximing Lu, Jian Hu, Xin Dong, Yejin Choi, Jan Kautz, Yi Dong</p><p><b>Upvotes:</b> 132</p><p><b>Summary:</b> Recent advances in reasoning-centric language models have highlighted reinforcement learning (RL) as a promising method for aligning models with verifiable rewards. However, it remains contentious whether RL truly expands a model's reasoning capabilities or merely amplifies high-reward outputs already latent in the base model's distribution, and whether continually scaling up RL compute reliably leads to improved reasoning performance. In this work, we challenge prevailing assumptions by demonstrating that prolonged RL (ProRL) training can uncover novel reasoning strategies that are inaccessible to base models, even under extensive sampling. We introduce ProRL, a novel training methodology that incorporates KL divergence control, reference policy resetting, and a diverse suite of tasks. Our empirical analysis reveals that RL-trained models consistently outperform base models across a wide range of pass@k evaluations, including scenarios where base models fail entirely regardless of the number of attempts. We further show that reasoning boundary improvements correlates strongly with task competence of base model and training duration, suggesting that RL can explore and populate new regions of solution space over time. These findings offer new insights into the conditions under which RL meaningfully expands reasoning boundaries in language models and establish a foundation for future work on long-horizon RL for reasoning. We release model weights to support further research: https://huggingface.co/nvidia/Nemotron-Research-Reasoning-Qwen-1.5B</p> + https://arxiv.org/abs/2505.24864 + Fri, 30 May 2025 17:59:01 +0000 - Bielik 11B v2 Technical Report - https://arxiv.org/abs/2505.02410 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.02410.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Krzysztof Ociepa, Łukasz Flis, Krzysztof Wróbel, Adrian Gwoździej, Remigiusz Kinas</p><p><b>Upvotes:</b> 52</p><p><b>Summary:</b> We present Bielik 11B v2, a state-of-the-art language model optimized for Polish text processing. Built on the Mistral 7B v0.2 architecture and scaled to 11B parameters using depth up-scaling, this model demonstrates exceptional performance across Polish language benchmarks while maintaining strong cross-lingual capabilities. We introduce two key technical innovations: Weighted Instruction Cross-Entropy Loss, which optimizes learning across diverse instruction types by assigning quality-based weights to training examples, and Adaptive Learning Rate, which dynamically adjusts based on context length. Comprehensive evaluation across multiple benchmarks demonstrates that Bielik 11B v2 outperforms many larger models, including those with 2-6 times more parameters, and significantly surpasses other specialized Polish language models on tasks ranging from linguistic understanding to complex reasoning. The model's parameter efficiency and extensive quantization options enable deployment across various hardware configurations, advancing Polish language AI capabilities and establishing new benchmarks for resource-efficient language modeling in less-represented languages.</p> - https://arxiv.org/abs/2505.02410 - Mon, 05 May 2025 07:03:41 +0000 + Time Blindness: Why Video-Language Models Can't See What Humans Can? + https://arxiv.org/abs/2505.24867 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.24867.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ujjwal Upadhyay, Mukul Ranjan, Zhiqiang Shen, Mohamed Elhoseiny</p><p><b>Upvotes:</b> 78</p><p><b>Summary:</b> Recent advances in vision-language models (VLMs) have made impressive strides in understanding spatio-temporal relationships in videos. However, when spatial information is obscured, these models struggle to capture purely temporal patterns. We introduce SpookyBench, a benchmark where information is encoded solely in temporal sequences of noise-like frames, mirroring natural phenomena from biological signaling to covert communication. Interestingly, while humans can recognize shapes, text, and patterns in these sequences with over 98% accuracy, state-of-the-art VLMs achieve 0% accuracy. This performance gap highlights a critical limitation: an over-reliance on frame-level spatial features and an inability to extract meaning from temporal cues. Furthermore, when trained in data sets with low spatial signal-to-noise ratios (SNR), temporal understanding of models degrades more rapidly than human perception, especially in tasks requiring fine-grained temporal reasoning. Overcoming this limitation will require novel architectures or training paradigms that decouple spatial dependencies from temporal processing. Our systematic analysis shows that this issue persists across model scales and architectures. We release SpookyBench to catalyze research in temporal pattern recognition and bridge the gap between human and machine video understanding. Dataset and code has been made available on our project website: https://timeblindness.github.io/.</p> + https://arxiv.org/abs/2505.24867 + Fri, 30 May 2025 17:59:12 +0000 - Bielik v3 Small: Technical Report - https://arxiv.org/abs/2505.02550 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.02550.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Krzysztof Ociepa, Łukasz Flis, Remigiusz Kinas, Krzysztof Wróbel, Adrian Gwoździej</p><p><b>Upvotes:</b> 63</p><p><b>Summary:</b> We introduce Bielik v3, a series of parameter-efficient generative text models (1.5B and 4.5B) optimized for Polish language processing. These models demonstrate that smaller, well-optimized architectures can achieve performance comparable to much larger counterparts while requiring substantially fewer computational resources. Our approach incorporates several key innovations: a custom Polish tokenizer (APT4) that significantly improves token efficiency, Weighted Instruction Cross-Entropy Loss to balance learning across instruction types, and Adaptive Learning Rate that dynamically adjusts based on training progress. Trained on a meticulously curated corpus of 292 billion tokens spanning 303 million documents, these models excel across multiple benchmarks, including the Open PL LLM Leaderboard, Complex Polish Text Understanding Benchmark, Polish EQ-Bench, and Polish Medical Leaderboard. The 4.5B parameter model achieves results competitive with models 2-3 times its size, while the 1.5B model delivers strong performance despite its extremely compact profile. These advances establish new benchmarks for parameter-efficient language modeling in less-represented languages, making high-quality Polish language AI more accessible for resource-constrained applications.</p> - https://arxiv.org/abs/2505.02550 - Mon, 05 May 2025 10:39:51 +0000 + Temporal In-Context Fine-Tuning for Versatile Control of Video Diffusion Models + https://arxiv.org/abs/2506.00996 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.00996.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Kinam Kim, Junha Hyung, Jaegul Choo</p><p><b>Upvotes:</b> 36</p><p><b>Summary:</b> Recent advances in text-to-video diffusion models have enabled high-quality video synthesis, but controllable generation remains challenging, particularly under limited data and compute. Existing fine-tuning methods for conditional generation often rely on external encoders or architectural modifications, which demand large datasets and are typically restricted to spatially aligned conditioning, limiting flexibility and scalability. In this work, we introduce Temporal In-Context Fine-Tuning (TIC-FT), an efficient and versatile approach for adapting pretrained video diffusion models to diverse conditional generation tasks. Our key idea is to concatenate condition and target frames along the temporal axis and insert intermediate buffer frames with progressively increasing noise levels. These buffer frames enable smooth transitions, aligning the fine-tuning process with the pretrained model's temporal dynamics. TIC-FT requires no architectural changes and achieves strong performance with as few as 10-30 training samples. We validate our method across a range of tasks, including image-to-video and video-to-video generation, using large-scale base models such as CogVideoX-5B and Wan-14B. Extensive experiments show that TIC-FT outperforms existing baselines in both condition fidelity and visual quality, while remaining highly efficient in both training and inference. For additional results, visit https://kinam0252.github.io/TIC-FT/</p> + https://arxiv.org/abs/2506.00996 + Sun, 01 Jun 2025 12:57:43 +0000 - Unified Multimodal Understanding and Generation Models: Advances, Challenges, and Opportunities - https://arxiv.org/abs/2505.02567 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.02567.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Xinjie Zhang, Jintao Guo, Shanshan Zhao, Minghao Fu, Lunhao Duan, Guo-Hua Wang, Qing-Guo Chen, Zhao Xu, Weihua Luo, Kaifu Zhang</p><p><b>Upvotes:</b> 73</p><p><b>Summary:</b> Recent years have seen remarkable progress in both multimodal understanding models and image generation models. Despite their respective successes, these two domains have evolved independently, leading to distinct architectural paradigms: While autoregressive-based architectures have dominated multimodal understanding, diffusion-based models have become the cornerstone of image generation. Recently, there has been growing interest in developing unified frameworks that integrate these tasks. The emergence of GPT-4o's new capabilities exemplifies this trend, highlighting the potential for unification. However, the architectural differences between the two domains pose significant challenges. To provide a clear overview of current efforts toward unification, we present a comprehensive survey aimed at guiding future research. First, we introduce the foundational concepts and recent advancements in multimodal understanding and text-to-image generation models. Next, we review existing unified models, categorizing them into three main architectural paradigms: diffusion-based, autoregressive-based, and hybrid approaches that fuse autoregressive and diffusion mechanisms. For each category, we analyze the structural designs and innovations introduced by related works. Additionally, we compile datasets and benchmarks tailored for unified models, offering resources for future exploration. Finally, we discuss the key challenges facing this nascent field, including tokenization strategy, cross-modal attention, and data. As this area is still in its early stages, we anticipate rapid advancements and will regularly update this survey. Our goal is to inspire further research and provide a valuable reference for the community. The references associated with this survey are available on GitHub (https://github.com/AIDC-AI/Awesome-Unified-Multimodal-Models).</p> - https://arxiv.org/abs/2505.02567 - Mon, 05 May 2025 11:18:03 +0000 + Taming LLMs by Scaling Learning Rates with Gradient Grouping + https://arxiv.org/abs/2506.01049 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.01049.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Siyuan Li, Juanxi Tian, Zedong Wang, Xin Jin, Zicheng Liu, Wentao Zhang, Dan Xu</p><p><b>Upvotes:</b> 36</p><p><b>Summary:</b> Training large language models (LLMs) poses challenges due to their massive scale and heterogeneous architectures. While adaptive optimizers like AdamW help address gradient variations, they still struggle with efficient and effective parameter-wise learning rate estimation, resulting in training instability, slow convergence, and poor compatibility with parameter-efficient fine-tuning (PEFT) techniques. This work introduces Scaling with Gradient Grouping (SGG), an optimizer wrapper that improves adaptive learning rate estimation by dynamic grouping and group-specific scaling. SGG first groups gradient statistics in each layer into clusters and then applies cluster-specific scaling to calibrate learning rates for each parameter, thus imposing collective group-wise constraints while maintaining precise per-parameter adaptation. Experiments on diverse (M)LLM benchmarks show that SGG integrates seamlessly with existing optimizers, and offers consistent gains and faster convergence over baselines, with various model sizes. Its stability across varying batch sizes and learning rates establishes SGG as a robust choice for LLM optimization.</p> + https://arxiv.org/abs/2506.01049 + Sun, 01 Jun 2025 15:30:37 +0000 - Voila: Voice-Language Foundation Models for Real-Time Autonomous Interaction and Voice Role-Play - https://arxiv.org/abs/2505.02707 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.02707.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yemin Shi, Yu Shu, Siwei Dong, Guangyi Liu, Jaward Sesay, Jingwen Li, Zhiting Hu</p><p><b>Upvotes:</b> 80</p><p><b>Summary:</b> A voice AI agent that blends seamlessly into daily life would interact with humans in an autonomous, real-time, and emotionally expressive manner. Rather than merely reacting to commands, it would continuously listen, reason, and respond proactively, fostering fluid, dynamic, and emotionally resonant interactions. We introduce Voila, a family of large voice-language foundation models that make a step towards this vision. Voila moves beyond traditional pipeline systems by adopting a new end-to-end architecture that enables full-duplex, low-latency conversations while preserving rich vocal nuances such as tone, rhythm, and emotion. It achieves a response latency of just 195 milliseconds, surpassing the average human response time. Its hierarchical multi-scale Transformer integrates the reasoning capabilities of large language models (LLMs) with powerful acoustic modeling, enabling natural, persona-aware voice generation -- where users can simply write text instructions to define the speaker's identity, tone, and other characteristics. Moreover, Voila supports over one million pre-built voices and efficient customization of new ones from brief audio samples as short as 10 seconds. Beyond spoken dialogue, Voila is designed as a unified model for a wide range of voice-based applications, including automatic speech recognition (ASR), Text-to-Speech (TTS), and, with minimal adaptation, multilingual speech translation. Voila is fully open-sourced to support open research and accelerate progress toward next-generation human-machine interactions.</p> - https://arxiv.org/abs/2505.02707 - Mon, 05 May 2025 15:05:01 +0000 + SRPO: Enhancing Multimodal LLM Reasoning via Reflection-Aware Reinforcement Learning + https://arxiv.org/abs/2506.01713 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.01713.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhongwei Wan, Zhihao Dou, Che Liu, Yu Zhang, Dongfei Cui, Qinjian Zhao, Hui Shen, Jing Xiong, Yi Xin, Yifan Jiang, Yangfan He, Mi Zhang, Shen Yan</p><p><b>Upvotes:</b> 46</p><p><b>Summary:</b> Multimodal large language models (MLLMs) have shown promising capabilities in reasoning tasks, yet still struggle with complex problems requiring explicit self-reflection and self-correction, especially compared to their unimodal text-based counterparts. Existing reflection methods are simplistic and struggle to generate meaningful and instructive feedback, as the reasoning ability and knowledge limits of pre-trained models are largely fixed during initial training. To overcome these challenges, we propose Multimodal Self-Reflection enhanced reasoning with Group Relative Policy Optimization (SRPO), a two-stage reflection-aware reinforcement learning (RL) framework explicitly designed to enhance multimodal LLM reasoning. In the first stage, we construct a high-quality, reflection-focused dataset under the guidance of an advanced MLLM, which generates reflections based on initial responses to help the policy model learn both reasoning and self-reflection. In the second stage, we introduce a novel reward mechanism within the GRPO framework that encourages concise and cognitively meaningful reflection while avoiding redundancy. Extensive experiments across multiple multimodal reasoning benchmarks, including MathVista, MathVision, MathVerse, and MMMU-Pro, using Qwen-2.5-VL-7B and Qwen-2.5-VL-32B demonstrate that SRPO significantly outperforms state-of-the-art models, achieving notable improvements in both reasoning accuracy and reflection quality.</p> + https://arxiv.org/abs/2506.01713 + Mon, 02 Jun 2025 14:21:44 +0000 - Unified Multimodal Chain-of-Thought Reward Model through Reinforcement Fine-Tuning - https://arxiv.org/abs/2505.03318 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.03318.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yibin Wang, Zhimin Li, Yuhang Zang, Chunyu Wang, Qinglin Lu, Cheng Jin, Jiaqi Wang</p><p><b>Upvotes:</b> 91</p><p><b>Summary:</b> Recent advances in multimodal Reward Models (RMs) have shown significant promise in delivering reward signals to align vision models with human preferences. However, current RMs are generally restricted to providing direct responses or engaging in shallow reasoning processes with limited depth, often leading to inaccurate reward signals. We posit that incorporating explicit long chains of thought (CoT) into the reward reasoning process can significantly strengthen their reliability and robustness. Furthermore, we believe that once RMs internalize CoT reasoning, their direct response accuracy can also be improved through implicit reasoning capabilities. To this end, this paper proposes UnifiedReward-Think, the first unified multimodal CoT-based reward model, capable of multi-dimensional, step-by-step long-chain reasoning for both visual understanding and generation reward tasks. Specifically, we adopt an exploration-driven reinforcement fine-tuning approach to elicit and incentivize the model's latent complex reasoning ability: (1) We first use a small amount of image generation preference data to distill the reasoning process of GPT-4o, which is then used for the model's cold start to learn the format and structure of CoT reasoning. (2) Subsequently, by leveraging the model's prior knowledge and generalization capabilities, we prepare large-scale unified multimodal preference data to elicit the model's reasoning process across various vision tasks. During this phase, correct reasoning outputs are retained for rejection sampling to refine the model (3) while incorrect predicted samples are finally used for Group Relative Policy Optimization (GRPO) based reinforcement fine-tuning, enabling the model to explore diverse reasoning paths and optimize for correct and robust solutions. Extensive experiments across various vision reward tasks demonstrate the superiority of our model.</p> - https://arxiv.org/abs/2505.03318 - Tue, 06 May 2025 08:46:41 +0000 + SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Robotics + https://arxiv.org/abs/2506.01844 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.01844.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Mustafa Shukor, Dana Aubakirova, Francesco Capuano, Pepijn Kooijmans, Steven Palma, Adil Zouitine, Michel Aractingi, Caroline Pascal, Martino Russi, Andres Marafioti, Simon Alibert, Matthieu Cord, Thomas Wolf, Remi Cadene</p><p><b>Upvotes:</b> 106</p><p><b>Summary:</b> Vision-language models (VLMs) pretrained on large-scale multimodal datasets encode rich visual and linguistic knowledge, making them a strong foundation for robotics. Rather than training robotic policies from scratch, recent approaches adapt VLMs into vision-language-action (VLA) models that enable natural language-driven perception and control. However, existing VLAs are typically massive--often with billions of parameters--leading to high training costs and limited real-world deployability. Moreover, they rely on academic and industrial datasets, overlooking the growing availability of community-collected data from affordable robotic platforms. In this work, we present SmolVLA, a small, efficient, and community-driven VLA that drastically reduces both training and inference costs, while retaining competitive performance. SmolVLA is designed to be trained on a single GPU and deployed on consumer-grade GPUs or even CPUs. To further improve responsiveness, we introduce an asynchronous inference stack decoupling perception and action prediction from action execution, allowing higher control rates with chunked action generation. Despite its compact size, SmolVLA achieves performance comparable to VLAs that are 10x larger. We evaluate SmolVLA on a range of both simulated as well as real-world robotic benchmarks and release all code, pretrained models, and training data.</p> + https://arxiv.org/abs/2506.01844 + Mon, 02 Jun 2025 16:30:19 +0000 - Absolute Zero: Reinforced Self-play Reasoning with Zero Data - https://arxiv.org/abs/2505.03335 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.03335.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Andrew Zhao, Yiran Wu, Yang Yue, Tong Wu, Quentin Xu, Yang Yue, Matthieu Lin, Shenzhi Wang, Qingyun Wu, Zilong Zheng, Gao Huang</p><p><b>Upvotes:</b> 165</p><p><b>Summary:</b> Reinforcement learning with verifiable rewards (RLVR) has shown promise in enhancing the reasoning capabilities of large language models by learning directly from outcome-based rewards. Recent RLVR works that operate under the zero setting avoid supervision in labeling the reasoning process, but still depend on manually curated collections of questions and answers for training. The scarcity of high-quality, human-produced examples raises concerns about the long-term scalability of relying on human supervision, a challenge already evident in the domain of language model pretraining. Furthermore, in a hypothetical future where AI surpasses human intelligence, tasks provided by humans may offer limited learning potential for a superintelligent system. To address these concerns, we propose a new RLVR paradigm called Absolute Zero, in which a single model learns to propose tasks that maximize its own learning progress and improves reasoning by solving them, without relying on any external data. Under this paradigm, we introduce the Absolute Zero Reasoner (AZR), a system that self-evolves its training curriculum and reasoning ability by using a code executor to both validate proposed code reasoning tasks and verify answers, serving as an unified source of verifiable reward to guide open-ended yet grounded learning. Despite being trained entirely without external data, AZR achieves overall SOTA performance on coding and mathematical reasoning tasks, outperforming existing zero-setting models that rely on tens of thousands of in-domain human-curated examples. Furthermore, we demonstrate that AZR can be effectively applied across different model scales and is compatible with various model classes.</p> - https://arxiv.org/abs/2505.03335 - Tue, 06 May 2025 09:08:00 +0000 + SynthRL: Scaling Visual Reasoning with Verifiable Data Synthesis + https://arxiv.org/abs/2506.02096 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.02096.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zijian Wu, Jinjie Ni, Xiangyan Liu, Zichen Liu, Hang Yan, Michael Qizhe Shieh</p><p><b>Upvotes:</b> 51</p><p><b>Summary:</b> Vision-language models (VLMs) trained via reinforcement learning with verifiable reward (RLVR) have shown notable progress in scaling test-time compute effectively. In this work, we investigate how synthesized RL data can further improve RLVR. To this end, we propose SynthRL-a scalable and guaranteed pipeline for automatic data scaling in reasoning-oriented RL training. SynthRL comprises three key stages: (1) selecting seed questions with appropriate distribution, (2) augmenting them into more challenging variants while preserving the original answers, and (3) a guaranteed verification stage that ensures near-perfect correctness and difficulty enhancement. Our empirical experiments demonstrate SynthRL's scalability and effectiveness. When applied to the MMK12 dataset, SynthRL synthesizes over 3.3K additional verifiable, challenging questions from approximately 8K seed samples. Models trained with our synthesized data achieve consistent gains across five out-of-domain visual math reasoning benchmarks, with a significant improvement over baseline models trained on seed data alone. Notably, detailed analysis reveals that the gains are more pronounced on the most challenging evaluation samples, highlighting SynthRL's effectiveness in eliciting deeper and more complex reasoning patterns.</p> + https://arxiv.org/abs/2506.02096 + Mon, 02 Jun 2025 17:45:16 +0000 - DeCLIP: Decoupled Learning for Open-Vocabulary Dense Perception - https://arxiv.org/abs/2505.04410 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.04410.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Junjie Wang, Bin Chen, Yulin Li, Bin Kang, Yichi Chen, Zhuotao Tian</p><p><b>Upvotes:</b> 43</p><p><b>Summary:</b> Dense visual prediction tasks have been constrained by their reliance on predefined categories, limiting their applicability in real-world scenarios where visual concepts are unbounded. While Vision-Language Models (VLMs) like CLIP have shown promise in open-vocabulary tasks, their direct application to dense prediction often leads to suboptimal performance due to limitations in local feature representation. In this work, we present our observation that CLIP's image tokens struggle to effectively aggregate information from spatially or semantically related regions, resulting in features that lack local discriminability and spatial consistency. To address this issue, we propose DeCLIP, a novel framework that enhances CLIP by decoupling the self-attention module to obtain ``content'' and ``context'' features respectively. The ``content'' features are aligned with image crop representations to improve local discriminability, while ``context'' features learn to retain the spatial correlations under the guidance of vision foundation models, such as DINO. Extensive experiments demonstrate that DeCLIP significantly outperforms existing methods across multiple open-vocabulary dense prediction tasks, including object detection and semantic segmentation. Code is available at magenta{https://github.com/xiaomoguhz/DeCLIP}.</p> - https://arxiv.org/abs/2505.04410 - Wed, 07 May 2025 13:46:34 +0000 + Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoning + https://arxiv.org/abs/2506.01939 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.01939.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Shenzhi Wang, Le Yu, Chang Gao, Chujie Zheng, Shixuan Liu, Rui Lu, Kai Dang, Xionghui Chen, Jianxin Yang, Zhenru Zhang, Yuqiong Liu, An Yang, Andrew Zhao, Yang Yue, Shiji Song, Bowen Yu, Gao Huang, Junyang Lin</p><p><b>Upvotes:</b> 165</p><p><b>Summary:</b> Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a powerful approach to enhancing the reasoning capabilities of Large Language Models (LLMs), while its mechanisms are not yet well understood. In this work, we undertake a pioneering exploration of RLVR through the novel perspective of token entropy patterns, comprehensively analyzing how different tokens influence reasoning performance. By examining token entropy patterns in Chain-of-Thought (CoT) reasoning, we observe that only a small fraction of tokens exhibit high entropy, and these tokens act as critical forks that steer the model toward diverse reasoning pathways. Furthermore, studying how entropy patterns evolve during RLVR training reveals that RLVR largely adheres to the base model's entropy patterns, primarily adjusting the entropy of high-entropy tokens. These findings highlight the significance of high-entropy tokens (i.e., forking tokens) to RLVR. We ultimately improve RLVR by restricting policy gradient updates to forking tokens and uncover a finding even beyond the 80/20 rule: utilizing only 20% of the tokens while maintaining performance comparable to full-gradient updates on the Qwen3-8B base model and significantly surpassing full-gradient updates on the Qwen3-32B (+11.04 on AIME'25 and +7.71 on AIME'24) and Qwen3-14B (+4.79 on AIME'25 and +5.21 on AIME'24) base models, highlighting a strong scaling trend. In contrast, training exclusively on the 80% lowest-entropy tokens leads to a marked decline in performance. These findings indicate that the efficacy of RLVR primarily arises from optimizing the high-entropy tokens that decide reasoning directions. Collectively, our results highlight the potential to understand RLVR through a token-entropy perspective and optimize RLVR by leveraging high-entropy minority tokens to further improve LLM reasoning.</p> + https://arxiv.org/abs/2506.01939 + Mon, 02 Jun 2025 17:54:39 +0000 - ZeroSearch: Incentivize the Search Capability of LLMs without Searching - https://arxiv.org/abs/2505.04588 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.04588.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Hao Sun, Zile Qiao, Jiayan Guo, Xuanbo Fan, Yingyan Hou, Yong Jiang, Pengjun Xie, Fei Huang, Yan Zhang</p><p><b>Upvotes:</b> 63</p><p><b>Summary:</b> Effective information searching is essential for enhancing the reasoning and generation capabilities of large language models (LLMs). Recent research has explored using reinforcement learning (RL) to improve LLMs' search capabilities by interacting with live search engines in real-world environments. While these approaches show promising results, they face two major challenges: (1) Uncontrolled Document Quality: The quality of documents returned by search engines is often unpredictable, introducing noise and instability into the training process. (2) Prohibitively High API Costs: RL training requires frequent rollouts, potentially involving hundreds of thousands of search requests, which incur substantial API expenses and severely constrain scalability. To address these challenges, we introduce ZeroSearch, a reinforcement learning framework that incentivizes the search capabilities of LLMs without interacting with real search engines. Our approach begins with lightweight supervised fine-tuning to transform the LLM into a retrieval module capable of generating both relevant and noisy documents in response to a query. During RL training, we employ a curriculum-based rollout strategy that incrementally degrades the quality of generated documents, progressively eliciting the model's reasoning ability by exposing it to increasingly challenging retrieval scenarios. Extensive experiments demonstrate that ZeroSearch effectively incentivizes the search capabilities of LLMs using a 3B LLM as the retrieval module. Remarkably, a 7B retrieval module achieves comparable performance to the real search engine, while a 14B retrieval module even surpasses it. Furthermore, it generalizes well across both base and instruction-tuned models of various parameter sizes and is compatible with a wide range of RL algorithms.</p> - https://arxiv.org/abs/2505.04588 - Wed, 07 May 2025 17:30:22 +0000 + VS-Bench: Evaluating VLMs for Strategic Reasoning and Decision-Making in Multi-Agent Environments + https://arxiv.org/abs/2506.02387 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.02387.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zelai Xu, Zhexuan Xu, Xiangmin Yi, Huining Yuan, Xinlei Chen, Yi Wu, Chao Yu, Yu Wang</p><p><b>Upvotes:</b> 57</p><p><b>Summary:</b> Recent advancements in Vision Language Models (VLMs) have expanded their capabilities to interactive agent tasks, yet existing benchmarks remain limited to single-agent or text-only environments. In contrast, real-world scenarios often involve multiple agents interacting within rich visual and linguistic contexts, posing challenges with both multimodal observations and strategic interactions. To bridge this gap, we introduce Visual Strategic Bench (VS-Bench), a multimodal benchmark that evaluates VLMs for strategic reasoning and decision-making in multi-agent environments. VS-Bench comprises eight vision-grounded environments spanning cooperative, competitive, and mixed-motive interactions, designed to assess agents' ability to predict others' future moves and optimize for long-term objectives. We consider two complementary evaluation dimensions, including offline evaluation of strategic reasoning by next-action prediction accuracy and online evaluation of decision-making by normalized episode return. Extensive experiments of fourteen leading VLMs reveal a significant gap between current models and optimal performance, with the best models attaining 47.8% prediction accuracy and 24.3% normalized return. We further conduct in-depth analyses on multimodal observations, test-time scaling, social behaviors, and failure cases of VLM agents. By standardizing the evaluation and highlighting the limitations of existing models, we envision VS-Bench as a foundation for future research on strategic multimodal agents. Code and data are available at https://vs-bench.github.io.</p> + https://arxiv.org/abs/2506.02387 + Tue, 03 Jun 2025 02:57:38 +0000 - On Path to Multimodal Generalist: General-Level and General-Bench - https://arxiv.org/abs/2505.04620 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.04620.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Hao Fei, Yuan Zhou, Juncheng Li, Xiangtai Li, Qingshan Xu, Bobo Li, Shengqiong Wu, Yaoting Wang, Junbao Zhou, Jiahao Meng, Qingyu Shi, Zhiyuan Zhou, Liangtao Shi, Minghe Gao, Daoan Zhang, Zhiqi Ge, Weiming Wu, Siliang Tang, Kaihang Pan, Yaobo Ye, Haobo Yuan, Tao Zhang, Tianjie Ju, Zixiang Meng, Shilin Xu, Liyu Jia, Wentao Hu, Meng Luo, Jiebo Luo, Tat-Seng Chua, Shuicheng Yan, Hanwang Zhang</p><p><b>Upvotes:</b> 78</p><p><b>Summary:</b> The Multimodal Large Language Model (MLLM) is currently experiencing rapid growth, driven by the advanced capabilities of LLMs. Unlike earlier specialists, existing MLLMs are evolving towards a Multimodal Generalist paradigm. Initially limited to understanding multiple modalities, these models have advanced to not only comprehend but also generate across modalities. Their capabilities have expanded from coarse-grained to fine-grained multimodal understanding and from supporting limited modalities to arbitrary ones. While many benchmarks exist to assess MLLMs, a critical question arises: Can we simply assume that higher performance across tasks indicates a stronger MLLM capability, bringing us closer to human-level AI? We argue that the answer is not as straightforward as it seems. This project introduces General-Level, an evaluation framework that defines 5-scale levels of MLLM performance and generality, offering a methodology to compare MLLMs and gauge the progress of existing systems towards more robust multimodal generalists and, ultimately, towards AGI. At the core of the framework is the concept of Synergy, which measures whether models maintain consistent capabilities across comprehension and generation, and across multiple modalities. To support this evaluation, we present General-Bench, which encompasses a broader spectrum of skills, modalities, formats, and capabilities, including over 700 tasks and 325,800 instances. The evaluation results that involve over 100 existing state-of-the-art MLLMs uncover the capability rankings of generalists, highlighting the challenges in reaching genuine AI. We expect this project to pave the way for future research on next-generation multimodal foundation models, providing a robust infrastructure to accelerate the realization of AGI. Project page: https://generalist.top/</p> - https://arxiv.org/abs/2505.04620 - Wed, 07 May 2025 17:59:32 +0000 + OThink-R1: Intrinsic Fast/Slow Thinking Mode Switching for Over-Reasoning Mitigation + https://arxiv.org/abs/2506.02397 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.02397.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Shengjia Zhang, Junjie Wu, Jiawei Chen, Changwang Zhang, Xingyu Lou, Wangchunshu Zhou, Sheng Zhou, Can Wang, Jun Wang</p><p><b>Upvotes:</b> 36</p><p><b>Summary:</b> Recent advanced large reasoning models (LRMs) leverage extended chain-of-thought (CoT) reasoning to solve complex tasks, achieving state-of-the-art performance. Despite their success, we identify a critical issue: a substantial portion of simple tasks solved by LRMs can also be addressed by non-reasoning LLMs using significantly fewer tokens, indicating the complex reasoning may not always be necessary. To address this, we systematically analyze the reasoning trajectories of LRMs and present a method utilizing identified paradigms and LLM-Judge to classify these trajectories as either Redundant Reasoning or Essential Reasoning. And we introduce OThink-R1, a method that prunes redundant reasoning steps while preserving logical validity. OThink-R1 dynamically employs the non-thinking mode (fast-thinking) for straightforward problems while engaging in deliberate thinking (slow-thinking) for complex problems. Experiments across mathematical and question-answering tasks demonstrate that OThink-R1 reduces reasoning redundancy by almost 23\% on average without compromising accuracy, offering practical guidelines for efficient reasoning models. The code is available at https://github.com/AgenticIR-Lab/OThink-R1.</p> + https://arxiv.org/abs/2506.02397 + Tue, 03 Jun 2025 03:31:30 +0000 - Perception, Reason, Think, and Plan: A Survey on Large Multimodal Reasoning Models - https://arxiv.org/abs/2505.04921 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.04921.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yunxin Li, Zhenyu Liu, Zitao Li, Xuanyu Zhang, Zhenran Xu, Xinyu Chen, Haoyuan Shi, Shenyuan Jiang, Xintong Wang, Jifang Wang, Shouzheng Huang, Xinping Zhao, Borui Jiang, Lanqing Hong, Longyue Wang, Zhuotao Tian, Baoxing Huai, Wenhan Luo, Weihua Luo, Zheng Zhang, Baotian Hu, Min Zhang</p><p><b>Upvotes:</b> 170</p><p><b>Summary:</b> Reasoning lies at the heart of intelligence, shaping the ability to make decisions, draw conclusions, and generalize across domains. In artificial intelligence, as systems increasingly operate in open, uncertain, and multimodal environments, reasoning becomes essential for enabling robust and adaptive behavior. Large Multimodal Reasoning Models (LMRMs) have emerged as a promising paradigm, integrating modalities such as text, images, audio, and video to support complex reasoning capabilities and aiming to achieve comprehensive perception, precise understanding, and deep reasoning. As research advances, multimodal reasoning has rapidly evolved from modular, perception-driven pipelines to unified, language-centric frameworks that offer more coherent cross-modal understanding. While instruction tuning and reinforcement learning have improved model reasoning, significant challenges remain in omni-modal generalization, reasoning depth, and agentic behavior. To address these issues, we present a comprehensive and structured survey of multimodal reasoning research, organized around a four-stage developmental roadmap that reflects the field's shifting design philosophies and emerging capabilities. First, we review early efforts based on task-specific modules, where reasoning was implicitly embedded across stages of representation, alignment, and fusion. Next, we examine recent approaches that unify reasoning into multimodal LLMs, with advances such as Multimodal Chain-of-Thought (MCoT) and multimodal reinforcement learning enabling richer and more structured reasoning chains. Finally, drawing on empirical insights from challenging benchmarks and experimental cases of OpenAI O3 and O4-mini, we discuss the conceptual direction of native large multimodal reasoning models (N-LMRMs), which aim to support scalable, agentic, and adaptive reasoning and planning in complex, real-world environments.</p> - https://arxiv.org/abs/2505.04921 - Thu, 08 May 2025 03:35:23 +0000 + OmniSpatial: Towards Comprehensive Spatial Reasoning Benchmark for Vision Language Models + https://arxiv.org/abs/2506.03135 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.03135.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Mengdi Jia, Zekun Qi, Shaochen Zhang, Wenyao Zhang, Xinqiang Yu, Jiawei He, He Wang, Li Yi</p><p><b>Upvotes:</b> 37</p><p><b>Summary:</b> Spatial reasoning is a key aspect of cognitive psychology and remains a major bottleneck for current vision-language models (VLMs). While extensive research has aimed to evaluate or improve VLMs' understanding of basic spatial relations, such as distinguishing left from right, near from far, and object counting, these tasks represent only the most fundamental level of spatial reasoning. In this work, we introduce OmniSpatial, a comprehensive and challenging benchmark for spatial reasoning, grounded in cognitive psychology. OmniSpatial covers four major categories: dynamic reasoning, complex spatial logic, spatial interaction, and perspective-taking, with 50 fine-grained subcategories. Through Internet data crawling and careful manual annotation, we construct over 1.5K question-answer pairs. Extensive experiments show that both open- and closed-source VLMs, as well as existing reasoning and spatial understanding models, exhibit significant limitations in comprehensive spatial understanding. We further analyze failure cases and propose potential directions for future research.</p> + https://arxiv.org/abs/2506.03135 + Tue, 03 Jun 2025 17:58:29 +0000 - Flow-GRPO: Training Flow Matching Models via Online RL - https://arxiv.org/abs/2505.05470 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.05470.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jie Liu, Gongye Liu, Jiajun Liang, Yangguang Li, Jiaheng Liu, Xintao Wang, Pengfei Wan, Di Zhang, Wanli Ouyang</p><p><b>Upvotes:</b> 76</p><p><b>Summary:</b> We propose Flow-GRPO, the first method integrating online reinforcement learning (RL) into flow matching models. Our approach uses two key strategies: (1) an ODE-to-SDE conversion that transforms a deterministic Ordinary Differential Equation (ODE) into an equivalent Stochastic Differential Equation (SDE) that matches the original model's marginal distribution at all timesteps, enabling statistical sampling for RL exploration; and (2) a Denoising Reduction strategy that reduces training denoising steps while retaining the original inference timestep number, significantly improving sampling efficiency without performance degradation. Empirically, Flow-GRPO is effective across multiple text-to-image tasks. For complex compositions, RL-tuned SD3.5 generates nearly perfect object counts, spatial relations, and fine-grained attributes, boosting GenEval accuracy from 63% to 95%. In visual text rendering, its accuracy improves from 59% to 92%, significantly enhancing text generation. Flow-GRPO also achieves substantial gains in human preference alignment. Notably, little to no reward hacking occurred, meaning rewards did not increase at the cost of image quality or diversity, and both remained stable in our experiments.</p> - https://arxiv.org/abs/2505.05470 - Thu, 08 May 2025 17:58:45 +0000 + GUI-Actor: Coordinate-Free Visual Grounding for GUI Agents + https://arxiv.org/abs/2506.03143 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.03143.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Qianhui Wu, Kanzhi Cheng, Rui Yang, Chaoyun Zhang, Jianwei Yang, Huiqiang Jiang, Jian Mu, Baolin Peng, Bo Qiao, Reuben Tan, Si Qin, Lars Liden, Qingwei Lin, Huan Zhang, Tong Zhang, Jianbing Zhang, Dongmei Zhang, Jianfeng Gao</p><p><b>Upvotes:</b> 48</p><p><b>Summary:</b> One of the principal challenges in building VLM-powered GUI agents is visual grounding, i.e., localizing the appropriate screen region for action execution based on both the visual content and the textual plans. Most existing work formulates this as a text-based coordinate generation task. However, these approaches suffer from several limitations: weak spatial-semantic alignment, inability to handle ambiguous supervision targets, and a mismatch between the dense nature of screen coordinates and the coarse, patch-level granularity of visual features extracted by models like Vision Transformers. In this paper, we propose GUI-Actor, a VLM-based method for coordinate-free GUI grounding. At its core, GUI-Actor introduces an attention-based action head that learns to align a dedicated <ACTOR> token with all relevant visual patch tokens, enabling the model to propose one or more action regions in a single forward pass. In line with this, we further design a grounding verifier to evaluate and select the most plausible action region from the candidates proposed for action execution. Extensive experiments show that GUI-Actor outperforms prior state-of-the-art methods on multiple GUI action grounding benchmarks, with improved generalization to unseen screen resolutions and layouts. Notably, GUI-Actor-7B even surpasses UI-TARS-72B (38.1) on ScreenSpot-Pro, achieving scores of 40.7 with Qwen2-VL and 44.6 with Qwen2.5-VL as backbones. Furthermore, by incorporating the verifier, we find that fine-tuning only the newly introduced action head (~100M parameters for 7B model) while keeping the VLM backbone frozen is sufficient to achieve performance comparable to previous state-of-the-art models, highlighting that GUI-Actor can endow the underlying VLM with effective grounding capabilities without compromising its general-purpose strengths.</p> + https://arxiv.org/abs/2506.03143 + Tue, 03 Jun 2025 17:59:08 +0000 - Seed1.5-VL Technical Report - https://arxiv.org/abs/2505.07062 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.07062.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Dong Guo, Faming Wu, Feida Zhu, Fuxing Leng, Guang Shi, Haobin Chen, Haoqi Fan, Jian Wang, Jianyu Jiang, Jiawei Wang, Jingji Chen, Jingjia Huang, Kang Lei, Liping Yuan, Lishu Luo, Pengfei Liu, Qinghao Ye, Rui Qian, Shen Yan, Shixiong Zhao, Shuai Peng, Shuangye Li, Sihang Yuan, Sijin Wu, Tianheng Cheng, Weiwei Liu, Wenqian Wang, Xianhan Zeng, Xiao Liu, Xiaobo Qin, Xiaohan Ding, Xiaojun Xiao, Xiaoying Zhang, Xuanwei Zhang, Xuehan Xiong, Yanghua Peng, Yangrui Chen, Yanwei Li, Yanxu Hu, Yi Lin, Yiyuan Hu, Yiyuan Zhang, Youbin Wu, Yu Li, Yudong Liu, Yue Ling, Yujia Qin, Zanbo Wang, Zhiwu He, Aoxue Zhang, Bairen Yi, Bencheng Liao, Can Huang, Can Zhang, Chaorui Deng, Chaoyi Deng, Cheng Lin, Cheng Yuan, Chenggang Li, Chenhui Gou, Chenwei Lou, Chengzhi Wei, Chundian Liu, Chunyuan Li, Deyao Zhu, Donghong Zhong, Feng Li, Feng Zhang, Gang Wu, Guodong Li, Guohong Xiao, Haibin Lin, Haihua Yang, Haoming Wang, Heng Ji, Hongxiang Hao, Hui Shen, Huixia Li, Jiahao Li, Jialong Wu, Jianhua Zhu, Jianpeng Jiao, Jiashi Feng, Jiaze Chen, Jianhui Duan, Jihao Liu, Jin Zeng, Jingqun Tang, Jingyu Sun, Joya Chen, Jun Long, Junda Feng, Junfeng Zhan, Junjie Fang, Junting Lu, Kai Hua, Kai Liu, Kai Shen, Kaiyuan Zhang, Ke Shen, Ke Wang, Keyu Pan, Kun Zhang, Kunchang Li, Lanxin Li, Lei Li, Lei Shi, Li Han, Liang Xiang, Liangqiang Chen, Lin Chen, Lin Li, Lin Yan, Liying Chi, Longxiang Liu, Mengfei Du, Mingxuan Wang, Ningxin Pan, Peibin Chen, Pengfei Chen, Pengfei Wu, Qingqing Yuan, Qingyao Shuai, Qiuyan Tao, Renjie Zheng, Renrui Zhang, Ru Zhang, Rui Wang, Rui Yang, Rui Zhao, Shaoqiang Xu, Shihao Liang, Shipeng Yan, Shu Zhong, Shuaishuai Cao, Shuangzhi Wu, Shufan Liu, Shuhan Chang, Songhua Cai, Tenglong Ao, Tianhao Yang, Tingting Zhang, Wanjun Zhong, Wei Jia, Wei Weng, Weihao Yu, Wenhao Huang, Wenjia Zhu, Wenli Yang, Wenzhi Wang, Xiang Long, XiangRui Yin, Xiao Li, Xiaolei Zhu, Xiaoying Jia, Xijin Zhang, Xin Liu, Xinchen Zhang, Xinyu Yang, Xiongcai Luo, Xiuli Chen, Xuantong Zhong, Xuefeng Xiao, Xujing Li, Yan Wu, Yawei Wen, Yifan Du, Yihao Zhang, Yining Ye, Yonghui Wu, Yu Liu, Yu Yue, Yufeng Zhou, Yufeng Yuan, Yuhang Xu, Yuhong Yang, Yun Zhang, Yunhao Fang, Yuntao Li, Yurui Ren, Yuwen Xiong, Zehua Hong, Zehua Wang, Zewei Sun, Zeyu Wang, Zhao Cai, Zhaoyue Zha, Zhecheng An, Zhehui Zhao, Zhengzhuo Xu, Zhipeng Chen, Zhiyong Wu, Zhuofan Zheng, Zihao Wang, Zilong Huang, Ziyu Zhu, Zuquan Song</p><p><b>Upvotes:</b> 140</p><p><b>Summary:</b> We present Seed1.5-VL, a vision-language foundation model designed to advance general-purpose multimodal understanding and reasoning. Seed1.5-VL is composed with a 532M-parameter vision encoder and a Mixture-of-Experts (MoE) LLM of 20B active parameters. Despite its relatively compact architecture, it delivers strong performance across a wide spectrum of public VLM benchmarks and internal evaluation suites, achieving the state-of-the-art performance on 38 out of 60 public benchmarks. Moreover, in agent-centric tasks such as GUI control and gameplay, Seed1.5-VL outperforms leading multimodal systems, including OpenAI CUA and Claude 3.7. Beyond visual and video understanding, it also demonstrates strong reasoning abilities, making it particularly effective for multimodal reasoning challenges such as visual puzzles. We believe these capabilities will empower broader applications across diverse tasks. In this report, we mainly provide a comprehensive review of our experiences in building Seed1.5-VL across model design, data construction, and training at various stages, hoping that this report can inspire further research. Seed1.5-VL is now accessible at https://www.volcengine.com/ (Volcano Engine Model ID: doubao-1-5-thinking-vision-pro-250428)</p> - https://arxiv.org/abs/2505.07062 - Sun, 11 May 2025 17:28:30 +0000 + UniWorld: High-Resolution Semantic Encoders for Unified Visual Understanding and Generation + https://arxiv.org/abs/2506.03147 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.03147.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Bin Lin, Zongjian Li, Xinhua Cheng, Yuwei Niu, Yang Ye, Xianyi He, Shenghai Yuan, Wangbo Yu, Shaodong Wang, Yunyang Ge, Yatian Pang, Li Yuan</p><p><b>Upvotes:</b> 58</p><p><b>Summary:</b> Although existing unified models deliver strong performance on vision-language understanding and text-to-image generation, their models are limited in exploring image perception and manipulation tasks, which are urgently desired by users for wide applications. Recently, OpenAI released their powerful GPT-4o-Image model for comprehensive image perception and manipulation, achieving expressive capability and attracting community interests. By observing the performance of GPT-4o-Image in our carefully constructed experiments, we infer that GPT-4o-Image leverages features extracted by semantic encoders instead of VAE, while VAEs are considered essential components in many image manipulation models. Motivated by such inspiring observations, we present a unified generative framework named UniWorld based on semantic features provided by powerful visual-language models and contrastive semantic encoders. As a result, we build a strong unified model using only 1% amount of BAGEL's data, which consistently outperforms BAGEL on image editing benchmarks. UniWorld also maintains competitive image understanding and generation capabilities, achieving strong performance across multiple image perception tasks. We fully open-source our models, including model weights, training and evaluation scripts, and datasets.</p> + https://arxiv.org/abs/2506.03147 + Tue, 03 Jun 2025 17:59:33 +0000 - Unified Continuous Generative Models - https://arxiv.org/abs/2505.07447 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.07447.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Peng Sun, Yi Jiang, Tao Lin</p><p><b>Upvotes:</b> 41</p><p><b>Summary:</b> Recent advances in continuous generative models, including multi-step approaches like diffusion and flow-matching (typically requiring 8-1000 sampling steps) and few-step methods such as consistency models (typically 1-8 steps), have demonstrated impressive generative performance. However, existing work often treats these approaches as distinct paradigms, resulting in separate training and sampling methodologies. We introduce a unified framework for training, sampling, and analyzing these models. Our implementation, the Unified Continuous Generative Models Trainer and Sampler (UCGM-{T,S}), achieves state-of-the-art (SOTA) performance. For example, on ImageNet 256x256 using a 675M diffusion transformer, UCGM-T trains a multi-step model achieving 1.30 FID in 20 steps and a few-step model reaching 1.42 FID in just 2 steps. Additionally, applying UCGM-S to a pre-trained model (previously 1.26 FID at 250 steps) improves performance to 1.06 FID in only 40 steps. Code is available at: https://github.com/LINs-lab/UCGM.</p> - https://arxiv.org/abs/2505.07447 - Mon, 12 May 2025 11:15:39 +0000 + MiMo-VL Technical Report + https://arxiv.org/abs/2506.03569 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.03569.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Xiaomi LLM-Core Team, Zihao Yue, Zhenru Lin, Yifan Song, Weikun Wang, Shuhuai Ren, Shuhao Gu, Shicheng Li, Peidian Li, Liang Zhao, Lei Li, Kainan Bao, Hao Tian, Hailin Zhang, Gang Wang, Dawei Zhu, Cici, Chenhong He, Bowen Ye, Bowen Shen, Zihan Zhang, Zihan Jiang, Zhixian Zheng, Zhichao Song, Zhenbo Luo, Yue Yu, Yudong Wang, Yuanyuan Tian, Yu Tu, Yihan Yan, Yi Huang, Xu Wang, Xinzhe Xu, Xingchen Song, Xing Zhang, Xing Yong, Xin Zhang, Xiangwei Deng, Wenyu Yang, Wenhan Ma, Weiwei Lv, Weiji Zhuang, Wei Liu, Sirui Deng, Shuo Liu, Shimao Chen, Shihua Yu, Shaohui Liu, Shande Wang, Rui Ma, Qiantong Wang, Peng Wang, Nuo Chen, Menghang Zhu, Kangyang Zhou, Kang Zhou, Kai Fang, Jun Shi, Jinhao Dong, Jiebao Xiao, Jiaming Xu, Huaqiu Liu, Hongshen Xu, Heng Qu, Haochen Zhao, Hanglong Lv, Guoan Wang, Duo Zhang, Dong Zhang, Di Zhang, Chong Ma, Chang Liu, Can Cai, Bingquan Xia</p><p><b>Upvotes:</b> 73</p><p><b>Summary:</b> We open-source MiMo-VL-7B-SFT and MiMo-VL-7B-RL, two powerful vision-language models delivering state-of-the-art performance in both general visual understanding and multimodal reasoning. MiMo-VL-7B-RL outperforms Qwen2.5-VL-7B on 35 out of 40 evaluated tasks, and scores 59.4 on OlympiadBench, surpassing models with up to 78B parameters. For GUI grounding applications, it sets a new standard with 56.1 on OSWorld-G, even outperforming specialized models such as UI-TARS. Our training combines four-stage pre-training (2.4 trillion tokens) with Mixed On-policy Reinforcement Learning (MORL) integrating diverse reward signals. We identify the importance of incorporating high-quality reasoning data with long Chain-of-Thought into pre-training stages, and the benefits of mixed RL despite challenges in simultaneous multi-domain optimization. We also contribute a comprehensive evaluation suite covering 50+ tasks to promote reproducibility and advance the field. The model checkpoints and full evaluation suite are available at https://github.com/XiaomiMiMo/MiMo-VL.</p> + https://arxiv.org/abs/2506.03569 + Wed, 04 Jun 2025 04:32:54 +0000 - MiniMax-Speech: Intrinsic Zero-Shot Text-to-Speech with a Learnable Speaker Encoder - https://arxiv.org/abs/2505.07916 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.07916.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Bowen Zhang, Congchao Guo, Geng Yang, Hang Yu, Haozhe Zhang, Heidi Lei, Jialong Mai, Junjie Yan, Kaiyue Yang, Mingqi Yang, Peikai Huang, Ruiyang Jin, Sitan Jiang, Weihua Cheng, Yawei Li, Yichen Xiao, Yiying Zhou, Yongmao Zhang, Yuan Lu, Yucen He</p><p><b>Upvotes:</b> 119</p><p><b>Summary:</b> We introduce MiniMax-Speech, an autoregressive Transformer-based Text-to-Speech (TTS) model that generates high-quality speech. A key innovation is our learnable speaker encoder, which extracts timbre features from a reference audio without requiring its transcription. This enables MiniMax-Speech to produce highly expressive speech with timbre consistent with the reference in a zero-shot manner, while also supporting one-shot voice cloning with exceptionally high similarity to the reference voice. In addition, the overall quality of the synthesized audio is enhanced through the proposed Flow-VAE. Our model supports 32 languages and demonstrates excellent performance across multiple objective and subjective evaluations metrics. Notably, it achieves state-of-the-art (SOTA) results on objective voice cloning metrics (Word Error Rate and Speaker Similarity) and has secured the top position on the public TTS Arena leaderboard. Another key strength of MiniMax-Speech, granted by the robust and disentangled representations from the speaker encoder, is its extensibility without modifying the base model, enabling various applications such as: arbitrary voice emotion control via LoRA; text to voice (T2V) by synthesizing timbre features directly from text description; and professional voice cloning (PVC) by fine-tuning timbre features with additional data. We encourage readers to visit https://minimax-ai.github.io/tts_tech_report for more examples.</p> - https://arxiv.org/abs/2505.07916 - Mon, 12 May 2025 14:25:20 +0000 + AmbiK: Dataset of Ambiguous Tasks in Kitchen Environment + https://arxiv.org/abs/2506.04089 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.04089.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Anastasiia Ivanova, Eva Bakaeva, Zoya Volovikova, Alexey K. Kovalev, Aleksandr I. Panov</p><p><b>Upvotes:</b> 46</p><p><b>Summary:</b> As a part of an embodied agent, Large Language Models (LLMs) are typically used for behavior planning given natural language instructions from the user. However, dealing with ambiguous instructions in real-world environments remains a challenge for LLMs. Various methods for task ambiguity detection have been proposed. However, it is difficult to compare them because they are tested on different datasets and there is no universal benchmark. For this reason, we propose AmbiK (Ambiguous Tasks in Kitchen Environment), the fully textual dataset of ambiguous instructions addressed to a robot in a kitchen environment. AmbiK was collected with the assistance of LLMs and is human-validated. It comprises 1000 pairs of ambiguous tasks and their unambiguous counterparts, categorized by ambiguity type (Human Preferences, Common Sense Knowledge, Safety), with environment descriptions, clarifying questions and answers, user intents, and task plans, for a total of 2000 tasks. We hope that AmbiK will enable researchers to perform a unified comparison of ambiguity detection methods. AmbiK is available at https://github.com/cog-model/AmbiK-dataset.</p> + https://arxiv.org/abs/2506.04089 + Wed, 04 Jun 2025 15:47:07 +0000 - MiMo: Unlocking the Reasoning Potential of Language Model -- From Pretraining to Posttraining - https://arxiv.org/abs/2505.07608 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.07608.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Xiaomi LLM-Core Team, Bingquan Xia, Bowen Shen, Cici, Dawei Zhu, Di Zhang, Gang Wang, Hailin Zhang, Huaqiu Liu, Jiebao Xiao, Jinhao Dong, Liang Zhao, Peidian Li, Peng Wang, Shihua Yu, Shimao Chen, Weikun Wang, Wenhan Ma, Xiangwei Deng, Yi Huang, Yifan Song, Zihan Jiang, Bowen Ye, Can Cai, Chenhong He, Dong Zhang, Duo Zhang, Guoan Wang, Hao Tian, Haochen Zhao, Heng Qu, Hongshen Xu, Jun Shi, Kainan Bao, QingKai Fang, Kang Zhou, Kangyang Zhou, Lei Li, Menghang Zhu, Nuo Chen, Qiantong Wang, Shaohui Liu, Shicheng Li, Shuhao Gu, Shuhuai Ren, Shuo Liu, Sirui Deng, Weiji Zhuang, Weiwei Lv, Wenyu Yang, Xin Zhang, Xing Yong, Xing Zhang, Xingchen Song, Xinzhe Xu, Xu Wang, Yihan Yan, Yu Tu, Yuanyuan Tian, Yudong Wang, Yue Yu, Zhenru Lin, Zhichao Song, Zihao Yue</p><p><b>Upvotes:</b> 76</p><p><b>Summary:</b> We present MiMo-7B, a large language model born for reasoning tasks, with optimization across both pre-training and post-training stages. During pre-training, we enhance the data preprocessing pipeline and employ a three-stage data mixing strategy to strengthen the base model's reasoning potential. MiMo-7B-Base is pre-trained on 25 trillion tokens, with additional Multi-Token Prediction objective for enhanced performance and accelerated inference speed. During post-training, we curate a dataset of 130K verifiable mathematics and programming problems for reinforcement learning, integrating a test-difficulty-driven code-reward scheme to alleviate sparse-reward issues and employing strategic data resampling to stabilize training. Extensive evaluations show that MiMo-7B-Base possesses exceptional reasoning potential, outperforming even much larger 32B models. The final RL-tuned model, MiMo-7B-RL, achieves superior performance on mathematics, code and general reasoning tasks, surpassing the performance of OpenAI o1-mini. The model checkpoints are available at https://github.com/xiaomimimo/MiMo.</p> - https://arxiv.org/abs/2505.07608 - Mon, 12 May 2025 14:30:11 +0000 + OpenThoughts: Data Recipes for Reasoning Models + https://arxiv.org/abs/2506.04178 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.04178.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Etash Guha, Ryan Marten, Sedrick Keh, Negin Raoof, Georgios Smyrnis, Hritik Bansal, Marianna Nezhurina, Jean Mercat, Trung Vu, Zayne Sprague, Ashima Suvarna, Benjamin Feuer, Liangyu Chen, Zaid Khan, Eric Frankel, Sachin Grover, Caroline Choi, Niklas Muennighoff, Shiye Su, Wanjia Zhao, John Yang, Shreyas Pimpalgaonkar, Kartik Sharma, Charlie Cheng-Jie Ji, Yichuan Deng, Sarah Pratt, Vivek Ramanujan, Jon Saad-Falcon, Jeffrey Li, Achal Dave, Alon Albalak, Kushal Arora, Blake Wulfe, Chinmay Hegde, Greg Durrett, Sewoong Oh, Mohit Bansal, Saadia Gabriel, Aditya Grover, Kai-Wei Chang, Vaishaal Shankar, Aaron Gokaslan, Mike A. Merrill, Tatsunori Hashimoto, Yejin Choi, Jenia Jitsev, Reinhard Heckel, Maheswaran Sathiamoorthy, Alexandros G. Dimakis, Ludwig Schmidt</p><p><b>Upvotes:</b> 41</p><p><b>Summary:</b> Reasoning models have made rapid progress on many benchmarks involving math, code, and science. Yet, there are still many open questions about the best training recipes for reasoning since state-of-the-art models often rely on proprietary datasets with little to no public information available. To address this, the goal of the OpenThoughts project is to create open-source datasets for training reasoning models. After initial explorations, our OpenThoughts2-1M dataset led to OpenThinker2-32B, the first model trained on public reasoning data to match DeepSeek-R1-Distill-32B on standard reasoning benchmarks such as AIME and LiveCodeBench. We then improve our dataset further by systematically investigating each step of our data generation pipeline with 1,000+ controlled experiments, which led to OpenThoughts3. Scaling the pipeline to 1.2M examples and using QwQ-32B as teacher yields our OpenThinker3-7B model, which achieves state-of-the-art results: 53% on AIME 2025, 51% on LiveCodeBench 06/24-01/25, and 54% on GPQA Diamond. All of our datasets and models are available on https://openthoughts.ai.</p> + https://arxiv.org/abs/2506.04178 + Wed, 04 Jun 2025 17:25:39 +0000 - Step1X-3D: Towards High-Fidelity and Controllable Generation of Textured 3D Assets - https://arxiv.org/abs/2505.07747 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.07747.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Weiyu Li, Xuanyang Zhang, Zheng Sun, Di Qi, Hao Li, Wei Cheng, Weiwei Cai, Shihao Wu, Jiarui Liu, Zihao Wang, Xiao Chen, Feipeng Tian, Jianxiong Pan, Zeming Li, Gang Yu, Xiangyu Zhang, Daxin Jiang, Ping Tan</p><p><b>Upvotes:</b> 59</p><p><b>Summary:</b> While generative artificial intelligence has advanced significantly across text, image, audio, and video domains, 3D generation remains comparatively underdeveloped due to fundamental challenges such as data scarcity, algorithmic limitations, and ecosystem fragmentation. To this end, we present Step1X-3D, an open framework addressing these challenges through: (1) a rigorous data curation pipeline processing >5M assets to create a 2M high-quality dataset with standardized geometric and textural properties; (2) a two-stage 3D-native architecture combining a hybrid VAE-DiT geometry generator with an diffusion-based texture synthesis module; and (3) the full open-source release of models, training code, and adaptation modules. For geometry generation, the hybrid VAE-DiT component produces TSDF representations by employing perceiver-based latent encoding with sharp edge sampling for detail preservation. The diffusion-based texture synthesis module then ensures cross-view consistency through geometric conditioning and latent-space synchronization. Benchmark results demonstrate state-of-the-art performance that exceeds existing open-source methods, while also achieving competitive quality with proprietary solutions. Notably, the framework uniquely bridges the 2D and 3D generation paradigms by supporting direct transfer of 2D control techniques~(e.g., LoRA) to 3D synthesis. By simultaneously advancing data quality, algorithmic fidelity, and reproducibility, Step1X-3D aims to establish new standards for open research in controllable 3D asset generation.</p> - https://arxiv.org/abs/2505.07747 - Mon, 12 May 2025 16:56:30 +0000 + SuperWriter: Reflection-Driven Long-Form Generation with Large Language Models + https://arxiv.org/abs/2506.04180 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.04180.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yuhao Wu, Yushi Bai, Zhiqiang Hu, Juanzi Li, Roy Ka-Wei Lee</p><p><b>Upvotes:</b> 32</p><p><b>Summary:</b> Long-form text generation remains a significant challenge for large language models (LLMs), particularly in maintaining coherence, ensuring logical consistency, and preserving text quality as sequence length increases. To address these limitations, we propose SuperWriter-Agent, an agent-based framework designed to enhance the quality and consistency of long-form text generation. SuperWriter-Agent introduces explicit structured thinking-through planning and refinement stages into the generation pipeline, guiding the model to follow a more deliberate and cognitively grounded process akin to that of a professional writer. Based on this framework, we construct a supervised fine-tuning dataset to train a 7B SuperWriter-LM. We further develop a hierarchical Direct Preference Optimization (DPO) procedure that uses Monte Carlo Tree Search (MCTS) to propagate final quality assessments and optimize each generation step accordingly. Empirical results across diverse benchmarks demonstrate that SuperWriter-LM achieves state-of-the-art performance, surpassing even larger-scale baseline models in both automatic evaluation and human evaluation. Furthermore, comprehensive ablation studies demonstrate the effectiveness of hierarchical DPO and underscore the value of incorporating structured thinking steps to improve the quality of long-form text generation.</p> + https://arxiv.org/abs/2506.04180 + Wed, 04 Jun 2025 17:27:42 +0000 - Learning from Peers in Reasoning Models - https://arxiv.org/abs/2505.07787 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.07787.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Tongxu Luo, Wenyu Du, Jiaxi Bi, Stephen Chung, Zhengyang Tang, Hao Yang, Min Zhang, Benyou Wang</p><p><b>Upvotes:</b> 44</p><p><b>Summary:</b> Large Reasoning Models (LRMs) have the ability to self-correct even when they make mistakes in their reasoning paths. However, our study reveals that when the reasoning process starts with a short but poor beginning, it becomes difficult for the model to recover. We refer to this phenomenon as the "Prefix Dominance Trap". Inspired by psychological findings that peer interaction can promote self-correction without negatively impacting already accurate individuals, we propose **Learning from Peers** (LeaP) to address this phenomenon. Specifically, every tokens, each reasoning path summarizes its intermediate reasoning and shares it with others through a routing mechanism, enabling paths to incorporate peer insights during inference. However, we observe that smaller models sometimes fail to follow summarization and reflection instructions effectively. To address this, we fine-tune them into our **LeaP-T** model series. Experiments on AIME 2024, AIME 2025, AIMO 2025, and GPQA Diamond show that LeaP provides substantial improvements. For instance, QwQ-32B with LeaP achieves nearly 5 absolute points higher than the baseline on average, and surpasses DeepSeek-R1-671B on three math benchmarks with an average gain of 3.3 points. Notably, our fine-tuned LeaP-T-7B matches the performance of DeepSeek-R1-Distill-Qwen-14B on AIME 2024. In-depth analysis reveals LeaP's robust error correction by timely peer insights, showing strong error tolerance and handling varied task difficulty. LeaP marks a milestone by enabling LRMs to collaborate during reasoning. Our code, datasets, and models are available at https://learning-from-peers.github.io/ .</p> - https://arxiv.org/abs/2505.07787 - Mon, 12 May 2025 17:39:56 +0000 + Advancing Multimodal Reasoning: From Optimized Cold Start to Staged Reinforcement Learning + https://arxiv.org/abs/2506.04207 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.04207.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Shuang Chen, Yue Guo, Zhaochen Su, Yafu Li, Yulun Wu, Jiacheng Chen, Jiayu Chen, Weijie Wang, Xiaoye Qu, Yu Cheng</p><p><b>Upvotes:</b> 45</p><p><b>Summary:</b> Inspired by the remarkable reasoning capabilities of Deepseek-R1 in complex textual tasks, many works attempt to incentivize similar capabilities in Multimodal Large Language Models (MLLMs) by directly applying reinforcement learning (RL). However, they still struggle to activate complex reasoning. In this paper, rather than examining multimodal RL in isolation, we delve into current training pipelines and identify three crucial phenomena: 1) Effective cold start initialization is critical for enhancing MLLM reasoning. Intriguingly, we find that initializing with carefully selected text data alone can lead to performance surpassing many recent multimodal reasoning models, even before multimodal RL. 2) Standard GRPO applied to multimodal RL suffers from gradient stagnation, which degrades training stability and performance. 3) Subsequent text-only RL training, following the multimodal RL phase, further enhances multimodal reasoning. This staged training approach effectively balances perceptual grounding and cognitive reasoning development. By incorporating the above insights and addressing multimodal RL issues, we introduce ReVisual-R1, achieving a new state-of-the-art among open-source 7B MLLMs on challenging benchmarks including MathVerse, MathVision, WeMath, LogicVista, DynaMath, and challenging AIME2024 and AIME2025.</p> + https://arxiv.org/abs/2506.04207 + Wed, 04 Jun 2025 17:51:08 +0000 - OpenThinkIMG: Learning to Think with Images via Visual Tool Reinforcement Learning - https://arxiv.org/abs/2505.08617 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.08617.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhaochen Su, Linjie Li, Mingyang Song, Yunzhuo Hao, Zhengyuan Yang, Jun Zhang, Guanjie Chen, Jiawei Gu, Juntao Li, Xiaoye Qu, Yu Cheng</p><p><b>Upvotes:</b> 40</p><p><b>Summary:</b> While humans can flexibly leverage interactive visual cognition for complex problem-solving, enabling Large Vision-Language Models (LVLMs) to learn similarly adaptive behaviors with visual tools remains challenging. A significant hurdle is the current lack of standardized infrastructure, which hinders integrating diverse tools, generating rich interaction data, and training robust agents effectively. To address these gaps, we introduce OpenThinkIMG, the first open-source, comprehensive end-to-end framework for tool-augmented LVLMs. It features standardized vision tool interfaces, scalable trajectory generation for policy initialization, and a flexible training environment. Furthermore, considering supervised fine-tuning (SFT) on static demonstrations offers limited policy generalization for dynamic tool invocation, we propose a novel reinforcement learning (RL) framework V-ToolRL to train LVLMs to learn adaptive policies for invoking external vision tools. V-ToolRL enables LVLMs to autonomously discover optimal tool-usage strategies by directly optimizing for task success using feedback from tool interactions. We empirically validate V-ToolRL on challenging chart reasoning tasks. Our RL-trained agent, built upon a Qwen2-VL-2B, significantly outperforms its SFT-initialized counterpart (+28.83 points) and surpasses established supervised tool-learning baselines like Taco and CogCom by an average of +12.7 points. Notably, it also surpasses prominent closed-source models like GPT-4.1 by +8.68 accuracy points. We hope OpenThinkIMG can serve as a foundational framework for advancing dynamic, tool-augmented visual reasoning, helping the community develop AI agents that can genuinely "think with images".</p> - https://arxiv.org/abs/2505.08617 - Tue, 13 May 2025 14:35:51 +0000 + RoboRefer: Towards Spatial Referring with Reasoning in Vision-Language Models for Robotics + https://arxiv.org/abs/2506.04308 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.04308.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Enshen Zhou, Jingkun An, Cheng Chi, Yi Han, Shanyu Rong, Chi Zhang, Pengwei Wang, Zhongyuan Wang, Tiejun Huang, Lu Sheng, Shanghang Zhang</p><p><b>Upvotes:</b> 40</p><p><b>Summary:</b> Spatial referring is a fundamental capability of embodied robots to interact with the 3D physical world. However, even with the powerful pretrained vision language models (VLMs), recent approaches are still not qualified to accurately understand the complex 3D scenes and dynamically reason about the instruction-indicated locations for interaction. To this end, we propose RoboRefer, a 3D-aware VLM that can first achieve precise spatial understanding by integrating a disentangled but dedicated depth encoder via supervised fine-tuning (SFT). Moreover, RoboRefer advances generalized multi-step spatial reasoning via reinforcement fine-tuning (RFT), with metric-sensitive process reward functions tailored for spatial referring tasks. To support SFT and RFT training, we introduce RefSpatial, a large-scale dataset of 20M QA pairs (2x prior), covering 31 spatial relations (vs. 15 prior) and supporting complex reasoning processes (up to 5 steps). In addition, we introduce RefSpatial-Bench, a challenging benchmark filling the gap in evaluating spatial referring with multi-step reasoning. Experiments show that SFT-trained RoboRefer achieves state-of-the-art spatial understanding, with an average success rate of 89.6%. RFT-trained RoboRefer further outperforms all other baselines by a large margin, even surpassing Gemini-2.5-Pro by 17.4% in average accuracy on RefSpatial-Bench. Notably, RoboRefer can be integrated with various control policies to execute long-horizon, dynamic tasks across diverse robots (e,g., UR5, G1 humanoid) in cluttered real-world scenes.</p> + https://arxiv.org/abs/2506.04308 + Wed, 04 Jun 2025 17:59:27 +0000 - Insights into DeepSeek-V3: Scaling Challenges and Reflections on Hardware for AI Architectures - https://arxiv.org/abs/2505.09343 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.09343.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Chenggang Zhao, Chengqi Deng, Chong Ruan, Damai Dai, Huazuo Gao, Jiashi Li, Liyue Zhang, Panpan Huang, Shangyan Zhou, Shirong Ma, Wenfeng Liang, Ying He, Yuqing Wang, Yuxuan Liu, Y. X. Wei</p><p><b>Upvotes:</b> 61</p><p><b>Summary:</b> The rapid scaling of large language models (LLMs) has unveiled critical limitations in current hardware architectures, including constraints in memory capacity, computational efficiency, and interconnection bandwidth. DeepSeek-V3, trained on 2,048 NVIDIA H800 GPUs, demonstrates how hardware-aware model co-design can effectively address these challenges, enabling cost-efficient training and inference at scale. This paper presents an in-depth analysis of the DeepSeek-V3/R1 model architecture and its AI infrastructure, highlighting key innovations such as Multi-head Latent Attention (MLA) for enhanced memory efficiency, Mixture of Experts (MoE) architectures for optimized computation-communication trade-offs, FP8 mixed-precision training to unlock the full potential of hardware capabilities, and a Multi-Plane Network Topology to minimize cluster-level network overhead. Building on the hardware bottlenecks encountered during DeepSeek-V3's development, we engage in a broader discussion with academic and industry peers on potential future hardware directions, including precise low-precision computation units, scale-up and scale-out convergence, and innovations in low-latency communication fabrics. These insights underscore the critical role of hardware and model co-design in meeting the escalating demands of AI workloads, offering a practical blueprint for innovation in next-generation AI systems.</p> - https://arxiv.org/abs/2505.09343 - Wed, 14 May 2025 12:39:03 +0000 + Truth in the Few: High-Value Data Selection for Efficient Multi-Modal Reasoning + https://arxiv.org/abs/2506.04755 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.04755.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Shenshen Li, Kaiyuan Deng, Lei Wang, Hao Yang, Chong Peng, Peng Yan, Fumin Shen, Heng Tao Shen, Xing Xu</p><p><b>Upvotes:</b> 36</p><p><b>Summary:</b> While multi-modal large language models (MLLMs) have made significant progress in complex reasoning tasks via reinforcement learning, it is commonly believed that extensive training data is necessary for improving multi-modal reasoning ability, inevitably leading to data redundancy and substantial computational costs. However, can smaller high-value datasets match or outperform full corpora for multi-modal reasoning in MLLMs? In this work, we challenge this assumption through a key observation: meaningful multi-modal reasoning is triggered by only a sparse subset of training samples, termed cognitive samples, whereas the majority contribute marginally. Building on this insight, we propose a novel data selection paradigm termed Reasoning Activation Potential (RAP), which identifies cognitive samples by estimating each sample's potential to stimulate genuine multi-modal reasoning by two complementary estimators: 1) Causal Discrepancy Estimator (CDE) based on the potential outcome model principle, eliminates samples that overly rely on language priors by comparing outputs between multi-modal and text-only inputs; 2) Attention Confidence Estimator (ACE), which exploits token-level self-attention to discard samples dominated by irrelevant but over-emphasized tokens in intermediate reasoning stages. Moreover, we introduce a Difficulty-aware Replacement Module (DRM) to substitute trivial instances with cognitively challenging ones, thereby ensuring complexity for robust multi-modal reasoning. Experiments on six datasets show that our RAP method consistently achieves superior performance using only 9.3% of the training data, while reducing computational costs by over 43%. Our code is available at https://github.com/Leo-ssl/RAP.</p> + https://arxiv.org/abs/2506.04755 + Thu, 05 Jun 2025 08:40:24 +0000 - Qwen3 Technical Report - https://arxiv.org/abs/2505.09388 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.09388.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, Chujie Zheng, Dayiheng Liu, Fan Zhou, Fei Huang, Feng Hu, Hao Ge, Haoran Wei, Huan Lin, Jialong Tang, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jing Zhou, Jingren Zhou, Junyang Lin, Kai Dang, Keqin Bao, Kexin Yang, Le Yu, Lianghao Deng, Mei Li, Mingfeng Xue, Mingze Li, Pei Zhang, Peng Wang, Qin Zhu, Rui Men, Ruize Gao, Shixuan Liu, Shuang Luo, Tianhao Li, Tianyi Tang, Wenbiao Yin, Xingzhang Ren, Xinyu Wang, Xinyu Zhang, Xuancheng Ren, Yang Fan, Yang Su, Yichang Zhang, Yinger Zhang, Yu Wan, Yuqiong Liu, Zekun Wang, Zeyu Cui, Zhenru Zhang, Zhipeng Zhou, Zihan Qiu</p><p><b>Upvotes:</b> 169</p><p><b>Summary:</b> In this work, we present Qwen3, the latest version of the Qwen model family. Qwen3 comprises a series of large language models (LLMs) designed to advance performance, efficiency, and multilingual capabilities. The Qwen3 series includes models of both dense and Mixture-of-Expert (MoE) architectures, with parameter scales ranging from 0.6 to 235 billion. A key innovation in Qwen3 is the integration of thinking mode (for complex, multi-step reasoning) and non-thinking mode (for rapid, context-driven responses) into a unified framework. This eliminates the need to switch between different models--such as chat-optimized models (e.g., GPT-4o) and dedicated reasoning models (e.g., QwQ-32B)--and enables dynamic mode switching based on user queries or chat templates. Meanwhile, Qwen3 introduces a thinking budget mechanism, allowing users to allocate computational resources adaptively during inference, thereby balancing latency and performance based on task complexity. Moreover, by leveraging the knowledge from the flagship models, we significantly reduce the computational resources required to build smaller-scale models, while ensuring their highly competitive performance. Empirical evaluations demonstrate that Qwen3 achieves state-of-the-art results across diverse benchmarks, including tasks in code generation, mathematical reasoning, agent tasks, etc., competitive against larger MoE models and proprietary models. Compared to its predecessor Qwen2.5, Qwen3 expands multilingual support from 29 to 119 languages and dialects, enhancing global accessibility through improved cross-lingual understanding and generation capabilities. To facilitate reproducibility and community-driven research and development, all Qwen3 models are publicly accessible under Apache 2.0.</p> - https://arxiv.org/abs/2505.09388 - Wed, 14 May 2025 13:41:34 +0000 + ComfyUI-Copilot: An Intelligent Assistant for Automated Workflow Development + https://arxiv.org/abs/2506.05010 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.05010.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhenran Xu, Xue Yang, Yiyu Wang, Qingli Hu, Zijiao Wu, Longyue Wang, Weihua Luo, Kaifu Zhang, Baotian Hu, Min Zhang</p><p><b>Upvotes:</b> 70</p><p><b>Summary:</b> We introduce ComfyUI-Copilot, a large language model-powered plugin designed to enhance the usability and efficiency of ComfyUI, an open-source platform for AI-driven art creation. Despite its flexibility and user-friendly interface, ComfyUI can present challenges to newcomers, including limited documentation, model misconfigurations, and the complexity of workflow design. ComfyUI-Copilot addresses these challenges by offering intelligent node and model recommendations, along with automated one-click workflow construction. At its core, the system employs a hierarchical multi-agent framework comprising a central assistant agent for task delegation and specialized worker agents for different usages, supported by our curated ComfyUI knowledge bases to streamline debugging and deployment. We validate the effectiveness of ComfyUI-Copilot through both offline quantitative evaluations and online user feedback, showing that it accurately recommends nodes and accelerates workflow development. Additionally, use cases illustrate that ComfyUI-Copilot lowers entry barriers for beginners and enhances workflow efficiency for experienced users. The ComfyUI-Copilot installation package and a demo video are available at https://github.com/AIDC-AI/ComfyUI-Copilot.</p> + https://arxiv.org/abs/2506.05010 + Thu, 05 Jun 2025 13:20:50 +0000 - System Prompt Optimization with Meta-Learning - https://arxiv.org/abs/2505.09666 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.09666.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yumin Choi, Jinheon Baek, Sung Ju Hwang</p><p><b>Upvotes:</b> 68</p><p><b>Summary:</b> Large Language Models (LLMs) have shown remarkable capabilities, with optimizing their input prompts playing a pivotal role in maximizing their performance. However, while LLM prompts consist of both the task-agnostic system prompts and task-specific user prompts, existing work on prompt optimization has focused on user prompts specific to individual queries or tasks, and largely overlooked the system prompt that is, once optimized, applicable across different tasks and domains. Motivated by this, we introduce the novel problem of bilevel system prompt optimization, whose objective is to design system prompts that are robust to diverse user prompts and transferable to unseen tasks. To tackle this problem, we then propose a meta-learning framework, which meta-learns the system prompt by optimizing it over various user prompts across multiple datasets, while simultaneously updating the user prompts in an iterative manner to ensure synergy between them. We conduct experiments on 14 unseen datasets spanning 5 different domains, on which we show that our approach produces system prompts that generalize effectively to diverse user prompts. Also, our findings reveal that the optimized system prompt enables rapid adaptation even to unseen tasks, requiring fewer optimization steps for test-time user prompts while achieving improved performance.</p> - https://arxiv.org/abs/2505.09666 - Wed, 14 May 2025 16:46:15 +0000 + Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models + https://arxiv.org/abs/2506.05176 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.05176.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yanzhao Zhang, Mingxin Li, Dingkun Long, Xin Zhang, Huan Lin, Baosong Yang, Pengjun Xie, An Yang, Dayiheng Liu, Junyang Lin, Fei Huang, Jingren Zhou</p><p><b>Upvotes:</b> 62</p><p><b>Summary:</b> In this work, we introduce the Qwen3 Embedding series, a significant advancement over its predecessor, the GTE-Qwen series, in text embedding and reranking capabilities, built upon the Qwen3 foundation models. Leveraging the Qwen3 LLMs' robust capabilities in multilingual text understanding and generation, our innovative multi-stage training pipeline combines large-scale unsupervised pre-training with supervised fine-tuning on high-quality datasets. Effective model merging strategies further ensure the robustness and adaptability of the Qwen3 Embedding series. During the training process, the Qwen3 LLMs serve not only as backbone models but also play a crucial role in synthesizing high-quality, rich, and diverse training data across multiple domains and languages, thus enhancing the training pipeline. The Qwen3 Embedding series offers a spectrum of model sizes (0.6B, 4B, 8B) for both embedding and reranking tasks, addressing diverse deployment scenarios where users can optimize for either efficiency or effectiveness. Empirical evaluations demonstrate that the Qwen3 Embedding series achieves state-of-the-art results across diverse benchmarks. Notably, it excels on the multilingual evaluation benchmark MTEB for text embedding, as well as in various retrieval tasks, including code retrieval, cross-lingual retrieval and multilingual retrieval. To facilitate reproducibility and promote community-driven research and development, the Qwen3 Embedding models are publicly available under the Apache 2.0 license.</p> + https://arxiv.org/abs/2506.05176 + Thu, 05 Jun 2025 15:49:48 +0000 - BLIP3-o: A Family of Fully Open Unified Multimodal Models-Architecture, Training and Dataset - https://arxiv.org/abs/2505.09568 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.09568.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jiuhai Chen, Zhiyang Xu, Xichen Pan, Yushi Hu, Can Qin, Tom Goldstein, Lifu Huang, Tianyi Zhou, Saining Xie, Silvio Savarese, Le Xue, Caiming Xiong, Ran Xu</p><p><b>Upvotes:</b> 85</p><p><b>Summary:</b> Unifying image understanding and generation has gained growing attention in recent research on multimodal models. Although design choices for image understanding have been extensively studied, the optimal model architecture and training recipe for a unified framework with image generation remain underexplored. Motivated by the strong potential of autoregressive and diffusion models for high-quality generation and scalability, we conduct a comprehensive study of their use in unified multimodal settings, with emphasis on image representations, modeling objectives, and training strategies. Grounded in these investigations, we introduce a novel approach that employs a diffusion transformer to generate semantically rich CLIP image features, in contrast to conventional VAE-based representations. This design yields both higher training efficiency and improved generative quality. Furthermore, we demonstrate that a sequential pretraining strategy for unified models-first training on image understanding and subsequently on image generation-offers practical advantages by preserving image understanding capability while developing strong image generation ability. Finally, we carefully curate a high-quality instruction-tuning dataset BLIP3o-60k for image generation by prompting GPT-4o with a diverse set of captions covering various scenes, objects, human gestures, and more. Building on our innovative model design, training recipe, and datasets, we develop BLIP3-o, a suite of state-of-the-art unified multimodal models. BLIP3-o achieves superior performance across most of the popular benchmarks spanning both image understanding and generation tasks. To facilitate future research, we fully open-source our models, including code, model weights, training scripts, and pretraining and instruction tuning datasets.</p> - https://arxiv.org/abs/2505.09568 - Wed, 14 May 2025 17:11:07 +0000 + The Common Pile v0.1: An 8TB Dataset of Public Domain and Openly Licensed Text + https://arxiv.org/abs/2506.05209 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.05209.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Nikhil Kandpal, Brian Lester, Colin Raffel, Sebastian Majstorovic, Stella Biderman, Baber Abbasi, Luca Soldaini, Enrico Shippole, A. Feder Cooper, Aviya Skowron, John Kirchenbauer, Shayne Longpre, Lintang Sutawika, Alon Albalak, Zhenlin Xu, Guilherme Penedo, Loubna Ben Allal, Elie Bakouch, John David Pressman, Honglu Fan, Dashiell Stander, Guangyu Song, Aaron Gokaslan, Tom Goldstein, Brian R. Bartoldson, Bhavya Kailkhura, Tyler Murray</p><p><b>Upvotes:</b> 42</p><p><b>Summary:</b> Large language models (LLMs) are typically trained on enormous quantities of unlicensed text, a practice that has led to scrutiny due to possible intellectual property infringement and ethical concerns. Training LLMs on openly licensed text presents a first step towards addressing these issues, but prior data collection efforts have yielded datasets too small or low-quality to produce performant LLMs. To address this gap, we collect, curate, and release the Common Pile v0.1, an eight terabyte collection of openly licensed text designed for LLM pretraining. The Common Pile comprises content from 30 sources that span diverse domains including research papers, code, books, encyclopedias, educational materials, audio transcripts, and more. Crucially, we validate our efforts by training two 7 billion parameter LLMs on text from the Common Pile: Comma v0.1-1T and Comma v0.1-2T, trained on 1 and 2 trillion tokens respectively. Both models attain competitive performance to LLMs trained on unlicensed text with similar computational budgets, such as Llama 1 and 2 7B. In addition to releasing the Common Pile v0.1 itself, we also release the code used in its creation as well as the training mixture and checkpoints for the Comma v0.1 models.</p> + https://arxiv.org/abs/2506.05209 + Thu, 05 Jun 2025 16:21:30 +0000 - Parallel Scaling Law for Language Models - https://arxiv.org/abs/2505.10475 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.10475.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Mouxiang Chen, Binyuan Hui, Zeyu Cui, Jiaxi Yang, Dayiheng Liu, Jianling Sun, Junyang Lin, Zhongxin Liu</p><p><b>Upvotes:</b> 75</p><p><b>Summary:</b> It is commonly believed that scaling language models should commit a significant space or time cost, by increasing the parameters (parameter scaling) or output tokens (inference-time scaling). We introduce the third and more inference-efficient scaling paradigm: increasing the model's parallel computation during both training and inference time. We apply P diverse and learnable transformations to the input, execute forward passes of the model in parallel, and dynamically aggregate the P outputs. This method, namely parallel scaling (ParScale), scales parallel computation by reusing existing parameters and can be applied to any model structure, optimization procedure, data, or task. We theoretically propose a new scaling law and validate it through large-scale pre-training, which shows that a model with P parallel streams is similar to scaling the parameters by O(log P) while showing superior inference efficiency. For example, ParScale can use up to 22times less memory increase and 6times less latency increase compared to parameter scaling that achieves the same performance improvement. It can also recycle an off-the-shelf pre-trained model into a parallelly scaled one by post-training on a small amount of tokens, further reducing the training budget. The new scaling law we discovered potentially facilitates the deployment of more powerful models in low-resource scenarios, and provides an alternative perspective for the role of computation in machine learning.</p> - https://arxiv.org/abs/2505.10475 - Thu, 15 May 2025 16:24:45 +0000 + Diagonal Batching Unlocks Parallelism in Recurrent Memory Transformers for Long Contexts + https://arxiv.org/abs/2506.05229 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.05229.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Danil Sivtsov, Ivan Rodkin, Gleb Kuzmin, Yuri Kuratov, Ivan Oseledets</p><p><b>Upvotes:</b> 37</p><p><b>Summary:</b> Transformer models struggle with long-context inference due to their quadratic time and linear memory complexity. Recurrent Memory Transformers (RMTs) offer a solution by reducing the asymptotic cost to linear time and constant memory usage. However, their memory update mechanism leads to sequential execution, causing a performance bottleneck. We introduce Diagonal Batching, a scheduling scheme that unlocks parallelism across segments in RMTs while preserving exact recurrence. This approach eliminates the sequential constraint, enabling efficient GPU inference even for single long-context inputs without complex batching and pipelining techniques. Because the technique is purely a run-time computation reordering, existing RMT models adopt it with no retraining. Applied to a LLaMA-1B ARMT model, Diagonal Batching yields a 3.3x speedup over standard full-attention LLaMA-1B and a 1.8x speedup over the sequential RMT implementation on 131,072-token sequences. By removing sequential bottleneck, Diagonal Batching reduces inference cost and latency, thereby strengthening RMTs as a practical solution for real-world, long-context applications.</p> + https://arxiv.org/abs/2506.05229 + Thu, 05 Jun 2025 16:43:48 +0000 - MMLongBench: Benchmarking Long-Context Vision-Language Models Effectively and Thoroughly - https://arxiv.org/abs/2505.10610 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.10610.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhaowei Wang, Wenhao Yu, Xiyu Ren, Jipeng Zhang, Yu Zhao, Rohit Saxena, Liang Cheng, Ginny Wong, Simon See, Pasquale Minervini, Yangqiu Song, Mark Steedman</p><p><b>Upvotes:</b> 52</p><p><b>Summary:</b> The rapid extension of context windows in large vision-language models has given rise to long-context vision-language models (LCVLMs), which are capable of handling hundreds of images with interleaved text tokens in a single forward pass. In this work, we introduce MMLongBench, the first benchmark covering a diverse set of long-context vision-language tasks, to evaluate LCVLMs effectively and thoroughly. MMLongBench is composed of 13,331 examples spanning five different categories of downstream tasks, such as Visual RAG and Many-Shot ICL. It also provides broad coverage of image types, including various natural and synthetic images. To assess the robustness of the models to different input lengths, all examples are delivered at five standardized input lengths (8K-128K tokens) via a cross-modal tokenization scheme that combines vision patches and text tokens. Through a thorough benchmarking of 46 closed-source and open-source LCVLMs, we provide a comprehensive analysis of the current models' vision-language long-context ability. Our results show that: i) performance on a single task is a weak proxy for overall long-context capability; ii) both closed-source and open-source models face challenges in long-context vision-language tasks, indicating substantial room for future improvement; iii) models with stronger reasoning ability tend to exhibit better long-context performance. By offering wide task coverage, various image types, and rigorous length control, MMLongBench provides the missing foundation for diagnosing and advancing the next generation of LCVLMs.</p> - https://arxiv.org/abs/2505.10610 - Thu, 15 May 2025 17:52:54 +0000 + Video World Models with Long-term Spatial Memory + https://arxiv.org/abs/2506.05284 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.05284.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Tong Wu, Shuai Yang, Ryan Po, Yinghao Xu, Ziwei Liu, Dahua Lin, Gordon Wetzstein</p><p><b>Upvotes:</b> 53</p><p><b>Summary:</b> Emerging world models autoregressively generate video frames in response to actions, such as camera movements and text prompts, among other control signals. Due to limited temporal context window sizes, these models often struggle to maintain scene consistency during revisits, leading to severe forgetting of previously generated environments. Inspired by the mechanisms of human memory, we introduce a novel framework to enhancing long-term consistency of video world models through a geometry-grounded long-term spatial memory. Our framework includes mechanisms to store and retrieve information from the long-term spatial memory and we curate custom datasets to train and evaluate world models with explicitly stored 3D memory mechanisms. Our evaluations show improved quality, consistency, and context length compared to relevant baselines, paving the way towards long-term consistent world generation.</p> + https://arxiv.org/abs/2506.05284 + Thu, 05 Jun 2025 17:42:34 +0000 - Beyond 'Aha!': Toward Systematic Meta-Abilities Alignment in Large Reasoning Models - https://arxiv.org/abs/2505.10554 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.10554.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhiyuan Hu, Yibo Wang, Hanze Dong, Yuhui Xu, Amrita Saha, Caiming Xiong, Bryan Hooi, Junnan Li</p><p><b>Upvotes:</b> 114</p><p><b>Summary:</b> Large reasoning models (LRMs) already possess a latent capacity for long chain-of-thought reasoning. Prior work has shown that outcome-based reinforcement learning (RL) can incidentally elicit advanced reasoning behaviors such as self-correction, backtracking, and verification phenomena often referred to as the model's "aha moment". However, the timing and consistency of these emergent behaviors remain unpredictable and uncontrollable, limiting the scalability and reliability of LRMs' reasoning capabilities. To address these limitations, we move beyond reliance on prompts and coincidental "aha moments". Instead, we explicitly align models with three meta-abilities: deduction, induction, and abduction, using automatically generated, self-verifiable tasks. Our three stage-pipeline individual alignment, parameter-space merging, and domain-specific reinforcement learning, boosting performance by over 10\% relative to instruction-tuned baselines. Furthermore, domain-specific RL from the aligned checkpoint yields an additional 2\% average gain in the performance ceiling across math, coding, and science benchmarks, demonstrating that explicit meta-ability alignment offers a scalable and dependable foundation for reasoning. Code is available at: https://github.com/zhiyuanhubj/Meta-Ability-Alignment</p> - https://arxiv.org/abs/2505.10554 - Thu, 15 May 2025 17:58:33 +0000 + SeedVR2: One-Step Video Restoration via Diffusion Adversarial Post-Training + https://arxiv.org/abs/2506.05301 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.05301.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jianyi Wang, Shanchuan Lin, Zhijie Lin, Yuxi Ren, Meng Wei, Zongsheng Yue, Shangchen Zhou, Hao Chen, Yang Zhao, Ceyuan Yang, Xuefeng Xiao, Chen Change Loy, Lu Jiang</p><p><b>Upvotes:</b> 55</p><p><b>Summary:</b> Recent advances in diffusion-based video restoration (VR) demonstrate significant improvement in visual quality, yet yield a prohibitive computational cost during inference. While several distillation-based approaches have exhibited the potential of one-step image restoration, extending existing approaches to VR remains challenging and underexplored, particularly when dealing with high-resolution video in real-world settings. In this work, we propose a one-step diffusion-based VR model, termed as SeedVR2, which performs adversarial VR training against real data. To handle the challenging high-resolution VR within a single step, we introduce several enhancements to both model architecture and training procedures. Specifically, an adaptive window attention mechanism is proposed, where the window size is dynamically adjusted to fit the output resolutions, avoiding window inconsistency observed under high-resolution VR using window attention with a predefined window size. To stabilize and improve the adversarial post-training towards VR, we further verify the effectiveness of a series of losses, including a proposed feature matching loss without significantly sacrificing training efficiency. Extensive experiments show that SeedVR2 can achieve comparable or even better performance compared with existing VR approaches in a single step.</p> + https://arxiv.org/abs/2506.05301 + Thu, 05 Jun 2025 17:51:05 +0000 - MathCoder-VL: Bridging Vision and Code for Enhanced Multimodal Mathematical Reasoning - https://arxiv.org/abs/2505.10557 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.10557.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ke Wang, Junting Pan, Linda Wei, Aojun Zhou, Weikang Shi, Zimu Lu, Han Xiao, Yunqiao Yang, Houxing Ren, Mingjie Zhan, Hongsheng Li</p><p><b>Upvotes:</b> 44</p><p><b>Summary:</b> Natural language image-caption datasets, widely used for training Large Multimodal Models, mainly focus on natural scenarios and overlook the intricate details of mathematical figures that are critical for problem-solving, hindering the advancement of current LMMs in multimodal mathematical reasoning. To this end, we propose leveraging code as supervision for cross-modal alignment, since code inherently encodes all information needed to generate corresponding figures, establishing a precise connection between the two modalities. Specifically, we co-develop our image-to-code model and dataset with model-in-the-loop approach, resulting in an image-to-code model, FigCodifier and ImgCode-8.6M dataset, the largest image-code dataset to date. Furthermore, we utilize FigCodifier to synthesize novel mathematical figures and then construct MM-MathInstruct-3M, a high-quality multimodal math instruction fine-tuning dataset. Finally, we present MathCoder-VL, trained with ImgCode-8.6M for cross-modal alignment and subsequently fine-tuned on MM-MathInstruct-3M for multimodal math problem solving. Our model achieves a new open-source SOTA across all six metrics. Notably, it surpasses GPT-4o and Claude 3.5 Sonnet in the geometry problem-solving subset of MathVista, achieving improvements of 8.9% and 9.2%. The dataset and models will be released at https://github.com/mathllm/MathCoder.</p> - https://arxiv.org/abs/2505.10557 - Thu, 15 May 2025 17:59:21 +0000 + MORSE-500: A Programmatically Controllable Video Benchmark to Stress-Test Multimodal Reasoning + https://arxiv.org/abs/2506.05523 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.05523.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zikui Cai, Andrew Wang, Anirudh Satheesh, Ankit Nakhawa, Hyunwoo Jae, Keenan Powell, Minghui Liu, Neel Jay, Sungbin Oh, Xiyao Wang, Yongyuan Liang, Tom Goldstein, Furong Huang</p><p><b>Upvotes:</b> 33</p><p><b>Summary:</b> Despite rapid advances in vision-language models (VLMs), current benchmarks for multimodal reasoning fall short in three key dimensions. First, they overwhelmingly rely on static images, failing to capture the temporal complexity of real-world environments. Second, they narrowly focus on mathematical problem-solving, neglecting the broader spectrum of reasoning skills -- including abstract, physical, planning, spatial, and temporal capabilities -- required for robust multimodal intelligence. Third, many benchmarks quickly saturate, offering limited headroom for diagnosing failure modes or measuring continued progress. We introduce MORSE-500 (Multimodal Reasoning Stress-test Environment), a video benchmark composed of 500 fully scripted clips with embedded questions spanning six complementary reasoning categories. Each instance is programmatically generated using deterministic Python scripts (via Manim, Matplotlib, MoviePy), generative video models, and curated real footage. This script-driven design allows fine-grained control over visual complexity, distractor density, and temporal dynamics -- enabling difficulty to be scaled systematically as models improve. Unlike static benchmarks that become obsolete once saturated, MORSE-500 is built to evolve: its controllable generation pipeline supports the creation of arbitrarily challenging new instances, making it ideally suited for stress-testing next-generation models. Initial experiments with state-of-the-art systems -- including various Gemini 2.5 Pro and OpenAI o3 which represent the strongest available at the time, alongside strong open-source models -- reveal substantial performance gaps across all categories, with particularly large deficits in abstract and planning tasks. We release the full dataset, generation scripts, and evaluation harness to support transparent, reproducible, and forward-looking multimodal reasoning research.</p> + https://arxiv.org/abs/2506.05523 + Thu, 05 Jun 2025 19:12:45 +0000 - GuardReasoner-VL: Safeguarding VLMs via Reinforced Reasoning - https://arxiv.org/abs/2505.11049 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.11049.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yue Liu, Shengfang Zhai, Mingzhe Du, Yulin Chen, Tri Cao, Hongcheng Gao, Cheng Wang, Xinfeng Li, Kun Wang, Junfeng Fang, Jiaheng Zhang, Bryan Hooi</p><p><b>Upvotes:</b> 58</p><p><b>Summary:</b> To enhance the safety of VLMs, this paper introduces a novel reasoning-based VLM guard model dubbed GuardReasoner-VL. The core idea is to incentivize the guard model to deliberatively reason before making moderation decisions via online RL. First, we construct GuardReasoner-VLTrain, a reasoning corpus with 123K samples and 631K reasoning steps, spanning text, image, and text-image inputs. Then, based on it, we cold-start our model's reasoning ability via SFT. In addition, we further enhance reasoning regarding moderation through online RL. Concretely, to enhance diversity and difficulty of samples, we conduct rejection sampling followed by data augmentation via the proposed safety-aware data concatenation. Besides, we use a dynamic clipping parameter to encourage exploration in early stages and exploitation in later stages. To balance performance and token efficiency, we design a length-aware safety reward that integrates accuracy, format, and token cost. Extensive experiments demonstrate the superiority of our model. Remarkably, it surpasses the runner-up by 19.27% F1 score on average. We release data, code, and models (3B/7B) of GuardReasoner-VL at https://github.com/yueliu1999/GuardReasoner-VL/</p> - https://arxiv.org/abs/2505.11049 - Fri, 16 May 2025 09:46:10 +0000 + Confidence Is All You Need: Few-Shot RL Fine-Tuning of Language Models + https://arxiv.org/abs/2506.06395 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.06395.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Pengyi Li, Matvey Skripkin, Alexander Zubrey, Andrey Kuznetsov, Ivan Oseledets</p><p><b>Upvotes:</b> 124</p><p><b>Summary:</b> Large language models (LLMs) excel at reasoning, yet post-training remains critical for aligning their behavior with task goals. Existing reinforcement learning (RL) methods often depend on costly human annotations or external reward models. We propose Reinforcement Learning via Self-Confidence (RLSC), which uses the model's own confidence as reward signals-eliminating the need for labels, preference models, or reward engineering. Applied to Qwen2.5-Math-7B with only 16 samples per question and 10 or 20 training steps, RLSC improves accuracy by +13.4% on AIME2024, +21.2% on MATH500, +21.7% on Minerva Math, +20.8% on Olympiadbench, and +9.7% on AMC23. RLSC provides a simple, scalable post-training method for inference models, requiring only a small number of samples and unlabelled supervision.</p> + https://arxiv.org/abs/2506.06395 + Thu, 05 Jun 2025 19:55:15 +0000 - Delta Attention: Fast and Accurate Sparse Attention Inference by Delta Correction - https://arxiv.org/abs/2505.11254 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.11254.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jeffrey Willette, Heejun Lee, Sung Ju Hwang</p><p><b>Upvotes:</b> 47</p><p><b>Summary:</b> The attention mechanism of a transformer has a quadratic complexity, leading to high inference costs and latency for long sequences. However, attention matrices are mostly sparse, which implies that many entries may be omitted from computation for efficient inference. Sparse attention inference methods aim to reduce this computational burden; however, they also come with a troublesome performance degradation. We discover that one reason for this degradation is that the sparse calculation induces a distributional shift in the attention outputs. The distributional shift causes decoding-time queries to fail to align well with the appropriate keys from the prefill stage, leading to a drop in performance. We propose a simple, novel, and effective procedure for correcting this distributional shift, bringing the distribution of sparse attention outputs closer to that of quadratic attention. Our method can be applied on top of any sparse attention method, and results in an average 36%pt performance increase, recovering 88% of quadratic attention accuracy on the 131K RULER benchmark when applied on top of sliding window attention with sink tokens while only adding a small overhead. Our method can maintain approximately 98.5% sparsity over full quadratic attention, making our model 32 times faster than Flash Attention 2 when processing 1M token prefills.</p> - https://arxiv.org/abs/2505.11254 - Fri, 16 May 2025 13:48:33 +0000 + PartCrafter: Structured 3D Mesh Generation via Compositional Latent Diffusion Transformers + https://arxiv.org/abs/2506.05573 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.05573.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yuchen Lin, Chenguo Lin, Panwang Pan, Honglei Yan, Yiqiang Feng, Yadong Mu, Katerina Fragkiadaki</p><p><b>Upvotes:</b> 69</p><p><b>Summary:</b> We introduce PartCrafter, the first structured 3D generative model that jointly synthesizes multiple semantically meaningful and geometrically distinct 3D meshes from a single RGB image. Unlike existing methods that either produce monolithic 3D shapes or follow two-stage pipelines, i.e., first segmenting an image and then reconstructing each segment, PartCrafter adopts a unified, compositional generation architecture that does not rely on pre-segmented inputs. Conditioned on a single image, it simultaneously denoises multiple 3D parts, enabling end-to-end part-aware generation of both individual objects and complex multi-object scenes. PartCrafter builds upon a pretrained 3D mesh diffusion transformer (DiT) trained on whole objects, inheriting the pretrained weights, encoder, and decoder, and introduces two key innovations: (1) A compositional latent space, where each 3D part is represented by a set of disentangled latent tokens; (2) A hierarchical attention mechanism that enables structured information flow both within individual parts and across all parts, ensuring global coherence while preserving part-level detail during generation. To support part-level supervision, we curate a new dataset by mining part-level annotations from large-scale 3D object datasets. Experiments show that PartCrafter outperforms existing approaches in generating decomposable 3D meshes, including parts that are not directly visible in input images, demonstrating the strength of part-aware generative priors for 3D understanding and synthesis. Code and training data will be released.</p> + https://arxiv.org/abs/2506.05573 + Thu, 05 Jun 2025 20:30:28 +0000 - Visual Planning: Let's Think Only with Images - https://arxiv.org/abs/2505.11409 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.11409.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yi Xu, Chengzu Li, Han Zhou, Xingchen Wan, Caiqi Zhang, Anna Korhonen, Ivan Vulić</p><p><b>Upvotes:</b> 51</p><p><b>Summary:</b> Recent advancements in Large Language Models (LLMs) and their multimodal extensions (MLLMs) have substantially enhanced machine reasoning across diverse tasks. However, these models predominantly rely on pure text as the medium for both expressing and structuring reasoning, even when visual information is present. In this work, we argue that language may not always be the most natural or effective modality for reasoning, particularly in tasks involving spatial and geometrical information. Motivated by this, we propose a new paradigm, Visual Planning, which enables planning through purely visual representations, independent of text. In this paradigm, planning is executed via sequences of images that encode step-by-step inference in the visual domain, akin to how humans sketch or visualize future actions. We introduce a novel reinforcement learning framework, Visual Planning via Reinforcement Learning (VPRL), empowered by GRPO for post-training large vision models, leading to substantial improvements in planning in a selection of representative visual navigation tasks, FrozenLake, Maze, and MiniBehavior. Our visual planning paradigm outperforms all other planning variants that conduct reasoning in the text-only space. Our results establish Visual Planning as a viable and promising alternative to language-based reasoning, opening new avenues for tasks that benefit from intuitive, image-based inference.</p> - https://arxiv.org/abs/2505.11409 - Fri, 16 May 2025 16:17:22 +0000 + Leveraging Self-Attention for Input-Dependent Soft Prompting in LLMs + https://arxiv.org/abs/2506.05629 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.05629.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ananth Muppidi, Abhilash Nandy, Sambaran Bandyopadhyay</p><p><b>Upvotes:</b> 34</p><p><b>Summary:</b> The performance of large language models in domain-specific tasks necessitates fine-tuning, which is computationally expensive and technically challenging. This paper focuses on parameter-efficient fine-tuning using soft prompting, a promising approach that adapts pre-trained models to downstream tasks by learning a small set of parameters. We propose a novel Input Dependent Soft Prompting technique with a self-Attention Mechanism (ID-SPAM) that generates soft prompts based on the input tokens and attends different tokens with varying importance. Our method is simple and efficient, keeping the number of trainable parameters small. We show the merits of the proposed approach compared to state-of-the-art techniques on various tasks and show the improved zero shot domain transfer capability.</p> + https://arxiv.org/abs/2506.05629 + Thu, 05 Jun 2025 23:13:22 +0000 - SageAttention3: Microscaling FP4 Attention for Inference and An Exploration of 8-Bit Training - https://arxiv.org/abs/2505.11594 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.11594.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jintao Zhang, Jia Wei, Pengle Zhang, Xiaoming Xu, Haofeng Huang, Haoxu Wang, Kai Jiang, Jun Zhu, Jianfei Chen</p><p><b>Upvotes:</b> 67</p><p><b>Summary:</b> The efficiency of attention is important due to its quadratic time complexity. We enhance the efficiency of attention through two key contributions: First, we leverage the new FP4 Tensor Cores in Blackwell GPUs to accelerate attention computation. Our implementation achieves 1038 TOPS on RTX5090, which is a 5x speedup over the fastest FlashAttention on RTX5090. Experiments show that our FP4 attention can accelerate inference of various models in a plug-and-play way. Second, we pioneer low-bit attention to training tasks. Existing low-bit attention works like FlashAttention3 and SageAttention focus only on inference. However, the efficiency of training large models is also important. To explore whether low-bit attention can be effectively applied to training tasks, we design an accurate and efficient 8-bit attention for both forward and backward propagation. Experiments indicate that 8-bit attention achieves lossless performance in fine-tuning tasks but exhibits slower convergence in pretraining tasks. The code will be available at https://github.com/thu-ml/SageAttention.</p> - https://arxiv.org/abs/2505.11594 - Fri, 16 May 2025 18:01:54 +0000 + Saffron-1: Towards an Inference Scaling Paradigm for LLM Safety Assurance + https://arxiv.org/abs/2506.06444 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.06444.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ruizhong Qiu, Gaotang Li, Tianxin Wei, Jingrui He, Hanghang Tong</p><p><b>Upvotes:</b> 72</p><p><b>Summary:</b> Existing safety assurance research has primarily focused on training-phase alignment to instill safe behaviors into LLMs. However, recent studies have exposed these methods' susceptibility to diverse jailbreak attacks. Concurrently, inference scaling has significantly advanced LLM reasoning capabilities but remains unexplored in the context of safety assurance. Addressing this gap, our work pioneers inference scaling for robust and effective LLM safety against emerging threats. We reveal that conventional inference scaling techniques, despite their success in reasoning tasks, perform poorly in safety contexts, even falling short of basic approaches like Best-of-N Sampling. We attribute this inefficiency to a newly identified challenge, the exploration--efficiency dilemma, arising from the high computational overhead associated with frequent process reward model (PRM) evaluations. To overcome this dilemma, we propose SAFFRON, a novel inference scaling paradigm tailored explicitly for safety assurance. Central to our approach is the introduction of a multifurcation reward model (MRM) that significantly reduces the required number of reward model evaluations. To operationalize this paradigm, we further propose: (i) a partial supervision training objective for MRM, (ii) a conservative exploration constraint to prevent out-of-distribution explorations, and (iii) a Trie-based key--value caching strategy that facilitates cache sharing across sequences during tree search. Extensive experiments validate the effectiveness of our method. Additionally, we publicly release our trained multifurcation reward model (Saffron-1) and the accompanying token-level safety reward dataset (Safety4M) to accelerate future research in LLM safety. Our code, model, and data are publicly available at https://github.com/q-rz/saffron , and our project homepage is at https://q-rz.github.io/p/saffron .</p> + https://arxiv.org/abs/2506.06444 + Fri, 06 Jun 2025 18:05:45 +0000 - Chain-of-Model Learning for Language Model - https://arxiv.org/abs/2505.11820 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.11820.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Kaitao Song, Xiaohua Wang, Xu Tan, Huiqiang Jiang, Chengruidong Zhang, Yongliang Shen, Cen LU, Zihao Li, Zifan Song, Caihua Shan, Yansen Wang, Kan Ren, Xiaoqing Zheng, Tao Qin, Yuqing Yang, Dongsheng Li, Lili Qiu</p><p><b>Upvotes:</b> 109</p><p><b>Summary:</b> In this paper, we propose a novel learning paradigm, termed Chain-of-Model (CoM), which incorporates the causal relationship into the hidden states of each layer as a chain style, thereby introducing great scaling efficiency in model training and inference flexibility in deployment. We introduce the concept of Chain-of-Representation (CoR), which formulates the hidden states at each layer as a combination of multiple sub-representations (i.e., chains) at the hidden dimension level. In each layer, each chain from the output representations can only view all of its preceding chains in the input representations. Consequently, the model built upon CoM framework can progressively scale up the model size by increasing the chains based on the previous models (i.e., chains), and offer multiple sub-models at varying sizes for elastic inference by using different chain numbers. Based on this principle, we devise Chain-of-Language-Model (CoLM), which incorporates the idea of CoM into each layer of Transformer architecture. Based on CoLM, we further introduce CoLM-Air by introducing a KV sharing mechanism, that computes all keys and values within the first chain and then shares across all chains. This design demonstrates additional extensibility, such as enabling seamless LM switching, prefilling acceleration and so on. Experimental results demonstrate our CoLM family can achieve comparable performance to the standard Transformer, while simultaneously enabling greater flexiblity, such as progressive scaling to improve training efficiency and offer multiple varying model sizes for elastic inference, paving a a new way toward building language models. Our code will be released in the future at: https://github.com/microsoft/CoLM.</p> - https://arxiv.org/abs/2505.11820 - Sat, 17 May 2025 04:06:12 +0000 + Geopolitical biases in LLMs: what are the "good" and the "bad" countries according to contemporary language models + https://arxiv.org/abs/2506.06751 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.06751.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Mikhail Salnikov, Dmitrii Korzh, Ivan Lazichny, Elvir Karimov, Artyom Iudin, Ivan Oseledets, Oleg Y. Rogov, Alexander Panchenko, Natalia Loukachevitch, Elena Tutubalina</p><p><b>Upvotes:</b> 72</p><p><b>Summary:</b> This paper evaluates geopolitical biases in LLMs with respect to various countries though an analysis of their interpretation of historical events with conflicting national perspectives (USA, UK, USSR, and China). We introduce a novel dataset with neutral event descriptions and contrasting viewpoints from different countries. Our findings show significant geopolitical biases, with models favoring specific national narratives. Additionally, simple debiasing prompts had a limited effect in reducing these biases. Experiments with manipulated participant labels reveal models' sensitivity to attribution, sometimes amplifying biases or recognizing inconsistencies, especially with swapped labels. This work highlights national narrative biases in LLMs, challenges the effectiveness of simple debiasing methods, and offers a framework and dataset for future geopolitical bias research.</p> + https://arxiv.org/abs/2506.06751 + Sat, 07 Jun 2025 10:45:17 +0000 - AdaCoT: Pareto-Optimal Adaptive Chain-of-Thought Triggering via Reinforcement Learning - https://arxiv.org/abs/2505.11896 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.11896.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Chenwei Lou, Zewei Sun, Xinnian Liang, Meng Qu, Wei Shen, Wenqi Wang, Yuntao Li, Qingping Yang, Shuangzhi Wu</p><p><b>Upvotes:</b> 55</p><p><b>Summary:</b> Large Language Models (LLMs) have demonstrated remarkable capabilities but often face challenges with tasks requiring sophisticated reasoning. While Chain-of-Thought (CoT) prompting significantly enhances reasoning, it indiscriminately generates lengthy reasoning steps for all queries, leading to substantial computational costs and inefficiency, especially for simpler inputs. To address this critical issue, we introduce AdaCoT (Adaptive Chain-of-Thought), a novel framework enabling LLMs to adaptively decide when to invoke CoT. AdaCoT framed adaptive reasoning as a Pareto optimization problem that seeks to balance model performance with the costs associated with CoT invocation (both frequency and computational overhead). We propose a reinforcement learning (RL) based method, specifically utilizing Proximal Policy Optimization (PPO), to dynamically control the CoT triggering decision boundary by adjusting penalty coefficients, thereby allowing the model to determine CoT necessity based on implicit query complexity. A key technical contribution is Selective Loss Masking (SLM), designed to counteract decision boundary collapse during multi-stage RL training, ensuring robust and stable adaptive triggering. Experimental results demonstrate that AdaCoT successfully navigates the Pareto frontier, achieving substantial reductions in CoT usage for queries not requiring elaborate reasoning. For instance, on our production traffic testset, AdaCoT reduced CoT triggering rates to as low as 3.18\% and decreased average response tokens by 69.06%, while maintaining high performance on complex tasks.</p> - https://arxiv.org/abs/2505.11896 - Sat, 17 May 2025 08:27:00 +0000 + Lingshu: A Generalist Foundation Model for Unified Multimodal Medical Understanding and Reasoning + https://arxiv.org/abs/2506.07044 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.07044.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> LASA Team, Weiwen Xu, Hou Pong Chan, Long Li, Mahani Aljunied, Ruifeng Yuan, Jianyu Wang, Chenghao Xiao, Guizhen Chen, Chaoqun Liu, Zhaodonghui Li, Yu Sun, Junao Shen, Chaojun Wang, Jie Tan, Deli Zhao, Tingyang Xu, Hao Zhang, Yu Rong</p><p><b>Upvotes:</b> 105</p><p><b>Summary:</b> Multimodal Large Language Models (MLLMs) have demonstrated impressive capabilities in understanding common visual elements, largely due to their large-scale datasets and advanced training strategies. However, their effectiveness in medical applications remains limited due to the inherent discrepancies between data and tasks in medical scenarios and those in the general domain. Concretely, existing medical MLLMs face the following critical limitations: (1) limited coverage of medical knowledge beyond imaging, (2) heightened susceptibility to hallucinations due to suboptimal data curation processes, (3) lack of reasoning capabilities tailored for complex medical scenarios. To address these challenges, we first propose a comprehensive data curation procedure that (1) efficiently acquires rich medical knowledge data not only from medical imaging but also from extensive medical texts and general-domain data; and (2) synthesizes accurate medical captions, visual question answering (VQA), and reasoning samples. As a result, we build a multimodal dataset enriched with extensive medical knowledge. Building on the curated data, we introduce our medical-specialized MLLM: Lingshu. Lingshu undergoes multi-stage training to embed medical expertise and enhance its task-solving capabilities progressively. Besides, we preliminarily explore the potential of applying reinforcement learning with verifiable rewards paradigm to enhance Lingshu's medical reasoning ability. Additionally, we develop MedEvalKit, a unified evaluation framework that consolidates leading multimodal and textual medical benchmarks for standardized, fair, and efficient model assessment. We evaluate the performance of Lingshu on three fundamental medical tasks, multimodal QA, text-based QA, and medical report generation. The results show that Lingshu consistently outperforms the existing open-source multimodal models on most tasks ...</p> + https://arxiv.org/abs/2506.07044 + Sun, 08 Jun 2025 08:47:30 +0000 - Model Merging in Pre-training of Large Language Models - https://arxiv.org/abs/2505.12082 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.12082.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yunshui Li, Yiyuan Ma, Shen Yan, Chaoyi Zhang, Jing Liu, Jianqiao Lu, Ziwen Xu, Mengzhao Chen, Minrui Wang, Shiyi Zhan, Jin Ma, Xunhao Lai, Yao Luo, Xingyan Bin, Hongbin Ren, Mingji Han, Wenhao Hao, Bairen Yi, LingJun Liu, Bole Ma, Xiaoying Jia, Zhou Xun, Liang Xiang, Yonghui Wu</p><p><b>Upvotes:</b> 35</p><p><b>Summary:</b> Model merging has emerged as a promising technique for enhancing large language models, though its application in large-scale pre-training remains relatively unexplored. In this paper, we present a comprehensive investigation of model merging techniques during the pre-training process. Through extensive experiments with both dense and Mixture-of-Experts (MoE) architectures ranging from millions to over 100 billion parameters, we demonstrate that merging checkpoints trained with constant learning rates not only achieves significant performance improvements but also enables accurate prediction of annealing behavior. These improvements lead to both more efficient model development and significantly lower training costs. Our detailed ablation studies on merging strategies and hyperparameters provide new insights into the underlying mechanisms while uncovering novel applications. Through comprehensive experimental analysis, we offer the open-source community practical pre-training guidelines for effective model merging.</p> - https://arxiv.org/abs/2505.12082 - Sat, 17 May 2025 16:53:14 +0000 + SpatialLM: Training Large Language Models for Structured Indoor Modeling + https://arxiv.org/abs/2506.07491 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.07491.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yongsen Mao, Junhao Zhong, Chuan Fang, Jia Zheng, Rui Tang, Hao Zhu, Ping Tan, Zihan Zhou</p><p><b>Upvotes:</b> 38</p><p><b>Summary:</b> SpatialLM is a large language model designed to process 3D point cloud data and generate structured 3D scene understanding outputs. These outputs include architectural elements like walls, doors, windows, and oriented object boxes with their semantic categories. Unlike previous methods which exploit task-specific network designs, our model adheres to the standard multimodal LLM architecture and is fine-tuned directly from open-source LLMs. To train SpatialLM, we collect a large-scale, high-quality synthetic dataset consisting of the point clouds of 12,328 indoor scenes (54,778 rooms) with ground-truth 3D annotations, and conduct a careful study on various modeling and training decisions. On public benchmarks, our model gives state-of-the-art performance in layout estimation and competitive results in 3D object detection. With that, we show a feasible path for enhancing the spatial understanding capabilities of modern LLMs for applications in augmented reality, embodied robotics, and more.</p> + https://arxiv.org/abs/2506.07491 + Mon, 09 Jun 2025 07:10:58 +0000 - Scaling Computer-Use Grounding via User Interface Decomposition and Synthesis - https://arxiv.org/abs/2505.13227 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.13227.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Tianbao Xie, Jiaqi Deng, Xiaochuan Li, Junlin Yang, Haoyuan Wu, Jixuan Chen, Wenjing Hu, Xinyuan Wang, Yuhui Xu, Zekun Wang, Yiheng Xu, Junli Wang, Doyen Sahoo, Tao Yu, Caiming Xiong</p><p><b>Upvotes:</b> 44</p><p><b>Summary:</b> Graphical user interface (GUI) grounding, the ability to map natural language instructions to specific actions on graphical user interfaces, remains a critical bottleneck in computer use agent development. Current benchmarks oversimplify grounding tasks as short referring expressions, failing to capture the complexity of real-world interactions that require software commonsense, layout understanding, and fine-grained manipulation capabilities. To address these limitations, we introduce OSWorld-G, a comprehensive benchmark comprising 564 finely annotated samples across diverse task types including text matching, element recognition, layout understanding, and precise manipulation. Additionally, we synthesize and release the largest computer use grounding dataset Jedi, which contains 4 million examples through multi-perspective decoupling of tasks. Our multi-scale models trained on Jedi demonstrate its effectiveness by outperforming existing approaches on ScreenSpot-v2, ScreenSpot-Pro, and our OSWorld-G. Furthermore, we demonstrate that improved grounding with Jedi directly enhances agentic capabilities of general foundation models on complex computer tasks, improving from 5% to 27% on OSWorld. Through detailed ablation studies, we identify key factors contributing to grounding performance and verify that combining specialized data for different interface elements enables compositional generalization to novel interfaces. All benchmark, data, checkpoints, and code are open-sourced and available at https://osworld-grounding.github.io.</p> - https://arxiv.org/abs/2505.13227 - Mon, 19 May 2025 15:09:23 +0000 + MiniCPM4: Ultra-Efficient LLMs on End Devices + https://arxiv.org/abs/2506.07900 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.07900.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> MiniCPM Team, Chaojun Xiao, Yuxuan Li, Xu Han, Yuzhuo Bai, Jie Cai, Haotian Chen, Wentong Chen, Xin Cong, Ganqu Cui, Ning Ding, Shengdan Fan, Yewei Fang, Zixuan Fu, Wenyu Guan, Yitong Guan, Junshao Guo, Yufeng Han, Bingxiang He, Yuxiang Huang, Cunliang Kong, Qiuzuo Li, Siyuan Li, Wenhao Li, Yanghao Li, Yishan Li, Zhen Li, Dan Liu, Biyuan Lin, Yankai Lin, Xiang Long, Quanyu Lu, Yaxi Lu, Peiyan Luo, Hongya Lyu, Litu Ou, Yinxu Pan, Zekai Qu, Qundong Shi, Zijun Song, Jiayuan Su, Zhou Su, Ao Sun, Xianghui Sun, Peijun Tang, Fangzheng Wang, Feng Wang, Shuo Wang, Yudong Wang, Yesai Wu, Zhenyu Xiao, Jie Xie, Zihao Xie, Yukun Yan, Jiarui Yuan, Kaihuo Zhang, Lei Zhang, Linyue Zhang, Xueren Zhang, Yudi Zhang, Hengyu Zhao, Weilin Zhao, Weilun Zhao, Yuanqian Zhao, Zhi Zheng, Ge Zhou, Jie Zhou, Wei Zhou, Zihan Zhou, Zixuan Zhou, Zhiyuan Liu, Guoyang Zeng, Chao Jia, Dahai Li, Maosong Sun</p><p><b>Upvotes:</b> 81</p><p><b>Summary:</b> This paper introduces MiniCPM4, a highly efficient large language model (LLM) designed explicitly for end-side devices. We achieve this efficiency through systematic innovation in four key dimensions: model architecture, training data, training algorithms, and inference systems. Specifically, in terms of model architecture, we propose InfLLM v2, a trainable sparse attention mechanism that accelerates both prefilling and decoding phases for long-context processing. Regarding training data, we propose UltraClean, an efficient and accurate pre-training data filtering and generation strategy, and UltraChat v2, a comprehensive supervised fine-tuning dataset. These datasets enable satisfactory model performance to be achieved using just 8 trillion training tokens. Regarding training algorithms, we propose ModelTunnel v2 for efficient pre-training strategy search, and improve existing post-training methods by introducing chunk-wise rollout for load-balanced reinforcement learning and data-efficient tenary LLM, BitCPM. Regarding inference systems, we propose CPM.cu that integrates sparse attention, model quantization, and speculative sampling to achieve efficient prefilling and decoding. To meet diverse on-device requirements, MiniCPM4 is available in two versions, with 0.5B and 8B parameters, respectively. Sufficient evaluation results show that MiniCPM4 outperforms open-source models of similar size across multiple benchmarks, highlighting both its efficiency and effectiveness. Notably, MiniCPM4-8B demonstrates significant speed improvements over Qwen3-8B when processing long sequences. Through further adaptation, MiniCPM4 successfully powers diverse applications, including trustworthy survey generation and tool use with model context protocol, clearly showcasing its broad usability.</p> + https://arxiv.org/abs/2506.07900 + Mon, 09 Jun 2025 16:16:50 +0000 - Thinkless: LLM Learns When to Think - https://arxiv.org/abs/2505.13379 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.13379.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Gongfan Fang, Xinyin Ma, Xinchao Wang</p><p><b>Upvotes:</b> 47</p><p><b>Summary:</b> Reasoning Language Models, capable of extended chain-of-thought reasoning, have demonstrated remarkable performance on tasks requiring complex logical inference. However, applying elaborate reasoning for all queries often results in substantial computational inefficiencies, particularly when many problems admit straightforward solutions. This motivates an open question: Can LLMs learn when to think? To answer this, we propose Thinkless, a learnable framework that empowers an LLM to adaptively select between short-form and long-form reasoning, based on both task complexity and the model's ability. Thinkless is trained under a reinforcement learning paradigm and employs two control tokens, <short> for concise responses and <think> for detailed reasoning. At the core of our method is a Decoupled Group Relative Policy Optimization (DeGRPO) algorithm, which decomposes the learning objective of hybrid reasoning into two components: (1) a control token loss that governs the selection of the reasoning mode, and (2) a response loss that improves the accuracy of the generated answers. This decoupled formulation enables fine-grained control over the contributions of each objective, stabilizing training and effectively preventing collapse observed in vanilla GRPO. Empirically, on several benchmarks such as Minerva Algebra, MATH-500, and GSM8K, Thinkless is able to reduce the usage of long-chain thinking by 50% - 90%, significantly improving the efficiency of Reasoning Language Models. The code is available at https://github.com/VainF/Thinkless</p> - https://arxiv.org/abs/2505.13379 - Mon, 19 May 2025 17:24:16 +0000 + OneIG-Bench: Omni-dimensional Nuanced Evaluation for Image Generation + https://arxiv.org/abs/2506.07977 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.07977.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jingjing Chang, Yixiao Fang, Peng Xing, Shuhan Wu, Wei Cheng, Rui Wang, Xianfang Zeng, Gang Yu, Hai-Bao Chen</p><p><b>Upvotes:</b> 39</p><p><b>Summary:</b> Text-to-image (T2I) models have garnered significant attention for generating high-quality images aligned with text prompts. However, rapid T2I model advancements reveal limitations in early benchmarks, lacking comprehensive evaluations, for example, the evaluation on reasoning, text rendering and style. Notably, recent state-of-the-art models, with their rich knowledge modeling capabilities, show promising results on the image generation problems requiring strong reasoning ability, yet existing evaluation systems have not adequately addressed this frontier. To systematically address these gaps, we introduce OneIG-Bench, a meticulously designed comprehensive benchmark framework for fine-grained evaluation of T2I models across multiple dimensions, including prompt-image alignment, text rendering precision, reasoning-generated content, stylization, and diversity. By structuring the evaluation, this benchmark enables in-depth analysis of model performance, helping researchers and practitioners pinpoint strengths and bottlenecks in the full pipeline of image generation. Specifically, OneIG-Bench enables flexible evaluation by allowing users to focus on a particular evaluation subset. Instead of generating images for the entire set of prompts, users can generate images only for the prompts associated with the selected dimension and complete the corresponding evaluation accordingly. Our codebase and dataset are now publicly available to facilitate reproducible evaluation studies and cross-model comparisons within the T2I research community.</p> + https://arxiv.org/abs/2506.07977 + Mon, 09 Jun 2025 17:50:21 +0000 - AdaptThink: Reasoning Models Can Learn When to Think - https://arxiv.org/abs/2505.13417 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.13417.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jiajie Zhang, Nianyi Lin, Lei Hou, Ling Feng, Juanzi Li</p><p><b>Upvotes:</b> 74</p><p><b>Summary:</b> Recently, large reasoning models have achieved impressive performance on various tasks by employing human-like deep thinking. However, the lengthy thinking process substantially increases inference overhead, making efficiency a critical bottleneck. In this work, we first demonstrate that NoThinking, which prompts the reasoning model to skip thinking and directly generate the final solution, is a better choice for relatively simple tasks in terms of both performance and efficiency. Motivated by this, we propose AdaptThink, a novel RL algorithm to teach reasoning models to choose the optimal thinking mode adaptively based on problem difficulty. Specifically, AdaptThink features two core components: (1) a constrained optimization objective that encourages the model to choose NoThinking while maintaining the overall performance; (2) an importance sampling strategy that balances Thinking and NoThinking samples during on-policy training, thereby enabling cold start and allowing the model to explore and exploit both thinking modes throughout the training process. Our experiments indicate that AdaptThink significantly reduces the inference costs while further enhancing performance. Notably, on three math datasets, AdaptThink reduces the average response length of DeepSeek-R1-Distill-Qwen-1.5B by 53% and improves its accuracy by 2.4%, highlighting the promise of adaptive thinking-mode selection for optimizing the balance between reasoning quality and efficiency. Our codes and models are available at https://github.com/THU-KEG/AdaptThink.</p> - https://arxiv.org/abs/2505.13417 - Mon, 19 May 2025 17:50:52 +0000 + Reinforcement Pre-Training + https://arxiv.org/abs/2506.08007 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.08007.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Qingxiu Dong, Li Dong, Yao Tang, Tianzhu Ye, Yutao Sun, Zhifang Sui, Furu Wei</p><p><b>Upvotes:</b> 236</p><p><b>Summary:</b> In this work, we introduce Reinforcement Pre-Training (RPT) as a new scaling paradigm for large language models and reinforcement learning (RL). Specifically, we reframe next-token prediction as a reasoning task trained using RL, where it receives verifiable rewards for correctly predicting the next token for a given context. RPT offers a scalable method to leverage vast amounts of text data for general-purpose RL, rather than relying on domain-specific annotated answers. By incentivizing the capability of next-token reasoning, RPT significantly improves the language modeling accuracy of predicting the next tokens. Moreover, RPT provides a strong pre-trained foundation for further reinforcement fine-tuning. The scaling curves show that increased training compute consistently improves the next-token prediction accuracy. The results position RPT as an effective and promising scaling paradigm to advance language model pre-training.</p> + https://arxiv.org/abs/2506.08007 + Mon, 09 Jun 2025 17:59:53 +0000 - Optimizing Anytime Reasoning via Budget Relative Policy Optimization - https://arxiv.org/abs/2505.13438 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.13438.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Penghui Qi, Zichen Liu, Tianyu Pang, Chao Du, Wee Sun Lee, Min Lin</p><p><b>Upvotes:</b> 35</p><p><b>Summary:</b> Scaling test-time compute is crucial for enhancing the reasoning capabilities of large language models (LLMs). Existing approaches typically employ reinforcement learning (RL) to maximize a verifiable reward obtained at the end of reasoning traces. However, such methods optimize only the final performance under a large and fixed token budget, which hinders efficiency in both training and deployment. In this work, we present a novel framework, AnytimeReasoner, to optimize anytime reasoning performance, which aims to improve token efficiency and the flexibility of reasoning under varying token budget constraints. To achieve this, we truncate the complete thinking process to fit within sampled token budgets from a prior distribution, compelling the model to summarize the optimal answer for each truncated thinking for verification. This introduces verifiable dense rewards into the reasoning process, facilitating more effective credit assignment in RL optimization. We then optimize the thinking and summary policies in a decoupled manner to maximize the cumulative reward. Additionally, we introduce a novel variance reduction technique, Budget Relative Policy Optimization (BRPO), to enhance the robustness and efficiency of the learning process when reinforcing the thinking policy. Empirical results in mathematical reasoning tasks demonstrate that our method consistently outperforms GRPO across all thinking budgets under various prior distributions, enhancing both training and token efficiency.</p> - https://arxiv.org/abs/2505.13438 - Mon, 19 May 2025 17:58:44 +0000 + Wait, We Don't Need to "Wait"! Removing Thinking Tokens Improves Reasoning Efficiency + https://arxiv.org/abs/2506.08343 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.08343.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Chenlong Wang, Yuanning Feng, Dongping Chen, Zhaoyang Chu, Ranjay Krishna, Tianyi Zhou</p><p><b>Upvotes:</b> 46</p><p><b>Summary:</b> Recent advances in large reasoning models have enabled complex, step-by-step reasoning but often introduce significant overthinking, resulting in verbose and redundant outputs that hinder efficiency. In this study, we examine whether explicit self-reflection, signaled by tokens such as "Wait" and "Hmm", is necessary for advanced reasoning. We propose NoWait, a simple yet effective approach that disables explicit self-reflection by suppressing these tokens during inference. Extensive experiments on ten benchmarks across textual, visual, and video reasoning tasks show that NoWait reduces chain-of-thought trajectory length by up to 27%-51% in five R1-style model series, without compromising model utility. NoWait thus offers a plug-and-play solution for efficient and utility-preserving multimodal reasoning.</p> + https://arxiv.org/abs/2506.08343 + Tue, 10 Jun 2025 01:54:04 +0000 - Efficient Agent Training for Computer Use - https://arxiv.org/abs/2505.13909 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.13909.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yanheng He, Jiahe Jin, Pengfei Liu</p><p><b>Upvotes:</b> 42</p><p><b>Summary:</b> Scaling up high-quality trajectory data has long been a critical bottleneck for developing human-like computer use agents. We introduce PC Agent-E, an efficient agent training framework that significantly reduces reliance on large-scale human demonstrations. Starting with just 312 human-annotated computer use trajectories, we further improved data quality by synthesizing diverse action decisions with Claude 3.7 Sonnet. Trained on these enriched trajectories, our PC Agent-E model achieved a remarkable 141% relative improvement, surpassing the strong Claude 3.7 Sonnet with extended thinking on WindowsAgentArena-V2, an improved benchmark we also released. Furthermore, PC Agent-E demonstrates strong generalizability to different operating systems on OSWorld. Our findings suggest that strong computer use capabilities can be stimulated from a small amount of high-quality trajectory data.</p> - https://arxiv.org/abs/2505.13909 - Tue, 20 May 2025 04:20:18 +0000 + Seedance 1.0: Exploring the Boundaries of Video Generation Models + https://arxiv.org/abs/2506.09113 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.09113.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yu Gao, Haoyuan Guo, Tuyen Hoang, Weilin Huang, Lu Jiang, Fangyuan Kong, Huixia Li, Jiashi Li, Liang Li, Xiaojie Li, Xunsong Li, Yifu Li, Shanchuan Lin, Zhijie Lin, Jiawei Liu, Shu Liu, Xiaonan Nie, Zhiwu Qing, Yuxi Ren, Li Sun, Zhi Tian, Rui Wang, Sen Wang, Guoqiang Wei, Guohong Wu, Jie Wu, Ruiqi Xia, Fei Xiao, Xuefeng Xiao, Jiangqiao Yan, Ceyuan Yang, Jianchao Yang, Runkai Yang, Tao Yang, Yihang Yang, Zilyu Ye, Xuejiao Zeng, Yan Zeng, Heng Zhang, Yang Zhao, Xiaozheng Zheng, Peihao Zhu, Jiaxin Zou, Feilong Zuo</p><p><b>Upvotes:</b> 90</p><p><b>Summary:</b> Notable breakthroughs in diffusion modeling have propelled rapid improvements in video generation, yet current foundational model still face critical challenges in simultaneously balancing prompt following, motion plausibility, and visual quality. In this report, we introduce Seedance 1.0, a high-performance and inference-efficient video foundation generation model that integrates several core technical improvements: (i) multi-source data curation augmented with precision and meaningful video captioning, enabling comprehensive learning across diverse scenarios; (ii) an efficient architecture design with proposed training paradigm, which allows for natively supporting multi-shot generation and jointly learning of both text-to-video and image-to-video tasks. (iii) carefully-optimized post-training approaches leveraging fine-grained supervised fine-tuning, and video-specific RLHF with multi-dimensional reward mechanisms for comprehensive performance improvements; (iv) excellent model acceleration achieving ~10x inference speedup through multi-stage distillation strategies and system-level optimizations. Seedance 1.0 can generate a 5-second video at 1080p resolution only with 41.4 seconds (NVIDIA-L20). Compared to state-of-the-art video generation models, Seedance 1.0 stands out with high-quality and fast video generation having superior spatiotemporal fluidity with structural stability, precise instruction adherence in complex multi-subject contexts, native multi-shot narrative coherence with consistent subject representation.</p> + https://arxiv.org/abs/2506.09113 + Tue, 10 Jun 2025 17:56:11 +0000 - UniVG-R1: Reasoning Guided Universal Visual Grounding with Reinforcement Learning - https://arxiv.org/abs/2505.14231 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.14231.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Sule Bai, Mingxing Li, Yong Liu, Jing Tang, Haoji Zhang, Lei Sun, Xiangxiang Chu, Yansong Tang</p><p><b>Upvotes:</b> 50</p><p><b>Summary:</b> Traditional visual grounding methods primarily focus on single-image scenarios with simple textual references. However, extending these methods to real-world scenarios that involve implicit and complex instructions, particularly in conjunction with multiple images, poses significant challenges, which is mainly due to the lack of advanced reasoning ability across diverse multi-modal contexts. In this work, we aim to address the more practical universal grounding task, and propose UniVG-R1, a reasoning guided multimodal large language model (MLLM) for universal visual grounding, which enhances reasoning capabilities through reinforcement learning (RL) combined with cold-start data. Specifically, we first construct a high-quality Chain-of-Thought (CoT) grounding dataset, annotated with detailed reasoning chains, to guide the model towards correct reasoning paths via supervised fine-tuning. Subsequently, we perform rule-based reinforcement learning to encourage the model to identify correct reasoning chains, thereby incentivizing its reasoning capabilities. In addition, we identify a difficulty bias arising from the prevalence of easy samples as RL training progresses, and we propose a difficulty-aware weight adjustment strategy to further strengthen the performance. Experimental results demonstrate the effectiveness of UniVG-R1, which achieves state-of-the-art performance on MIG-Bench with a 9.1% improvement over the previous method. Furthermore, our model exhibits strong generalizability, achieving an average improvement of 23.4% in zero-shot performance across four image and video reasoning grounding benchmarks. The project page can be accessed at https://amap-ml.github.io/UniVG-R1-page/.</p> - https://arxiv.org/abs/2505.14231 - Tue, 20 May 2025 11:40:43 +0000 + Autoregressive Semantic Visual Reconstruction Helps VLMs Understand Better + https://arxiv.org/abs/2506.09040 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.09040.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Dianyi Wang, Wei Song, Yikun Wang, Siyuan Wang, Kaicheng Yu, Zhongyu Wei, Jiaqi Wang</p><p><b>Upvotes:</b> 35</p><p><b>Summary:</b> Typical large vision-language models (LVLMs) apply autoregressive supervision solely to textual sequences, without fully incorporating the visual modality into the learning process. This results in three key limitations: (1) an inability to utilize images without accompanying captions, (2) the risk that captions omit critical visual details, and (3) the challenge that certain vision-centric content cannot be adequately conveyed through text. As a result, current LVLMs often prioritize vision-to-language alignment while potentially overlooking fine-grained visual information. While some prior works have explored autoregressive image generation, effectively leveraging autoregressive visual supervision to enhance image understanding remains an open challenge. In this paper, we introduce Autoregressive Semantic Visual Reconstruction (ASVR), which enables joint learning of visual and textual modalities within a unified autoregressive framework. We show that autoregressively reconstructing the raw visual appearance of images does not enhance and may even impair multimodal understanding. In contrast, autoregressively reconstructing the semantic representation of images consistently improves comprehension. Notably, we find that even when models are given continuous image features as input, they can effectively reconstruct discrete semantic tokens, resulting in stable and consistent improvements across a wide range of multimodal understanding benchmarks. Our approach delivers significant performance gains across varying data scales (556k-2M) and types of LLM bacbones. Specifically, ASVR improves LLaVA-1.5 by 5% in average scores across 14 multimodal benchmarks. The code is available at https://github.com/AlenjandroWang/ASVR.</p> + https://arxiv.org/abs/2506.09040 + Tue, 10 Jun 2025 17:57:50 +0000 - Scaling Law for Quantization-Aware Training - https://arxiv.org/abs/2505.14302 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.14302.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Mengzhao Chen, Chaoyi Zhang, Jing Liu, Yutao Zeng, Zeyue Xue, Zhiheng Liu, Yunshui Li, Jin Ma, Jie Huang, Xun Zhou, Ping Luo</p><p><b>Upvotes:</b> 70</p><p><b>Summary:</b> Large language models (LLMs) demand substantial computational and memory resources, creating deployment challenges. Quantization-aware training (QAT) addresses these challenges by reducing model precision while maintaining performance. However, the scaling behavior of QAT, especially at 4-bit precision (W4A4), is not well understood. Existing QAT scaling laws often ignore key factors such as the number of training tokens and quantization granularity, which limits their applicability. This paper proposes a unified scaling law for QAT that models quantization error as a function of model size, training data volume, and quantization group size. Through 268 QAT experiments, we show that quantization error decreases as model size increases, but rises with more training tokens and coarser quantization granularity. To identify the sources of W4A4 quantization error, we decompose it into weight and activation components. Both components follow the overall trend of W4A4 quantization error, but with different sensitivities. Specifically, weight quantization error increases more rapidly with more training tokens. Further analysis shows that the activation quantization error in the FC2 layer, caused by outliers, is the primary bottleneck of W4A4 QAT quantization error. By applying mixed-precision quantization to address this bottleneck, we demonstrate that weight and activation quantization errors can converge to similar levels. Additionally, with more training data, weight quantization error eventually exceeds activation quantization error, suggesting that reducing weight quantization error is also important in such scenarios. These findings offer key insights for improving QAT research and development.</p> - https://arxiv.org/abs/2505.14302 - Tue, 20 May 2025 12:54:43 +0000 + VIKI-R: Coordinating Embodied Multi-Agent Cooperation via Reinforcement Learning + https://arxiv.org/abs/2506.09049 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.09049.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Li Kang, Xiufeng Song, Heng Zhou, Yiran Qin, Jie Yang, Xiaohong Liu, Philip Torr, Lei Bai, Zhenfei Yin</p><p><b>Upvotes:</b> 33</p><p><b>Summary:</b> Coordinating multiple embodied agents in dynamic environments remains a core challenge in artificial intelligence, requiring both perception-driven reasoning and scalable cooperation strategies. While recent works have leveraged large language models (LLMs) for multi-agent planning, a few have begun to explore vision-language models (VLMs) for visual reasoning. However, these VLM-based approaches remain limited in their support for diverse embodiment types. In this work, we introduce VIKI-Bench, the first hierarchical benchmark tailored for embodied multi-agent cooperation, featuring three structured levels: agent activation, task planning, and trajectory perception. VIKI-Bench includes diverse robot embodiments, multi-view visual observations, and structured supervision signals to evaluate reasoning grounded in visual inputs. To demonstrate the utility of VIKI-Bench, we propose VIKI-R, a two-stage framework that fine-tunes a pretrained vision-language model (VLM) using Chain-of-Thought annotated demonstrations, followed by reinforcement learning under multi-level reward signals. Our extensive experiments show that VIKI-R significantly outperforms baselines method across all task levels. Furthermore, we show that reinforcement learning enables the emergence of compositional cooperation patterns among heterogeneous agents. Together, VIKI-Bench and VIKI-R offer a unified testbed and method for advancing multi-agent, visual-driven cooperation in embodied AI systems.</p> + https://arxiv.org/abs/2506.09049 + Tue, 10 Jun 2025 17:59:44 +0000 - This Time is Different: An Observability Perspective on Time Series Foundation Models - https://arxiv.org/abs/2505.14766 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.14766.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ben Cohen, Emaad Khwaja, Youssef Doubli, Salahidine Lemaachi, Chris Lettieri, Charles Masson, Hugo Miccinilli, Elise Ramé, Qiqi Ren, Afshin Rostamizadeh, Jean Ogier du Terrail, Anna-Monica Toon, Kan Wang, Stephan Xie, David Asker, Ameet Talwalkar, Othmane Abou-Amal</p><p><b>Upvotes:</b> 36</p><p><b>Summary:</b> We introduce Toto, a time series forecasting foundation model with 151 million parameters. Toto uses a modern decoder-only architecture coupled with architectural innovations designed to account for specific challenges found in multivariate observability time series data. Toto's pre-training corpus is a mixture of observability data, open datasets, and synthetic data, and is 4-10times larger than those of leading time series foundation models. Additionally, we introduce BOOM, a large-scale benchmark consisting of 350 million observations across 2,807 real-world time series. For both Toto and BOOM, we source observability data exclusively from Datadog's own telemetry and internal observability metrics. Extensive evaluations demonstrate that Toto achieves state-of-the-art performance on both BOOM and on established general purpose time series forecasting benchmarks. Toto's model weights, inference code, and evaluation scripts, as well as BOOM's data and evaluation code, are all available as open source under the Apache 2.0 License available at https://huggingface.co/Datadog/Toto-Open-Base-1.0 and https://github.com/DataDog/toto.</p> - https://arxiv.org/abs/2505.14766 - Tue, 20 May 2025 17:48:13 +0000 + Autoregressive Adversarial Post-Training for Real-Time Interactive Video Generation + https://arxiv.org/abs/2506.09350 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.09350.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Shanchuan Lin, Ceyuan Yang, Hao He, Jianwen Jiang, Yuxi Ren, Xin Xia, Yang Zhao, Xuefeng Xiao, Lu Jiang</p><p><b>Upvotes:</b> 48</p><p><b>Summary:</b> Existing large-scale video generation models are computationally intensive, preventing adoption in real-time and interactive applications. In this work, we propose autoregressive adversarial post-training (AAPT) to transform a pre-trained latent video diffusion model into a real-time, interactive video generator. Our model autoregressively generates a latent frame at a time using a single neural function evaluation (1NFE). The model can stream the result to the user in real time and receive interactive responses as controls to generate the next latent frame. Unlike existing approaches, our method explores adversarial training as an effective paradigm for autoregressive generation. This not only allows us to design an architecture that is more efficient for one-step generation while fully utilizing the KV cache, but also enables training the model in a student-forcing manner that proves to be effective in reducing error accumulation during long video generation. Our experiments demonstrate that our 8B model achieves real-time, 24fps, streaming video generation at 736x416 resolution on a single H100, or 1280x720 on 8xH100 up to a minute long (1440 frames). Visit our research website at https://seaweed-apt.com/2</p> + https://arxiv.org/abs/2506.09350 + Wed, 11 Jun 2025 03:04:23 +0000 - Quartet: Native FP4 Training Can Be Optimal for Large Language Models - https://arxiv.org/abs/2505.14669 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.14669.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Roberto L. Castro, Andrei Panferov, Soroush Tabesh, Oliver Sieberling, Jiale Chen, Mahdi Nikdan, Saleh Ashkboos, Dan Alistarh</p><p><b>Upvotes:</b> 72</p><p><b>Summary:</b> The rapid advancement of large language models (LLMs) has been paralleled by unprecedented increases in computational demands, with training costs for state-of-the-art models doubling every few months. Training models directly in low-precision arithmetic offers a solution, by improving both computational throughput and energy efficiency. Specifically, NVIDIA's recent Blackwell architecture facilitates extremely low-precision operations, specifically FP4 variants, promising substantial efficiency gains. Yet, current algorithms for training LLMs in FP4 precision face significant accuracy degradation and often rely on mixed-precision fallbacks. In this paper, we systematically investigate hardware-supported FP4 training and introduce Quartet, a new approach enabling accurate, end-to-end FP4 training with all the major computations (in e.g. linear layers) being performed in low precision. Through extensive evaluations on Llama-type models, we reveal a new low-precision scaling law that quantifies performance trade-offs across varying bit-widths and allows us to identify a "near-optimal" low-precision training technique in terms of accuracy-vs-computation, called Quartet. We implement Quartet using optimized CUDA kernels tailored for NVIDIA Blackwell GPUs, and show that it can achieve state-of-the-art accuracy for FP4 precision, successfully training billion-scale models. Our method demonstrates that fully FP4-based training is a competitive alternative to standard-precision and FP8 training. Our code is available at https://github.com/IST-DASLab/Quartet.</p> - https://arxiv.org/abs/2505.14669 - Tue, 20 May 2025 17:55:50 +0000 + ReasonMed: A 370K Multi-Agent Generated Dataset for Advancing Medical Reasoning + https://arxiv.org/abs/2506.09513 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.09513.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yu Sun, Xingyu Qian, Weiwen Xu, Hao Zhang, Chenghao Xiao, Long Li, Yu Rong, Wenbing Huang, Qifeng Bai, Tingyang Xu</p><p><b>Upvotes:</b> 93</p><p><b>Summary:</b> Though reasoning-based large language models (LLMs) have excelled in mathematics and programming, their capabilities in knowledge-intensive medical question answering remain underexplored. To address this, we introduce ReasonMed, the largest medical reasoning dataset, comprising 370k high-quality examples distilled from 1.7 million initial reasoning paths generated by various LLMs. ReasonMed is constructed through a multi-agent verification and refinement process, where we design an Error Refiner to enhance the reasoning paths by identifying and correcting error-prone steps flagged by a verifier. Leveraging ReasonMed, we systematically investigate best practices for training medical reasoning models and find that combining detailed Chain-of-Thought (CoT) reasoning with concise answer summaries yields the most effective fine-tuning strategy. Based on this strategy, we train ReasonMed-7B, which sets a new benchmark for sub-10B models, outperforming the prior best by 4.17\% and even exceeding LLaMA3.1-70B on PubMedQA by 4.60\%.</p> + https://arxiv.org/abs/2506.09513 + Wed, 11 Jun 2025 08:36:55 +0000 - Emerging Properties in Unified Multimodal Pretraining - https://arxiv.org/abs/2505.14683 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.14683.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Chaorui Deng, Deyao Zhu, Kunchang Li, Chenhui Gou, Feng Li, Zeyu Wang, Shu Zhong, Weihao Yu, Xiaonan Nie, Ziang Song, Guang Shi, Haoqi Fan</p><p><b>Upvotes:</b> 124</p><p><b>Summary:</b> Unifying multimodal understanding and generation has shown impressive capabilities in cutting-edge proprietary systems. In this work, we introduce BAGEL, an open0source foundational model that natively supports multimodal understanding and generation. BAGEL is a unified, decoder0only model pretrained on trillions of tokens curated from large0scale interleaved text, image, video, and web data. When scaled with such diverse multimodal interleaved data, BAGEL exhibits emerging capabilities in complex multimodal reasoning. As a result, it significantly outperforms open-source unified models in both multimodal generation and understanding across standard benchmarks, while exhibiting advanced multimodal reasoning abilities such as free-form image manipulation, future frame prediction, 3D manipulation, and world navigation. In the hope of facilitating further opportunities for multimodal research, we share the key findings, pretraining details, data creation protocal, and release our code and checkpoints to the community. The project page is at https://bagel-ai.org/</p> - https://arxiv.org/abs/2505.14683 - Tue, 20 May 2025 17:59:30 +0000 + Effective Red-Teaming of Policy-Adherent Agents + https://arxiv.org/abs/2506.09600 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.09600.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Itay Nakash, George Kour, Koren Lazar, Matan Vetzler, Guy Uziel, Ateret Anaby-Tavor</p><p><b>Upvotes:</b> 37</p><p><b>Summary:</b> Task-oriented LLM-based agents are increasingly used in domains with strict policies, such as refund eligibility or cancellation rules. The challenge lies in ensuring that the agent consistently adheres to these rules and policies, appropriately refusing any request that would violate them, while still maintaining a helpful and natural interaction. This calls for the development of tailored design and evaluation methodologies to ensure agent resilience against malicious user behavior. We propose a novel threat model that focuses on adversarial users aiming to exploit policy-adherent agents for personal benefit. To address this, we present CRAFT, a multi-agent red-teaming system that leverages policy-aware persuasive strategies to undermine a policy-adherent agent in a customer-service scenario, outperforming conventional jailbreak methods such as DAN prompts, emotional manipulation, and coercive. Building upon the existing tau-bench benchmark, we introduce tau-break, a complementary benchmark designed to rigorously assess the agent's robustness against manipulative user behavior. Finally, we evaluate several straightforward yet effective defense strategies. While these measures provide some protection, they fall short, highlighting the need for stronger, research-driven safeguards to protect policy-adherent agents from adversarial attacks</p> + https://arxiv.org/abs/2506.09600 + Wed, 11 Jun 2025 10:59:47 +0000 - Scaling Reasoning, Losing Control: Evaluating Instruction Following in Large Reasoning Models - https://arxiv.org/abs/2505.14810 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.14810.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Tingchen Fu, Jiawei Gu, Yafu Li, Xiaoye Qu, Yu Cheng</p><p><b>Upvotes:</b> 58</p><p><b>Summary:</b> Instruction-following is essential for aligning large language models (LLMs) with user intent. While recent reasoning-oriented models exhibit impressive performance on complex mathematical problems, their ability to adhere to natural language instructions remains underexplored. In this work, we introduce MathIF, a dedicated benchmark for evaluating instruction-following in mathematical reasoning tasks. Our empirical analysis reveals a consistent tension between scaling up reasoning capacity and maintaining controllability, as models that reason more effectively often struggle to comply with user directives. We find that models tuned on distilled long chains-of-thought or trained with reasoning-oriented reinforcement learning often degrade in instruction adherence, especially when generation length increases. Furthermore, we show that even simple interventions can partially recover obedience, though at the cost of reasoning performance. These findings highlight a fundamental tension in current LLM training paradigms and motivate the need for more instruction-aware reasoning models. We release the code and data at https://github.com/TingchenFu/MathIF.</p> - https://arxiv.org/abs/2505.14810 - Tue, 20 May 2025 18:18:01 +0000 + ComfyUI-R1: Exploring Reasoning Models for Workflow Generation + https://arxiv.org/abs/2506.09790 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.09790.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhenran Xu, Yiyu Wang, Xue Yang, Longyue Wang, Weihua Luo, Kaifu Zhang, Baotian Hu, Min Zhang</p><p><b>Upvotes:</b> 52</p><p><b>Summary:</b> AI-generated content has evolved from monolithic models to modular workflows, particularly on platforms like ComfyUI, enabling customization in creative pipelines. However, crafting effective workflows requires great expertise to orchestrate numerous specialized components, presenting a steep learning curve for users. To address this challenge, we introduce ComfyUI-R1, the first large reasoning model for automated workflow generation. Starting with our curated dataset of 4K workflows, we construct long chain-of-thought (CoT) reasoning data, including node selection, workflow planning, and code-level workflow representation. ComfyUI-R1 is trained through a two-stage framework: (1) CoT fine-tuning for cold start, adapting models to the ComfyUI domain; (2) reinforcement learning for incentivizing reasoning capability, guided by a fine-grained rule-metric hybrid reward, ensuring format validity, structural integrity, and node-level fidelity. Experiments show that our 7B-parameter model achieves a 97\% format validity rate, along with high pass rate, node-level and graph-level F1 scores, significantly surpassing prior state-of-the-art methods that employ leading closed-source models such as GPT-4o and Claude series. Further analysis highlights the critical role of the reasoning process and the advantage of transforming workflows into code. Qualitative comparison reveals our strength in synthesizing intricate workflows with diverse nodes, underscoring the potential of long CoT reasoning in AI art creation.</p> + https://arxiv.org/abs/2506.09790 + Wed, 11 Jun 2025 14:35:15 +0000 - Diffusion vs. Autoregressive Language Models: A Text Embedding Perspective - https://arxiv.org/abs/2505.15045 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.15045.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Siyue Zhang, Yilun Zhao, Liyuan Geng, Arman Cohan, Anh Tuan Luu, Chen Zhao</p><p><b>Upvotes:</b> 52</p><p><b>Summary:</b> Large language model (LLM)-based embedding models, benefiting from large scale pre-training and post-training, have begun to surpass BERT and T5-based models on general-purpose text embedding tasks such as document retrieval. However, a fundamental limitation of LLM embeddings lies in the unidirectional attention used during autoregressive pre-training, which misaligns with the bidirectional nature of text embedding tasks. To this end, We propose adopting diffusion language models for text embeddings, motivated by their inherent bidirectional architecture and recent success in matching or surpassing LLMs especially on reasoning tasks. We present the first systematic study of the diffusion language embedding model, which outperforms the LLM-based embedding model by 20% on long-document retrieval, 8% on reasoning-intensive retrieval, 2% on instruction-following retrieval, and achieve competitive performance on traditional text embedding benchmarks. Our analysis verifies that bidirectional attention is crucial for encoding global context in long and complex text.</p> - https://arxiv.org/abs/2505.15045 - Wed, 21 May 2025 02:59:14 +0000 + Multiverse: Your Language Models Secretly Decide How to Parallelize and Merge Generation + https://arxiv.org/abs/2506.09991 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.09991.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Xinyu Yang, Yuwei An, Hongyi Liu, Tianqi Chen, Beidi Chen</p><p><b>Upvotes:</b> 56</p><p><b>Summary:</b> Autoregressive Large Language Models (AR-LLMs) frequently exhibit implicit parallelism in sequential generation. Inspired by this, we introduce Multiverse, a new generative model that enables natively parallel generation. Multiverse internalizes a MapReduce paradigm, generating automatically through three stages: (i) a Map stage for adaptive task decomposition, (ii) a Process stage for parallel subtask execution, and (iii) a Reduce stage for lossless result synthesis. Next, we build a real-world Multiverse reasoning model with co-design of data, algorithm, and system, enabling rapid and seamless transfer from frontier AR-LLMs. Starting from sequential reasoning chains, we create Multiverse 1K by converting them into structured training data using an automated LLM-assisted pipeline, avoiding costly human annotations. Algorithmically, we design Multiverse Attention to separate parallel reasoning steps while keeping compatibility with causal attention for efficient training. Systematically, we implement Multiverse Engine to enable parallel inference. It features a dedicated scheduler that dynamically switches between sequential and parallel generation, triggered directly by the model. After a 3-hour fine-tuning with 1K examples, our Multiverse-32B stands as the only open-sourced non-AR model achieving performance on par with leading AR-LLMs of the same scale, evidenced by AIME24 & 25 scores of 54% and 46%, respectively. Moreover, our budget control experiments show that Multiverse-32B exhibits superior scaling, outperforming AR-LLMs by 1.87% on average using the same context length. Such scaling further leads to practical efficiency gain, achieving up to 2x speedup across varying batch sizes. We have open-sourced the entire Multiverse ecosystem, including data, model weights, engine, supporting tools, as well as complete data curation prompts and detailed training and evaluation recipes.</p> + https://arxiv.org/abs/2506.09991 + Wed, 11 Jun 2025 17:59:23 +0000 - Web-Shepherd: Advancing PRMs for Reinforcing Web Agents - https://arxiv.org/abs/2505.15277 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.15277.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Hyungjoo Chae, Sunghwan Kim, Junhee Cho, Seungone Kim, Seungjun Moon, Gyeom Hwangbo, Dongha Lim, Minjin Kim, Yeonjun Hwang, Minju Gwak, Dongwook Choi, Minseok Kang, Gwanhoon Im, ByeongUng Cho, Hyojun Kim, Jun Hee Han, Taeyoon Kwon, Minju Kim, Beong-woo Kwak, Dongjin Kang, Jinyoung Yeo</p><p><b>Upvotes:</b> 97</p><p><b>Summary:</b> Web navigation is a unique domain that can automate many repetitive real-life tasks and is challenging as it requires long-horizon sequential decision making beyond typical multimodal large language model (MLLM) tasks. Yet, specialized reward models for web navigation that can be utilized during both training and test-time have been absent until now. Despite the importance of speed and cost-effectiveness, prior works have utilized MLLMs as reward models, which poses significant constraints for real-world deployment. To address this, in this work, we propose the first process reward model (PRM) called Web-Shepherd which could assess web navigation trajectories in a step-level. To achieve this, we first construct the WebPRM Collection, a large-scale dataset with 40K step-level preference pairs and annotated checklists spanning diverse domains and difficulty levels. Next, we also introduce the WebRewardBench, the first meta-evaluation benchmark for evaluating PRMs. In our experiments, we observe that our Web-Shepherd achieves about 30 points better accuracy compared to using GPT-4o on WebRewardBench. Furthermore, when testing on WebArena-lite by using GPT-4o-mini as the policy and Web-Shepherd as the verifier, we achieve 10.9 points better performance, in 10 less cost compared to using GPT-4o-mini as the verifier. Our model, dataset, and code are publicly available at LINK.</p> - https://arxiv.org/abs/2505.15277 - Wed, 21 May 2025 08:56:55 +0000 + Text-Aware Image Restoration with Diffusion Models + https://arxiv.org/abs/2506.09993 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.09993.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jaewon Min, Jin Hyeon Kim, Paul Hyunbin Cho, Jaeeun Lee, Jihye Park, Minkyu Park, Sangpil Kim, Hyunhee Park, Seungryong Kim</p><p><b>Upvotes:</b> 40</p><p><b>Summary:</b> Image restoration aims to recover degraded images. However, existing diffusion-based restoration methods, despite great success in natural image restoration, often struggle to faithfully reconstruct textual regions in degraded images. Those methods frequently generate plausible but incorrect text-like patterns, a phenomenon we refer to as text-image hallucination. In this paper, we introduce Text-Aware Image Restoration (TAIR), a novel restoration task that requires the simultaneous recovery of visual contents and textual fidelity. To tackle this task, we present SA-Text, a large-scale benchmark of 100K high-quality scene images densely annotated with diverse and complex text instances. Furthermore, we propose a multi-task diffusion framework, called TeReDiff, that integrates internal features from diffusion models into a text-spotting module, enabling both components to benefit from joint training. This allows for the extraction of rich text representations, which are utilized as prompts in subsequent denoising steps. Extensive experiments demonstrate that our approach consistently outperforms state-of-the-art restoration methods, achieving significant gains in text recognition accuracy. See our project page: https://cvlab-kaist.github.io/TAIR/</p> + https://arxiv.org/abs/2506.09993 + Wed, 11 Jun 2025 17:59:46 +0000 - MMaDA: Multimodal Large Diffusion Language Models - https://arxiv.org/abs/2505.15809 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.15809.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ling Yang, Ye Tian, Bowen Li, Xinchen Zhang, Ke Shen, Yunhai Tong, Mengdi Wang</p><p><b>Upvotes:</b> 83</p><p><b>Summary:</b> We introduce MMaDA, a novel class of multimodal diffusion foundation models designed to achieve superior performance across diverse domains such as textual reasoning, multimodal understanding, and text-to-image generation. The approach is distinguished by three key innovations: (i) MMaDA adopts a unified diffusion architecture with a shared probabilistic formulation and a modality-agnostic design, eliminating the need for modality-specific components. This architecture ensures seamless integration and processing across different data types. (ii) We implement a mixed long chain-of-thought (CoT) fine-tuning strategy that curates a unified CoT format across modalities. By aligning reasoning processes between textual and visual domains, this strategy facilitates cold-start training for the final reinforcement learning (RL) stage, thereby enhancing the model's ability to handle complex tasks from the outset. (iii) We propose UniGRPO, a unified policy-gradient-based RL algorithm specifically tailored for diffusion foundation models. Utilizing diversified reward modeling, UniGRPO unifies post-training across both reasoning and generation tasks, ensuring consistent performance improvements. Experimental results demonstrate that MMaDA-8B exhibits strong generalization capabilities as a unified multimodal foundation model. It surpasses powerful models like LLaMA-3-7B and Qwen2-7B in textual reasoning, outperforms Show-o and SEED-X in multimodal understanding, and excels over SDXL and Janus in text-to-image generation. These achievements highlight MMaDA's effectiveness in bridging the gap between pretraining and post-training within unified diffusion architectures, providing a comprehensive framework for future research and development. We open-source our code and trained models at: https://github.com/Gen-Verse/MMaDA</p> - https://arxiv.org/abs/2505.15809 - Wed, 21 May 2025 17:59:05 +0000 + PlayerOne: Egocentric World Simulator + https://arxiv.org/abs/2506.09995 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.09995.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yuanpeng Tu, Hao Luo, Xi Chen, Xiang Bai, Fan Wang, Hengshuang Zhao</p><p><b>Upvotes:</b> 33</p><p><b>Summary:</b> We introduce PlayerOne, the first egocentric realistic world simulator, facilitating immersive and unrestricted exploration within vividly dynamic environments. Given an egocentric scene image from the user, PlayerOne can accurately construct the corresponding world and generate egocentric videos that are strictly aligned with the real scene human motion of the user captured by an exocentric camera. PlayerOne is trained in a coarse-to-fine pipeline that first performs pretraining on large-scale egocentric text-video pairs for coarse-level egocentric understanding, followed by finetuning on synchronous motion-video data extracted from egocentric-exocentric video datasets with our automatic construction pipeline. Besides, considering the varying importance of different components, we design a part-disentangled motion injection scheme, enabling precise control of part-level movements. In addition, we devise a joint reconstruction framework that progressively models both the 4D scene and video frames, ensuring scene consistency in the long-form video generation. Experimental results demonstrate its great generalization ability in precise control of varying human movements and worldconsistent modeling of diverse scenarios. It marks the first endeavor into egocentric real-world simulation and can pave the way for the community to delve into fresh frontiers of world modeling and its diverse applications.</p> + https://arxiv.org/abs/2506.09995 + Wed, 11 Jun 2025 17:59:53 +0000 - Pixel Reasoner: Incentivizing Pixel-Space Reasoning with Curiosity-Driven Reinforcement Learning - https://arxiv.org/abs/2505.15966 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.15966.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Alex Su, Haozhe Wang, Weimin Ren, Fangzhen Lin, Wenhu Chen</p><p><b>Upvotes:</b> 48</p><p><b>Summary:</b> Chain-of-thought reasoning has significantly improved the performance of Large Language Models (LLMs) across various domains. However, this reasoning process has been confined exclusively to textual space, limiting its effectiveness in visually intensive tasks. To address this limitation, we introduce the concept of reasoning in the pixel-space. Within this novel framework, Vision-Language Models (VLMs) are equipped with a suite of visual reasoning operations, such as zoom-in and select-frame. These operations enable VLMs to directly inspect, interrogate, and infer from visual evidences, thereby enhancing reasoning fidelity for visual tasks. Cultivating such pixel-space reasoning capabilities in VLMs presents notable challenges, including the model's initially imbalanced competence and its reluctance to adopt the newly introduced pixel-space operations. We address these challenges through a two-phase training approach. The first phase employs instruction tuning on synthesized reasoning traces to familiarize the model with the novel visual operations. Following this, a reinforcement learning (RL) phase leverages a curiosity-driven reward scheme to balance exploration between pixel-space reasoning and textual reasoning. With these visual operations, VLMs can interact with complex visual inputs, such as information-rich images or videos to proactively gather necessary information. We demonstrate that this approach significantly improves VLM performance across diverse visual reasoning benchmarks. Our 7B model, \model, achieves 84\% on V* bench, 74\% on TallyQA-Complex, and 84\% on InfographicsVQA, marking the highest accuracy achieved by any open-source model to date. These results highlight the importance of pixel-space reasoning and the effectiveness of our framework.</p> - https://arxiv.org/abs/2505.15966 - Wed, 21 May 2025 19:35:08 +0000 + Scientists' First Exam: Probing Cognitive Abilities of MLLM via Perception, Understanding, and Reasoning + https://arxiv.org/abs/2506.10521 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.10521.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yuhao Zhou, Yiheng Wang, Xuming He, Ruoyao Xiao, Zhiwei Li, Qiantai Feng, Zijie Guo, Yuejin Yang, Hao Wu, Wenxuan Huang, Jiaqi Wei, Dan Si, Xiuqi Yao, Jia Bu, Haiwen Huang, Tianfan Fu, Shixiang Tang, Ben Fei, Dongzhan Zhou, Fenghua Ling, Yan Lu, Siqi Sun, Chenhui Li, Guanjie Zheng, Jiancheng Lv, Wenlong Zhang, Lei Bai</p><p><b>Upvotes:</b> 65</p><p><b>Summary:</b> Scientific discoveries increasingly rely on complex multimodal reasoning based on information-intensive scientific data and domain-specific expertise. Empowered by expert-level scientific benchmarks, scientific Multimodal Large Language Models (MLLMs) hold the potential to significantly enhance this discovery process in realistic workflows. However, current scientific benchmarks mostly focus on evaluating the knowledge understanding capabilities of MLLMs, leading to an inadequate assessment of their perception and reasoning abilities. To address this gap, we present the Scientists' First Exam (SFE) benchmark, designed to evaluate the scientific cognitive capacities of MLLMs through three interconnected levels: scientific signal perception, scientific attribute understanding, scientific comparative reasoning. Specifically, SFE comprises 830 expert-verified VQA pairs across three question types, spanning 66 multimodal tasks across five high-value disciplines. Extensive experiments reveal that current state-of-the-art GPT-o3 and InternVL-3 achieve only 34.08% and 26.52% on SFE, highlighting significant room for MLLMs to improve in scientific realms. We hope the insights obtained in SFE will facilitate further developments in AI-enhanced scientific discoveries.</p> + https://arxiv.org/abs/2506.10521 + Thu, 12 Jun 2025 09:29:16 +0000 - QuickVideo: Real-Time Long Video Understanding with System Algorithm Co-Design - https://arxiv.org/abs/2505.16175 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.16175.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Benjamin Schneider, Dongfu Jiang, Chao Du, Tianyu Pang, Wenhu Chen</p><p><b>Upvotes:</b> 38</p><p><b>Summary:</b> Long-video understanding has emerged as a crucial capability in real-world applications such as video surveillance, meeting summarization, educational lecture analysis, and sports broadcasting. However, it remains computationally prohibitive for VideoLLMs, primarily due to two bottlenecks: 1) sequential video decoding, the process of converting the raw bit stream to RGB frames can take up to a minute for hour-long video inputs, and 2) costly prefilling of up to several million tokens for LLM inference, resulting in high latency and memory use. To address these challenges, we propose QuickVideo, a system-algorithm co-design that substantially accelerates long-video understanding to support real-time downstream applications. It comprises three key innovations: QuickDecoder, a parallelized CPU-based video decoder that achieves 2-3 times speedup by splitting videos into keyframe-aligned intervals processed concurrently; QuickPrefill, a memory-efficient prefilling method using KV-cache pruning to support more frames with less GPU memory; and an overlapping scheme that overlaps CPU video decoding with GPU inference. Together, these components infernece time reduce by a minute on long video inputs, enabling scalable, high-quality video understanding even on limited hardware. Experiments show that QuickVideo generalizes across durations and sampling rates, making long video processing feasible in practice.</p> - https://arxiv.org/abs/2505.16175 - Thu, 22 May 2025 03:26:50 +0000 + AniMaker: Automated Multi-Agent Animated Storytelling with MCTS-Driven Clip Generation + https://arxiv.org/abs/2506.10540 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.10540.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Haoyuan Shi, Yunxin Li, Xinyu Chen, Longyue Wang, Baotian Hu, Min Zhang</p><p><b>Upvotes:</b> 37</p><p><b>Summary:</b> Despite rapid advancements in video generation models, generating coherent storytelling videos that span multiple scenes and characters remains challenging. Current methods often rigidly convert pre-generated keyframes into fixed-length clips, resulting in disjointed narratives and pacing issues. Furthermore, the inherent instability of video generation models means that even a single low-quality clip can significantly degrade the entire output animation's logical coherence and visual continuity. To overcome these obstacles, we introduce AniMaker, a multi-agent framework enabling efficient multi-candidate clip generation and storytelling-aware clip selection, thus creating globally consistent and story-coherent animation solely from text input. The framework is structured around specialized agents, including the Director Agent for storyboard generation, the Photography Agent for video clip generation, the Reviewer Agent for evaluation, and the Post-Production Agent for editing and voiceover. Central to AniMaker's approach are two key technical components: MCTS-Gen in Photography Agent, an efficient Monte Carlo Tree Search (MCTS)-inspired strategy that intelligently navigates the candidate space to generate high-potential clips while optimizing resource usage; and AniEval in Reviewer Agent, the first framework specifically designed for multi-shot animation evaluation, which assesses critical aspects such as story-level consistency, action completion, and animation-specific features by considering each clip in the context of its preceding and succeeding clips. Experiments demonstrate that AniMaker achieves superior quality as measured by popular metrics including VBench and our proposed AniEval framework, while significantly improving the efficiency of multi-candidate generation, pushing AI-generated storytelling animation closer to production standards.</p> + https://arxiv.org/abs/2506.10540 + Thu, 12 Jun 2025 10:06:21 +0000 - Embodied Agents Meet Personalization: Exploring Memory Utilization for Personalized Assistance - https://arxiv.org/abs/2505.16348 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.16348.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Taeyoon Kwon, Dongwook Choi, Sunghwan Kim, Hyojun Kim, Seungjun Moon, Beong-woo Kwak, Kuan-Hao Huang, Jinyoung Yeo</p><p><b>Upvotes:</b> 43</p><p><b>Summary:</b> Embodied agents empowered by large language models (LLMs) have shown strong performance in household object rearrangement tasks. However, these tasks primarily focus on single-turn interactions with simplified instructions, which do not truly reflect the challenges of providing meaningful assistance to users. To provide personalized assistance, embodied agents must understand the unique semantics that users assign to the physical world (e.g., favorite cup, breakfast routine) by leveraging prior interaction history to interpret dynamic, real-world instructions. Yet, the effectiveness of embodied agents in utilizing memory for personalized assistance remains largely underexplored. To address this gap, we present MEMENTO, a personalized embodied agent evaluation framework designed to comprehensively assess memory utilization capabilities to provide personalized assistance. Our framework consists of a two-stage memory evaluation process design that enables quantifying the impact of memory utilization on task performance. This process enables the evaluation of agents' understanding of personalized knowledge in object rearrangement tasks by focusing on its role in goal interpretation: (1) the ability to identify target objects based on personal meaning (object semantics), and (2) the ability to infer object-location configurations from consistent user patterns, such as routines (user patterns). Our experiments across various LLMs reveal significant limitations in memory utilization, with even frontier models like GPT-4o experiencing a 30.5% performance drop when required to reference multiple memories, particularly in tasks involving user patterns. These findings, along with our detailed analyses and case studies, provide valuable insights for future research in developing more effective personalized embodied agents. Project website: https://connoriginal.github.io/MEMENTO</p> - https://arxiv.org/abs/2505.16348 - Thu, 22 May 2025 08:00:10 +0000 + The Diffusion Duality + https://arxiv.org/abs/2506.10892 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.10892.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Subham Sekhar Sahoo, Justin Deschenaux, Aaron Gokaslan, Guanghan Wang, Justin Chiu, Volodymyr Kuleshov</p><p><b>Upvotes:</b> 36</p><p><b>Summary:</b> Uniform-state discrete diffusion models hold the promise of fast text generation due to their inherent ability to self-correct. However, they are typically outperformed by autoregressive models and masked diffusion models. In this work, we narrow this performance gap by leveraging a key insight: Uniform-state diffusion processes naturally emerge from an underlying Gaussian diffusion. Our method, Duo, transfers powerful techniques from Gaussian diffusion to improve both training and sampling. First, we introduce a curriculum learning strategy guided by the Gaussian process, doubling training speed by reducing variance. Models trained with curriculum learning surpass autoregressive models in zero-shot perplexity on 3 of 7 benchmarks. Second, we present Discrete Consistency Distillation, which adapts consistency distillation from the continuous to the discrete setting. This algorithm unlocks few-step generation in diffusion language models by accelerating sampling by two orders of magnitude. We provide the code and model checkpoints on the project page: http://s-sahoo.github.io/duo</p> + https://arxiv.org/abs/2506.10892 + Thu, 12 Jun 2025 16:55:35 +0000 - Tool-Star: Empowering LLM-Brained Multi-Tool Reasoner via Reinforcement Learning - https://arxiv.org/abs/2505.16410 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.16410.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Guanting Dong, Yifei Chen, Xiaoxi Li, Jiajie Jin, Hongjin Qian, Yutao Zhu, Hangyu Mao, Guorui Zhou, Zhicheng Dou, Ji-Rong Wen</p><p><b>Upvotes:</b> 53</p><p><b>Summary:</b> Recently, large language models (LLMs) have shown remarkable reasoning capabilities via large-scale reinforcement learning (RL). However, leveraging the RL algorithm to empower effective multi-tool collaborative reasoning in LLMs remains an open challenge. In this paper, we introduce Tool-Star, an RL-based framework designed to empower LLMs to autonomously invoke multiple external tools during stepwise reasoning. Tool-Star integrates six types of tools and incorporates systematic designs in both data synthesis and training. To address the scarcity of tool-use data, we propose a general tool-integrated reasoning data synthesis pipeline, which combines tool-integrated prompting with hint-based sampling to automatically and scalably generate tool-use trajectories. A subsequent quality normalization and difficulty-aware classification process filters out low-quality samples and organizes the dataset from easy to hard. Furthermore, we propose a two-stage training framework to enhance multi-tool collaborative reasoning by: (1) cold-start fine-tuning, which guides LLMs to explore reasoning patterns via tool-invocation feedback; and (2) a multi-tool self-critic RL algorithm with hierarchical reward design, which reinforces reward understanding and promotes effective tool collaboration. Experimental analyses on over 10 challenging reasoning benchmarks highlight the effectiveness and efficiency of Tool-Star. The code is available at https://github.com/dongguanting/Tool-Star.</p> - https://arxiv.org/abs/2505.16410 - Thu, 22 May 2025 09:00:19 +0000 + Magistral + https://arxiv.org/abs/2506.10910 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.10910.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Mistral-AI, Abhinav Rastogi, Albert Q. Jiang, Andy Lo, Gabrielle Berrada, Guillaume Lample, Jason Rute, Joep Barmentlo, Karmesh Yadav, Kartik Khandelwal, Khyathi Raghavi Chandu, Léonard Blier, Lucile Saulnier, Matthieu Dinot, Maxime Darrin, Neha Gupta, Roman Soletskyi, Sagar Vaze, Teven Le Scao, Yihan Wang, Adam Yang, Alexander H. Liu, Alexandre Sablayrolles, Amélie Héliou, Amélie Martin, Andy Ehrenberg, Anmol Agarwal, Antoine Roux, Arthur Darcet, Arthur Mensch, Baptiste Bout, Baptiste Rozière, Baudouin De Monicault, Chris Bamford, Christian Wallenwein, Christophe Renaudin, Clémence Lanfranchi, Darius Dabert, Devon Mizelle, Diego de las Casas, Elliot Chane-Sane, Emilien Fugier, Emma Bou Hanna, Gauthier Delerce, Gauthier Guinet, Georgii Novikov, Guillaume Martin, Himanshu Jaju, Jan Ludziejewski, Jean-Hadrien Chabran, Jean-Malo Delignon, Joachim Studnia, Jonas Amar, Josselin Somerville Roberts, Julien Denize, Karan Saxena, Kush Jain, Lingxiao Zhao, Louis Martin, Luyu Gao, Lélio Renard Lavaud, Marie Pellat, Mathilde Guillaumin, Mathis Felardos, Maximilian Augustin, Mickaël Seznec, Nikhil Raghuraman, Olivier Duchenne, Patricia Wang, Patrick von Platen, Patryk Saffer, Paul Jacob, Paul Wambergue, Paula Kurylowicz, Pavankumar Reddy Muddireddy, Philomène Chagniot, Pierre Stock, Pravesh Agrawal, Romain Sauvestre, Rémi Delacourt, Sanchit Gandhi, Sandeep Subramanian, Shashwat Dalal, Siddharth Gandhi, Soham Ghosh, Srijan Mishra, Sumukh Aithal, Szymon Antoniak, Thibault Schueller, Thibaut Lavril, Thomas Robert, Thomas Wang, Timothée Lacroix, Valeriia Nemychnikova, Victor Paltz, Virgile Richard, Wen-Ding Li, William Marshall, Xuanyu Zhang, Yunhao Tang</p><p><b>Upvotes:</b> 60</p><p><b>Summary:</b> We introduce Magistral, Mistral's first reasoning model and our own scalable reinforcement learning (RL) pipeline. Instead of relying on existing implementations and RL traces distilled from prior models, we follow a ground up approach, relying solely on our own models and infrastructure. Notably, we demonstrate a stack that enabled us to explore the limits of pure RL training of LLMs, present a simple method to force the reasoning language of the model, and show that RL on text data alone maintains most of the initial checkpoint's capabilities. We find that RL on text maintains or improves multimodal understanding, instruction following and function calling. We present Magistral Medium, trained for reasoning on top of Mistral Medium 3 with RL alone, and we open-source Magistral Small (Apache 2.0) which further includes cold-start data from Magistral Medium.</p> + https://arxiv.org/abs/2506.10910 + Thu, 12 Jun 2025 17:22:37 +0000 - MMMR: Benchmarking Massive Multi-Modal Reasoning Tasks - https://arxiv.org/abs/2505.16459 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.16459.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Guiyao Tie, Xueyang Zhou, Tianhe Gu, Ruihang Zhang, Chaoran Hu, Sizhe Zhang, Mengqu Sun, Yan Zhang, Pan Zhou, Lichao Sun</p><p><b>Upvotes:</b> 44</p><p><b>Summary:</b> Recent advances in Multi-Modal Large Language Models (MLLMs) have enabled unified processing of language, vision, and structured inputs, opening the door to complex tasks such as logical deduction, spatial reasoning, and scientific analysis. Despite their promise, the reasoning capabilities of MLLMs, particularly those augmented with intermediate thinking traces (MLLMs-T), remain poorly understood and lack standardized evaluation benchmarks. Existing work focuses primarily on perception or final answer correctness, offering limited insight into how models reason or fail across modalities. To address this gap, we introduce the MMMR, a new benchmark designed to rigorously evaluate multi-modal reasoning with explicit thinking. The MMMR comprises 1) a high-difficulty dataset of 1,083 questions spanning six diverse reasoning types with symbolic depth and multi-hop demands and 2) a modular Reasoning Trace Evaluation Pipeline (RTEP) for assessing reasoning quality beyond accuracy through metrics like relevance, consistency, and structured error annotations. Empirical results show that MLLMs-T overall outperform non-thinking counterparts, but even top models like Claude-3.7-Sonnet and Gemini-2.5 Pro suffer from reasoning pathologies such as inconsistency and overthinking. This benchmark reveals persistent gaps between accuracy and reasoning quality and provides an actionable evaluation pipeline for future model development. Overall, the MMMR offers a scalable foundation for evaluating, comparing, and improving the next generation of multi-modal reasoning systems.</p> - https://arxiv.org/abs/2505.16459 - Thu, 22 May 2025 09:41:55 +0000 + SWE-Factory: Your Automated Factory for Issue Resolution Training Data and Evaluation Benchmarks + https://arxiv.org/abs/2506.10954 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.10954.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Lianghong Guo, Yanlin Wang, Caihua Li, Pengyu Yang, Jiachi Chen, Wei Tao, Yingtian Zou, Duyu Tang, Zibin Zheng</p><p><b>Upvotes:</b> 51</p><p><b>Summary:</b> Constructing large-scale datasets for the GitHub issue resolution task is crucial for both training and evaluating the software engineering capabilities of Large Language Models (LLMs). However, the traditional process for creating such benchmarks is notoriously challenging and labor-intensive, particularly in the stages of setting up evaluation environments, grading test outcomes, and validating task instances. In this paper, we propose SWE-Factory, an automated pipeline designed to address these challenges. To tackle these issues, our pipeline integrates three core automated components. First, we introduce SWE-Builder, a multi-agent system that automates evaluation environment construction, which employs four specialized agents that work in a collaborative, iterative loop and leverages an environment memory pool to enhance efficiency. Second, we introduce a standardized, exit-code-based grading method that eliminates the need for manually writing custom parsers. Finally, we automate the fail2pass validation process using these reliable exit code signals. Experiments on 671 issues across four programming languages show that our pipeline can effectively construct valid task instances; for example, with GPT-4.1-mini, our SWE-Builder constructs 269 valid instances at 0.045 per instance, while with Gemini-2.5-flash, it achieves comparable performance at the lowest cost of 0.024 per instance. We also demonstrate that our exit-code-based grading achieves 100% accuracy compared to manual inspection, and our automated fail2pass validation reaches a precision of 0.92 and a recall of 1.00. We hope our automated pipeline will accelerate the collection of large-scale, high-quality GitHub issue resolution datasets for both training and evaluation. Our code and datasets are released at https://github.com/DeepSoftwareAnalytics/swe-factory.</p> + https://arxiv.org/abs/2506.10954 + Thu, 12 Jun 2025 17:54:17 +0000 - KRIS-Bench: Benchmarking Next-Level Intelligent Image Editing Models - https://arxiv.org/abs/2505.16707 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.16707.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yongliang Wu, Zonghui Li, Xinting Hu, Xinyu Ye, Xianfang Zeng, Gang Yu, Wenbo Zhu, Bernt Schiele, Ming-Hsuan Yang, Xu Yang</p><p><b>Upvotes:</b> 41</p><p><b>Summary:</b> Recent advances in multi-modal generative models have enabled significant progress in instruction-based image editing. However, while these models produce visually plausible outputs, their capacity for knowledge-based reasoning editing tasks remains under-explored. In this paper, we introduce KRIS-Bench (Knowledge-based Reasoning in Image-editing Systems Benchmark), a diagnostic benchmark designed to assess models through a cognitively informed lens. Drawing from educational theory, KRIS-Bench categorizes editing tasks across three foundational knowledge types: Factual, Conceptual, and Procedural. Based on this taxonomy, we design 22 representative tasks spanning 7 reasoning dimensions and release 1,267 high-quality annotated editing instances. To support fine-grained evaluation, we propose a comprehensive protocol that incorporates a novel Knowledge Plausibility metric, enhanced by knowledge hints and calibrated through human studies. Empirical results on 10 state-of-the-art models reveal significant gaps in reasoning performance, highlighting the need for knowledge-centric benchmarks to advance the development of intelligent image editing systems.</p> - https://arxiv.org/abs/2505.16707 - Thu, 22 May 2025 14:08:59 +0000 + DeepResearch Bench: A Comprehensive Benchmark for Deep Research Agents + https://arxiv.org/abs/2506.11763 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.11763.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Mingxuan Du, Benfeng Xu, Chiwei Zhu, Xiaorui Wang, Zhendong Mao</p><p><b>Upvotes:</b> 60</p><p><b>Summary:</b> Deep Research Agents are a prominent category of LLM-based agents. By autonomously orchestrating multistep web exploration, targeted retrieval, and higher-order synthesis, they transform vast amounts of online information into analyst-grade, citation-rich reports--compressing hours of manual desk research into minutes. However, a comprehensive benchmark for systematically evaluating the capabilities of these agents remains absent. To bridge this gap, we present DeepResearch Bench, a benchmark consisting of 100 PhD-level research tasks, each meticulously crafted by domain experts across 22 distinct fields. Evaluating DRAs is inherently complex and labor-intensive. We therefore propose two novel methodologies that achieve strong alignment with human judgment. The first is a reference-based method with adaptive criteria to assess the quality of generated research reports. The other framework is introduced to evaluate DRA's information retrieval and collection capabilities by assessing its effective citation count and overall citation accuracy. We have open-sourced DeepResearch Bench and key components of these frameworks at https://github.com/Ayanami0730/deep_research_bench to accelerate the development of practical LLM-based agents.</p> + https://arxiv.org/abs/2506.11763 + Fri, 13 Jun 2025 13:17:32 +0000 - NovelSeek: When Agent Becomes the Scientist -- Building Closed-Loop System from Hypothesis to Verification - https://arxiv.org/abs/2505.16938 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.16938.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> NovelSeek Team, Bo Zhang, Shiyang Feng, Xiangchao Yan, Jiakang Yuan, Zhiyin Yu, Xiaohan He, Songtao Huang, Shaowei Hou, Zheng Nie, Zhilong Wang, Jinyao Liu, Runmin Ma, Tianshuo Peng, Peng Ye, Dongzhan Zhou, Shufei Zhang, Xiaosong Wang, Yilan Zhang, Meng Li, Zhongying Tu, Xiangyu Yue, Wangli Ouyang, Bowen Zhou, Lei Bai</p><p><b>Upvotes:</b> 110</p><p><b>Summary:</b> Artificial Intelligence (AI) is accelerating the transformation of scientific research paradigms, not only enhancing research efficiency but also driving innovation. We introduce NovelSeek, a unified closed-loop multi-agent framework to conduct Autonomous Scientific Research (ASR) across various scientific research fields, enabling researchers to tackle complicated problems in these fields with unprecedented speed and precision. NovelSeek highlights three key advantages: 1) Scalability: NovelSeek has demonstrated its versatility across 12 scientific research tasks, capable of generating innovative ideas to enhance the performance of baseline code. 2) Interactivity: NovelSeek provides an interface for human expert feedback and multi-agent interaction in automated end-to-end processes, allowing for the seamless integration of domain expert knowledge. 3) Efficiency: NovelSeek has achieved promising performance gains in several scientific fields with significantly less time cost compared to human efforts. For instance, in reaction yield prediction, it increased from 27.6% to 35.4% in just 12 hours; in enhancer activity prediction, accuracy rose from 0.52 to 0.79 with only 4 hours of processing; and in 2D semantic segmentation, precision advanced from 78.8% to 81.0% in a mere 30 hours.</p> - https://arxiv.org/abs/2505.16938 - Thu, 22 May 2025 17:27:43 +0000 + Aligned Novel View Image and Geometry Synthesis via Cross-modal Attention Instillation + https://arxiv.org/abs/2506.11924 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.11924.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Min-Seop Kwak, Junho Kim, Sangdoo Yun, Dongyoon Han, Taekyoung Kim, Seungryong Kim, Jin-Hwa Kim</p><p><b>Upvotes:</b> 32</p><p><b>Summary:</b> We introduce a diffusion-based framework that performs aligned novel view image and geometry generation via a warping-and-inpainting methodology. Unlike prior methods that require dense posed images or pose-embedded generative models limited to in-domain views, our method leverages off-the-shelf geometry predictors to predict partial geometries viewed from reference images, and formulates novel-view synthesis as an inpainting task for both image and geometry. To ensure accurate alignment between generated images and geometry, we propose cross-modal attention distillation, where attention maps from the image diffusion branch are injected into a parallel geometry diffusion branch during both training and inference. This multi-task approach achieves synergistic effects, facilitating geometrically robust image synthesis as well as well-defined geometry prediction. We further introduce proximity-based mesh conditioning to integrate depth and normal cues, interpolating between point cloud and filtering erroneously predicted geometry from influencing the generation process. Empirically, our method achieves high-fidelity extrapolative view synthesis on both image and geometry across a range of unseen scenes, delivers competitive reconstruction quality under interpolation settings, and produces geometrically aligned colored point clouds for comprehensive 3D completion. Project page is available at https://cvlab-kaist.github.io/MoAI.</p> + https://arxiv.org/abs/2506.11924 + Fri, 13 Jun 2025 16:19:00 +0000 - Reasoning Model is Stubborn: Diagnosing Instruction Overriding in Reasoning Models - https://arxiv.org/abs/2505.17225 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.17225.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Doohyuk Jang, Yoonjeon Kim, Chanjae Park, Hyun Ryu, Eunho Yang</p><p><b>Upvotes:</b> 62</p><p><b>Summary:</b> Large language models have demonstrated remarkable proficiency in long and complex reasoning tasks. However, they frequently exhibit a problematic reliance on familiar reasoning patterns, a phenomenon we term reasoning rigidity. Despite explicit instructions from users, these models often override clearly stated conditions and default to habitual reasoning trajectories, leading to incorrect conclusions. This behavior presents significant challenges, particularly in domains such as mathematics and logic puzzle, where precise adherence to specified constraints is critical. To systematically investigate reasoning rigidity, a behavior largely unexplored in prior work, we introduce a expert-curated diagnostic set, . Our dataset includes specially modified variants of existing mathematical benchmarks, namely AIME and MATH500, as well as well-known puzzles deliberately redesigned to require deviation from familiar reasoning strategies. Using this dataset, we identify recurring contamination patterns that occur when models default to ingrained reasoning. Specifically, we categorize this contamination into three distinctive modes: (i) Interpretation Overload, (ii) Input Distrust, and (iii) Partial Instruction Attention, each causing models to ignore or distort provided instructions. We publicly release our diagnostic set to facilitate future research on mitigating reasoning rigidity in language models.</p> - https://arxiv.org/abs/2505.17225 - Thu, 22 May 2025 19:00:01 +0000 + Feedback Friction: LLMs Struggle to Fully Incorporate External Feedback + https://arxiv.org/abs/2506.11930 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.11930.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Dongwei Jiang, Alvin Zhang, Andrew Wang, Nicholas Andrews, Daniel Khashabi</p><p><b>Upvotes:</b> 53</p><p><b>Summary:</b> Recent studies have shown LLMs possess some ability to improve their responses when given external feedback. However, it remains unclear how effectively and thoroughly these models can incorporate extrinsic feedback. In an ideal scenario, if LLMs receive near-perfect and complete feedback, we would expect them to fully integrate the feedback and change their incorrect answers to correct ones. In this paper, we systematically investigate LLMs' ability to incorporate feedback by designing a controlled experimental environment. For each problem, a solver model attempts a solution, then a feedback generator with access to near-complete ground-truth answers produces targeted feedback, after which the solver tries again. We evaluate this pipeline across a diverse range of tasks, including math reasoning, knowledge reasoning, scientific reasoning, and general multi-domain evaluations with state-of-the-art language models including Claude 3.7 (with and without extended thinking). Surprisingly, even under these near-ideal conditions, solver models consistently show resistance to feedback, a limitation that we term FEEDBACK FRICTION. To mitigate this limitation, we experiment with sampling-based strategies like progressive temperature increases and explicit rejection of previously attempted incorrect answers, which yield improvements but still fail to help models achieve target performance. We also perform a rigorous exploration of potential causes of FEEDBACK FRICTION, ruling out factors such as model overconfidence and data familiarity. We hope that highlighting this issue in LLMs and ruling out several apparent causes will help future research in self-improvement.</p> + https://arxiv.org/abs/2506.11930 + Fri, 13 Jun 2025 16:31:51 +0000 - Distilling LLM Agent into Small Models with Retrieval and Code Tools - https://arxiv.org/abs/2505.17612 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.17612.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Minki Kang, Jongwon Jeong, Seanie Lee, Jaewoong Cho, Sung Ju Hwang</p><p><b>Upvotes:</b> 75</p><p><b>Summary:</b> Large language models (LLMs) excel at complex reasoning tasks but remain computationally expensive, limiting their practical deployment. To address this, recent works have focused on distilling reasoning capabilities into smaller language models (sLMs) using chain-of-thought (CoT) traces from teacher LLMs. However, this approach struggles in scenarios requiring rare factual knowledge or precise computation, where sLMs often hallucinate due to limited capability. In this work, we propose Agent Distillation, a framework for transferring not only reasoning capability but full task-solving behavior from LLM-based agents into sLMs with retrieval and code tools. We improve agent distillation along two complementary axes: (1) we introduce a prompting method called first-thought prefix to enhance the quality of teacher-generated trajectories; and (2) we propose a self-consistent action generation for improving test-time robustness of small agents. We evaluate our method on eight reasoning tasks across factual and mathematical domains, covering both in-domain and out-of-domain generalization. Our results show that sLMs as small as 0.5B, 1.5B, 3B parameters can achieve performance competitive with next-tier larger 1.5B, 3B, 7B models fine-tuned using CoT distillation, demonstrating the potential of agent distillation for building practical, tool-using small agents. Our code is available at https://github.com/Nardien/agent-distillation.</p> - https://arxiv.org/abs/2505.17612 - Fri, 23 May 2025 08:20:15 +0000 + CMI-Bench: A Comprehensive Benchmark for Evaluating Music Instruction Following + https://arxiv.org/abs/2506.12285 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.12285.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yinghao Ma, Siyou Li, Juntao Yu, Emmanouil Benetos, Akira Maezawa</p><p><b>Upvotes:</b> 53</p><p><b>Summary:</b> Recent advances in audio-text large language models (LLMs) have opened new possibilities for music understanding and generation. However, existing benchmarks are limited in scope, often relying on simplified tasks or multi-choice evaluations that fail to reflect the complexity of real-world music analysis. We reinterpret a broad range of traditional MIR annotations as instruction-following formats and introduce CMI-Bench, a comprehensive music instruction following benchmark designed to evaluate audio-text LLMs on a diverse set of music information retrieval (MIR) tasks. These include genre classification, emotion regression, emotion tagging, instrument classification, pitch estimation, key detection, lyrics transcription, melody extraction, vocal technique recognition, instrument performance technique detection, music tagging, music captioning, and (down)beat tracking: reflecting core challenges in MIR research. Unlike previous benchmarks, CMI-Bench adopts standardized evaluation metrics consistent with previous state-of-the-art MIR models, ensuring direct comparability with supervised approaches. We provide an evaluation toolkit supporting all open-source audio-textual LLMs, including LTU, Qwen-audio, SALMONN, MusiLingo, etc. Experiment results reveal significant performance gaps between LLMs and supervised models, along with their culture, chronological and gender bias, highlighting the potential and limitations of current models in addressing MIR tasks. CMI-Bench establishes a unified foundation for evaluating music instruction following, driving progress in music-aware LLMs.</p> + https://arxiv.org/abs/2506.12285 + Sat, 14 Jun 2025 00:18:44 +0000 - Scaling Image and Video Generation via Test-Time Evolutionary Search - https://arxiv.org/abs/2505.17618 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.17618.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Haoran He, Jiajun Liang, Xintao Wang, Pengfei Wan, Di Zhang, Kun Gai, Ling Pan</p><p><b>Upvotes:</b> 38</p><p><b>Summary:</b> As the marginal cost of scaling computation (data and parameters) during model pre-training continues to increase substantially, test-time scaling (TTS) has emerged as a promising direction for improving generative model performance by allocating additional computation at inference time. While TTS has demonstrated significant success across multiple language tasks, there remains a notable gap in understanding the test-time scaling behaviors of image and video generative models (diffusion-based or flow-based models). Although recent works have initiated exploration into inference-time strategies for vision tasks, these approaches face critical limitations: being constrained to task-specific domains, exhibiting poor scalability, or falling into reward over-optimization that sacrifices sample diversity. In this paper, we propose Evolutionary Search (EvoSearch), a novel, generalist, and efficient TTS method that effectively enhances the scalability of both image and video generation across diffusion and flow models, without requiring additional training or model expansion. EvoSearch reformulates test-time scaling for diffusion and flow models as an evolutionary search problem, leveraging principles from biological evolution to efficiently explore and refine the denoising trajectory. By incorporating carefully designed selection and mutation mechanisms tailored to the stochastic differential equation denoising process, EvoSearch iteratively generates higher-quality offspring while preserving population diversity. Through extensive evaluation across both diffusion and flow architectures for image and video generation tasks, we demonstrate that our method consistently outperforms existing approaches, achieves higher diversity, and shows strong generalizability to unseen evaluation metrics. Our project is available at the website https://tinnerhrhe.github.io/evosearch.</p> - https://arxiv.org/abs/2505.17618 - Fri, 23 May 2025 08:25:46 +0000 + Scaling Test-time Compute for LLM Agents + https://arxiv.org/abs/2506.12928 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.12928.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> King Zhu, Hanhao Li, Siwei Wu, Tianshun Xing, Dehua Ma, Xiangru Tang, Minghao Liu, Jian Yang, Jiaheng Liu, Yuchen Eleanor Jiang, Changwang Zhang, Chenghua Lin, Jun Wang, Ge Zhang, Wangchunshu Zhou</p><p><b>Upvotes:</b> 59</p><p><b>Summary:</b> Scaling test time compute has shown remarkable success in improving the reasoning abilities of large language models (LLMs). In this work, we conduct the first systematic exploration of applying test-time scaling methods to language agents and investigate the extent to which it improves their effectiveness. Specifically, we explore different test-time scaling strategies, including: (1) parallel sampling algorithms; (2) sequential revision strategies; (3) verifiers and merging methods; (4)strategies for diversifying rollouts.We carefully analyze and ablate the impact of different design strategies on applying test-time scaling on language agents, and have follow findings: 1. Scaling test time compute could improve the performance of agents. 2. Knowing when to reflect is important for agents. 3. Among different verification and result merging approaches, the list-wise method performs best. 4. Increasing diversified rollouts exerts a positive effect on the agent's task performance.</p> + https://arxiv.org/abs/2506.12928 + Sun, 15 Jun 2025 17:59:47 +0000 - QwenLong-L1: Towards Long-Context Large Reasoning Models with Reinforcement Learning - https://arxiv.org/abs/2505.17667 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.17667.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Fanqi Wan, Weizhou Shen, Shengyi Liao, Yingcheng Shi, Chenliang Li, Ziyi Yang, Ji Zhang, Fei Huang, Jingren Zhou, Ming Yan</p><p><b>Upvotes:</b> 80</p><p><b>Summary:</b> Recent large reasoning models (LRMs) have demonstrated strong reasoning capabilities through reinforcement learning (RL). These improvements have primarily been observed within the short-context reasoning tasks. In contrast, extending LRMs to effectively process and reason on long-context inputs via RL remains a critical unsolved challenge. To bridge this gap, we first formalize the paradigm of long-context reasoning RL, and identify key challenges in suboptimal training efficiency and unstable optimization process. To address these issues, we propose QwenLong-L1, a framework that adapts short-context LRMs to long-context scenarios via progressive context scaling. Specifically, we utilize a warm-up supervised fine-tuning (SFT) stage to establish a robust initial policy, followed by a curriculum-guided phased RL technique to stabilize the policy evolution, and enhanced with a difficulty-aware retrospective sampling strategy to incentivize the policy exploration. Experiments on seven long-context document question-answering benchmarks demonstrate that QwenLong-L1-32B outperforms flagship LRMs like OpenAI-o3-mini and Qwen3-235B-A22B, achieving performance on par with Claude-3.7-Sonnet-Thinking, demonstrating leading performance among state-of-the-art LRMs. This work advances the development of practical long-context LRMs capable of robust reasoning across information-intensive environments.</p> - https://arxiv.org/abs/2505.17667 - Fri, 23 May 2025 09:31:55 +0000 + MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attention + https://arxiv.org/abs/2506.13585 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.13585.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> MiniMax, Aili Chen, Aonian Li, Bangwei Gong, Binyang Jiang, Bo Fei, Bo Yang, Boji Shan, Changqing Yu, Chao Wang, Cheng Zhu, Chengjun Xiao, Chengyu Du, Chi Zhang, Chu Qiao, Chunhao Zhang, Chunhui Du, Congchao Guo, Da Chen, Deming Ding, Dianjun Sun, Dong Li, Enwei Jiao, Haigang Zhou, Haimo Zhang, Han Ding, Haohai Sun, Haoyu Feng, Huaiguang Cai, Haichao Zhu, Jian Sun, Jiaqi Zhuang, Jiaren Cai, Jiayuan Song, Jin Zhu, Jingyang Li, Jinhao Tian, Jinli Liu, Junhao Xu, Junjie Yan, Junteng Liu, Junxian He, Kaiyi Feng, Ke Yang, Kecheng Xiao, Le Han, Leyang Wang, Lianfei Yu, Liheng Feng, Lin Li, Lin Zheng, Linge Du, Lingyu Yang, Lunbin Zeng, Minghui Yu, Mingliang Tao, Mingyuan Chi, Mozhi Zhang, Mujie Lin, Nan Hu, Nongyu Di, Peng Gao, Pengfei Li, Pengyu Zhao, Qibing Ren, Qidi Xu, Qile Li, Qin Wang, Rong Tian, Ruitao Leng, Shaoxiang Chen, Shaoyu Chen, Shengmin Shi, Shitong Weng, Shuchang Guan, Shuqi Yu, Sichen Li, Songquan Zhu, Tengfei Li, Tianchi Cai, Tianrun Liang, Weiyu Cheng, Weize Kong, Wenkai Li, Xiancai Chen, Xiangjun Song, Xiao Luo, Xiao Su, Xiaobo Li, Xiaodong Han, Xinzhu Hou, Xuan Lu, Xun Zou, Xuyang Shen, Yan Gong, Yan Ma, Yang Wang, Yiqi Shi, Yiran Zhong, Yonghong Duan, Yongxiang Fu, Yongyi Hu, Yu Gao, Yuanxiang Fan, Yufeng Yang, Yuhao Li, Yulin Hu, Yunan Huang, Yunji Li, Yunzhi Xu, Yuxin Mao, Yuxuan Shi, Yuze Wenren, Zehan Li, Zelin Li, Zhanxu Tian, Zhengmao Zhu, Zhenhua Fan, Zhenzhen Wu, Zhichao Xu, Zhihang Yu, Zhiheng Lyu, Zhuo Jiang, Zibo Gao, Zijia Wu, Zijian Song, Zijun Sun</p><p><b>Upvotes:</b> 248</p><p><b>Summary:</b> We introduce MiniMax-M1, the world's first open-weight, large-scale hybrid-attention reasoning model. MiniMax-M1 is powered by a hybrid Mixture-of-Experts (MoE) architecture combined with a lightning attention mechanism. The model is developed based on our previous MiniMax-Text-01 model, which contains a total of 456 billion parameters with 45.9 billion parameters activated per token. The M1 model natively supports a context length of 1 million tokens, 8x the context size of DeepSeek R1. Furthermore, the lightning attention mechanism in MiniMax-M1 enables efficient scaling of test-time compute. These properties make M1 particularly suitable for complex tasks that require processing long inputs and thinking extensively. MiniMax-M1 is trained using large-scale reinforcement learning (RL) on diverse problems including sandbox-based, real-world software engineering environments. In addition to M1's inherent efficiency advantage for RL training, we propose CISPO, a novel RL algorithm to further enhance RL efficiency. CISPO clips importance sampling weights rather than token updates, outperforming other competitive RL variants. Combining hybrid-attention and CISPO enables MiniMax-M1's full RL training on 512 H800 GPUs to complete in only three weeks, with a rental cost of just $534,700. We release two versions of MiniMax-M1 models with 40K and 80K thinking budgets respectively, where the 40K model represents an intermediate phase of the 80K training. Experiments on standard benchmarks show that our models are comparable or superior to strong open-weight models such as the original DeepSeek-R1 and Qwen3-235B, with particular strengths in complex software engineering, tool utilization, and long-context tasks. We publicly release MiniMax-M1 at https://github.com/MiniMax-AI/MiniMax-M1.</p> + https://arxiv.org/abs/2506.13585 + Mon, 16 Jun 2025 15:08:02 +0000 - Don't Overthink it. Preferring Shorter Thinking Chains for Improved LLM Reasoning - https://arxiv.org/abs/2505.17813 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.17813.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Michael Hassid, Gabriel Synnaeve, Yossi Adi, Roy Schwartz</p><p><b>Upvotes:</b> 50</p><p><b>Summary:</b> Reasoning large language models (LLMs) heavily rely on scaling test-time compute to perform complex reasoning tasks by generating extensive "thinking" chains. While demonstrating impressive results, this approach incurs significant computational costs and inference time. In this work, we challenge the assumption that long thinking chains results in better reasoning capabilities. We first demonstrate that shorter reasoning chains within individual questions are significantly more likely to yield correct answers - up to 34.5% more accurate than the longest chain sampled for the same question. Based on these results, we suggest short-m@k, a novel reasoning LLM inference method. Our method executes k independent generations in parallel and halts computation once the first m thinking processes are done. The final answer is chosen using majority voting among these m chains. Basic short-1@k demonstrates similar or even superior performance over standard majority voting in low-compute settings - using up to 40% fewer thinking tokens. short-3@k, while slightly less efficient than short-1@k, consistently surpasses majority voting across all compute budgets, while still being substantially faster (up to 33% wall time reduction). Inspired by our results, we finetune an LLM using short, long, and randomly selected reasoning chains. We then observe that training on the shorter ones leads to better performance. Our findings suggest rethinking current methods of test-time compute in reasoning LLMs, emphasizing that longer "thinking" does not necessarily translate to improved performance and can, counter-intuitively, lead to degraded results.</p> - https://arxiv.org/abs/2505.17813 - Fri, 23 May 2025 12:29:06 +0000 + Ego-R1: Chain-of-Tool-Thought for Ultra-Long Egocentric Video Reasoning + https://arxiv.org/abs/2506.13654 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.13654.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Shulin Tian, Ruiqi Wang, Hongming Guo, Penghao Wu, Yuhao Dong, Xiuying Wang, Jingkang Yang, Hao Zhang, Hongyuan Zhu, Ziwei Liu</p><p><b>Upvotes:</b> 42</p><p><b>Summary:</b> We introduce Ego-R1, a novel framework for reasoning over ultra-long (i.e., in days and weeks) egocentric videos, which leverages a structured Chain-of-Tool-Thought (CoTT) process, orchestrated by an Ego-R1 Agent trained via reinforcement learning (RL). Inspired by human problem-solving strategies, CoTT decomposes complex reasoning into modular steps, with the RL agent invoking specific tools, one per step, to iteratively and collaboratively answer sub-questions tackling such tasks as temporal retrieval and multi-modal understanding. We design a two-stage training paradigm involving supervised finetuning (SFT) of a pretrained language model using CoTT data and RL to enable our agent to dynamically propose step-by-step tools for long-range reasoning. To facilitate training, we construct a dataset called Ego-R1 Data, which consists of Ego-CoTT-25K for SFT and Ego-QA-4.4K for RL. Furthermore, our Ego-R1 agent is evaluated on a newly curated week-long video QA benchmark, Ego-R1 Bench, which contains human-verified QA pairs from hybrid sources. Extensive results demonstrate that the dynamic, tool-augmented chain-of-thought reasoning by our Ego-R1 Agent can effectively tackle the unique challenges of understanding ultra-long egocentric videos, significantly extending the time coverage from few hours to a week.</p> + https://arxiv.org/abs/2506.13654 + Mon, 16 Jun 2025 16:17:08 +0000 - Mutarjim: Advancing Bidirectional Arabic-English Translation with a Small Language Model - https://arxiv.org/abs/2505.17894 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.17894.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Khalil Hennara, Muhammad Hreden, Mohamed Motaism Hamed, Zeina Aldallal, Sara Chrouf, Safwan AlModhayan</p><p><b>Upvotes:</b> 205</p><p><b>Summary:</b> We introduce Mutarjim, a compact yet powerful language model for bidirectional Arabic-English translation. While large-scale LLMs have shown impressive progress in natural language processing tasks, including machine translation, smaller models. Leveraging this insight, we developed Mutarjim based on Kuwain-1.5B , a language model tailored for both Arabic and English. Despite its modest size, Mutarjim outperforms much larger models on several established benchmarks, achieved through an optimized two-phase training approach and a carefully curated, high-quality training corpus.. Experimental results show that Mutarjim rivals models up to 20 times larger while significantly reducing computational costs and training requirements. We also introduce Tarjama-25, a new benchmark designed to overcome limitations in existing Arabic-English benchmarking datasets, such as domain narrowness, short sentence lengths, and English-source bias. Tarjama-25 comprises 5,000 expert-reviewed sentence pairs and spans a wide range of domains, offering a more comprehensive and balanced evaluation framework. Notably, Mutarjim achieves state-of-the-art performance on the English-to-Arabic task in Tarjama-25, surpassing even significantly larger and proprietary models like GPT-4o mini. We publicly release Tarjama-25 to support future research and advance the evaluation of Arabic-English translation systems.</p> - https://arxiv.org/abs/2505.17894 - Fri, 23 May 2025 13:42:21 +0000 + Discrete Diffusion in Large Language and Multimodal Models: A Survey + https://arxiv.org/abs/2506.13759 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.13759.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Runpeng Yu, Qi Li, Xinchao Wang</p><p><b>Upvotes:</b> 41</p><p><b>Summary:</b> In this work, we provide a systematic survey of Discrete Diffusion Language Models (dLLMs) and Discrete Diffusion Multimodal Language Models (dMLLMs). Unlike autoregressive (AR) models, dLLMs and dMLLMs adopt a multi-token, parallel decoding paradigm using full attention and a denoising-based generation strategy. This paradigm naturally enables parallel generation, fine-grained output controllability, and dynamic, response-aware perception. These capabilities are previously difficult to achieve with AR models. Recently, a growing number of industrial-scale proprietary d(M)LLMs, as well as a large number of open-source academic d(M)LLMs, have demonstrated performance comparable to their autoregressive counterparts, while achieving up to 10x acceleration in inference speed. The advancement of discrete diffusion LLMs and MLLMs has been largely driven by progress in two domains. The first is the development of autoregressive LLMs and MLLMs, which has accumulated vast amounts of data, benchmarks, and foundational infrastructure for training and inference. The second contributing domain is the evolution of the mathematical models underlying discrete diffusion. Together, these advancements have catalyzed a surge in dLLMs and dMLLMs research in early 2025. In this work, we present a comprehensive overview of the research in the dLLM and dMLLM domains. We trace the historical development of dLLMs and dMLLMs, formalize the underlying mathematical frameworks, and categorize representative models. We further analyze key techniques for training and inference, and summarize emerging applications across language, vision-language, and biological domains. We conclude by discussing future directions for research and deployment. Paper collection: https://github.com/LiQiiiii/DLLM-Survey</p> + https://arxiv.org/abs/2506.13759 + Mon, 16 Jun 2025 17:59:08 +0000 - QwenLong-CPRS: Towards infty-LLMs with Dynamic Context Optimization - https://arxiv.org/abs/2505.18092 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.18092.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Weizhou Shen, Chenliang Li, Fanqi Wan, Shengyi Liao, Shaopeng Lai, Bo Zhang, Yingcheng Shi, Yuning Wu, Gang Fu, Zhansheng Li, Bin Yang, Ji Zhang, Fei Huang, Jingren Zhou, Ming Yan</p><p><b>Upvotes:</b> 39</p><p><b>Summary:</b> This technical report presents QwenLong-CPRS, a context compression framework designed for explicit long-context optimization, addressing prohibitive computation overhead during the prefill stage and the "lost in the middle" performance degradation of large language models (LLMs) during long sequence processing. Implemented through a novel dynamic context optimization mechanism, QwenLong-CPRS enables multi-granularity context compression guided by natural language instructions, achieving both efficiency gains and improved performance. Evolved from the Qwen architecture series, QwenLong-CPRS introduces four key innovations: (1) Natural language-guided dynamic optimization, (2) Bidirectional reasoning layers for enhanced boundary awareness, (3) Token critic mechanisms with language modeling heads, and (4) Window-parallel inference. Comprehensive evaluations across five benchmarks (4K-2M word contexts) demonstrate QwenLong-CPRS's threefold effectiveness: (1) Consistent superiority over other context management methods like RAG and sparse attention in both accuracy and efficiency. (2) Architecture-agnostic integration with all flagship LLMs, including GPT-4o, Gemini2.0-pro, Claude3.7-sonnet, DeepSeek-v3, and Qwen2.5-max, achieves 21.59times context compression alongside 19.15-point average performance gains; (3) Deployed with Qwen2.5-32B-Instruct, QwenLong-CPRS surpasses leading proprietary LLMs by 4.85 and 10.88 points on Ruler-128K and InfiniteBench, establishing new SOTA performance.</p> - https://arxiv.org/abs/2505.18092 - Fri, 23 May 2025 16:47:00 +0000 + MultiFinBen: A Multilingual, Multimodal, and Difficulty-Aware Benchmark for Financial LLM Evaluation + https://arxiv.org/abs/2506.14028 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.14028.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Xueqing Peng, Lingfei Qian, Yan Wang, Ruoyu Xiang, Yueru He, Yang Ren, Mingyang Jiang, Jeff Zhao, Huan He, Yi Han, Yun Feng, Yuechen Jiang, Yupeng Cao, Haohang Li, Yangyang Yu, Xiaoyu Wang, Penglei Gao, Shengyuan Lin, Keyi Wang, Shanshan Yang, Yilun Zhao, Zhiwei Liu, Peng Lu, Jerry Huang, Suyuchen Wang, Triantafillos Papadopoulos, Polydoros Giannouris, Efstathia Soufleri, Nuo Chen, Guojun Xiong, Zhiyang Deng, Yijia Zhao, Mingquan Lin, Meikang Qiu, Kaleb E Smith, Arman Cohan, Xiao-Yang Liu, Jimin Huang, Alejandro Lopez-Lira, Xi Chen, Junichi Tsujii, Jian-Yun Nie, Sophia Ananiadou, Qianqian Xie</p><p><b>Upvotes:</b> 90</p><p><b>Summary:</b> Recent advances in large language models (LLMs) have accelerated progress in financial NLP and applications, yet existing benchmarks remain limited to monolingual and unimodal settings, often over-relying on simple tasks and failing to reflect the complexity of real-world financial communication. We introduce MultiFinBen, the first multilingual and multimodal benchmark tailored to the global financial domain, evaluating LLMs across modalities (text, vision, audio) and linguistic settings (monolingual, bilingual, multilingual) on domain-specific tasks. We introduce two novel tasks, including PolyFiQA-Easy and PolyFiQA-Expert, the first multilingual financial benchmarks requiring models to perform complex reasoning over mixed-language inputs; and EnglishOCR and SpanishOCR, the first OCR-embedded financial QA tasks challenging models to extract and reason over information from visual-text financial documents. Moreover, we propose a dynamic, difficulty-aware selection mechanism and curate a compact, balanced benchmark rather than simple aggregation existing datasets. Extensive evaluation of 22 state-of-the-art models reveals that even the strongest models, despite their general multimodal and multilingual capabilities, struggle dramatically when faced with complex cross-lingual and multimodal tasks in financial domain. MultiFinBen is publicly released to foster transparent, reproducible, and inclusive progress in financial studies and applications.</p> + https://arxiv.org/abs/2506.14028 + Mon, 16 Jun 2025 22:01:49 +0000 - TabSTAR: A Foundation Tabular Model With Semantically Target-Aware Representations - https://arxiv.org/abs/2505.18125 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.18125.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Alan Arazi, Eilam Shapira, Roi Reichart</p><p><b>Upvotes:</b> 106</p><p><b>Summary:</b> While deep learning has achieved remarkable success across many domains, it has historically underperformed on tabular learning tasks, which remain dominated by gradient boosting decision trees (GBDTs). However, recent advancements are paving the way for Tabular Foundation Models, which can leverage real-world knowledge and generalize across diverse datasets, particularly when the data contains free-text. Although incorporating language model capabilities into tabular tasks has been explored, most existing methods utilize static, target-agnostic textual representations, limiting their effectiveness. We introduce TabSTAR: a Foundation Tabular Model with Semantically Target-Aware Representations. TabSTAR is designed to enable transfer learning on tabular data with textual features, with an architecture free of dataset-specific parameters. It unfreezes a pretrained text encoder and takes as input target tokens, which provide the model with the context needed to learn task-specific embeddings. TabSTAR achieves state-of-the-art performance for both medium- and large-sized datasets across known benchmarks of classification tasks with text features, and its pretraining phase exhibits scaling laws in the number of datasets, offering a pathway for further performance improvements.</p> - https://arxiv.org/abs/2505.18125 - Fri, 23 May 2025 17:34:28 +0000 + Essential-Web v1.0: 24T tokens of organized web data + https://arxiv.org/abs/2506.14111 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.14111.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Essential AI, Andrew Hojel, Michael Pust, Tim Romanski, Yash Vanjani, Ritvik Kapila, Mohit Parmar, Adarsh Chaluvaraju, Alok Tripathy, Anil Thomas, Ashish Tanwer, Darsh J Shah, Ishaan Shah, Karl Stratos, Khoi Nguyen, Kurt Smith, Michael Callahan, Peter Rushton, Philip Monk, Platon Mazarakis, Saad Jamal, Saurabh Srivastava, Somanshu Singla, Ashish Vaswani</p><p><b>Upvotes:</b> 39</p><p><b>Summary:</b> Data plays the most prominent role in how language models acquire skills and knowledge. The lack of massive, well-organized pre-training datasets results in costly and inaccessible data pipelines. We present Essential-Web v1.0, a 24-trillion-token dataset in which every document is annotated with a twelve-category taxonomy covering topic, format, content complexity, and quality. Taxonomy labels are produced by EAI-Distill-0.5b, a fine-tuned 0.5b-parameter model that achieves an annotator agreement within 3% of Qwen2.5-32B-Instruct. With nothing more than SQL-style filters, we obtain competitive web-curated datasets in math (-8.0% relative to SOTA), web code (+14.3%), STEM (+24.5%) and medical (+8.6%). Essential-Web v1.0 is available on HuggingFace: https://huggingface.co/datasets/EssentialAI/essential-web-v1.0</p> + https://arxiv.org/abs/2506.14111 + Tue, 17 Jun 2025 02:03:36 +0000 - One RL to See Them All: Visual Triple Unified Reinforcement Learning - https://arxiv.org/abs/2505.18129 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.18129.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yan Ma, Linge Du, Xuyang Shen, Shaoxiang Chen, Pengfei Li, Qibing Ren, Lizhuang Ma, Yuchao Dai, Pengfei Liu, Junjie Yan</p><p><b>Upvotes:</b> 56</p><p><b>Summary:</b> Reinforcement learning (RL) has significantly advanced the reasoning capabilities of vision-language models (VLMs). However, the use of RL beyond reasoning tasks remains largely unexplored, especially for perceptionintensive tasks like object detection and grounding. We propose V-Triune, a Visual Triple Unified Reinforcement Learning system that enables VLMs to jointly learn visual reasoning and perception tasks within a single training pipeline. V-Triune comprises triple complementary components: Sample-Level Data Formatting (to unify diverse task inputs), Verifier-Level Reward Computation (to deliver custom rewards via specialized verifiers) , and Source-Level Metric Monitoring (to diagnose problems at the data-source level). We further introduce a novel Dynamic IoU reward, which provides adaptive, progressive, and definite feedback for perception tasks handled by V-Triune. Our approach is instantiated within off-the-shelf RL training framework using open-source 7B and 32B backbone models. The resulting model, dubbed Orsta (One RL to See Them All), demonstrates consistent improvements across both reasoning and perception tasks. This broad capability is significantly shaped by its training on a diverse dataset, constructed around four representative visual reasoning tasks (Math, Puzzle, Chart, and Science) and four visual perception tasks (Grounding, Detection, Counting, and OCR). Subsequently, Orsta achieves substantial gains on MEGA-Bench Core, with improvements ranging from +2.1 to an impressive +14.1 across its various 7B and 32B model variants, with performance benefits extending to a wide range of downstream tasks. These results highlight the effectiveness and scalability of our unified RL approach for VLMs. The V-Triune system, along with the Orsta models, is publicly available at https://github.com/MiniMax-AI.</p> - https://arxiv.org/abs/2505.18129 - Fri, 23 May 2025 17:41:14 +0000 + Xolver: Multi-Agent Reasoning with Holistic Experience Learning Just Like an Olympiad Team + https://arxiv.org/abs/2506.14234 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.14234.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Md Tanzib Hosain, Salman Rahman, Md Kishor Morol, Md Rizwan Parvez</p><p><b>Upvotes:</b> 37</p><p><b>Summary:</b> Despite impressive progress on complex reasoning, current large language models (LLMs) typically operate in isolation - treating each problem as an independent attempt, without accumulating or integrating experiential knowledge. In contrast, expert problem solvers - such as Olympiad or programming contest teams - leverage a rich tapestry of experiences: absorbing mentorship from coaches, developing intuition from past problems, leveraging knowledge of tool usage and library functionality, adapting strategies based on the expertise and experiences of peers, continuously refining their reasoning through trial and error, and learning from other related problems even during competition. We introduce Xolver, a training-free multi-agent reasoning framework that equips a black-box LLM with a persistent, evolving memory of holistic experience. Xolver integrates diverse experience modalities, including external and self-retrieval, tool use, collaborative interactions, agent-driven evaluation, and iterative refinement. By learning from relevant strategies, code fragments, and abstract reasoning patterns at inference time, Xolver avoids generating solutions from scratch - marking a transition from isolated inference toward experience-aware language agents. Built on both open-weight and proprietary models, Xolver consistently outperforms specialized reasoning agents. Even with lightweight backbones (e.g., QWQ-32B), it often surpasses advanced models including Qwen3-235B, Gemini 2.5 Pro, o3, and o4-mini-high. With o3-mini-high, it achieves new best results on GSM8K (98.1%), AIME'24 (94.4%), AIME'25 (93.7%), Math-500 (99.8%), and LiveCodeBench-V5 (91.6%) - highlighting holistic experience learning as a key step toward generalist agents capable of expert-level reasoning. Code and data are available at https://kagnlp.github.io/xolver.github.io/.</p> + https://arxiv.org/abs/2506.14234 + Tue, 17 Jun 2025 06:47:19 +0000 - Guided by Gut: Efficient Test-Time Scaling with Reinforced Intrinsic Confidence - https://arxiv.org/abs/2505.20325 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.20325.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Amirhosein Ghasemabadi, Keith G. Mills, Baochun Li, Di Niu</p><p><b>Upvotes:</b> 42</p><p><b>Summary:</b> Test-Time Scaling (TTS) methods for enhancing Large Language Model (LLM) reasoning often incur substantial computational costs, primarily due to extensive reliance on external Process Reward Models (PRMs) or sampling methods like Best-of-N (BoN). This paper introduces Guided by Gut (GG), an efficient self-guided TTS framework that achieves PRM-level performance without costly external verifier models. Our method employs a lightweight tree search guided solely by intrinsic LLM signals, token-level confidence and step novelty. One critical innovation is improving the reliability of internal confidence estimates via a targeted reinforcement learning fine-tuning phase. Empirical evaluations on challenging mathematical reasoning benchmarks demonstrate that GG enables smaller models (e.g., 1.5B parameters) to achieve accuracy matching or surpassing significantly larger models (e.g., 32B-70B parameters), while reducing GPU memory usage by up to 10x. Compared to PRM-based methods, GG achieves comparable accuracy with 8x faster inference speeds and 4-5x lower memory usage. Additionally, GG reduces KV cache memory usage by approximately 50% compared to the BoN strategy, facilitating more efficient and practical deployment of TTS techniques.</p> - https://arxiv.org/abs/2505.20325 - Fri, 23 May 2025 18:19:09 +0000 + Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMs + https://arxiv.org/abs/2506.14245 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.14245.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Xumeng Wen, Zihan Liu, Shun Zheng, Zhijian Xu, Shengyu Ye, Zhirong Wu, Xiao Liang, Yang Wang, Junjie Li, Ziming Miao, Jiang Bian, Mao Yang</p><p><b>Upvotes:</b> 36</p><p><b>Summary:</b> Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a promising paradigm for advancing the reasoning capabilities of Large Language Models (LLMs). However, a critical paradox clouds its efficacy: RLVR-tuned models often underperform their base models on the Pass@K metric for solution-finding, leading to the hypothesis that RLVR merely re-weights existing reasoning paths at the cost of reasoning diversity. In this work, we resolve this contradiction by identifying the source of the problem: the Pass@K metric itself is a flawed measure of reasoning, as it credits correct final answers that probably arise from inaccurate or incomplete chains of thought (CoTs). To address this, we introduce a more precise evaluation metric, CoT-Pass@K, which mandates that both the reasoning path and the final answer be correct. We provide a new theoretical foundation that formalizes how RLVR, unlike traditional RL, is uniquely structured to incentivize logical integrity. Our empirical results are supportive: using CoT-Pass@K, we observe that RLVR can incentivize the generalization of correct reasoning for all values of K. Furthermore, by analyzing the training dynamics, we find that this enhanced reasoning capability emerges early in the training process and smoothly generalizes. Our work provides a clear perspective on the role of RLVR, offers a more reliable method for its evaluation, and confirms its potential to genuinely advance machine reasoning.</p> + https://arxiv.org/abs/2506.14245 + Tue, 17 Jun 2025 07:06:56 +0000 - OmniConsistency: Learning Style-Agnostic Consistency from Paired Stylization Data - https://arxiv.org/abs/2505.18445 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.18445.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yiren Song, Cheng Liu, Mike Zheng Shou</p><p><b>Upvotes:</b> 62</p><p><b>Summary:</b> Diffusion models have advanced image stylization significantly, yet two core challenges persist: (1) maintaining consistent stylization in complex scenes, particularly identity, composition, and fine details, and (2) preventing style degradation in image-to-image pipelines with style LoRAs. GPT-4o's exceptional stylization consistency highlights the performance gap between open-source methods and proprietary models. To bridge this gap, we propose OmniConsistency, a universal consistency plugin leveraging large-scale Diffusion Transformers (DiTs). OmniConsistency contributes: (1) an in-context consistency learning framework trained on aligned image pairs for robust generalization; (2) a two-stage progressive learning strategy decoupling style learning from consistency preservation to mitigate style degradation; and (3) a fully plug-and-play design compatible with arbitrary style LoRAs under the Flux framework. Extensive experiments show that OmniConsistency significantly enhances visual coherence and aesthetic quality, achieving performance comparable to commercial state-of-the-art model GPT-4o.</p> - https://arxiv.org/abs/2505.18445 - Sat, 24 May 2025 01:00:20 +0000 + LongLLaDA: Unlocking Long Context Capabilities in Diffusion LLMs + https://arxiv.org/abs/2506.14429 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.14429.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Xiaoran Liu, Zhigeng Liu, Zengfeng Huang, Qipeng Guo, Ziwei He, Xipeng Qiu</p><p><b>Upvotes:</b> 43</p><p><b>Summary:</b> Large Language Diffusion Models, or diffusion LLMs, have emerged as a significant focus in NLP research, with substantial effort directed toward understanding their scalability and downstream task performance. However, their long-context capabilities remain unexplored, lacking systematic analysis or methods for context extension. In this work, we present the first systematic investigation comparing the long-context performance of diffusion LLMs and traditional auto-regressive LLMs. We first identify a unique characteristic of diffusion LLMs, unlike auto-regressive LLMs, they maintain remarkably \textit{stable perplexity} during direct context extrapolation. Furthermore, where auto-regressive models fail outright during the Needle-In-A-Haystack task with context exceeding their pretrained length, we discover diffusion LLMs exhibit a distinct \textit{local perception} phenomenon, enabling successful retrieval from recent context segments. We explain both phenomena through the lens of Rotary Position Embedding (RoPE) scaling theory. Building on these observations, we propose LongLLaDA, a training-free method that integrates LLaDA with the NTK-based RoPE extrapolation. Our results validate that established extrapolation scaling laws remain effective for extending the context windows of diffusion LLMs. Furthermore, we identify long-context tasks where diffusion LLMs outperform auto-regressive LLMs and others where they fall short. Consequently, this study establishes the first context extrapolation method for diffusion LLMs while providing essential theoretical insights and empirical benchmarks critical for advancing future research on long-context diffusion LLMs.</p> + https://arxiv.org/abs/2506.14429 + Tue, 17 Jun 2025 11:45:37 +0000 - Sparse VideoGen2: Accelerate Video Generation with Sparse Attention via Semantic-Aware Permutation - https://arxiv.org/abs/2505.18875 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.18875.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Shuo Yang, Haocheng Xi, Yilong Zhao, Muyang Li, Jintao Zhang, Han Cai, Yujun Lin, Xiuyu Li, Chenfeng Xu, Kelly Peng, Jianfei Chen, Song Han, Kurt Keutzer, Ion Stoica</p><p><b>Upvotes:</b> 37</p><p><b>Summary:</b> Diffusion Transformers (DiTs) are essential for video generation but suffer from significant latency due to the quadratic complexity of attention. By computing only critical tokens, sparse attention reduces computational costs and offers a promising acceleration approach. However, we identify that existing methods fail to approach optimal generation quality under the same computation budget for two reasons: (1) Inaccurate critical token identification: current methods cluster tokens based on position rather than semantics, leading to imprecise aggregated representations. (2) Excessive computation waste: critical tokens are scattered among non-critical ones, leading to wasted computation on GPUs, which are optimized for processing contiguous tokens. In this paper, we propose SVG2, a training-free framework that maximizes identification accuracy and minimizes computation waste, achieving a Pareto frontier trade-off between generation quality and efficiency. The core of SVG2 is semantic-aware permutation, which clusters and reorders tokens based on semantic similarity using k-means. This approach ensures both a precise cluster representation, improving identification accuracy, and a densified layout of critical tokens, enabling efficient computation without padding. Additionally, SVG2 integrates top-p dynamic budget control and customized kernel implementations, achieving up to 2.30x and 1.89x speedup while maintaining a PSNR of up to 30 and 26 on HunyuanVideo and Wan 2.1, respectively.</p> - https://arxiv.org/abs/2505.18875 - Sat, 24 May 2025 21:30:29 +0000 + OAgents: An Empirical Study of Building Effective Agents + https://arxiv.org/abs/2506.15741 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.15741.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> He Zhu, Tianrui Qin, King Zhu, Heyuan Huang, Yeyi Guan, Jinxiang Xia, Yi Yao, Hanhao Li, Ningning Wang, Pai Liu, Tianhao Peng, Xin Gui, Xiaowan Li, Yuhui Liu, Yuchen Eleanor Jiang, Jun Wang, Changwang Zhang, Xiangru Tang, Ge Zhang, Jian Yang, Minghao Liu, Xitong Gao, Jiaheng Liu, Wangchunshu Zhou</p><p><b>Upvotes:</b> 34</p><p><b>Summary:</b> Recently, Agentic AI has become an increasingly popular research field. However, we argue that current agent research practices lack standardization and scientific rigor, making it hard to conduct fair comparisons among methods. As a result, it is still unclear how different design choices in agent frameworks affect effectiveness, and measuring their progress remains challenging. In this work, we conduct a systematic empirical study on GAIA benchmark and BrowseComp to examine the impact of popular design choices in key agent components in a fair and rigorous manner. We find that the lack of a standard evaluation protocol makes previous works, even open-sourced ones, non-reproducible, with significant variance between random runs. Therefore, we introduce a more robust evaluation protocol to stabilize comparisons. Our study reveals which components and designs are crucial for effective agents, while others are redundant, despite seeming logical. Based on our findings, we build and open-source OAgents, a new foundation agent framework that achieves state-of-the-art performance among open-source projects. OAgents offers a modular design for various agent components, promoting future research in Agentic AI.</p> + https://arxiv.org/abs/2506.15741 + Tue, 17 Jun 2025 17:59:02 +0000 - VerIPO: Cultivating Long Reasoning in Video-LLMs via Verifier-Gudied Iterative Policy Optimization - https://arxiv.org/abs/2505.19000 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.19000.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yunxin Li, Xinyu Chen, Zitao Li, Zhenyu Liu, Longyue Wang, Wenhan Luo, Baotian Hu, Min Zhang</p><p><b>Upvotes:</b> 38</p><p><b>Summary:</b> Applying Reinforcement Learning (RL) to Video Large Language Models (Video-LLMs) shows significant promise for complex video reasoning. However, popular Reinforcement Fine-Tuning (RFT) methods, such as outcome-based Group Relative Policy Optimization (GRPO), are limited by data preparation bottlenecks (e.g., noise or high cost) and exhibit unstable improvements in the quality of long chain-of-thoughts (CoTs) and downstream performance.To address these limitations, we propose VerIPO, a Verifier-guided Iterative Policy Optimization method designed to gradually improve video LLMs' capacity for generating deep, long-term reasoning chains. The core component is Rollout-Aware Verifier, positioned between the GRPO and Direct Preference Optimization (DPO) training phases to form the GRPO-Verifier-DPO training loop. This verifier leverages small LLMs as a judge to assess the reasoning logic of rollouts, enabling the construction of high-quality contrastive data, including reflective and contextually consistent CoTs. These curated preference samples drive the efficient DPO stage (7x faster than GRPO), leading to marked improvements in reasoning chain quality, especially in terms of length and contextual consistency. This training loop benefits from GRPO's expansive search and DPO's targeted optimization. Experimental results demonstrate: 1) Significantly faster and more effective optimization compared to standard GRPO variants, yielding superior performance; 2) Our trained models exceed the direct inference of large-scale instruction-tuned Video-LLMs, producing long and contextually consistent CoTs on diverse video reasoning tasks; and 3) Our model with one iteration outperforms powerful LMMs (e.g., Kimi-VL) and long reasoning models (e.g., Video-R1), highlighting its effectiveness and stability.</p> - https://arxiv.org/abs/2505.19000 - Sun, 25 May 2025 06:41:28 +0000 + Revisiting Reinforcement Learning for LLM Reasoning from A Cross-Domain Perspective + https://arxiv.org/abs/2506.14965 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.14965.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhoujun Cheng, Shibo Hao, Tianyang Liu, Fan Zhou, Yutao Xie, Feng Yao, Yuexin Bian, Yonghao Zhuang, Nilabjo Dey, Yuheng Zha, Yi Gu, Kun Zhou, Yuqi Wang, Yuan Li, Richard Fan, Jianshu She, Chengqian Gao, Abulhair Saparov, Haonan Li, Taylor W. Killian, Mikhail Yurochkin, Zhengzhong Liu, Eric P. Xing, Zhiting Hu</p><p><b>Upvotes:</b> 43</p><p><b>Summary:</b> Reinforcement learning (RL) has emerged as a promising approach to improve large language model (LLM) reasoning, yet most open efforts focus narrowly on math and code, limiting our understanding of its broader applicability to general reasoning. A key challenge lies in the lack of reliable, scalable RL reward signals across diverse reasoning domains. We introduce Guru, a curated RL reasoning corpus of 92K verifiable examples spanning six reasoning domains--Math, Code, Science, Logic, Simulation, and Tabular--each built through domain-specific reward design, deduplication, and filtering to ensure reliability and effectiveness for RL training. Based on Guru, we systematically revisit established findings in RL for LLM reasoning and observe significant variation across domains. For example, while prior work suggests that RL primarily elicits existing knowledge from pretrained models, our results reveal a more nuanced pattern: domains frequently seen during pretraining (Math, Code, Science) easily benefit from cross-domain RL training, while domains with limited pretraining exposure (Logic, Simulation, and Tabular) require in-domain training to achieve meaningful performance gains, suggesting that RL is likely to facilitate genuine skill acquisition. Finally, we present Guru-7B and Guru-32B, two models that achieve state-of-the-art performance among open models RL-trained with publicly available data, outperforming best baselines by 7.9% and 6.7% on our 17-task evaluation suite across six reasoning domains. We also show that our models effectively improve the Pass@k performance of their base models, particularly on complex tasks less likely to appear in pretraining data. We release data, models, training and evaluation code to facilitate general-purpose reasoning at: https://github.com/LLM360/Reasoning360</p> + https://arxiv.org/abs/2506.14965 + Tue, 17 Jun 2025 20:24:00 +0000 - Shifting AI Efficiency From Model-Centric to Data-Centric Compression - https://arxiv.org/abs/2505.19147 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.19147.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Xuyang Liu, Zichen Wen, Shaobo Wang, Junjie Chen, Zhishan Tao, Yubo Wang, Xiangqi Jin, Chang Zou, Yiyu Wang, Chenfei Liao, Xu Zheng, Honggang Chen, Weijia Li, Xuming Hu, Conghui He, Linfeng Zhang</p><p><b>Upvotes:</b> 139</p><p><b>Summary:</b> The rapid advancement of large language models (LLMs) and multi-modal LLMs (MLLMs) has historically relied on model-centric scaling through increasing parameter counts from millions to hundreds of billions to drive performance gains. However, as we approach hardware limits on model size, the dominant computational bottleneck has fundamentally shifted to the quadratic cost of self-attention over long token sequences, now driven by ultra-long text contexts, high-resolution images, and extended videos. In this position paper, we argue that the focus of research for efficient AI is shifting from model-centric compression to data-centric compression. We position token compression as the new frontier, which improves AI efficiency via reducing the number of tokens during model training or inference. Through comprehensive analysis, we first examine recent developments in long-context AI across various domains and establish a unified mathematical framework for existing model efficiency strategies, demonstrating why token compression represents a crucial paradigm shift in addressing long-context overhead. Subsequently, we systematically review the research landscape of token compression, analyzing its fundamental benefits and identifying its compelling advantages across diverse scenarios. Furthermore, we provide an in-depth analysis of current challenges in token compression research and outline promising future directions. Ultimately, our work aims to offer a fresh perspective on AI efficiency, synthesize existing research, and catalyze innovative developments to address the challenges that increasing context lengths pose to the AI community's advancement.</p> - https://arxiv.org/abs/2505.19147 - Sun, 25 May 2025 13:51:17 +0000 + ProtoReasoning: Prototypes as the Foundation for Generalizable Reasoning in LLMs + https://arxiv.org/abs/2506.15211 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.15211.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Feng He, Zijun Chen, Xinnian Liang, Tingting Ma, Yunqi Qiu, Shuangzhi Wu, Junchi Yan</p><p><b>Upvotes:</b> 33</p><p><b>Summary:</b> Recent advances in Large Reasoning Models (LRMs) trained with Long Chain-of-Thought (Long CoT) reasoning have demonstrated remarkable cross-domain generalization capabilities. However, the underlying mechanisms supporting such transfer remain poorly understood. We hypothesize that cross-domain generalization arises from shared abstract reasoning prototypes -- fundamental reasoning patterns that capture the essence of problems across domains. These prototypes minimize the nuances of the representation, revealing that seemingly diverse tasks are grounded in shared reasoning structures.Based on this hypothesis, we propose ProtoReasoning, a framework that enhances the reasoning ability of LLMs by leveraging scalable and verifiable prototypical representations (Prolog for logical reasoning, PDDL for planning).ProtoReasoning features: (1) an automated prototype construction pipeline that transforms problems into corresponding prototype representations; (2) a comprehensive verification system providing reliable feedback through Prolog/PDDL interpreters; (3) the scalability to synthesize problems arbitrarily within prototype space while ensuring correctness. Extensive experiments show that ProtoReasoning achieves 4.7% improvement over baseline models on logical reasoning (Enigmata-Eval), 6.3% improvement on planning tasks, 4.0% improvement on general reasoning (MMLU) and 1.0% on mathematics (AIME24). Significantly, our ablation studies confirm that learning in prototype space also demonstrates enhanced generalization to structurally similar problems compared to training solely on natural language representations, validating our hypothesis that reasoning prototypes serve as the foundation for generalizable reasoning in large language models.</p> + https://arxiv.org/abs/2506.15211 + Wed, 18 Jun 2025 07:44:09 +0000 - PATS: Process-Level Adaptive Thinking Mode Switching - https://arxiv.org/abs/2505.19250 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.19250.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yi Wang, Junxiao Liu, Shimao Zhang, Jiajun Chen, Shujian Huang</p><p><b>Upvotes:</b> 45</p><p><b>Summary:</b> Current large-language models (LLMs) typically adopt a fixed reasoning strategy, either simple or complex, for all questions, regardless of their difficulty. This neglect of variation in task and reasoning process complexity leads to an imbalance between performance and efficiency. Existing methods attempt to implement training-free fast-slow thinking system switching to handle problems of varying difficulty, but are limited by coarse-grained solution-level strategy adjustments. To address this issue, we propose a novel reasoning paradigm: Process-Level Adaptive Thinking Mode Switching (PATS), which enables LLMs to dynamically adjust their reasoning strategy based on the difficulty of each step, optimizing the balance between accuracy and computational efficiency. Our approach integrates Process Reward Models (PRMs) with Beam Search, incorporating progressive mode switching and bad-step penalty mechanisms. Experiments on diverse mathematical benchmarks demonstrate that our methodology achieves high accuracy while maintaining moderate token usage. This study emphasizes the significance of process-level, difficulty-aware reasoning strategy adaptation, offering valuable insights into efficient inference for LLMs.</p> - https://arxiv.org/abs/2505.19250 - Sun, 25 May 2025 17:58:50 +0000 + All is Not Lost: LLM Recovery without Checkpoints + https://arxiv.org/abs/2506.15461 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.15461.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Nikolay Blagoev, Oğuzhan Ersoy, Lydia Yiyu Chen</p><p><b>Upvotes:</b> 36</p><p><b>Summary:</b> Training LLMs on decentralized and wimpy computation nodes, e.g., multiple on-spot instances, lowers the training cost and enables model democratization. The inevitable challenge here is the churn of nodes due to failures and the operator's scheduling policies, leading to losing a stage - a part of the model. The conventional approaches to recover from failures are to either use checkpointing, where periodically a copy of the entire model is sent to an additional storage, or redundant computation. These approaches yield significant communication and/or computation overhead even in non-failure cases and scale poorly in settings with large models. In this paper, we propose, CheckFree, an efficient recovery method where a failing stage is substituted by a weighted average of the closest neighboring stages. In contrast to the state of the art, CheckFree requires no additional computation or storage. However, because of the nature of averaging neighbouring stages, it can only recover failures of intermediate stages. We further extend our method to CheckFree+ with out-of-order pipeline execution to tolerate crashes of the first and last stages. Thanks to out-of-order pipelining, behaviour of those stages is mimicked by their neighboring ones, which allows CheckFree+ to recover them by simply copying the weights from the immediate neighbour. To be able to recover the (de)embedding layers, CheckFree+ copies those layers to the neighboring stages, which requires relatively small storage overhead. We extensively evaluate our method on LLaMa models of model sizes from 124M to 1.5B with varying failure frequencies. In the case of low and medium failure rates (5-10%), CheckFree and CheckFree+ outperform both checkpointing and redundant computation in terms of convergence in wall-clock time by over 12%. Both of our proposals can be run via our code available at: https://github.com/gensyn-ai/CheckFree.</p> + https://arxiv.org/abs/2506.15461 + Wed, 18 Jun 2025 13:48:33 +0000 - Alchemist: Turning Public Text-to-Image Data into Generative Gold - https://arxiv.org/abs/2505.19297 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.19297.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Valerii Startsev, Alexander Ustyuzhanin, Alexey Kirillov, Dmitry Baranchuk, Sergey Kastryulin</p><p><b>Upvotes:</b> 72</p><p><b>Summary:</b> Pre-training equips text-to-image (T2I) models with broad world knowledge, but this alone is often insufficient to achieve high aesthetic quality and alignment. Consequently, supervised fine-tuning (SFT) is crucial for further refinement. However, its effectiveness highly depends on the quality of the fine-tuning dataset. Existing public SFT datasets frequently target narrow domains (e.g., anime or specific art styles), and the creation of high-quality, general-purpose SFT datasets remains a significant challenge. Current curation methods are often costly and struggle to identify truly impactful samples. This challenge is further complicated by the scarcity of public general-purpose datasets, as leading models often rely on large, proprietary, and poorly documented internal data, hindering broader research progress. This paper introduces a novel methodology for creating general-purpose SFT datasets by leveraging a pre-trained generative model as an estimator of high-impact training samples. We apply this methodology to construct and release Alchemist, a compact (3,350 samples) yet highly effective SFT dataset. Experiments demonstrate that Alchemist substantially improves the generative quality of five public T2I models while preserving diversity and style. Additionally, we release the fine-tuned models' weights to the public.</p> - https://arxiv.org/abs/2505.19297 - Sun, 25 May 2025 20:08:20 +0000 + Sekai: A Video Dataset towards World Exploration + https://arxiv.org/abs/2506.15675 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.15675.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhen Li, Chuanhao Li, Xiaofeng Mao, Shaoheng Lin, Ming Li, Shitian Zhao, Zhaopan Xu, Xinyue Li, Yukang Feng, Jianwen Sun, Zizhen Li, Fanrui Zhang, Jiaxin Ai, Zhixiang Wang, Yuwei Wu, Tong He, Jiangmiao Pang, Yu Qiao, Yunde Jia, Kaipeng Zhang</p><p><b>Upvotes:</b> 61</p><p><b>Summary:</b> Video generation techniques have made remarkable progress, promising to be the foundation of interactive world exploration. However, existing video generation datasets are not well-suited for world exploration training as they suffer from some limitations: limited locations, short duration, static scenes, and a lack of annotations about exploration and the world. In this paper, we introduce Sekai (meaning ``world'' in Japanese), a high-quality first-person view worldwide video dataset with rich annotations for world exploration. It consists of over 5,000 hours of walking or drone view (FPV and UVA) videos from over 100 countries and regions across 750 cities. We develop an efficient and effective toolbox to collect, pre-process and annotate videos with location, scene, weather, crowd density, captions, and camera trajectories. Experiments demonstrate the quality of the dataset. And, we use a subset to train an interactive video world exploration model, named YUME (meaning ``dream'' in Japanese). We believe Sekai will benefit the area of video generation and world exploration, and motivate valuable applications.</p> + https://arxiv.org/abs/2506.15675 + Wed, 18 Jun 2025 17:57:06 +0000 - BizFinBench: A Business-Driven Real-World Financial Benchmark for Evaluating LLMs - https://arxiv.org/abs/2505.19457 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.19457.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Guilong Lu, Xuntao Guo, Rongjunchen Zhang, Wenqiao Zhu, Ji Liu</p><p><b>Upvotes:</b> 60</p><p><b>Summary:</b> Large language models excel in general tasks, yet assessing their reliability in logic-heavy, precision-critical domains like finance, law, and healthcare remains challenging. To address this, we introduce BizFinBench, the first benchmark specifically designed to evaluate LLMs in real-world financial applications. BizFinBench consists of 6,781 well-annotated queries in Chinese, spanning five dimensions: numerical calculation, reasoning, information extraction, prediction recognition, and knowledge-based question answering, grouped into nine fine-grained categories. The benchmark includes both objective and subjective metrics. We also introduce IteraJudge, a novel LLM evaluation method that reduces bias when LLMs serve as evaluators in objective metrics. We benchmark 25 models, including both proprietary and open-source systems. Extensive experiments show that no model dominates across all tasks. Our evaluation reveals distinct capability patterns: (1) In Numerical Calculation, Claude-3.5-Sonnet (63.18) and DeepSeek-R1 (64.04) lead, while smaller models like Qwen2.5-VL-3B (15.92) lag significantly; (2) In Reasoning, proprietary models dominate (ChatGPT-o3: 83.58, Gemini-2.0-Flash: 81.15), with open-source models trailing by up to 19.49 points; (3) In Information Extraction, the performance spread is the largest, with DeepSeek-R1 scoring 71.46, while Qwen3-1.7B scores 11.23; (4) In Prediction Recognition, performance variance is minimal, with top models scoring between 39.16 and 50.00. We find that while current LLMs handle routine finance queries competently, they struggle with complex scenarios requiring cross-concept reasoning. BizFinBench offers a rigorous, business-aligned benchmark for future research. The code and dataset are available at https://github.com/HiThink-Research/BizFinBench.</p> - https://arxiv.org/abs/2505.19457 - Mon, 26 May 2025 03:23:02 +0000 + GenRecal: Generation after Recalibration from Large to Small Vision-Language Models + https://arxiv.org/abs/2506.15681 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.15681.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Byung-Kwan Lee, Ryo Hachiuma, Yong Man Ro, Yu-Chiang Frank Wang, Yueh-Hua Wu</p><p><b>Upvotes:</b> 36</p><p><b>Summary:</b> Recent advancements in vision-language models (VLMs) have leveraged large language models (LLMs) to achieve performance on par with closed-source systems like GPT-4V. However, deploying these models in real-world scenarios, particularly on resource-constrained devices, remains challenging due to their substantial computational demands. This has spurred interest in distilling knowledge from large VLMs into smaller, more efficient counterparts. A key challenge arises here from the diversity of VLM architectures, which are built on different LLMs and employ varying token types-differing in vocabulary size, token splits, and token index ordering. To address this challenge of limitation to a specific VLM type, we present Generation after Recalibration (GenRecal), a novel, general-purpose distillation framework for VLMs. GenRecal incorporates a Recalibrator that aligns and adapts feature representations between heterogeneous VLMs, enabling effective knowledge transfer across different types of VLMs. Through extensive experiments on multiple challenging benchmarks, we demonstrate that GenRecal significantly improves baseline performances, eventually outperforming large-scale open- and closed-source VLMs.</p> + https://arxiv.org/abs/2506.15681 + Wed, 18 Jun 2025 17:59:49 +0000 - GraLoRA: Granular Low-Rank Adaptation for Parameter-Efficient Fine-Tuning - https://arxiv.org/abs/2505.20355 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.20355.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yeonjoon Jung, Daehyun Ahn, Hyungjun Kim, Taesu Kim, Eunhyeok Park</p><p><b>Upvotes:</b> 35</p><p><b>Summary:</b> Low-Rank Adaptation (LoRA) is a popular method for parameter-efficient fine-tuning (PEFT) of generative models, valued for its simplicity and effectiveness. Despite recent enhancements, LoRA still suffers from a fundamental limitation: overfitting when the bottleneck is widened. It performs best at ranks 32-64, yet its accuracy stagnates or declines at higher ranks, still falling short of full fine-tuning (FFT) performance. We identify the root cause as LoRA's structural bottleneck, which introduces gradient entanglement to the unrelated input channels and distorts gradient propagation. To address this, we introduce a novel structure, Granular Low-Rank Adaptation (GraLoRA) that partitions weight matrices into sub-blocks, each with its own low-rank adapter. With negligible computational or storage cost, GraLoRA overcomes LoRA's limitations, effectively increases the representational capacity, and more closely approximates FFT behavior. Experiments on code generation and commonsense reasoning benchmarks show that GraLoRA consistently outperforms LoRA and other baselines, achieving up to +8.5% absolute gain in Pass@1 on HumanEval+. These improvements hold across model sizes and rank settings, making GraLoRA a scalable and robust solution for PEFT. Code, data, and scripts are available at https://github.com/SqueezeBits/GraLoRA.git</p> - https://arxiv.org/abs/2505.20355 - Mon, 26 May 2025 06:48:20 +0000 + Vision-Guided Chunking Is All You Need: Enhancing RAG with Multimodal Document Understanding + https://arxiv.org/abs/2506.16035 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.16035.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Vishesh Tripathi, Tanmay Odapally, Indraneel Das, Uday Allu, Biddwan Ahmed</p><p><b>Upvotes:</b> 82</p><p><b>Summary:</b> Retrieval-Augmented Generation (RAG) systems have revolutionized information retrieval and question answering, but traditional text-based chunking methods struggle with complex document structures, multi-page tables, embedded figures, and contextual dependencies across page boundaries. We present a novel multimodal document chunking approach that leverages Large Multimodal Models (LMMs) to process PDF documents in batches while maintaining semantic coherence and structural integrity. Our method processes documents in configurable page batches with cross-batch context preservation, enabling accurate handling of tables spanning multiple pages, embedded visual elements, and procedural content. We evaluate our approach on a curated dataset of PDF documents with manually crafted queries, demonstrating improvements in chunk quality and downstream RAG performance. Our vision-guided approach achieves better accuracy compared to traditional vanilla RAG systems, with qualitative analysis showing superior preservation of document structure and semantic coherence.</p> + https://arxiv.org/abs/2506.16035 + Thu, 19 Jun 2025 05:11:43 +0000 - SynLogic: Synthesizing Verifiable Reasoning Data at Scale for Learning Logical Reasoning and Beyond - https://arxiv.org/abs/2505.19641 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.19641.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Junteng Liu, Yuanxiang Fan, Zhuo Jiang, Han Ding, Yongyi Hu, Chi Zhang, Yiqi Shi, Shitong Weng, Aili Chen, Shiqi Chen, Yunan Huang, Mozhi Zhang, Pengyu Zhao, Junjie Yan, Junxian He</p><p><b>Upvotes:</b> 57</p><p><b>Summary:</b> Recent advances such as OpenAI-o1 and DeepSeek R1 have demonstrated the potential of Reinforcement Learning (RL) to enhance reasoning abilities in Large Language Models (LLMs). While open-source replication efforts have primarily focused on mathematical and coding domains, methods and resources for developing general reasoning capabilities remain underexplored. This gap is partly due to the challenge of collecting diverse and verifiable reasoning data suitable for RL. We hypothesize that logical reasoning is critical for developing general reasoning capabilities, as logic forms a fundamental building block of reasoning. In this work, we present SynLogic, a data synthesis framework and dataset that generates diverse logical reasoning data at scale, encompassing 35 diverse logical reasoning tasks. The SynLogic approach enables controlled synthesis of data with adjustable difficulty and quantity. Importantly, all examples can be verified by simple rules, making them ideally suited for RL with verifiable rewards. In our experiments, we validate the effectiveness of RL training on the SynLogic dataset based on 7B and 32B models. SynLogic leads to state-of-the-art logical reasoning performance among open-source datasets, surpassing DeepSeek-R1-Distill-Qwen-32B by 6 points on BBEH. Furthermore, mixing SynLogic data with mathematical and coding tasks improves the training efficiency of these domains and significantly enhances reasoning generalization. Notably, our mixed training model outperforms DeepSeek-R1-Zero-Qwen-32B across multiple benchmarks. These findings position SynLogic as a valuable resource for advancing the broader reasoning capabilities of LLMs. We open-source both the data synthesis pipeline and the SynLogic dataset at https://github.com/MiniMax-AI/SynLogic.</p> - https://arxiv.org/abs/2505.19641 - Mon, 26 May 2025 07:59:36 +0000 + PAROAttention: Pattern-Aware ReOrdering for Efficient Sparse and Quantized Attention in Visual Generation Models + https://arxiv.org/abs/2506.16054 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.16054.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Tianchen Zhao, Ke Hong, Xinhao Yang, Xuefeng Xiao, Huixia Li, Feng Ling, Ruiqi Xie, Siqi Chen, Hongyu Zhu, Yichong Zhang, Yu Wang</p><p><b>Upvotes:</b> 57</p><p><b>Summary:</b> In visual generation, the quadratic complexity of attention mechanisms results in high memory and computational costs, especially for longer token sequences required in high-resolution image or multi-frame video generation. To address this, prior research has explored techniques such as sparsification and quantization. However, these techniques face significant challenges under low density and reduced bitwidths. Through systematic analysis, we identify that the core difficulty stems from the dispersed and irregular characteristics of visual attention patterns. Therefore, instead of introducing specialized sparsification and quantization design to accommodate such patterns, we propose an alternative strategy: *reorganizing* the attention pattern to alleviate the challenges. Inspired by the local aggregation nature of visual feature extraction, we design a novel **Pattern-Aware token ReOrdering (PARO)** technique, which unifies the diverse attention patterns into a hardware-friendly block-wise pattern. This unification substantially simplifies and enhances both sparsification and quantization. We evaluate the performance-efficiency trade-offs of various design choices and finalize a methodology tailored for the unified pattern. Our approach, **PAROAttention**, achieves video and image generation with lossless metrics, and nearly identical results from full-precision (FP) baselines, while operating at notably lower density (~20%-30%) and bitwidth (**INT8/INT4**), achieving a **1.9x** to **2.7x** end-to-end latency speedup.</p> + https://arxiv.org/abs/2506.16054 + Thu, 19 Jun 2025 06:25:02 +0000 - Deciphering Trajectory-Aided LLM Reasoning: An Optimization Perspective - https://arxiv.org/abs/2505.19815 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.19815.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Junnan Liu, Hongwei Liu, Linchen Xiao, Shudong Liu, Taolin Zhang, Zihan Ma, Songyang Zhang, Kai Chen</p><p><b>Upvotes:</b> 36</p><p><b>Summary:</b> We propose a novel framework for comprehending the reasoning capabilities of large language models (LLMs) through the perspective of meta-learning. By conceptualizing reasoning trajectories as pseudo-gradient descent updates to the LLM's parameters, we identify parallels between LLM reasoning and various meta-learning paradigms. We formalize the training process for reasoning tasks as a meta-learning setup, with each question treated as an individual task, and reasoning trajectories serving as the inner loop optimization for adapting model parameters. Once trained on a diverse set of questions, the LLM develops fundamental reasoning capabilities that can generalize to previously unseen questions. Extensive empirical evaluations substantiate the strong connection between LLM reasoning and meta-learning, exploring several issues of significant interest from a meta-learning standpoint. Our work not only enhances the understanding of LLM reasoning but also provides practical insights for improving these models through established meta-learning techniques.</p> - https://arxiv.org/abs/2505.19815 - Mon, 26 May 2025 10:52:17 +0000 + Drag-and-Drop LLMs: Zero-Shot Prompt-to-Weights + https://arxiv.org/abs/2506.16406 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.16406.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zhiyuan Liang, Dongwen Tang, Yuhao Zhou, Xuanlei Zhao, Mingjia Shi, Wangbo Zhao, Zekai Li, Peihao Wang, Konstantin Schürholt, Damian Borth, Michael M. Bronstein, Yang You, Zhangyang Wang, Kai Wang</p><p><b>Upvotes:</b> 113</p><p><b>Summary:</b> Modern Parameter-Efficient Fine-Tuning (PEFT) methods such as low-rank adaptation (LoRA) reduce the cost of customizing large language models (LLMs), yet still require a separate optimization run for every downstream dataset. We introduce Drag-and-Drop LLMs (\textit{DnD)}, a prompt-conditioned parameter generator that eliminates per-task training by mapping a handful of unlabeled task prompts directly to LoRA weight updates. A lightweight text encoder distills each prompt batch into condition embeddings, which are then transformed by a cascaded hyper-convolutional decoder into the full set of LoRA matrices. Once trained in a diverse collection of prompt-checkpoint pairs, DnD produces task-specific parameters in seconds, yielding i) up to 12,000times lower overhead than full fine-tuning, ii) average gains up to 30\% in performance over the strongest training LoRAs on unseen common-sense reasoning, math, coding, and multimodal benchmarks, and iii) robust cross-domain generalization despite never seeing the target data or labels. Our results demonstrate that prompt-conditioned parameter generation is a viable alternative to gradient-based adaptation for rapidly specializing LLMs. Our project is available at https://jerryliang24.github.io/DnD{https://jerryliang24.github.io/DnD}.</p> + https://arxiv.org/abs/2506.16406 + Thu, 19 Jun 2025 15:38:21 +0000 - Enigmata: Scaling Logical Reasoning in Large Language Models with Synthetic Verifiable Puzzles - https://arxiv.org/abs/2505.19914 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.19914.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jiangjie Chen, Qianyu He, Siyu Yuan, Aili Chen, Zhicheng Cai, Weinan Dai, Hongli Yu, Qiying Yu, Xuefeng Li, Jiaze Chen, Hao Zhou, Mingxuan Wang</p><p><b>Upvotes:</b> 39</p><p><b>Summary:</b> Large Language Models (LLMs), such as OpenAI's o1 and DeepSeek's R1, excel at advanced reasoning tasks like math and coding via Reinforcement Learning with Verifiable Rewards (RLVR), but still struggle with puzzles solvable by humans without domain knowledge. We introduce Enigmata, the first comprehensive suite tailored for improving LLMs with puzzle reasoning skills. It includes 36 tasks across seven categories, each with 1) a generator that produces unlimited examples with controllable difficulty and 2) a rule-based verifier for automatic evaluation. This generator-verifier design supports scalable, multi-task RL training, fine-grained analysis, and seamless RLVR integration. We further propose Enigmata-Eval, a rigorous benchmark, and develop optimized multi-task RLVR strategies. Our trained model, Qwen2.5-32B-Enigmata, consistently surpasses o3-mini-high and o1 on the puzzle reasoning benchmarks like Enigmata-Eval, ARC-AGI (32.8%), and ARC-AGI 2 (0.6%). It also generalizes well to out-of-domain puzzle benchmarks and mathematical reasoning, with little multi-tasking trade-off. When trained on larger models like Seed1.5-Thinking (20B activated parameters and 200B total parameters), puzzle data from Enigmata further boosts SoTA performance on advanced math and STEM reasoning tasks such as AIME (2024-2025), BeyondAIME and GPQA (Diamond), showing nice generalization benefits of Enigmata. This work offers a unified, controllable framework for advancing logical reasoning in LLMs. Resources of this work can be found at https://seed-enigmata.github.io.</p> - https://arxiv.org/abs/2505.19914 - Mon, 26 May 2025 12:40:31 +0000 + Hunyuan-GameCraft: High-dynamic Interactive Game Video Generation with Hybrid History Condition + https://arxiv.org/abs/2506.17201 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.17201.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jiaqi Li, Junshu Tang, Zhiyong Xu, Longhuang Wu, Yuan Zhou, Shuai Shao, Tianbao Yu, Zhiguo Cao, Qinglin Lu</p><p><b>Upvotes:</b> 50</p><p><b>Summary:</b> Recent advances in diffusion-based and controllable video generation have enabled high-quality and temporally coherent video synthesis, laying the groundwork for immersive interactive gaming experiences. However, current methods face limitations in dynamics, generality, long-term consistency, and efficiency, which limit the ability to create various gameplay videos. To address these gaps, we introduce Hunyuan-GameCraft, a novel framework for high-dynamic interactive video generation in game environments. To achieve fine-grained action control, we unify standard keyboard and mouse inputs into a shared camera representation space, facilitating smooth interpolation between various camera and movement operations. Then we propose a hybrid history-conditioned training strategy that extends video sequences autoregressively while preserving game scene information. Additionally, to enhance inference efficiency and playability, we achieve model distillation to reduce computational overhead while maintaining consistency across long temporal sequences, making it suitable for real-time deployment in complex interactive environments. The model is trained on a large-scale dataset comprising over one million gameplay recordings across over 100 AAA games, ensuring broad coverage and diversity, then fine-tuned on a carefully annotated synthetic dataset to enhance precision and control. The curated game scene data significantly improves the visual fidelity, realism and action controllability. Extensive experiments demonstrate that Hunyuan-GameCraft significantly outperforms existing models, advancing the realism and playability of interactive game video generation.</p> + https://arxiv.org/abs/2506.17201 + Fri, 20 Jun 2025 17:50:37 +0000 - ARM: Adaptive Reasoning Model - https://arxiv.org/abs/2505.20258 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.20258.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Siye Wu, Jian Xie, Yikai Zhang, Aili Chen, Kai Zhang, Yu Su, Yanghua Xiao</p><p><b>Upvotes:</b> 42</p><p><b>Summary:</b> While large reasoning models demonstrate strong performance on complex tasks, they lack the ability to adjust reasoning token usage based on task difficulty. This often leads to the "overthinking" problem -- excessive and unnecessary reasoning -- which, although potentially mitigated by human intervention to control the token budget, still fundamentally contradicts the goal of achieving fully autonomous AI. In this work, we propose Adaptive Reasoning Model (ARM), a reasoning model capable of adaptively selecting appropriate reasoning formats based on the task at hand. These formats include three efficient ones -- Direct Answer, Short CoT, and Code -- as well as a more elaborate format, Long CoT. To train ARM, we introduce Ada-GRPO, an adaptation of Group Relative Policy Optimization (GRPO), which addresses the format collapse issue in traditional GRPO. Ada-GRPO enables ARM to achieve high token efficiency, reducing tokens by an average of 30%, and up to 70%, while maintaining performance comparable to the model that relies solely on Long CoT. Furthermore, not only does it improve inference efficiency through reduced token generation, but it also brings a 2x speedup in training. In addition to the default Adaptive Mode, ARM supports two additional reasoning modes: 1) Instruction-Guided Mode, which allows users to explicitly specify the reasoning format via special tokens -- ideal when the appropriate format is known for a batch of tasks. 2) Consensus-Guided Mode, which aggregates the outputs of the three efficient formats and resorts to Long CoT in case of disagreement, prioritizing performance with higher token usage.</p> - https://arxiv.org/abs/2505.20258 - Mon, 26 May 2025 17:38:50 +0000 + BlenderFusion: 3D-Grounded Visual Editing and Generative Compositing + https://arxiv.org/abs/2506.17450 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.17450.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jiacheng Chen, Ramin Mehran, Xuhui Jia, Saining Xie, Sanghyun Woo</p><p><b>Upvotes:</b> 45</p><p><b>Summary:</b> We present BlenderFusion, a generative visual compositing framework that synthesizes new scenes by recomposing objects, camera, and background. It follows a layering-editing-compositing pipeline: (i) segmenting and converting visual inputs into editable 3D entities (layering), (ii) editing them in Blender with 3D-grounded control (editing), and (iii) fusing them into a coherent scene using a generative compositor (compositing). Our generative compositor extends a pre-trained diffusion model to process both the original (source) and edited (target) scenes in parallel. It is fine-tuned on video frames with two key training strategies: (i) source masking, enabling flexible modifications like background replacement; (ii) simulated object jittering, facilitating disentangled control over objects and camera. BlenderFusion significantly outperforms prior methods in complex compositional scene editing tasks.</p> + https://arxiv.org/abs/2506.17450 + Fri, 20 Jun 2025 19:38:34 +0000 - OpenS2V-Nexus: A Detailed Benchmark and Million-Scale Dataset for Subject-to-Video Generation - https://arxiv.org/abs/2505.20292 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.20292.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Shenghai Yuan, Xianyi He, Yufan Deng, Yang Ye, Jinfa Huang, Bin Lin, Chongyang Ma, Jiebo Luo, Li Yuan</p><p><b>Upvotes:</b> 51</p><p><b>Summary:</b> Subject-to-Video (S2V) generation aims to create videos that faithfully incorporate reference content, providing enhanced flexibility in the production of videos. To establish the infrastructure for S2V generation, we propose OpenS2V-Nexus, consisting of (i) OpenS2V-Eval, a fine-grained benchmark, and (ii) OpenS2V-5M, a million-scale dataset. In contrast to existing S2V benchmarks inherited from VBench that focus on global and coarse-grained assessment of generated videos, OpenS2V-Eval focuses on the model's ability to generate subject-consistent videos with natural subject appearance and identity fidelity. For these purposes, OpenS2V-Eval introduces 180 prompts from seven major categories of S2V, which incorporate both real and synthetic test data. Furthermore, to accurately align human preferences with S2V benchmarks, we propose three automatic metrics, NexusScore, NaturalScore and GmeScore, to separately quantify subject consistency, naturalness, and text relevance in generated videos. Building on this, we conduct a comprehensive evaluation of 16 representative S2V models, highlighting their strengths and weaknesses across different content. Moreover, we create the first open-source large-scale S2V generation dataset OpenS2V-5M, which consists of five million high-quality 720P subject-text-video triples. Specifically, we ensure subject-information diversity in our dataset by (1) segmenting subjects and building pairing information via cross-video associations and (2) prompting GPT-Image-1 on raw frames to synthesize multi-view representations. Through OpenS2V-Nexus, we deliver a robust infrastructure to accelerate future S2V generation research.</p> - https://arxiv.org/abs/2505.20292 - Mon, 26 May 2025 17:59:46 +0000 + JarvisArt: Liberating Human Artistic Creativity via an Intelligent Photo Retouching Agent + https://arxiv.org/abs/2506.17612 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.17612.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yunlong Lin, Zixu Lin, Kunjie Lin, Jinbin Bai, Panwang Pan, Chenxin Li, Haoyu Chen, Zhongdao Wang, Xinghao Ding, Wenbo Li, Shuicheng Yan</p><p><b>Upvotes:</b> 59</p><p><b>Summary:</b> Photo retouching has become integral to contemporary visual storytelling, enabling users to capture aesthetics and express creativity. While professional tools such as Adobe Lightroom offer powerful capabilities, they demand substantial expertise and manual effort. In contrast, existing AI-based solutions provide automation but often suffer from limited adjustability and poor generalization, failing to meet diverse and personalized editing needs. To bridge this gap, we introduce JarvisArt, a multi-modal large language model (MLLM)-driven agent that understands user intent, mimics the reasoning process of professional artists, and intelligently coordinates over 200 retouching tools within Lightroom. JarvisArt undergoes a two-stage training process: an initial Chain-of-Thought supervised fine-tuning to establish basic reasoning and tool-use skills, followed by Group Relative Policy Optimization for Retouching (GRPO-R) to further enhance its decision-making and tool proficiency. We also propose the Agent-to-Lightroom Protocol to facilitate seamless integration with Lightroom. To evaluate performance, we develop MMArt-Bench, a novel benchmark constructed from real-world user edits. JarvisArt demonstrates user-friendly interaction, superior generalization, and fine-grained control over both global and local adjustments, paving a new avenue for intelligent photo retouching. Notably, it outperforms GPT-4o with a 60% improvement in average pixel-level metrics on MMArt-Bench for content fidelity, while maintaining comparable instruction-following capabilities. Project Page: https://jarvisart.vercel.app/.</p> + https://arxiv.org/abs/2506.17612 + Sat, 21 Jun 2025 06:36:00 +0000 - SWE-rebench: An Automated Pipeline for Task Collection and Decontaminated Evaluation of Software Engineering Agents - https://arxiv.org/abs/2505.20411 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.20411.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ibragim Badertdinov, Alexander Golubev, Maksim Nekrashevich, Anton Shevtsov, Simon Karasik, Andrei Andriushchenko, Maria Trofimova, Daria Litvintseva, Boris Yangel</p><p><b>Upvotes:</b> 78</p><p><b>Summary:</b> LLM-based agents have shown promising capabilities in a growing range of software engineering (SWE) tasks. However, advancing this field faces two critical challenges. First, high-quality training data is scarce, especially data that reflects real-world SWE scenarios, where agents must interact with development environments, execute code and adapt behavior based on the outcomes of their actions. Existing datasets are either limited to one-shot code generation or comprise small, manually curated collections of interactive tasks, lacking both scale and diversity. Second, the lack of fresh interactive SWE tasks affects evaluation of rapidly improving models, as static benchmarks quickly become outdated due to contamination issues. To address these limitations, we introduce a novel, automated, and scalable pipeline to continuously extract real-world interactive SWE tasks from diverse GitHub repositories. Using this pipeline, we construct SWE-rebench, a public dataset comprising over 21,000 interactive Python-based SWE tasks, suitable for reinforcement learning of SWE agents at scale. Additionally, we use continuous supply of fresh tasks collected using SWE-rebench methodology to build a contamination-free benchmark for agentic software engineering. We compare results of various LLMs on this benchmark to results on SWE-bench Verified and show that performance of some language models might be inflated due to contamination issues.</p> - https://arxiv.org/abs/2505.20411 - Mon, 26 May 2025 18:01:00 +0000 + ShareGPT-4o-Image: Aligning Multimodal Models with GPT-4o-Level Image Generation + https://arxiv.org/abs/2506.18095 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.18095.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Junying Chen, Zhenyang Cai, Pengcheng Chen, Shunian Chen, Ke Ji, Xidong Wang, Yunjin Yang, Benyou Wang</p><p><b>Upvotes:</b> 61</p><p><b>Summary:</b> Recent advances in multimodal generative models have unlocked photorealistic, instruction-aligned image generation, yet leading systems like GPT-4o-Image remain proprietary and inaccessible. To democratize these capabilities, we present ShareGPT-4o-Image, the first dataset comprising 45K text-to-image and 46K text-and-image-to-image data, all synthesized using GPT-4o's image generation capabilities for distilling its advanced image generation abilities. Leveraging this dataset, we develop Janus-4o, a multimodal large language model capable of both text-to-image and text-and-image-to-image generation. Janus-4o not only significantly improves text-to-image generation over its predecessor, Janus-Pro, but also newly supports text-and-image-to-image generation. Notably, it achieves impressive performance in text-and-image-to-image generation from scratch, using only 91K synthetic samples and 6 hours of training on an 8 A800-GPU machine. We hope the release of ShareGPT-4o-Image and Janus-4o will foster open research in photorealistic, instruction-aligned image generation.</p> + https://arxiv.org/abs/2506.18095 + Sun, 22 Jun 2025 16:51:09 +0000 - SageAttention2++: A More Efficient Implementation of SageAttention2 - https://arxiv.org/abs/2505.21136 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.21136.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jintao Zhang, Xiaoming Xu, Jia Wei, Haofeng Huang, Pengle Zhang, Chendong Xiang, Jun Zhu, Jianfei Chen</p><p><b>Upvotes:</b> 35</p><p><b>Summary:</b> The efficiency of attention is critical because its time complexity grows quadratically with sequence length. SageAttention2 addresses this by utilizing quantization to accelerate matrix multiplications (Matmul) in attention. To further accelerate SageAttention2, we propose to utilize the faster instruction of FP8 Matmul accumulated in FP16. The instruction is 2x faster than the FP8 Matmul used in SageAttention2. Our experiments show that SageAttention2++ achieves a 3.9x speedup over FlashAttention while maintaining the same attention accuracy as SageAttention2. This means SageAttention2++ effectively accelerates various models, including those for language, image, and video generation, with negligible end-to-end metrics loss. The code will be available at https://github.com/thu-ml/SageAttention.</p> - https://arxiv.org/abs/2505.21136 - Tue, 27 May 2025 12:50:36 +0000 + Chain-of-Experts: Unlocking the Communication Power of Mixture-of-Experts Models + https://arxiv.org/abs/2506.18945 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.18945.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zihan Wang, Rui Pan, Jiarui Yao, Robert Csordas, Linjie Li, Lu Yin, Jiajun Wu, Tong Zhang, Manling Li, Shiwei Liu</p><p><b>Upvotes:</b> 39</p><p><b>Summary:</b> We propose Chain-of-Experts (CoE), a new Mixture-of-Experts (MoE) architecture that introduces sequential expert communication within each layer. Unlike traditional MoE models, where experts operate independently in parallel, CoE processes tokens iteratively across a chain of experts inside a layer. To support dynamic expert selection across iterations, CoE employs a dedicated router at each iteration step within a layer. This design allows tokens to re-evaluate and select different experts during each iteration, rather than being statically assigned. As a result, CoE introduces a flexible routing mechanism that increases the diversity of expert combinations and enriches the model's representational capacity. CoE demonstrates improved performance under fixed compute: on math reasoning tasks, it reduces validation loss from 1.20 to 1.12 compared to a standard MoE. Beyond performance, CoE offers a new scaling axis: depth through expert iteration, which complements conventional width/depth scaling. For example, using 2x iterations matches the performance of 3x expert selections (in width), while reducing memory usage by 17.6-42% relative to other scaling strategies. Our analysis reveals that CoE's benefits stem from its iterative residual structure and enhanced expert specialization empowered by iterative routing, which together unlock more expressive representations. Code is available at https://github.com/ZihanWang314/coe.</p> + https://arxiv.org/abs/2506.18945 + Mon, 23 Jun 2025 02:15:43 +0000 - Exploring the Latent Capacity of LLMs for One-Step Text Generation - https://arxiv.org/abs/2505.21189 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.21189.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Gleb Mezentsev, Ivan Oseledets</p><p><b>Upvotes:</b> 56</p><p><b>Summary:</b> A recent study showed that large language models (LLMs) can reconstruct surprisingly long texts - up to thousands of tokens - via autoregressive generation from just one specially trained input embedding. In this work, we explore whether such reconstruction is possible without autoregression. We show that frozen LLMs can generate hundreds of accurate tokens in just one forward pass, when provided with only two learned embeddings. This reveals a surprising and underexplored capability of LLMs - multi-token generation without iterative decoding. We investigate the behaviour of these embeddings and provide insight into the type of information they encode. We also empirically show that although these representations are not unique for a given text, they form connected and local regions in embedding space - a property that suggests the potential of learning a dedicated encoder into that space.</p> - https://arxiv.org/abs/2505.21189 - Tue, 27 May 2025 13:39:24 +0000 + Matrix-Game: Interactive World Foundation Model + https://arxiv.org/abs/2506.18701 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.18701.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yifan Zhang, Chunli Peng, Boyang Wang, Puyi Wang, Qingcheng Zhu, Fei Kang, Biao Jiang, Zedong Gao, Eric Li, Yang Liu, Yahui Zhou</p><p><b>Upvotes:</b> 58</p><p><b>Summary:</b> We introduce Matrix-Game, an interactive world foundation model for controllable game world generation. Matrix-Game is trained using a two-stage pipeline that first performs large-scale unlabeled pretraining for environment understanding, followed by action-labeled training for interactive video generation. To support this, we curate Matrix-Game-MC, a comprehensive Minecraft dataset comprising over 2,700 hours of unlabeled gameplay video clips and over 1,000 hours of high-quality labeled clips with fine-grained keyboard and mouse action annotations. Our model adopts a controllable image-to-world generation paradigm, conditioned on a reference image, motion context, and user actions. With over 17 billion parameters, Matrix-Game enables precise control over character actions and camera movements, while maintaining high visual quality and temporal coherence. To evaluate performance, we develop GameWorld Score, a unified benchmark measuring visual quality, temporal quality, action controllability, and physical rule understanding for Minecraft world generation. Extensive experiments show that Matrix-Game consistently outperforms prior open-source Minecraft world models (including Oasis and MineWorld) across all metrics, with particularly strong gains in controllability and physical consistency. Double-blind human evaluations further confirm the superiority of Matrix-Game, highlighting its ability to generate perceptually realistic and precisely controllable videos across diverse game scenarios. To facilitate future research on interactive image-to-world generation, we will open-source the Matrix-Game model weights and the GameWorld Score benchmark at https://github.com/SkyworkAI/Matrix-Game.</p> + https://arxiv.org/abs/2506.18701 + Mon, 23 Jun 2025 14:40:49 +0000 - MME-Reasoning: A Comprehensive Benchmark for Logical Reasoning in MLLMs - https://arxiv.org/abs/2505.21327 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.21327.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jiakang Yuan, Tianshuo Peng, Yilei Jiang, Yiting Lu, Renrui Zhang, Kaituo Feng, Chaoyou Fu, Tao Chen, Lei Bai, Bo Zhang, Xiangyu Yue</p><p><b>Upvotes:</b> 79</p><p><b>Summary:</b> Logical reasoning is a fundamental aspect of human intelligence and an essential capability for multimodal large language models (MLLMs). Despite the significant advancement in multimodal reasoning, existing benchmarks fail to comprehensively evaluate their reasoning abilities due to the lack of explicit categorization for logical reasoning types and an unclear understanding of reasoning. To address these issues, we introduce MME-Reasoning, a comprehensive benchmark designed to evaluate the reasoning ability of MLLMs, which covers all three types of reasoning (i.e., inductive, deductive, and abductive) in its questions. We carefully curate the data to ensure that each question effectively evaluates reasoning ability rather than perceptual skills or knowledge breadth, and extend the evaluation protocols to cover the evaluation of diverse questions. Our evaluation reveals substantial limitations of state-of-the-art MLLMs when subjected to holistic assessments of logical reasoning capabilities. Even the most advanced MLLMs show limited performance in comprehensive logical reasoning, with notable performance imbalances across reasoning types. In addition, we conducted an in-depth analysis of approaches such as ``thinking mode'' and Rule-based RL, which are commonly believed to enhance reasoning abilities. These findings highlight the critical limitations and performance imbalances of current MLLMs in diverse logical reasoning scenarios, providing comprehensive and systematic insights into the understanding and evaluation of reasoning capabilities.</p> - https://arxiv.org/abs/2505.21327 - Tue, 27 May 2025 15:23:23 +0000 + LongWriter-Zero: Mastering Ultra-Long Text Generation via Reinforcement Learning + https://arxiv.org/abs/2506.18841 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.18841.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yuhao Wu, Yushi Bai, Zhiqiang Hu, Roy Ka-Wei Lee, Juanzi Li</p><p><b>Upvotes:</b> 51</p><p><b>Summary:</b> Ultra-long generation by large language models (LLMs) is a widely demanded scenario, yet it remains a significant challenge due to their maximum generation length limit and overall quality degradation as sequence length increases. Previous approaches, exemplified by LongWriter, typically rely on ''teaching'', which involves supervised fine-tuning (SFT) on synthetic long-form outputs. However, this strategy heavily depends on synthetic SFT data, which is difficult and costly to construct, often lacks coherence and consistency, and tends to be overly artificial and structurally monotonous. In this work, we propose an incentivization-based approach that, starting entirely from scratch and without relying on any annotated or synthetic data, leverages reinforcement learning (RL) to foster the emergence of ultra-long, high-quality text generation capabilities in LLMs. We perform RL training starting from a base model, similar to R1-Zero, guiding it to engage in reasoning that facilitates planning and refinement during the writing process. To support this, we employ specialized reward models that steer the LLM towards improved length control, writing quality, and structural formatting. Experimental evaluations show that our LongWriter-Zero model, trained from Qwen2.5-32B, consistently outperforms traditional SFT methods on long-form writing tasks, achieving state-of-the-art results across all metrics on WritingBench and Arena-Write, and even surpassing 100B+ models such as DeepSeek R1 and Qwen3-235B. We open-source our data and model checkpoints under https://huggingface.co/THU-KEG/LongWriter-Zero-32B</p> + https://arxiv.org/abs/2506.18841 + Mon, 23 Jun 2025 16:59:02 +0000 - MME-VideoOCR: Evaluating OCR-Based Capabilities of Multimodal LLMs in Video Scenarios - https://arxiv.org/abs/2505.21333 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.21333.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yang Shi, Huanqian Wang, Wulin Xie, Huanyao Zhang, Lijie Zhao, Yi-Fan Zhang, Xinfeng Li, Chaoyou Fu, Zhuoer Wen, Wenting Liu, Zhuoran Zhang, Xinlong Chen, Bohan Zeng, Sihan Yang, Yuanxing Zhang, Pengfei Wan, Haotian Wang, Wenjing Yang</p><p><b>Upvotes:</b> 36</p><p><b>Summary:</b> Multimodal Large Language Models (MLLMs) have achieved considerable accuracy in Optical Character Recognition (OCR) from static images. However, their efficacy in video OCR is significantly diminished due to factors such as motion blur, temporal variations, and visual effects inherent in video content. To provide clearer guidance for training practical MLLMs, we introduce the MME-VideoOCR benchmark, which encompasses a comprehensive range of video OCR application scenarios. MME-VideoOCR features 10 task categories comprising 25 individual tasks and spans 44 diverse scenarios. These tasks extend beyond text recognition to incorporate deeper comprehension and reasoning of textual content within videos. The benchmark consists of 1,464 videos with varying resolutions, aspect ratios, and durations, along with 2,000 meticulously curated, manually annotated question-answer pairs. We evaluate 18 state-of-the-art MLLMs on MME-VideoOCR, revealing that even the best-performing model (Gemini-2.5 Pro) achieves an accuracy of only 73.7%. Fine-grained analysis indicates that while existing MLLMs demonstrate strong performance on tasks where relevant texts are contained within a single or few frames, they exhibit limited capability in effectively handling tasks that demand holistic video comprehension. These limitations are especially evident in scenarios that require spatio-temporal reasoning, cross-frame information integration, or resistance to language prior bias. Our findings also highlight the importance of high-resolution visual input and sufficient temporal coverage for reliable OCR in dynamic video scenarios.</p> - https://arxiv.org/abs/2505.21333 - Tue, 27 May 2025 15:27:46 +0000 + OmniGen2: Exploration to Advanced Multimodal Generation + https://arxiv.org/abs/2506.18871 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.18871.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Chenyuan Wu, Pengfei Zheng, Ruiran Yan, Shitao Xiao, Xin Luo, Yueze Wang, Wanli Li, Xiyan Jiang, Yexin Liu, Junjie Zhou, Ze Liu, Ziyi Xia, Chaofan Li, Haoge Deng, Jiahao Wang, Kun Luo, Bo Zhang, Defu Lian, Xinlong Wang, Zhongyuan Wang, Tiejun Huang, Zheng Liu</p><p><b>Upvotes:</b> 68</p><p><b>Summary:</b> In this work, we introduce OmniGen2, a versatile and open-source generative model designed to provide a unified solution for diverse generation tasks, including text-to-image, image editing, and in-context generation. Unlike OmniGen v1, OmniGen2 features two distinct decoding pathways for text and image modalities, utilizing unshared parameters and a decoupled image tokenizer. This design enables OmniGen2 to build upon existing multimodal understanding models without the need to re-adapt VAE inputs, thereby preserving the original text generation capabilities. To facilitate the training of OmniGen2, we developed comprehensive data construction pipelines, encompassing image editing and in-context generation data. Additionally, we introduce a reflection mechanism tailored for image generation tasks and curate a dedicated reflection dataset based on OmniGen2. Despite its relatively modest parameter size, OmniGen2 achieves competitive results on multiple task benchmarks, including text-to-image and image editing. To further evaluate in-context generation, also referred to as subject-driven tasks, we introduce a new benchmark named OmniContext. OmniGen2 achieves state-of-the-art performance among open-source models in terms of consistency. We will release our models, training code, datasets, and data construction pipeline to support future research in this field. Project Page: https://vectorspacelab.github.io/OmniGen2; GitHub Link: https://github.com/VectorSpaceLab/OmniGen2</p> + https://arxiv.org/abs/2506.18871 + Mon, 23 Jun 2025 17:38:54 +0000 - R2R: Efficiently Navigating Divergent Reasoning Paths with Small-Large Model Token Routing - https://arxiv.org/abs/2505.21600 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.21600.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Tianyu Fu, Yi Ge, Yichen You, Enshu Liu, Zhihang Yuan, Guohao Dai, Shengen Yan, Huazhong Yang, Yu Wang</p><p><b>Upvotes:</b> 63</p><p><b>Summary:</b> Large Language Models (LLMs) achieve impressive reasoning capabilities at the cost of substantial inference overhead, posing substantial deployment challenges. Although distilled Small Language Models (SLMs) significantly enhance efficiency, their performance suffers as they fail to follow LLMs' reasoning paths. Luckily, we reveal that only a small fraction of tokens genuinely diverge reasoning paths between LLMs and SLMs. Most generated tokens are either identical or exhibit neutral differences, such as minor variations in abbreviations or expressions. Leveraging this insight, we introduce **Roads to Rome (R2R)**, a neural token routing method that selectively utilizes LLMs only for these critical, path-divergent tokens, while leaving the majority of token generation to the SLM. We also develop an automatic data generation pipeline that identifies divergent tokens and generates token-level routing labels to train the lightweight router. We apply R2R to combine R1-1.5B and R1-32B models from the DeepSeek family, and evaluate on challenging math, coding, and QA benchmarks. With an average activated parameter size of 5.6B, R2R surpasses the average accuracy of R1-7B by 1.6x, outperforming even the R1-14B model. Compared to R1-32B, it delivers a 2.8x wall-clock speedup with comparable performance, advancing the Pareto frontier of test-time scaling efficiency. Our code is available at https://github.com/thu-nics/R2R.</p> - https://arxiv.org/abs/2505.21600 - Tue, 27 May 2025 16:57:20 +0000 + Light of Normals: Unified Feature Representation for Universal Photometric Stereo + https://arxiv.org/abs/2506.18882 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.18882.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Hong Li, Houyuan Chen, Chongjie Ye, Zhaoxi Chen, Bohan Li, Shaocong Xu, Xianda Guo, Xuhui Liu, Yikai Wang, Baochang Zhang, Satoshi Ikehata, Boxin Shi, Anyi Rao, Hao Zhao</p><p><b>Upvotes:</b> 80</p><p><b>Summary:</b> Universal photometric stereo (PS) aims to recover high-quality surface normals from objects under arbitrary lighting conditions without relying on specific illumination models. Despite recent advances such as SDM-UniPS and Uni MS-PS, two fundamental challenges persist: 1) the deep coupling between varying illumination and surface normal features, where ambiguity in observed intensity makes it difficult to determine whether brightness variations stem from lighting changes or surface orientation; and 2) the preservation of high-frequency geometric details in complex surfaces, where intricate geometries create self-shadowing, inter-reflections, and subtle normal variations that conventional feature processing operations struggle to capture accurately.</p> + https://arxiv.org/abs/2506.18882 + Mon, 23 Jun 2025 17:53:11 +0000 - UI-Genie: A Self-Improving Approach for Iteratively Boosting MLLM-based Mobile GUI Agents - https://arxiv.org/abs/2505.21496 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.21496.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Han Xiao, Guozhi Wang, Yuxiang Chai, Zimu Lu, Weifeng Lin, Hao He, Lue Fan, Liuyang Bian, Rui Hu, Liang Liu, Shuai Ren, Yafei Wen, Xiaoxin Chen, Aojun Zhou, Hongsheng Li</p><p><b>Upvotes:</b> 35</p><p><b>Summary:</b> In this paper, we introduce UI-Genie, a self-improving framework addressing two key challenges in GUI agents: verification of trajectory outcome is challenging and high-quality training data are not scalable. These challenges are addressed by a reward model and a self-improving pipeline, respectively. The reward model, UI-Genie-RM, features an image-text interleaved architecture that efficiently pro- cesses historical context and unifies action-level and task-level rewards. To sup- port the training of UI-Genie-RM, we develop deliberately-designed data genera- tion strategies including rule-based verification, controlled trajectory corruption, and hard negative mining. To address the second challenge, a self-improvement pipeline progressively expands solvable complex GUI tasks by enhancing both the agent and reward models through reward-guided exploration and outcome verification in dynamic environments. For training the model, we generate UI- Genie-RM-517k and UI-Genie-Agent-16k, establishing the first reward-specific dataset for GUI agents while demonstrating high-quality synthetic trajectory gen- eration without manual annotation. Experimental results show that UI-Genie achieves state-of-the-art performance across multiple GUI agent benchmarks with three generations of data-model self-improvement. We open-source our complete framework implementation and generated datasets to facilitate further research in https://github.com/Euphoria16/UI-Genie.</p> - https://arxiv.org/abs/2505.21496 - Tue, 27 May 2025 17:58:06 +0000 + Inverse-and-Edit: Effective and Fast Image Editing by Cycle Consistency Models + https://arxiv.org/abs/2506.19103 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.19103.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ilia Beletskii, Andrey Kuznetsov, Aibek Alanov</p><p><b>Upvotes:</b> 40</p><p><b>Summary:</b> Recent advances in image editing with diffusion models have achieved impressive results, offering fine-grained control over the generation process. However, these methods are computationally intensive because of their iterative nature. While distilled diffusion models enable faster inference, their editing capabilities remain limited, primarily because of poor inversion quality. High-fidelity inversion and reconstruction are essential for precise image editing, as they preserve the structural and semantic integrity of the source image. In this work, we propose a novel framework that enhances image inversion using consistency models, enabling high-quality editing in just four steps. Our method introduces a cycle-consistency optimization strategy that significantly improves reconstruction accuracy and enables a controllable trade-off between editability and content preservation. We achieve state-of-the-art performance across various image editing tasks and datasets, demonstrating that our method matches or surpasses full-step diffusion models while being substantially more efficient. The code of our method is available on GitHub at https://github.com/ControlGenAI/Inverse-and-Edit.</p> + https://arxiv.org/abs/2506.19103 + Mon, 23 Jun 2025 20:34:43 +0000 - Paper2Poster: Towards Multimodal Poster Automation from Scientific Papers - https://arxiv.org/abs/2505.21497 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.21497.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Wei Pang, Kevin Qinghong Lin, Xiangru Jian, Xi He, Philip Torr</p><p><b>Upvotes:</b> 83</p><p><b>Summary:</b> Academic poster generation is a crucial yet challenging task in scientific communication, requiring the compression of long-context interleaved documents into a single, visually coherent page. To address this challenge, we introduce the first benchmark and metric suite for poster generation, which pairs recent conference papers with author-designed posters and evaluates outputs on (i)Visual Quality-semantic alignment with human posters, (ii)Textual Coherence-language fluency, (iii)Holistic Assessment-six fine-grained aesthetic and informational criteria scored by a VLM-as-judge, and notably (iv)PaperQuiz-the poster's ability to convey core paper content as measured by VLMs answering generated quizzes. Building on this benchmark, we propose PosterAgent, a top-down, visual-in-the-loop multi-agent pipeline: the (a)Parser distills the paper into a structured asset library; the (b)Planner aligns text-visual pairs into a binary-tree layout that preserves reading order and spatial balance; and the (c)Painter-Commenter loop refines each panel by executing rendering code and using VLM feedback to eliminate overflow and ensure alignment. In our comprehensive evaluation, we find that GPT-4o outputs-though visually appealing at first glance-often exhibit noisy text and poor PaperQuiz scores, and we find that reader engagement is the primary aesthetic bottleneck, as human-designed posters rely largely on visual semantics to convey meaning. Our fully open-source variants (e.g. based on the Qwen-2.5 series) outperform existing 4o-driven multi-agent systems across nearly all metrics, while using 87% fewer tokens. It transforms a 22-page paper into a finalized yet editable .pptx poster - all for just $0.005. These findings chart clear directions for the next generation of fully automated poster-generation models. The code and datasets are available at https://github.com/Paper2Poster/Paper2Poster.</p> - https://arxiv.org/abs/2505.21497 - Tue, 27 May 2025 17:58:49 +0000 + Skywork-SWE: Unveiling Data Scaling Laws for Software Engineering in LLMs + https://arxiv.org/abs/2506.19290 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.19290.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Liang Zeng, Yongcong Li, Yuzhen Xiao, Changshi Li, Chris Yuhao Liu, Rui Yan, Tianwen Wei, Jujie He, Xuchen Song, Yang Liu, Yahui Zhou</p><p><b>Upvotes:</b> 48</p><p><b>Summary:</b> Software engineering (SWE) has recently emerged as a crucial testbed for next-generation LLM agents, demanding inherent capabilities in two critical dimensions: sustained iterative problem-solving (e.g., >50 interaction rounds) and long-context dependency resolution (e.g., >32k tokens). However, the data curation process in SWE remains notoriously time-consuming, as it heavily relies on manual annotation for code file filtering and the setup of dedicated runtime environments to execute and validate unit tests. Consequently, most existing datasets are limited to only a few thousand GitHub-sourced instances. To this end, we propose an incremental, automated data-curation pipeline that systematically scales both the volume and diversity of SWE datasets. Our dataset comprises 10,169 real-world Python task instances from 2,531 distinct GitHub repositories, each accompanied by a task specified in natural language and a dedicated runtime-environment image for automated unit-test validation. We have carefully curated over 8,000 successfully runtime-validated training trajectories from our proposed SWE dataset. When fine-tuning the Skywork-SWE model on these trajectories, we uncover a striking data scaling phenomenon: the trained model's performance for software engineering capabilities in LLMs continues to improve as the data size increases, showing no signs of saturation. Notably, our Skywork-SWE model achieves 38.0% pass@1 accuracy on the SWE-bench Verified benchmark without using verifiers or multiple rollouts, establishing a new state-of-the-art (SOTA) among the Qwen2.5-Coder-32B-based LLMs built on the OpenHands agent framework. Furthermore, with the incorporation of test-time scaling techniques, the performance further improves to 47.0% accuracy, surpassing the previous SOTA results for sub-32B parameter models. We release the Skywork-SWE-32B model checkpoint to accelerate future research.</p> + https://arxiv.org/abs/2506.19290 + Tue, 24 Jun 2025 03:53:36 +0000 - Skywork Open Reasoner 1 Technical Report - https://arxiv.org/abs/2505.22312 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.22312.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jujie He, Jiacai Liu, Chris Yuhao Liu, Rui Yan, Chaojie Wang, Peng Cheng, Xiaoyu Zhang, Fuxiang Zhang, Jiacheng Xu, Wei Shen, Siyuan Li, Liang Zeng, Tianwen Wei, Cheng Cheng, Bo An, Yang Liu, Yahui Zhou</p><p><b>Upvotes:</b> 49</p><p><b>Summary:</b> The success of DeepSeek-R1 underscores the significant role of reinforcement learning (RL) in enhancing the reasoning capabilities of large language models (LLMs). In this work, we present Skywork-OR1, an effective and scalable RL implementation for long Chain-of-Thought (CoT) models. Building on the DeepSeek-R1-Distill model series, our RL approach achieves notable performance gains, increasing average accuracy across AIME24, AIME25, and LiveCodeBench from 57.8% to 72.8% (+15.0%) for the 32B model and from 43.6% to 57.5% (+13.9%) for the 7B model. Our Skywork-OR1-32B model surpasses both DeepSeek-R1 and Qwen3-32B on the AIME24 and AIME25 benchmarks, while achieving comparable results on LiveCodeBench. The Skywork-OR1-7B and Skywork-OR1-Math-7B models demonstrate competitive reasoning capabilities among models of similar size. We perform comprehensive ablation studies on the core components of our training pipeline to validate their effectiveness. Additionally, we thoroughly investigate the phenomenon of entropy collapse, identify key factors affecting entropy dynamics, and demonstrate that mitigating premature entropy collapse is critical for improved test performance. To support community research, we fully open-source our model weights, training code, and training datasets.</p> - https://arxiv.org/abs/2505.22312 - Wed, 28 May 2025 12:56:04 +0000 + Outlier-Safe Pre-Training for Robust 4-Bit Quantization of Large Language Models + https://arxiv.org/abs/2506.19697 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.19697.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jungwoo Park, Taewhoo Lee, Chanwoong Yoon, Hyeon Hwang, Jaewoo Kang</p><p><b>Upvotes:</b> 43</p><p><b>Summary:</b> Extreme activation outliers in Large Language Models (LLMs) critically degrade quantization performance, hindering efficient on-device deployment. While channel-wise operations and adaptive gradient scaling are recognized causes, practical mitigation remains challenging. We introduce Outlier-Safe Pre-Training (OSP), a practical guideline that proactively prevents outlier formation rather than relying on post-hoc mitigation. OSP combines three key innovations: (1) the Muon optimizer, eliminating privileged bases while maintaining training efficiency; (2) Single-Scale RMSNorm, preventing channel-wise amplification; and (3) a learnable embedding projection, redistributing activation magnitudes originating from embedding matrices. We validate OSP by training a 1.4B-parameter model on 1 trillion tokens, which is the first production-scale LLM trained without such outliers. Under aggressive 4-bit quantization, our OSP model achieves a 35.7 average score across 10 benchmarks (compared to 26.5 for an Adam-trained model), with only a 2% training overhead. Remarkably, OSP models exhibit near-zero excess kurtosis (0.04) compared to extreme values (1818.56) in standard models, fundamentally altering LLM quantization behavior. Our work demonstrates that outliers are not inherent to LLMs but are consequences of training strategies, paving the way for more efficient LLM deployment. The source code and pretrained checkpoints are available at https://github.com/dmis-lab/Outlier-Safe-Pre-Training.</p> + https://arxiv.org/abs/2506.19697 + Tue, 24 Jun 2025 15:03:57 +0000 - Unsupervised Post-Training for Multi-Modal LLM Reasoning via GRPO - https://arxiv.org/abs/2505.22453 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.22453.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Lai Wei, Yuting Li, Chen Wang, Yue Wang, Linghe Kong, Weiran Huang, Lichao Sun</p><p><b>Upvotes:</b> 42</p><p><b>Summary:</b> Improving Multi-modal Large Language Models (MLLMs) in the post-training stage typically relies on supervised fine-tuning (SFT) or reinforcement learning (RL). However, these supervised methods require expensive and manually annotated multi-modal data--an ultimately unsustainable resource. While recent efforts have explored unsupervised post-training, their methods are complex and difficult to iterate. In this work, we are the first to investigate the use of GRPO, a stable and scalable online RL algorithm, for enabling continual self-improvement without any external supervision. We propose MM-UPT, a simple yet effective framework for unsupervised post-training of MLLMs. MM-UPT builds upon GRPO, replacing traditional reward signals with a self-rewarding mechanism based on majority voting over multiple sampled responses. Our experiments demonstrate that MM-UPT significantly improves the reasoning ability of Qwen2.5-VL-7B (e.g., 66.3 %rightarrow72.9 % on MathVista, 62.9 %rightarrow68.7 % on We-Math), using standard dataset without ground truth labels. MM-UPT also outperforms prior unsupervised baselines and even approaches the results of supervised GRPO. Furthermore, we show that incorporating synthetic questions, generated solely by MLLM itself, can boost performance as well, highlighting a promising approach for scalable self-improvement. Overall, MM-UPT offers a new paradigm for continual, autonomous enhancement of MLLMs in the absence of external supervision. Our code is available at https://github.com/waltonfuture/MM-UPT.</p> - https://arxiv.org/abs/2505.22453 - Wed, 28 May 2025 15:11:16 +0000 + AnimaX: Animating the Inanimate in 3D with Joint Video-Pose Diffusion Models + https://arxiv.org/abs/2506.19851 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.19851.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zehuan Huang, Haoran Feng, Yangtian Sun, Yuanchen Guo, Yanpei Cao, Lu Sheng</p><p><b>Upvotes:</b> 52</p><p><b>Summary:</b> We present AnimaX, a feed-forward 3D animation framework that bridges the motion priors of video diffusion models with the controllable structure of skeleton-based animation. Traditional motion synthesis methods are either restricted to fixed skeletal topologies or require costly optimization in high-dimensional deformation spaces. In contrast, AnimaX effectively transfers video-based motion knowledge to the 3D domain, supporting diverse articulated meshes with arbitrary skeletons. Our method represents 3D motion as multi-view, multi-frame 2D pose maps, and enables joint video-pose diffusion conditioned on template renderings and a textual motion prompt. We introduce shared positional encodings and modality-aware embeddings to ensure spatial-temporal alignment between video and pose sequences, effectively transferring video priors to motion generation task. The resulting multi-view pose sequences are triangulated into 3D joint positions and converted into mesh animation via inverse kinematics. Trained on a newly curated dataset of 160,000 rigged sequences, AnimaX achieves state-of-the-art results on VBench in generalization, motion fidelity, and efficiency, offering a scalable solution for category-agnostic 3D animation. Project page: https://anima-x.github.io/{https://anima-x.github.io/}.</p> + https://arxiv.org/abs/2506.19851 + Tue, 24 Jun 2025 17:59:58 +0000 - The Entropy Mechanism of Reinforcement Learning for Reasoning Language Models - https://arxiv.org/abs/2505.22617 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.22617.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ganqu Cui, Yuchen Zhang, Jiacheng Chen, Lifan Yuan, Zhi Wang, Yuxin Zuo, Haozhan Li, Yuchen Fan, Huayu Chen, Weize Chen, Zhiyuan Liu, Hao Peng, Lei Bai, Wanli Ouyang, Yu Cheng, Bowen Zhou, Ning Ding</p><p><b>Upvotes:</b> 100</p><p><b>Summary:</b> This paper aims to overcome a major obstacle in scaling RL for reasoning with LLMs, namely the collapse of policy entropy. Such phenomenon is consistently observed across vast RL runs without entropy intervention, where the policy entropy dropped sharply at the early training stage, this diminished exploratory ability is always accompanied with the saturation of policy performance. In practice, we establish a transformation equation R=-a*e^H+b between entropy H and downstream performance R. This empirical law strongly indicates that, the policy performance is traded from policy entropy, thus bottlenecked by its exhaustion, and the ceiling is fully predictable H=0, R=-a+b. Our finding necessitates entropy management for continuous exploration toward scaling compute for RL. To this end, we investigate entropy dynamics both theoretically and empirically. Our derivation highlights that, the change in policy entropy is driven by the covariance between action probability and the change in logits, which is proportional to its advantage when using Policy Gradient-like algorithms. Empirical study shows that, the values of covariance term and entropy differences matched exactly, supporting the theoretical conclusion. Moreover, the covariance term stays mostly positive throughout training, further explaining why policy entropy would decrease monotonically. Through understanding the mechanism behind entropy dynamics, we motivate to control entropy by restricting the update of high-covariance tokens. Specifically, we propose two simple yet effective techniques, namely Clip-Cov and KL-Cov, which clip and apply KL penalty to tokens with high covariances respectively. Experiments show that these methods encourage exploration, thus helping policy escape entropy collapse and achieve better downstream performance.</p> - https://arxiv.org/abs/2505.22617 - Wed, 28 May 2025 17:38:45 +0000 + OctoThinker: Mid-training Incentivizes Reinforcement Learning Scaling + https://arxiv.org/abs/2506.20512 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.20512.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zengzhi Wang, Fan Zhou, Xuefeng Li, Pengfei Liu</p><p><b>Upvotes:</b> 38</p><p><b>Summary:</b> Different base language model families, such as Llama and Qwen, exhibit divergent behaviors during post-training with reinforcement learning (RL), especially on reasoning-intensive tasks. What makes a base language model suitable for reinforcement learning? Gaining deeper insight into this question is essential for developing RL-scalable foundation models of the next generation. In this work, we investigate how mid-training strategies shape RL dynamics, focusing on two representative model families: Qwen and Llama. Our study reveals that (1) high-quality mathematical corpora, such as MegaMath-Web-Pro, significantly improve both base model and RL performance, while existing alternatives (e.g., FineMath-4plus) fail to do so; (2) further adding QA-style data, particularly long chain-of-thought (CoT) reasoning examples, enhances RL outcomes, and instruction data further unlocks this effect; (3) while long-CoT improves reasoning depth, it can also induce verbosity of model responses and unstability of RL training, underscoring the importance of data formatting; (4) scaling mid-training consistently leads to stronger downstream RL performance. Building on these insights, we introduce a two-stage mid-training strategy, Stable-then-Decay, in which base models are first trained on 200B tokens with a constant learning rate, followed by 20B tokens across three CoT-focused branches with learning rate decay. This yields OctoThinker, a family of models demonstrating strong RL compatibility and closing the performance gap with more RL-friendly model families, i.e., Qwen. We hope our work will help shape pre-training strategies for foundation models in the RL era. To support further research, we release our open-source models along with a curated math reasoning-intensive corpus of over 70 billion tokens (i.e., MegaMath-Web-Pro-Max).</p> + https://arxiv.org/abs/2506.20512 + Wed, 25 Jun 2025 14:58:13 +0000 - Sherlock: Self-Correcting Reasoning in Vision-Language Models - https://arxiv.org/abs/2505.22651 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.22651.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Yi Ding, Ruqi Zhang</p><p><b>Upvotes:</b> 47</p><p><b>Summary:</b> Reasoning Vision-Language Models (VLMs) have shown promising performance on complex multimodal tasks. However, they still face significant challenges: they are highly sensitive to reasoning errors, require large volumes of annotated data or accurate verifiers, and struggle to generalize beyond specific domains. To address these limitations, we explore self-correction as a strategy to enhance reasoning VLMs. We first conduct an in-depth analysis of reasoning VLMs' self-correction abilities and identify key gaps. Based on our findings, we introduce Sherlock, a self-correction and self-improvement training framework. Sherlock introduces a trajectory-level self-correction objective, a preference data construction method based on visual perturbation, and a dynamic beta for preference tuning. Once the model acquires self-correction capabilities using only 20k randomly sampled annotated data, it continues to self-improve without external supervision. Built on the Llama3.2-Vision-11B model, Sherlock achieves remarkable results across eight benchmarks, reaching an average accuracy of 64.1 with direct generation and 65.4 after self-correction. It outperforms LLaVA-CoT (63.2), Mulberry (63.9), and LlamaV-o1 (63.4) while using less than 20% of the annotated data.</p> - https://arxiv.org/abs/2505.22651 - Wed, 28 May 2025 17:58:03 +0000 + MMSearch-R1: Incentivizing LMMs to Search + https://arxiv.org/abs/2506.20670 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.20670.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jinming Wu, Zihao Deng, Wei Li, Yiding Liu, Bo You, Bo Li, Zejun Ma, Ziwei Liu</p><p><b>Upvotes:</b> 51</p><p><b>Summary:</b> Robust deployment of large multimodal models (LMMs) in real-world scenarios requires access to external knowledge sources, given the complexity and dynamic nature of real-world information. Existing approaches such as retrieval-augmented generation (RAG) and prompt engineered search agents rely on rigid pipelines, often leading to inefficient or excessive search behaviors. We present MMSearch-R1, the first end-to-end reinforcement learning framework that enables LMMs to perform on-demand, multi-turn search in real-world Internet environments. Our framework integrates both image and text search tools, allowing the model to reason about when and how to invoke them guided by an outcome-based reward with a search penalty. To support training, We collect a multimodal search VQA dataset through a semi-automated pipeline that covers diverse visual and textual knowledge needs and curate a search-balanced subset with both search-required and search-free samples, which proves essential for shaping efficient and on-demand search behavior. Extensive experiments on knowledge-intensive and info-seeking VQA tasks show that our model not only outperforms RAG-based baselines of the same model size, but also matches the performance of a larger RAG-based model while reducing search calls by over 30%. We further analyze key empirical findings to offer actionable insights for advancing research in multimodal search.</p> + https://arxiv.org/abs/2506.20670 + Wed, 25 Jun 2025 17:59:42 +0000 - The Climb Carves Wisdom Deeper Than the Summit: On the Noisy Rewards in Learning to Reason - https://arxiv.org/abs/2505.22653 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.22653.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Ang Lv, Ruobing Xie, Xingwu Sun, Zhanhui Kang, Rui Yan</p><p><b>Upvotes:</b> 58</p><p><b>Summary:</b> Recent studies on post-training large language models (LLMs) for reasoning through reinforcement learning (RL) typically focus on tasks that can be accurately verified and rewarded, such as solving math problems. In contrast, our research investigates the impact of reward noise, a more practical consideration for real-world scenarios involving the post-training of LLMs using reward models. We found that LLMs demonstrate strong robustness to substantial reward noise. For example, manually flipping 40% of the reward function's outputs in math tasks still allows a Qwen-2.5-7B model to achieve rapid convergence, improving its performance on math tasks from 5% to 72%, compared to the 75% accuracy achieved by a model trained with noiseless rewards. Surprisingly, by only rewarding the appearance of key reasoning phrases (namely reasoning pattern reward, RPR), such as ``first, I need to''-without verifying the correctness of answers, the model achieved peak downstream performance (over 70% accuracy for Qwen-2.5-7B) comparable to models trained with strict correctness verification and accurate rewards. Recognizing the importance of the reasoning process over the final results, we combined RPR with noisy reward models. RPR helped calibrate the noisy reward models, mitigating potential false negatives and enhancing the LLM's performance on open-ended tasks. These findings suggest the importance of improving models' foundational abilities during the pre-training phase while providing insights for advancing post-training techniques. Our code and scripts are available at https://github.com/trestad/Noisy-Rewards-in-Learning-to-Reason.</p> - https://arxiv.org/abs/2505.22653 - Wed, 28 May 2025 17:59:03 +0000 + FaSTA^*: Fast-Slow Toolpath Agent with Subroutine Mining for Efficient Multi-turn Image Editing + https://arxiv.org/abs/2506.20911 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.20911.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Advait Gupta, Rishie Raj, Dang Nguyen, Tianyi Zhou</p><p><b>Upvotes:</b> 38</p><p><b>Summary:</b> We develop a cost-efficient neurosymbolic agent to address challenging multi-turn image editing tasks such as "Detect the bench in the image while recoloring it to pink. Also, remove the cat for a clearer view and recolor the wall to yellow.'' It combines the fast, high-level subtask planning by large language models (LLMs) with the slow, accurate, tool-use, and local A^* search per subtask to find a cost-efficient toolpath -- a sequence of calls to AI tools. To save the cost of A^* on similar subtasks, we perform inductive reasoning on previously successful toolpaths via LLMs to continuously extract/refine frequently used subroutines and reuse them as new tools for future tasks in an adaptive fast-slow planning, where the higher-level subroutines are explored first, and only when they fail, the low-level A^* search is activated. The reusable symbolic subroutines considerably save exploration cost on the same types of subtasks applied to similar images, yielding a human-like fast-slow toolpath agent "FaSTA^*'': fast subtask planning followed by rule-based subroutine selection per subtask is attempted by LLMs at first, which is expected to cover most tasks, while slow A^* search is only triggered for novel and challenging subtasks. By comparing with recent image editing approaches, we demonstrate FaSTA^* is significantly more computationally efficient while remaining competitive with the state-of-the-art baseline in terms of success rate.</p> + https://arxiv.org/abs/2506.20911 + Thu, 26 Jun 2025 00:33:43 +0000 - Table-R1: Inference-Time Scaling for Table Reasoning - https://arxiv.org/abs/2505.23621 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.23621.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Zheyuan Yang, Lyuhao Chen, Arman Cohan, Yilun Zhao</p><p><b>Upvotes:</b> 70</p><p><b>Summary:</b> In this work, we present the first study to explore inference-time scaling on table reasoning tasks. We develop and evaluate two post-training strategies to enable inference-time scaling: distillation from frontier model reasoning traces and reinforcement learning with verifiable rewards (RLVR). For distillation, we introduce a large-scale dataset of reasoning traces generated by DeepSeek-R1, which we use to fine-tune LLMs into the Table-R1-SFT model. For RLVR, we propose task-specific verifiable reward functions and apply the GRPO algorithm to obtain the Table-R1-Zero model. We evaluate our Table-R1-series models across diverse table reasoning tasks, including short-form QA, fact verification, and free-form QA. Notably, the Table-R1-Zero model matches or exceeds the performance of GPT-4.1 and DeepSeek-R1, while using only a 7B-parameter LLM. It also demonstrates strong generalization to out-of-domain datasets. Extensive ablation and qualitative analyses reveal the benefits of instruction tuning, model architecture choices, and cross-task generalization, as well as emergence of essential table reasoning skills during RL training.</p> - https://arxiv.org/abs/2505.23621 - Thu, 29 May 2025 16:28:50 +0000 + FineWeb2: One Pipeline to Scale Them All -- Adapting Pre-Training Data Processing to Every Language + https://arxiv.org/abs/2506.20920 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.20920.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Guilherme Penedo, Hynek Kydlíček, Vinko Sabolčec, Bettina Messmer, Negar Foroutan, Amir Hossein Kargaran, Colin Raffel, Martin Jaggi, Leandro Von Werra, Thomas Wolf</p><p><b>Upvotes:</b> 45</p><p><b>Summary:</b> Pre-training state-of-the-art large language models (LLMs) requires vast amounts of clean and diverse text data. While the open development of large high-quality English pre-training datasets has seen substantial recent progress, training performant multilingual LLMs remains a challenge, in large part due to the inherent difficulty of tailoring filtering and deduplication pipelines to a large number of languages. In this work, we introduce a new pre-training dataset curation pipeline based on FineWeb that can be automatically adapted to support any language. We extensively ablate our pipeline design choices on a set of nine diverse languages, guided by a set of meaningful and informative evaluation tasks that were chosen through a novel selection process based on measurable criteria. Ultimately, we show that our pipeline can be used to create non-English corpora that produce more performant models than prior datasets. We additionally introduce a straightforward and principled approach to rebalance datasets that takes into consideration both duplication count and quality, providing an additional performance uplift. Finally, we scale our pipeline to over 1000 languages using almost 100 Common Crawl snapshots to produce FineWeb2, a new 20 terabyte (5 billion document) multilingual dataset which we release along with our pipeline, training, and evaluation codebases.</p> + https://arxiv.org/abs/2506.20920 + Thu, 26 Jun 2025 01:01:47 +0000 - VF-Eval: Evaluating Multimodal LLMs for Generating Feedback on AIGC Videos - https://arxiv.org/abs/2505.23693 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.23693.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Tingyu Song, Tongyan Hu, Guo Gan, Yilun Zhao</p><p><b>Upvotes:</b> 51</p><p><b>Summary:</b> MLLMs have been widely studied for video question answering recently. However, most existing assessments focus on natural videos, overlooking synthetic videos, such as AI-generated content (AIGC). Meanwhile, some works in video generation rely on MLLMs to evaluate the quality of generated videos, but the capabilities of MLLMs on interpreting AIGC videos remain largely underexplored. To address this, we propose a new benchmark, VF-Eval, which introduces four tasks-coherence validation, error awareness, error type detection, and reasoning evaluation-to comprehensively evaluate the abilities of MLLMs on AIGC videos. We evaluate 13 frontier MLLMs on VF-Eval and find that even the best-performing model, GPT-4.1, struggles to achieve consistently good performance across all tasks. This highlights the challenging nature of our benchmark. Additionally, to investigate the practical applications of VF-Eval in improving video generation, we conduct an experiment, RePrompt, demonstrating that aligning MLLMs more closely with human feedback can benefit video generation.</p> - https://arxiv.org/abs/2505.23693 - Thu, 29 May 2025 17:31:13 +0000 + Mind2Web 2: Evaluating Agentic Search with Agent-as-a-Judge + https://arxiv.org/abs/2506.21506 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.21506.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Boyu Gou, Zanming Huang, Yuting Ning, Yu Gu, Michael Lin, Weijian Qi, Andrei Kopanev, Botao Yu, Bernal Jiménez Gutiérrez, Yiheng Shu, Chan Hee Song, Jiaman Wu, Shijie Chen, Hanane Nour Moussa, Tianshu Zhang, Jian Xie, Yifei Li, Tianci Xue, Zeyi Liao, Kai Zhang, Boyuan Zheng, Zhaowei Cai, Viktor Rozgic, Morteza Ziyadi, Huan Sun, Yu Su</p><p><b>Upvotes:</b> 41</p><p><b>Summary:</b> Agentic search such as Deep Research systems, where large language models autonomously browse the web, synthesize information, and return comprehensive citation-backed answers, represents a major shift in how users interact with web-scale information. While promising greater efficiency and cognitive offloading, the growing complexity and open-endedness of agentic search have outpaced existing evaluation benchmarks and methodologies, which largely assume short search horizons and static answers. In this paper, we introduce Mind2Web 2, a benchmark of 130 realistic, high-quality, and long-horizon tasks that require real-time web browsing and extensive information synthesis, constructed with over 1,000 hours of human labor. To address the challenge of evaluating time-varying and complex answers, we propose a novel Agent-as-a-Judge framework. Our method constructs task-specific judge agents based on a tree-structured rubric design to automatically assess both answer correctness and source attribution. We conduct a comprehensive evaluation of nine frontier agentic search systems and human performance, along with a detailed error analysis to draw insights for future development. The best-performing system, OpenAI Deep Research, can already achieve 50-70% of human performance while spending half the time, showing a great potential. Altogether, Mind2Web 2 provides a rigorous foundation for developing and benchmarking the next generation of agentic search systems.</p> + https://arxiv.org/abs/2506.21506 + Thu, 26 Jun 2025 17:32:50 +0000 - Spatial-MLLM: Boosting MLLM Capabilities in Visual-based Spatial Intelligence - https://arxiv.org/abs/2505.23747 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.23747.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Diankun Wu, Fangfu Liu, Yi-Hsin Hung, Yueqi Duan</p><p><b>Upvotes:</b> 57</p><p><b>Summary:</b> Recent advancements in Multimodal Large Language Models (MLLMs) have significantly enhanced performance on 2D visual tasks. However, improving their spatial intelligence remains a challenge. Existing 3D MLLMs always rely on additional 3D or 2.5D data to incorporate spatial awareness, restricting their utility in scenarios with only 2D inputs, such as images or videos. In this paper, we present Spatial-MLLM, a novel framework for visual-based spatial reasoning from purely 2D observations. Unlike conventional video MLLMs which rely on CLIP-based visual encoders optimized for semantic understanding, our key insight is to unleash the strong structure prior from the feed-forward visual geometry foundation model. Specifically, we propose a dual-encoder architecture: a pretrained 2D visual encoder to extract semantic features, and a spatial encoder-initialized from the backbone of the visual geometry model-to extract 3D structure features. A connector then integrates both features into unified visual tokens for enhanced spatial understanding. Furthermore, we propose a space-aware frame sampling strategy at inference time, which selects the spatially informative frames of a video sequence, ensuring that even under limited token length, the model focuses on frames critical for spatial reasoning. Beyond architecture improvements, we construct the Spatial-MLLM-120k dataset and train the model on it using supervised fine-tuning and GRPO. Extensive experiments on various real-world datasets demonstrate that our spatial-MLLM achieves state-of-the-art performance in a wide range of visual-based spatial understanding and reasoning tasks. Project page: https://diankun-wu.github.io/Spatial-MLLM/.</p> - https://arxiv.org/abs/2505.23747 - Thu, 29 May 2025 17:59:04 +0000 + MADrive: Memory-Augmented Driving Scene Modeling + https://arxiv.org/abs/2506.21520 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.21520.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Polina Karpikova, Daniil Selikhanovych, Kirill Struminsky, Ruslan Musaev, Maria Golitsyna, Dmitry Baranchuk</p><p><b>Upvotes:</b> 35</p><p><b>Summary:</b> Recent advances in scene reconstruction have pushed toward highly realistic modeling of autonomous driving (AD) environments using 3D Gaussian splatting. However, the resulting reconstructions remain closely tied to the original observations and struggle to support photorealistic synthesis of significantly altered or novel driving scenarios. This work introduces MADrive, a memory-augmented reconstruction framework designed to extend the capabilities of existing scene reconstruction methods by replacing observed vehicles with visually similar 3D assets retrieved from a large-scale external memory bank. Specifically, we release MAD-Cars, a curated dataset of {sim}70K 360{\deg} car videos captured in the wild and present a retrieval module that finds the most similar car instances in the memory bank, reconstructs the corresponding 3D assets from video, and integrates them into the target scene through orientation alignment and relighting. The resulting replacements provide complete multi-view representations of vehicles in the scene, enabling photorealistic synthesis of substantially altered configurations, as demonstrated in our experiments. Project page: https://yandex-research.github.io/madrive/</p> + https://arxiv.org/abs/2506.21520 + Thu, 26 Jun 2025 17:41:07 +0000 - ZeroGUI: Automating Online GUI Learning at Zero Human Cost - https://arxiv.org/abs/2505.23762 - <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2505.23762.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Chenyu Yang, Shiqian Su, Shi Liu, Xuan Dong, Yue Yu, Weijie Su, Xuehui Wang, Zhaoyang Liu, Jinguo Zhu, Hao Li, Wenhai Wang, Yu Qiao, Xizhou Zhu, Jifeng Dai</p><p><b>Upvotes:</b> 41</p><p><b>Summary:</b> The rapid advancement of large Vision-Language Models (VLMs) has propelled the development of pure-vision-based GUI Agents, capable of perceiving and operating Graphical User Interfaces (GUI) to autonomously fulfill user instructions. However, existing approaches usually adopt an offline learning framework, which faces two core limitations: (1) heavy reliance on high-quality manual annotations for element grounding and action supervision, and (2) limited adaptability to dynamic and interactive environments. To address these limitations, we propose ZeroGUI, a scalable, online learning framework for automating GUI Agent training at Zero human cost. Specifically, ZeroGUI integrates (i) VLM-based automatic task generation to produce diverse training goals from the current environment state, (ii) VLM-based automatic reward estimation to assess task success without hand-crafted evaluation functions, and (iii) two-stage online reinforcement learning to continuously interact with and learn from GUI environments. Experiments on two advanced GUI Agents (UI-TARS and Aguvis) demonstrate that ZeroGUI significantly boosts performance across OSWorld and AndroidLab environments. The code is available at https://github.com/OpenGVLab/ZeroGUI.</p> - https://arxiv.org/abs/2505.23762 - Thu, 29 May 2025 17:59:51 +0000 + WorldVLA: Towards Autoregressive Action World Model + https://arxiv.org/abs/2506.21539 + <p><img src="https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2506.21539.png" alt="Paper thumbnail" style="max-width: 300px; height: auto;" /></p><p><b>Authors:</b> Jun Cen, Chaohui Yu, Hangjie Yuan, Yuming Jiang, Siteng Huang, Jiayan Guo, Xin Li, Yibing Song, Hao Luo, Fan Wang, Deli Zhao, Hao Chen</p><p><b>Upvotes:</b> 34</p><p><b>Summary:</b> We present WorldVLA, an autoregressive action world model that unifies action and image understanding and generation. Our WorldVLA intergrates Vision-Language-Action (VLA) model and world model in one single framework. The world model predicts future images by leveraging both action and image understanding, with the purpose of learning the underlying physics of the environment to improve action generation. Meanwhile, the action model generates the subsequent actions based on image observations, aiding in visual understanding and in turn helps visual generation of the world model. We demonstrate that WorldVLA outperforms standalone action and world models, highlighting the mutual enhancement between the world model and the action model. In addition, we find that the performance of the action model deteriorates when generating sequences of actions in an autoregressive manner. This phenomenon can be attributed to the model's limited generalization capability for action prediction, leading to the propagation of errors from earlier actions to subsequent ones. To address this issue, we propose an attention mask strategy that selectively masks prior actions during the generation of the current action, which shows significant performance improvement in the action chunk generation task.</p> + https://arxiv.org/abs/2506.21539 + Thu, 26 Jun 2025 17:55:40 +0000