date
timestamp[ns]date 2023-05-05 00:00:00
2025-05-15 00:00:00
| arxiv_id
stringlengths 10
10
| title
stringlengths 8
202
| authors
sequencelengths 1
942
| github
stringlengths 0
116
| abstract
stringlengths 165
1.92k
|
|---|---|---|---|---|---|
2025-05-08T00:00:00 | 2505.04588 | ZeroSearch: Incentivize the Search Capability of LLMs without Searching | [
"Hao Sun",
"Zile Qiao",
"Jiayan Guo",
"Xuanbo Fan",
"Yingyan Hou",
"Yong Jiang",
"Pengjun Xie",
"Fei Huang",
"Yan Zhang"
] | 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. |
|
2025-05-08T00:00:00 | 2505.04364 | Benchmarking LLMs' Swarm intelligence | [
"Kai Ruan",
"Mowen Huang",
"Ji-Rong Wen",
"Hao Sun"
] | https://github.com/x66ccff/swarmbench | Large Language Models (LLMs) show potential for complex reasoning, yet their capacity for emergent coordination in Multi-Agent Systems (MAS) when operating under strict constraints-such as limited local perception and communication, characteristic of natural swarms-remains largely unexplored, particularly concerning the nuances of swarm intelligence. Existing benchmarks often do not fully capture the unique challenges of decentralized coordination that arise when agents operate with incomplete spatio-temporal information. To bridge this gap, we introduce SwarmBench, a novel benchmark designed to systematically evaluate the swarm intelligence capabilities of LLMs acting as decentralized agents. SwarmBench features five foundational MAS coordination tasks within a configurable 2D grid environment, forcing agents to rely primarily on local sensory input (k x k view) and local communication. We propose metrics for coordination effectiveness and analyze emergent group dynamics. Evaluating several leading LLMs in a zero-shot setting, we find significant performance variations across tasks, highlighting the difficulties posed by local information constraints. While some coordination emerges, results indicate limitations in robust planning and strategy formation under uncertainty in these decentralized scenarios. Assessing LLMs under swarm-like conditions is crucial for realizing their potential in future decentralized systems. We release SwarmBench as an open, extensible toolkit-built upon a customizable and scalable physical system with defined mechanical properties. It provides environments, prompts, evaluation scripts, and the comprehensive experimental datasets generated, aiming to foster reproducible research into LLM-based MAS coordination and the theoretical underpinnings of Embodied MAS. Our code repository is available at https://github.com/x66ccff/swarmbench. |
2025-05-08T00:00:00 | 2505.02393 | Uncertainty-Weighted Image-Event Multimodal Fusion for Video Anomaly Detection | [
"Sungheon Jeong",
"Jihong Park",
"Mohsen Imani"
] | https://github.com/EavnJeong/IEF-VAD | Most existing video anomaly detectors rely solely on RGB frames, which lack the temporal resolution needed to capture abrupt or transient motion cues, key indicators of anomalous events. To address this limitation, we propose Image-Event Fusion for Video Anomaly Detection (IEF-VAD), a framework that synthesizes event representations directly from RGB videos and fuses them with image features through a principled, uncertainty-aware process. The system (i) models heavy-tailed sensor noise with a Student`s-t likelihood, deriving value-level inverse-variance weights via a Laplace approximation; (ii) applies Kalman-style frame-wise updates to balance modalities over time; and (iii) iteratively refines the fused latent state to erase residual cross-modal noise. Without any dedicated event sensor or frame-level labels, IEF-VAD sets a new state of the art across multiple real-world anomaly detection benchmarks. These findings highlight the utility of synthetic event representations in emphasizing motion cues that are often underrepresented in RGB frames, enabling accurate and robust video understanding across diverse applications without requiring dedicated event sensors. Code and models are available at https://github.com/EavnJeong/IEF-VAD. |
2025-05-08T00:00:00 | 2505.04528 | Beyond Theorem Proving: Formulation, Framework and Benchmark for Formal Problem-Solving | [
"Qi Liu",
"Xinhao Zheng",
"Renqiu Xia",
"Xingzhi Qi",
"Qinxiang Cao",
"Junchi Yan"
] | As a seemingly self-explanatory task, problem-solving has been a significant component of science and engineering. However, a general yet concrete formulation of problem-solving itself is missing. With the recent development of AI-based problem-solving agents, the demand for process-level verifiability is rapidly increasing yet underexplored. To fill these gaps, we present a principled formulation of problem-solving as a deterministic Markov decision process; a novel framework, FPS (Formal Problem-Solving), which utilizes existing FTP (formal theorem proving) environments to perform process-verified problem-solving; and D-FPS (Deductive FPS), decoupling solving and answer verification for better human-alignment. The expressiveness, soundness and completeness of the frameworks are proven. We construct three benchmarks on problem-solving: FormalMath500, a formalization of a subset of the MATH500 benchmark; MiniF2F-Solving and PutnamBench-Solving, adaptations of FTP benchmarks MiniF2F and PutnamBench. For faithful, interpretable, and human-aligned evaluation, we propose RPE (Restricted Propositional Equivalence), a symbolic approach to determine the correctness of answers by formal verification. We evaluate four prevalent FTP models and two prompting methods as baselines, solving at most 23.77% of FormalMath500, 27.47% of MiniF2F-Solving, and 0.31% of PutnamBench-Solving. |
|
2025-05-08T00:00:00 | 2505.03418 | Knowledge Augmented Complex Problem Solving with Large Language Models: A Survey | [
"Da Zheng",
"Lun Du",
"Junwei Su",
"Yuchen Tian",
"Yuqi Zhu",
"Jintian Zhang",
"Lanning Wei",
"Ningyu Zhang",
"Huajun Chen"
] | Problem-solving has been a fundamental driver of human progress in numerous domains. With advancements in artificial intelligence, Large Language Models (LLMs) have emerged as powerful tools capable of tackling complex problems across diverse domains. Unlike traditional computational systems, LLMs combine raw computational power with an approximation of human reasoning, allowing them to generate solutions, make inferences, and even leverage external computational tools. However, applying LLMs to real-world problem-solving presents significant challenges, including multi-step reasoning, domain knowledge integration, and result verification. This survey explores the capabilities and limitations of LLMs in complex problem-solving, examining techniques including Chain-of-Thought (CoT) reasoning, knowledge augmentation, and various LLM-based and tool-based verification techniques. Additionally, we highlight domain-specific challenges in various domains, such as software engineering, mathematical reasoning and proving, data analysis and modeling, and scientific research. The paper further discusses the fundamental limitations of the current LLM solutions and the future directions of LLM-based complex problems solving from the perspective of multi-step reasoning, domain knowledge integration and result verification. |
|
2025-05-08T00:00:00 | 2505.04622 | PrimitiveAnything: Human-Crafted 3D Primitive Assembly Generation with Auto-Regressive Transformer | [
"Jingwen Ye",
"Yuze He",
"Yanning Zhou",
"Yiqin Zhu",
"Kaiwen Xiao",
"Yong-Jin Liu",
"Wei Yang",
"Xiao Han"
] | Shape primitive abstraction, which decomposes complex 3D shapes into simple geometric elements, plays a crucial role in human visual cognition and has broad applications in computer vision and graphics. While recent advances in 3D content generation have shown remarkable progress, existing primitive abstraction methods either rely on geometric optimization with limited semantic understanding or learn from small-scale, category-specific datasets, struggling to generalize across diverse shape categories. We present PrimitiveAnything, a novel framework that reformulates shape primitive abstraction as a primitive assembly generation task. PrimitiveAnything includes a shape-conditioned primitive transformer for auto-regressive generation and an ambiguity-free parameterization scheme to represent multiple types of primitives in a unified manner. The proposed framework directly learns the process of primitive assembly from large-scale human-crafted abstractions, enabling it to capture how humans decompose complex shapes into primitive elements. Through extensive experiments, we demonstrate that PrimitiveAnything can generate high-quality primitive assemblies that better align with human perception while maintaining geometric fidelity across diverse shape categories. It benefits various 3D applications and shows potential for enabling primitive-based user-generated content (UGC) in games. Project page: https://primitiveanything.github.io |
|
2025-05-08T00:00:00 | 2505.04512 | HunyuanCustom: A Multimodal-Driven Architecture for Customized Video Generation | [
"Teng Hu",
"Zhentao Yu",
"Zhengguang Zhou",
"Sen Liang",
"Yuan Zhou",
"Qin Lin",
"Qinglin Lu"
] | Customized video generation aims to produce videos featuring specific subjects under flexible user-defined conditions, yet existing methods often struggle with identity consistency and limited input modalities. In this paper, we propose HunyuanCustom, a multi-modal customized video generation framework that emphasizes subject consistency while supporting image, audio, video, and text conditions. Built upon HunyuanVideo, our model first addresses the image-text conditioned generation task by introducing a text-image fusion module based on LLaVA for enhanced multi-modal understanding, along with an image ID enhancement module that leverages temporal concatenation to reinforce identity features across frames. To enable audio- and video-conditioned generation, we further propose modality-specific condition injection mechanisms: an AudioNet module that achieves hierarchical alignment via spatial cross-attention, and a video-driven injection module that integrates latent-compressed conditional video through a patchify-based feature-alignment network. Extensive experiments on single- and multi-subject scenarios demonstrate that HunyuanCustom significantly outperforms state-of-the-art open- and closed-source methods in terms of ID consistency, realism, and text-video alignment. Moreover, we validate its robustness across downstream tasks, including audio and video-driven customized video generation. Our results highlight the effectiveness of multi-modal conditioning and identity-preserving strategies in advancing controllable video generation. All the code and models are available at https://hunyuancustom.github.io. |
|
2025-05-08T00:00:00 | 2505.03912 | OpenHelix: A Short Survey, Empirical Analysis, and Open-Source Dual-System VLA Model for Robotic Manipulation | [
"Can Cui",
"Pengxiang Ding",
"Wenxuan Song",
"Shuanghao Bai",
"Xinyang Tong",
"Zirui Ge",
"Runze Suo",
"Wanqi Zhou",
"Yang Liu",
"Bofang Jia",
"Han Zhao",
"Siteng Huang",
"Donglin Wang"
] | Dual-system VLA (Vision-Language-Action) architectures have become a hot topic in embodied intelligence research, but there is a lack of sufficient open-source work for further performance analysis and optimization. To address this problem, this paper will summarize and compare the structural designs of existing dual-system architectures, and conduct systematic empirical evaluations on the core design elements of existing dual-system architectures. Ultimately, it will provide a low-cost open-source model for further exploration. Of course, this project will continue to update with more experimental conclusions and open-source models with improved performance for everyone to choose from. Project page: https://openhelix-robot.github.io/. |
|
2025-05-08T00:00:00 | 2505.00358 | R&B: Domain Regrouping and Data Mixture Balancing for Efficient Foundation Model Training | [
"Albert Ge",
"Tzu-Heng Huang",
"John Cooper",
"Avi Trost",
"Ziyi Chu",
"Satya Sai Srinath Namburi GNVV",
"Ziyang Cai",
"Kendall Park",
"Nicholas Roberts",
"Frederic Sala"
] | Data mixing strategies have successfully reduced the costs involved in training language models. While promising, such methods suffer from two flaws. First, they rely on predetermined data domains (e.g., data sources, task types), which may fail to capture critical semantic nuances, leaving performance on the table. Second, these methods scale with the number of domains in a computationally prohibitive way. We address these challenges via R&B, a framework that re-partitions training data based on semantic similarity (Regroup) to create finer-grained domains, and efficiently optimizes the data composition (Balance) by leveraging a Gram matrix induced by domain gradients obtained throughout training. Unlike prior works, it removes the need for additional compute to obtain evaluation information such as losses or gradients. We analyze this technique under standard regularity conditions and provide theoretical insights that justify R&B's effectiveness compared to non-adaptive mixing approaches. Empirically, we demonstrate the effectiveness of R&B on five diverse datasets ranging from natural language to reasoning and multimodal tasks. With as little as 0.01% additional compute overhead, R&B matches or exceeds the performance of state-of-the-art data mixing strategies. |
|
2025-05-08T00:00:00 | 2505.02567 | Unified Multimodal Understanding and Generation Models: Advances, Challenges, and Opportunities | [
"Xinjie Zhang",
"Jintao Guo",
"Shanshan Zhao",
"Minghao Fu",
"Lunhao Duan",
"Guo-Hua Wang",
"Qing-Guo Chen",
"Zhao Xu",
"Weihua Luo",
"Kaifu Zhang"
] | https://github.com/AIDC-AI/Awesome-Unified-Multimodal-Models | 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). |
2025-05-08T00:00:00 | 2505.03570 | OSUniverse: Benchmark for Multimodal GUI-navigation AI Agents | [
"Mariya Davydova",
"Daniel Jeffries",
"Patrick Barker",
"Arturo Márquez Flores",
"Sinéad Ryan"
] | https://github.com/agentsea/osuniverse | In this paper, we introduce OSUniverse: a benchmark of complex, multimodal desktop-oriented tasks for advanced GUI-navigation AI agents that focuses on ease of use, extensibility, comprehensive coverage of test cases, and automated validation. We divide the tasks in increasing levels of complexity, from basic precision clicking to multistep, multiapplication tests requiring dexterity, precision, and clear thinking from the agent. In version one of the benchmark, presented here, we have calibrated the complexity of the benchmark test cases to ensure that the SOTA (State of the Art) agents (at the time of publication) do not achieve results higher than 50%, while the average white collar worker can perform all these tasks with perfect accuracy. The benchmark can be scored manually, but we also introduce an automated validation mechanism that has an average error rate less than 2%. Therefore, this benchmark presents solid ground for fully automated measuring of progress, capabilities and the effectiveness of GUI-navigation AI agents over the short and medium-term horizon. The source code of the benchmark is available at https://github.com/agentsea/osuniverse. |
2025-05-08T00:00:00 | 2505.03821 | Beyond Recognition: Evaluating Visual Perspective Taking in Vision Language Models | [
"Gracjan Góral",
"Alicja Ziarko",
"Piotr Miłoś",
"Michał Nauman",
"Maciej Wołczyk",
"Michał Kosiński"
] | We investigate the ability of Vision Language Models (VLMs) to perform visual perspective taking using a novel set of visual tasks inspired by established human tests. Our approach leverages carefully controlled scenes, in which a single humanoid minifigure is paired with a single object. By systematically varying spatial configurations - such as object position relative to the humanoid minifigure and the humanoid minifigure's orientation - and using both bird's-eye and surface-level views, we created 144 unique visual tasks. Each visual task is paired with a series of 7 diagnostic questions designed to assess three levels of visual cognition: scene understanding, spatial reasoning, and visual perspective taking. Our evaluation of several state-of-the-art models, including GPT-4-Turbo, GPT-4o, Llama-3.2-11B-Vision-Instruct, and variants of Claude Sonnet, reveals that while they excel in scene understanding, the performance declines significantly on spatial reasoning and further deteriorates on perspective-taking. Our analysis suggests a gap between surface-level object recognition and the deeper spatial and perspective reasoning required for complex visual tasks, pointing to the need for integrating explicit geometric representations and tailored training protocols in future VLM development. |
|
2025-05-08T00:00:00 | 2505.04606 | OmniGIRL: A Multilingual and Multimodal Benchmark for GitHub Issue Resolution | [
"Lianghong Guo",
"Wei Tao",
"Runhan Jiang",
"Yanlin Wang",
"Jiachi Chen",
"Xilin Liu",
"Yuchi Ma",
"Mingzhi Mao",
"Hongyu Zhang",
"Zibin Zheng"
] | The GitHub issue resolution task aims to resolve issues reported in repositories automatically. With advances in large language models (LLMs), this task has gained increasing attention, and several benchmarks are proposed to evaluate the issue resolution ability of LLMs. However, existing benchmarks have three main limitations. First, current benchmarks focus on a single programming language, limiting the evaluation of issues from repositories across different languages. Second, they usually cover a narrow range of domains, which may fail to represent the diversity of real-world issues. Third, existing benchmarks rely solely on textual information in issue descriptions, overlooking multimodal information such as images in issues. In this paper, we propose OmniGIRL, a GitHub Issue ResoLution benchmark that is multilingual, multimodal, and multi-domain. OmniGIRL includes 959 task instances, which are collected from repositories across four programming languages (i.e., Python, JavaScript, TypeScript, and Java) and eight different domains. Our evaluation shows that current LLMs show limited performances on OmniGIRL. Notably, the best-performing model, GPT-4o, resolves only 8.6% of the issues. Besides, we find that current LLMs struggle to resolve issues requiring understanding images. The best performance is achieved by Claude-3.5-Sonnet, which resolves only 10.5% of the issues with image information. Finally, we analyze the reasons behind current LLMs' failure on OmniGIRL, providing insights for future improvements. |
|
2025-05-08T00:00:00 | 2505.03105 | Cognitio Emergens: Agency, Dimensions, and Dynamics in Human-AI Knowledge Co-Creation | [
"Xule Lin"
] | Scientific knowledge creation is fundamentally transforming as humans and AI systems evolve beyond tool-user relationships into co-evolutionary epistemic partnerships. When AlphaFold revolutionized protein structure prediction, researchers described engaging with an epistemic partner that reshaped how they conceptualized fundamental relationships. This article introduces Cognitio Emergens (CE), a framework addressing critical limitations in existing models that focus on static roles or narrow metrics while failing to capture how scientific understanding emerges through recursive human-AI interaction over time. CE integrates three components addressing these limitations: Agency Configurations describing how authority distributes between humans and AI (Directed, Contributory, Partnership), with partnerships dynamically oscillating between configurations rather than following linear progression; Epistemic Dimensions capturing six specific capabilities emerging through collaboration across Discovery, Integration, and Projection axes, creating distinctive "capability signatures" that guide development; and Partnership Dynamics identifying forces shaping how these relationships evolve, particularly the risk of epistemic alienation where researchers lose interpretive control over knowledge they formally endorse. Drawing from autopoiesis theory, social systems theory, and organizational modularity, CE reveals how knowledge co-creation emerges through continuous negotiation of roles, values, and organizational structures. By reconceptualizing human-AI scientific collaboration as fundamentally co-evolutionary, CE offers a balanced perspective that neither uncritically celebrates nor unnecessarily fears AI's evolving role, instead providing conceptual tools for cultivating partnerships that maintain meaningful human participation while enabling transformative scientific breakthroughs. |
|
2025-05-08T00:00:00 | 2505.03538 | RAIL: Region-Aware Instructive Learning for Semi-Supervised Tooth Segmentation in CBCT | [
"Chuyu Zhao",
"Hao Huang",
"Jiashuo Guo",
"Ziyu Shen",
"Zhongwei Zhou",
"Jie Liu",
"Zekuan Yu"
] | https://github.com/Tournesol-Saturday/RAIL | Semi-supervised learning has become a compelling approach for 3D tooth segmentation from CBCT scans, where labeled data is minimal. However, existing methods still face two persistent challenges: limited corrective supervision in structurally ambiguous or mislabeled regions during supervised training and performance degradation caused by unreliable pseudo-labels on unlabeled data. To address these problems, we propose Region-Aware Instructive Learning (RAIL), a dual-group dual-student, semi-supervised framework. Each group contains two student models guided by a shared teacher network. By alternating training between the two groups, RAIL promotes intergroup knowledge transfer and collaborative region-aware instruction while reducing overfitting to the characteristics of any single model. Specifically, RAIL introduces two instructive mechanisms. Disagreement-Focused Supervision (DFS) Controller improves supervised learning by instructing predictions only within areas where student outputs diverge from both ground truth and the best student, thereby concentrating supervision on structurally ambiguous or mislabeled areas. In the unsupervised phase, Confidence-Aware Learning (CAL) Modulator reinforces agreement in regions with high model certainty while reducing the effect of low-confidence predictions during training. This helps prevent our model from learning unstable patterns and improves the overall reliability of pseudo-labels. Extensive experiments on four CBCT tooth segmentation datasets show that RAIL surpasses state-of-the-art methods under limited annotation. Our code will be available at https://github.com/Tournesol-Saturday/RAIL. |
2025-05-08T00:00:00 | 2505.04253 | LLM-Independent Adaptive RAG: Let the Question Speak for Itself | [
"Maria Marina",
"Nikolay Ivanov",
"Sergey Pletenev",
"Mikhail Salnikov",
"Daria Galimzianova",
"Nikita Krayko",
"Vasily Konovalov",
"Alexander Panchenko",
"Viktor Moskvoretskii"
] | Large Language Models~(LLMs) are prone to hallucinations, and Retrieval-Augmented Generation (RAG) helps mitigate this, but at a high computational cost while risking misinformation. Adaptive retrieval aims to retrieve only when necessary, but existing approaches rely on LLM-based uncertainty estimation, which remain inefficient and impractical. In this study, we introduce lightweight LLM-independent adaptive retrieval methods based on external information. We investigated 27 features, organized into 7 groups, and their hybrid combinations. We evaluated these methods on 6 QA datasets, assessing the QA performance and efficiency. The results show that our approach matches the performance of complex LLM-based methods while achieving significant efficiency gains, demonstrating the potential of external information for adaptive retrieval. |
|
2025-05-08T00:00:00 | 2505.02820 | AutoLibra: Agent Metric Induction from Open-Ended Feedback | [
"Hao Zhu",
"Phil Cuvin",
"Xinkai Yu",
"Charlotte Ka Yee Yan",
"Jason Zhang",
"Diyi Yang"
] | Agents are predominantly evaluated and optimized via task success metrics, which are coarse, rely on manual design from experts, and fail to reward intermediate emergent behaviors. We propose AutoLibra, a framework for agent evaluation, that transforms open-ended human feedback, e.g., "If you find that the button is disabled, don't click it again", or "This agent has too much autonomy to decide what to do on its own", into metrics for evaluating fine-grained behaviors in agent trajectories. AutoLibra accomplishes this by grounding feedback to an agent's behavior, clustering similar positive and negative behaviors, and creating concrete metrics with clear definitions and concrete examples, which can be used for prompting LLM-as-a-Judge as evaluators. We further propose two meta-metrics to evaluate the alignment of a set of (induced) metrics with open feedback: "coverage" and "redundancy". Through optimizing these meta-metrics, we experimentally demonstrate AutoLibra's ability to induce more concrete agent evaluation metrics than the ones proposed in previous agent evaluation benchmarks and discover new metrics to analyze agents. We also present two applications of AutoLibra in agent improvement: First, we show that AutoLibra-induced metrics serve as better prompt-engineering targets than the task success rate on a wide range of text game tasks, improving agent performance over baseline by a mean of 20%. Second, we show that AutoLibra can iteratively select high-quality fine-tuning data for web navigation agents. Our results suggest that AutoLibra is a powerful task-agnostic tool for evaluating and improving language agents. |
|
2025-05-08T00:00:00 | 2505.04620 | On Path to Multimodal Generalist: General-Level and General-Bench | [
"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"
] | 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/ |
|
2025-05-08T00:00:00 | 2505.01449 | COSMOS: Predictable and Cost-Effective Adaptation of LLMs | [
"Jiayu Wang",
"Aws Albarghouthi",
"Frederic Sala"
] | Large language models (LLMs) achieve remarkable performance across numerous tasks by using a diverse array of adaptation strategies. However, optimally selecting a model and adaptation strategy under resource constraints is challenging and often requires extensive experimentation. We investigate whether it is possible to accurately predict both performance and cost without expensive trials. We formalize the strategy selection problem for LLMs and introduce COSMOS, a unified prediction framework that efficiently estimates adaptation outcomes at minimal cost. We instantiate and study the capability of our framework via a pair of powerful predictors: embedding-augmented lightweight proxy models to predict fine-tuning performance, and low-sample scaling laws to forecast retrieval-augmented in-context learning. Extensive evaluation across eight representative benchmarks demonstrates that COSMOS achieves high prediction accuracy while reducing computational costs by 92.72% on average, and up to 98.71% in resource-intensive scenarios. Our results show that efficient prediction of adaptation outcomes is not only feasible but can substantially reduce the computational overhead of LLM deployment while maintaining performance standards. |
|
2025-05-08T00:00:00 | 2505.04601 | OpenVision: A Fully-Open, Cost-Effective Family of Advanced Vision Encoders for Multimodal Learning | [
"Xianhang Li",
"Yanqing Liu",
"Haoqin Tu",
"Hongru Zhu",
"Cihang Xie"
] | OpenAI's CLIP, released in early 2021, have long been the go-to choice of vision encoder for building multimodal foundation models. Although recent alternatives such as SigLIP have begun to challenge this status quo, to our knowledge none are fully open: their training data remains proprietary and/or their training recipes are not released. This paper fills this gap with OpenVision, a fully-open, cost-effective family of vision encoders that match or surpass the performance of OpenAI's CLIP when integrated into multimodal frameworks like LLaVA. OpenVision builds on existing works -- e.g., CLIPS for training framework and Recap-DataComp-1B for training data -- while revealing multiple key insights in enhancing encoder quality and showcasing practical benefits in advancing multimodal models. By releasing vision encoders spanning from 5.9M to 632.1M parameters, OpenVision offers practitioners a flexible trade-off between capacity and efficiency in building multimodal models: larger models deliver enhanced multimodal performance, while smaller versions enable lightweight, edge-ready multimodal deployments. |
|
2025-05-09T00:00:00 | 2505.02847 | Sentient Agent as a Judge: Evaluating Higher-Order Social Cognition in Large Language Models | [
"Bang Zhang",
"Ruotian Ma",
"Qingxuan Jiang",
"Peisong Wang",
"Jiaqi Chen",
"Zheng Xie",
"Xingyu Chen",
"Yue Wang",
"Fanghua Ye",
"Jian Li",
"Yifan Yang",
"Zhaopeng Tu",
"Xiaolong Li"
] | Assessing how well a large language model (LLM) understands human, rather than merely text, remains an open challenge. To bridge the gap, we introduce Sentient Agent as a Judge (SAGE), an automated evaluation framework that measures an LLM's higher-order social cognition. SAGE instantiates a Sentient Agent that simulates human-like emotional changes and inner thoughts during interaction, providing a more realistic evaluation of the tested model in multi-turn conversations. At every turn, the agent reasons about (i) how its emotion changes, (ii) how it feels, and (iii) how it should reply, yielding a numerical emotion trajectory and interpretable inner thoughts. Experiments on 100 supportive-dialogue scenarios show that the final Sentient emotion score correlates strongly with Barrett-Lennard Relationship Inventory (BLRI) ratings and utterance-level empathy metrics, validating psychological fidelity. We also build a public Sentient Leaderboard covering 18 commercial and open-source models that uncovers substantial gaps (up to 4x) between frontier systems (GPT-4o-Latest, Gemini2.5-Pro) and earlier baselines, gaps not reflected in conventional leaderboards (e.g., Arena). SAGE thus provides a principled, scalable and interpretable tool for tracking progress toward genuinely empathetic and socially adept language agents. |
|
2025-05-09T00:00:00 | 2505.05315 | Scalable Chain of Thoughts via Elastic Reasoning | [
"Yuhui Xu",
"Hanze Dong",
"Lei Wang",
"Doyen Sahoo",
"Junnan Li",
"Caiming Xiong"
] | Large reasoning models (LRMs) have achieved remarkable progress on complex tasks by generating extended chains of thought (CoT). However, their uncontrolled output lengths pose significant challenges for real-world deployment, where inference-time budgets on tokens, latency, or compute are strictly constrained. We propose Elastic Reasoning, a novel framework for scalable chain of thoughts that explicitly separates reasoning into two phases--thinking and solution--with independently allocated budgets. At test time, Elastic Reasoning prioritize that completeness of solution segments, significantly improving reliability under tight resource constraints. To train models that are robust to truncated thinking, we introduce a lightweight budget-constrained rollout strategy, integrated into GRPO, which teaches the model to reason adaptively when the thinking process is cut short and generalizes effectively to unseen budget constraints without additional training. Empirical results on mathematical (AIME, MATH500) and programming (LiveCodeBench, Codeforces) benchmarks demonstrate that Elastic Reasoning performs robustly under strict budget constraints, while incurring significantly lower training cost than baseline methods. Remarkably, our approach also produces more concise and efficient reasoning even in unconstrained settings. Elastic Reasoning offers a principled and practical solution to the pressing challenge of controllable reasoning at scale. |
|
2025-05-09T00:00:00 | 2505.05474 | 3D Scene Generation: A Survey | [
"Beichen Wen",
"Haozhe Xie",
"Zhaoxi Chen",
"Fangzhou Hong",
"Ziwei Liu"
] | https://github.com/hzxie/Awesome-3D-Scene-Generation | 3D scene generation seeks to synthesize spatially structured, semantically meaningful, and photorealistic environments for applications such as immersive media, robotics, autonomous driving, and embodied AI. Early methods based on procedural rules offered scalability but limited diversity. Recent advances in deep generative models (e.g., GANs, diffusion models) and 3D representations (e.g., NeRF, 3D Gaussians) have enabled the learning of real-world scene distributions, improving fidelity, diversity, and view consistency. Recent advances like diffusion models bridge 3D scene synthesis and photorealism by reframing generation as image or video synthesis problems. This survey provides a systematic overview of state-of-the-art approaches, organizing them into four paradigms: procedural generation, neural 3D-based generation, image-based generation, and video-based generation. We analyze their technical foundations, trade-offs, and representative results, and review commonly used datasets, evaluation protocols, and downstream applications. We conclude by discussing key challenges in generation capacity, 3D representation, data and annotations, and evaluation, and outline promising directions including higher fidelity, physics-aware and interactive generation, and unified perception-generation models. This review organizes recent advances in 3D scene generation and highlights promising directions at the intersection of generative AI, 3D vision, and embodied intelligence. To track ongoing developments, we maintain an up-to-date project page: https://github.com/hzxie/Awesome-3D-Scene-Generation. |
2025-05-09T00:00:00 | 2505.03981 | X-Reasoner: Towards Generalizable Reasoning Across Modalities and Domains | [
"Qianchu Liu",
"Sheng Zhang",
"Guanghui Qin",
"Timothy Ossowski",
"Yu Gu",
"Ying Jin",
"Sid Kiblawi",
"Sam Preston",
"Mu Wei",
"Paul Vozila",
"Tristan Naumann",
"Hoifung Poon"
] | Recent proprietary models (e.g., o3) have begun to demonstrate strong multimodal reasoning capabilities. Yet, most existing open-source research concentrates on training text-only reasoning models, with evaluations limited to mainly mathematical and general-domain tasks. Therefore, it remains unclear how to effectively extend reasoning capabilities beyond text input and general domains. This paper explores a fundamental research question: Is reasoning generalizable across modalities and domains? Our findings support an affirmative answer: General-domain text-based post-training can enable such strong generalizable reasoning. Leveraging this finding, we introduce X-Reasoner, a vision-language model post-trained solely on general-domain text for generalizable reasoning, using a two-stage approach: an initial supervised fine-tuning phase with distilled long chain-of-thoughts, followed by reinforcement learning with verifiable rewards. Experiments show that X-Reasoner successfully transfers reasoning capabilities to both multimodal and out-of-domain settings, outperforming existing state-of-the-art models trained with in-domain and multimodal data across various general and medical benchmarks (Figure 1). Additionally, we find that X-Reasoner's performance in specialized domains can be further enhanced through continued training on domain-specific text-only data. Building upon this, we introduce X-Reasoner-Med, a medical-specialized variant that achieves new state of the art on numerous text-only and multimodal medical benchmarks. |
|
2025-05-09T00:00:00 | 2504.19314 | BrowseComp-ZH: Benchmarking Web Browsing Ability of Large Language Models in Chinese | [
"Peilin Zhou",
"Bruce Leon",
"Xiang Ying",
"Can Zhang",
"Yifan Shao",
"Qichen Ye",
"Dading Chong",
"Zhiling Jin",
"Chenxuan Xie",
"Meng Cao",
"Yuxin Gu",
"Sixin Hong",
"Jing Ren",
"Jian Chen",
"Chao Liu",
"Yining Hua"
] | https://github.com/PALIN2018/BrowseComp-ZH | As large language models (LLMs) evolve into tool-using agents, the ability to browse the web in real-time has become a critical yardstick for measuring their reasoning and retrieval competence. Existing benchmarks such as BrowseComp concentrate on English and overlook the linguistic, infrastructural, and censorship-related complexities of other major information ecosystems -- most notably Chinese. To address this gap, we introduce BrowseComp-ZH, a high-difficulty benchmark purpose-built to comprehensively evaluate LLM agents on the Chinese web. BrowseComp-ZH consists of 289 multi-hop questions spanning 11 diverse domains. Each question is reverse-engineered from a short, objective, and easily verifiable answer (e.g., a date, number, or proper noun). A two-stage quality control protocol is applied to strive for high question difficulty and answer uniqueness. We benchmark over 20 state-of-the-art language models and agentic search systems on our proposed BrowseComp-ZH. Despite their strong conversational and retrieval capabilities, most models struggle severely: a large number achieve accuracy rates below 10%, and only a handful exceed 20%. Even the best-performing system, OpenAI's DeepResearch, reaches just 42.9%. These results demonstrate the considerable difficulty of BrowseComp-ZH, where success demands not only effective retrieval strategies, but also sophisticated reasoning and information reconciliation -- capabilities that current models still struggle to master. Our dataset, construction guidelines, and benchmark results have been publicly released at https://github.com/PALIN2018/BrowseComp-ZH. |
2025-05-09T00:00:00 | 2505.05467 | StreamBridge: Turning Your Offline Video Large Language Model into a Proactive Streaming Assistant | [
"Haibo Wang",
"Bo Feng",
"Zhengfeng Lai",
"Mingze Xu",
"Shiyu Li",
"Weifeng Ge",
"Afshin Dehghan",
"Meng Cao",
"Ping Huang"
] | We present StreamBridge, a simple yet effective framework that seamlessly transforms offline Video-LLMs into streaming-capable models. It addresses two fundamental challenges in adapting existing models into online scenarios: (1) limited capability for multi-turn real-time understanding, and (2) lack of proactive response mechanisms. Specifically, StreamBridge incorporates (1) a memory buffer combined with a round-decayed compression strategy, supporting long-context multi-turn interactions, and (2) a decoupled, lightweight activation model that can be effortlessly integrated into existing Video-LLMs, enabling continuous proactive responses. To further support StreamBridge, we construct Stream-IT, a large-scale dataset tailored for streaming video understanding, featuring interleaved video-text sequences and diverse instruction formats. Extensive experiments show that StreamBridge significantly improves the streaming understanding capabilities of offline Video-LLMs across various tasks, outperforming even proprietary models such as GPT-4o and Gemini 1.5 Pro. Simultaneously, it achieves competitive or superior performance on standard video understanding benchmarks. |
|
2025-05-09T00:00:00 | 2505.05288 | PlaceIt3D: Language-Guided Object Placement in Real 3D Scenes | [
"Ahmed Abdelreheem",
"Filippo Aleotti",
"Jamie Watson",
"Zawar Qureshi",
"Abdelrahman Eldesokey",
"Peter Wonka",
"Gabriel Brostow",
"Sara Vicente",
"Guillermo Garcia-Hernando"
] | We introduce the novel task of Language-Guided Object Placement in Real 3D Scenes. Our model is given a 3D scene's point cloud, a 3D asset, and a textual prompt broadly describing where the 3D asset should be placed. The task here is to find a valid placement for the 3D asset that respects the prompt. Compared with other language-guided localization tasks in 3D scenes such as grounding, this task has specific challenges: it is ambiguous because it has multiple valid solutions, and it requires reasoning about 3D geometric relationships and free space. We inaugurate this task by proposing a new benchmark and evaluation protocol. We also introduce a new dataset for training 3D LLMs on this task, as well as the first method to serve as a non-trivial baseline. We believe that this challenging task and our new benchmark could become part of the suite of benchmarks used to evaluate and compare generalist 3D LLM models. |
|
2025-05-09T00:00:00 | 2505.05470 | Flow-GRPO: Training Flow Matching Models via Online RL | [
"Jie Liu",
"Gongye Liu",
"Jiajun Liang",
"Yangguang Li",
"Jiaheng Liu",
"Xintao Wang",
"Pengfei Wan",
"Di Zhang",
"Wanli Ouyang"
] | 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. |
|
2025-05-09T00:00:00 | 2505.05071 | FG-CLIP: Fine-Grained Visual and Textual Alignment | [
"Chunyu Xie",
"Bin Wang",
"Fanjing Kong",
"Jincheng Li",
"Dawei Liang",
"Gengshen Zhang",
"Dawei Leng",
"Yuhui Yin"
] | https://github.com/360CVGroup/FG-CLIP | Contrastive Language-Image Pre-training (CLIP) excels in multimodal tasks such as image-text retrieval and zero-shot classification but struggles with fine-grained understanding due to its focus on coarse-grained short captions. To address this, we propose Fine-Grained CLIP (FG-CLIP), which enhances fine-grained understanding through three key innovations. First, we leverage large multimodal models to generate 1.6 billion long caption-image pairs for capturing global-level semantic details. Second, a high-quality dataset is constructed with 12 million images and 40 million region-specific bounding boxes aligned with detailed captions to ensure precise, context-rich representations. Third, 10 million hard fine-grained negative samples are incorporated to improve the model's ability to distinguish subtle semantic differences. Corresponding training methods are meticulously designed for these data. Extensive experiments demonstrate that FG-CLIP outperforms the original CLIP and other state-of-the-art methods across various downstream tasks, including fine-grained understanding, open-vocabulary object detection, image-text retrieval, and general multimodal benchmarks. These results highlight FG-CLIP's effectiveness in capturing fine-grained image details and improving overall model performance. The related data, code, and models are available at https://github.com/360CVGroup/FG-CLIP. |
2025-05-09T00:00:00 | 2505.05408 | Crosslingual Reasoning through Test-Time Scaling | [
"Zheng-Xin Yong",
"M. Farid Adilazuarda",
"Jonibek Mansurov",
"Ruochen Zhang",
"Niklas Muennighoff",
"Carsten Eickhoff",
"Genta Indra Winata",
"Julia Kreutzer",
"Stephen H. Bach",
"Alham Fikri Aji"
] | Reasoning capabilities of large language models are primarily studied for English, even when pretrained models are multilingual. In this work, we investigate to what extent English reasoning finetuning with long chain-of-thoughts (CoTs) can generalize across languages. First, we find that scaling up inference compute for English-centric reasoning language models (RLMs) improves multilingual mathematical reasoning across many languages including low-resource languages, to an extent where they outperform models twice their size. Second, we reveal that while English-centric RLM's CoTs are naturally predominantly English, they consistently follow a quote-and-think pattern to reason about quoted non-English inputs. Third, we discover an effective strategy to control the language of long CoT reasoning, and we observe that models reason better and more efficiently in high-resource languages. Finally, we observe poor out-of-domain reasoning generalization, in particular from STEM to cultural commonsense knowledge, even for English. Overall, we demonstrate the potentials, study the mechanisms and outline the limitations of crosslingual generalization of English reasoning test-time scaling. We conclude that practitioners should let English-centric RLMs reason in high-resource languages, while further work is needed to improve reasoning in low-resource languages and out-of-domain contexts. |
|
2025-05-09T00:00:00 | 2505.05327 | ICon: In-Context Contribution for Automatic Data Selection | [
"Yixin Yang",
"Qingxiu Dong",
"Linli Yao",
"Fangwei Zhu",
"Zhifang Sui"
] | Data selection for instruction tuning is essential for improving the performance of Large Language Models (LLMs) and reducing training cost. However, existing automated selection methods either depend on computationally expensive gradient-based measures or manually designed heuristics, which may fail to fully exploit the intrinsic attributes of data. In this paper, we propose In-context Learning for Contribution Measurement (ICon), a novel gradient-free method that takes advantage of the implicit fine-tuning nature of in-context learning (ICL) to measure sample contribution without gradient computation or manual indicators engineering. ICon offers a computationally efficient alternative to gradient-based methods and reduces human inductive bias inherent in heuristic-based approaches. ICon comprises three components and identifies high-contribution data by assessing performance shifts under implicit learning through ICL. Extensive experiments on three LLMs across 12 benchmarks and 5 pairwise evaluation sets demonstrate the effectiveness of ICon. Remarkably, on LLaMA3.1-8B, models trained on 15% of ICon-selected data outperform full datasets by 5.42% points and exceed the best performance of widely used selection methods by 2.06% points. We further analyze high-contribution samples selected by ICon, which show both diverse tasks and appropriate difficulty levels, rather than just the hardest ones. |
|
2025-05-09T00:00:00 | 2505.04921 | Perception, Reason, Think, and Plan: A Survey on Large Multimodal Reasoning Models | [
"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"
] | 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. |
|
2025-05-09T00:00:00 | 2505.05469 | Generating Physically Stable and Buildable LEGO Designs from Text | [
"Ava Pun",
"Kangle Deng",
"Ruixuan Liu",
"Deva Ramanan",
"Changliu Liu",
"Jun-Yan Zhu"
] | We introduce LegoGPT, the first approach for generating physically stable LEGO brick models from text prompts. To achieve this, we construct a large-scale, physically stable dataset of LEGO designs, along with their associated captions, and train an autoregressive large language model to predict the next brick to add via next-token prediction. To improve the stability of the resulting designs, we employ an efficient validity check and physics-aware rollback during autoregressive inference, which prunes infeasible token predictions using physics laws and assembly constraints. Our experiments show that LegoGPT produces stable, diverse, and aesthetically pleasing LEGO designs that align closely with the input text prompts. We also develop a text-based LEGO texturing method to generate colored and textured designs. We show that our designs can be assembled manually by humans and automatically by robotic arms. We also release our new dataset, StableText2Lego, containing over 47,000 LEGO structures of over 28,000 unique 3D objects accompanied by detailed captions, along with our code and models at the project website: https://avalovelace1.github.io/LegoGPT/. |
|
2025-05-09T00:00:00 | 2505.04955 | Chain-of-Thought Tokens are Computer Program Variables | [
"Fangwei Zhu",
"Peiyi Wang",
"Zhifang Sui"
] | https://github.com/solitaryzero/CoTs_are_Variables | Chain-of-thoughts (CoT) requires large language models (LLMs) to generate intermediate steps before reaching the final answer, and has been proven effective to help LLMs solve complex reasoning tasks. However, the inner mechanism of CoT still remains largely unclear. In this paper, we empirically study the role of CoT tokens in LLMs on two compositional tasks: multi-digit multiplication and dynamic programming. While CoT is essential for solving these problems, we find that preserving only tokens that store intermediate results would achieve comparable performance. Furthermore, we observe that storing intermediate results in an alternative latent form will not affect model performance. We also randomly intervene some values in CoT, and notice that subsequent CoT tokens and the final answer would change correspondingly. These findings suggest that CoT tokens may function like variables in computer programs but with potential drawbacks like unintended shortcuts and computational complexity limits between tokens. The code and data are available at https://github.com/solitaryzero/CoTs_are_Variables. |
2025-05-09T00:00:00 | 2505.03422 | LiftFeat: 3D Geometry-Aware Local Feature Matching | [
"Yepeng Liu",
"Wenpeng Lai",
"Zhou Zhao",
"Yuxuan Xiong",
"Jinchi Zhu",
"Jun Cheng",
"Yongchao Xu"
] | https://github.com/lyp-deeplearning/LiftFeat | Robust and efficient local feature matching plays a crucial role in applications such as SLAM and visual localization for robotics. Despite great progress, it is still very challenging to extract robust and discriminative visual features in scenarios with drastic lighting changes, low texture areas, or repetitive patterns. In this paper, we propose a new lightweight network called LiftFeat, which lifts the robustness of raw descriptor by aggregating 3D geometric feature. Specifically, we first adopt a pre-trained monocular depth estimation model to generate pseudo surface normal label, supervising the extraction of 3D geometric feature in terms of predicted surface normal. We then design a 3D geometry-aware feature lifting module to fuse surface normal feature with raw 2D descriptor feature. Integrating such 3D geometric feature enhances the discriminative ability of 2D feature description in extreme conditions. Extensive experimental results on relative pose estimation, homography estimation, and visual localization tasks, demonstrate that our LiftFeat outperforms some lightweight state-of-the-art methods. Code will be released at : https://github.com/lyp-deeplearning/LiftFeat. |
2025-05-09T00:00:00 | 2505.04769 | Vision-Language-Action Models: Concepts, Progress, Applications and Challenges | [
"Ranjan Sapkota",
"Yang Cao",
"Konstantinos I. Roumeliotis",
"Manoj Karkee"
] | Vision-Language-Action (VLA) models mark a transformative advancement in artificial intelligence, aiming to unify perception, natural language understanding, and embodied action within a single computational framework. This foundational review presents a comprehensive synthesis of recent advancements in Vision-Language-Action models, systematically organized across five thematic pillars that structure the landscape of this rapidly evolving field. We begin by establishing the conceptual foundations of VLA systems, tracing their evolution from cross-modal learning architectures to generalist agents that tightly integrate vision-language models (VLMs), action planners, and hierarchical controllers. Our methodology adopts a rigorous literature review framework, covering over 80 VLA models published in the past three years. Key progress areas include architectural innovations, parameter-efficient training strategies, and real-time inference accelerations. We explore diverse application domains such as humanoid robotics, autonomous vehicles, medical and industrial robotics, precision agriculture, and augmented reality navigation. The review further addresses major challenges across real-time control, multimodal action representation, system scalability, generalization to unseen tasks, and ethical deployment risks. Drawing from the state-of-the-art, we propose targeted solutions including agentic AI adaptation, cross-embodiment generalization, and unified neuro-symbolic planning. In our forward-looking discussion, we outline a future roadmap where VLA models, VLMs, and agentic AI converge to power socially aligned, adaptive, and general-purpose embodied agents. This work serves as a foundational reference for advancing intelligent, real-world robotics and artificial general intelligence. >Vision-language-action, Agentic AI, AI Agents, Vision-language Models |
|
2025-05-09T00:00:00 | 2505.05064 | WaterDrum: Watermarking for Data-centric Unlearning Metric | [
"Xinyang Lu",
"Xinyuan Niu",
"Gregory Kang Ruey Lau",
"Bui Thi Cam Nhung",
"Rachael Hwee Ling Sim",
"Fanyu Wen",
"Chuan-Sheng Foo",
"See-Kiong Ng",
"Bryan Kian Hsiang Low"
] | https://github.com/lululu008/WaterDrum | Large language model (LLM) unlearning is critical in real-world applications where it is necessary to efficiently remove the influence of private, copyrighted, or harmful data from some users. However, existing utility-centric unlearning metrics (based on model utility) may fail to accurately evaluate the extent of unlearning in realistic settings such as when (a) the forget and retain set have semantically similar content, (b) retraining the model from scratch on the retain set is impractical, and/or (c) the model owner can improve the unlearning metric without directly performing unlearning on the LLM. This paper presents the first data-centric unlearning metric for LLMs called WaterDrum that exploits robust text watermarking for overcoming these limitations. We also introduce new benchmark datasets for LLM unlearning that contain varying levels of similar data points and can be used to rigorously evaluate unlearning algorithms using WaterDrum. Our code is available at https://github.com/lululu008/WaterDrum and our new benchmark datasets are released at https://huggingface.co/datasets/Glow-AI/WaterDrum-Ax. |
2025-05-09T00:00:00 | 2505.02363 | SIMPLEMIX: Frustratingly Simple Mixing of Off- and On-policy Data in Language Model Preference Learning | [
"Tianjian Li",
"Daniel Khashabi"
] | Aligning language models with human preferences relies on pairwise preference datasets. While some studies suggest that on-policy data consistently outperforms off -policy data for preference learning, others indicate that the advantages of on-policy data may be task-dependent, highlighting the need for a systematic exploration of their interplay. In this work, we show that on-policy and off-policy data offer complementary strengths in preference optimization: on-policy data is particularly effective for reasoning tasks like math and coding, while off-policy data performs better on open-ended tasks such as creative writing and making personal recommendations. Guided by these findings, we introduce SIMPLEMIX, an approach to combine the complementary strengths of on-policy and off-policy preference learning by simply mixing these two data sources. Our empirical results across diverse tasks and benchmarks demonstrate that SIMPLEMIX substantially improves language model alignment. Specifically, SIMPLEMIX improves upon on-policy DPO and off-policy DPO by an average of 6.03% on Alpaca Eval 2.0. Moreover, it outperforms prior approaches that are much more complex in combining on- and off-policy data, such as HyPO and DPO-Mix-P, by an average of 3.05%. |
|
2025-05-09T00:00:00 | 2505.04842 | Putting the Value Back in RL: Better Test-Time Scaling by Unifying LLM Reasoners With Verifiers | [
"Kusha Sareen",
"Morgane M Moss",
"Alessandro Sordoni",
"Rishabh Agarwal",
"Arian Hosseini"
] | Prevalent reinforcement learning~(RL) methods for fine-tuning LLM reasoners, such as GRPO or Leave-one-out PPO, abandon the learned value function in favor of empirically estimated returns. This hinders test-time compute scaling that relies on using the value-function for verification. In this work, we propose RL^V that augments any ``value-free'' RL method by jointly training the LLM as both a reasoner and a generative verifier using RL-generated data, adding verification capabilities without significant overhead. Empirically, RL^V boosts MATH accuracy by over 20\% with parallel sampling and enables 8-32times efficient test-time compute scaling compared to the base RL method. RL^V also exhibits strong generalization capabilities for both easy-to-hard and out-of-domain tasks. Furthermore, RL^V achieves 1.2-1.6times higher performance when jointly scaling parallel and sequential test-time compute with a long reasoning R1 model. |
|
2025-05-12T00:00:00 | 2505.06111 | UniVLA: Learning to Act Anywhere with Task-centric Latent Actions | [
"Qingwen Bu",
"Yanting Yang",
"Jisong Cai",
"Shenyuan Gao",
"Guanghui Ren",
"Maoqing Yao",
"Ping Luo",
"Hongyang Li"
] | A generalist robot should perform effectively across various environments. However, most existing approaches heavily rely on scaling action-annotated data to enhance their capabilities. Consequently, they are often limited to single physical specification and struggle to learn transferable knowledge across different embodiments and environments. To confront these limitations, we propose UniVLA, a new framework for learning cross-embodiment vision-language-action (VLA) policies. Our key innovation is to derive task-centric action representations from videos with a latent action model. This enables us to exploit extensive data across a wide spectrum of embodiments and perspectives. To mitigate the effect of task-irrelevant dynamics, we incorporate language instructions and establish a latent action model within the DINO feature space. Learned from internet-scale videos, the generalist policy can be deployed to various robots through efficient latent action decoding. We obtain state-of-the-art results across multiple manipulation and navigation benchmarks, as well as real-robot deployments. UniVLA achieves superior performance over OpenVLA with less than 1/20 of pretraining compute and 1/10 of downstream data. Continuous performance improvements are observed as heterogeneous data, even including human videos, are incorporated into the training pipeline. The results underscore UniVLA's potential to facilitate scalable and efficient robot policy learning. |
|
2025-05-12T00:00:00 | 2505.05026 | G-FOCUS: Towards a Robust Method for Assessing UI Design Persuasiveness | [
"Jaehyun Jeon",
"Jang Han Yoon",
"Min Soo Kim",
"Sumin Shim",
"Yejin Choi",
"Hanbin Kim",
"Youngjae Yu"
] | Evaluating user interface (UI) design effectiveness extends beyond aesthetics to influencing user behavior, a principle central to Design Persuasiveness. A/B testing is the predominant method for determining which UI variations drive higher user engagement, but it is costly and time-consuming. While recent Vision-Language Models (VLMs) can process automated UI analysis, current approaches focus on isolated design attributes rather than comparative persuasiveness-the key factor in optimizing user interactions. To address this, we introduce WiserUI-Bench, a benchmark designed for Pairwise UI Design Persuasiveness Assessment task, featuring 300 real-world UI image pairs labeled with A/B test results and expert rationales. Additionally, we propose G-FOCUS, a novel inference-time reasoning strategy that enhances VLM-based persuasiveness assessment by reducing position bias and improving evaluation accuracy. Experimental results show that G-FOCUS surpasses existing inference strategies in consistency and accuracy for pairwise UI evaluation. Through promoting VLM-driven evaluation of UI persuasiveness, our work offers an approach to complement A/B testing, propelling progress in scalable UI preference modeling and design optimization. Code and data will be released publicly. |
|
2025-05-12T00:00:00 | 2505.02686 | Sailing AI by the Stars: A Survey of Learning from Rewards in Post-Training and Test-Time Scaling of Large Language Models | [
"Xiaobao Wu"
] | https://github.com/bobxwu/learning-from-rewards-llm-papers | Recent developments in Large Language Models (LLMs) have shifted from pre-training scaling to post-training and test-time scaling. Across these developments, a key unified paradigm has arisen: Learning from Rewards, where reward signals act as the guiding stars to steer LLM behavior. It has underpinned a wide range of prevalent techniques, such as reinforcement learning (in RLHF, DPO, and GRPO), reward-guided decoding, and post-hoc correction. Crucially, this paradigm enables the transition from passive learning from static data to active learning from dynamic feedback. This endows LLMs with aligned preferences and deep reasoning capabilities. In this survey, we present a comprehensive overview of the paradigm of learning from rewards. We categorize and analyze the strategies under this paradigm across training, inference, and post-inference stages. We further discuss the benchmarks for reward models and the primary applications. Finally we highlight the challenges and future directions. We maintain a paper collection at https://github.com/bobxwu/learning-from-rewards-llm-papers. |
2025-05-12T00:00:00 | 2505.02550 | Bielik v3 Small: Technical Report | [
"Krzysztof Ociepa",
"Łukasz Flis",
"Remigiusz Kinas",
"Krzysztof Wróbel",
"Adrian Gwoździej"
] | 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. |
|
2025-05-12T00:00:00 | 2505.02410 | Bielik 11B v2 Technical Report | [
"Krzysztof Ociepa",
"Łukasz Flis",
"Krzysztof Wróbel",
"Adrian Gwoździej",
"Remigiusz Kinas"
] | 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. |
|
2025-05-12T00:00:00 | 2505.06046 | Healthy LLMs? Benchmarking LLM Knowledge of UK Government Public Health Information | [
"Joshua Harris",
"Fan Grayson",
"Felix Feldman",
"Timothy Laurence",
"Toby Nonnenmacher",
"Oliver Higgins",
"Leo Loman",
"Selina Patel",
"Thomas Finnie",
"Samuel Collins",
"Michael Borowitz"
] | As Large Language Models (LLMs) become widely accessible, a detailed understanding of their knowledge within specific domains becomes necessary for successful real world use. This is particularly critical in public health, where failure to retrieve relevant, accurate, and current information could significantly impact UK residents. However, currently little is known about LLM knowledge of UK Government public health information. To address this issue, this paper introduces a new benchmark, PubHealthBench, with over 8000 questions for evaluating LLMs' Multiple Choice Question Answering (MCQA) and free form responses to public health queries, created via an automated pipeline. We also release a new dataset of the extracted UK Government public health guidance documents used as source text for PubHealthBench. Assessing 24 LLMs on PubHealthBench we find the latest private LLMs (GPT-4.5, GPT-4.1 and o1) have a high degree of knowledge, achieving >90% in the MCQA setup, and outperform humans with cursory search engine use. However, in the free form setup we see lower performance with no model scoring >75%. Therefore, whilst there are promising signs that state of the art (SOTA) LLMs are an increasingly accurate source of public health information, additional safeguards or tools may still be needed when providing free form responses on public health topics. |
|
2025-05-12T00:00:00 | 2505.05621 | A Preliminary Study for GPT-4o on Image Restoration | [
"Hao Yang",
"Yan Yang",
"Ruikun Zhang",
"Liyuan Pan"
] | OpenAI's GPT-4o model, integrating multi-modal inputs and outputs within an autoregressive architecture, has demonstrated unprecedented performance in image generation. In this work, we investigate its potential impact on the image restoration community. We present the first systematic evaluation of GPT-4o across diverse restoration tasks. Our experiments reveal that, although restoration outputs from GPT-4o are visually appealing, they often suffer from pixel-level structural fidelity when compared to ground-truth images. Common issues are variations in image proportions, shifts in object positions and quantities, and changes in viewpoint.To address it, taking image dehazing, derainning, and low-light enhancement as representative case studies, we show that GPT-4o's outputs can serve as powerful visual priors, substantially enhancing the performance of existing dehazing networks. It offers practical guidelines and a baseline framework to facilitate the integration of GPT-4o into future image restoration pipelines. We hope the study on GPT-4o image restoration will accelerate innovation in the broader field of image generation areas. To support further research, we will release GPT-4o-restored images from over 10 widely used image restoration datasets. |
|
2025-05-12T00:00:00 | 2504.21467 | Multiview Point Cloud Registration via Optimization in an Autoencoder Latent Space | [
"Luc Vedrenne",
"Sylvain Faisan",
"Denis Fortun"
] | Point cloud rigid registration is a fundamental problem in 3D computer vision. In the multiview case, we aim to find a set of 6D poses to align a set of objects. Methods based on pairwise registration rely on a subsequent synchronization algorithm, which makes them poorly scalable with the number of views. Generative approaches overcome this limitation, but are based on Gaussian Mixture Models and use an Expectation-Maximization algorithm. Hence, they are not well suited to handle large transformations. Moreover, most existing methods cannot handle high levels of degradations. In this paper, we introduce POLAR (POint cloud LAtent Registration), a multiview registration method able to efficiently deal with a large number of views, while being robust to a high level of degradations and large initial angles. To achieve this, we transpose the registration problem into the latent space of a pretrained autoencoder, design a loss taking degradations into account, and develop an efficient multistart optimization strategy. Our proposed method significantly outperforms state-of-the-art approaches on synthetic and real data. POLAR is available at github.com/pypolar/polar or as a standalone package which can be installed with pip install polaregistration. |
|
2025-05-13T00:00:00 | 2505.06548 | REFINE-AF: A Task-Agnostic Framework to Align Language Models via Self-Generated Instructions using Reinforcement Learning from Automated Feedback | [
"Aniruddha Roy",
"Pretam Ray",
"Abhilash Nandy",
"Somak Aditya",
"Pawan Goyal"
] | Instruction-based Large Language Models (LLMs) have proven effective in numerous few-shot or zero-shot Natural Language Processing (NLP) tasks. However, creating human-annotated instruction data is time-consuming, expensive, and often limited in quantity and task diversity. Previous research endeavors have attempted to address this challenge by proposing frameworks capable of generating instructions in a semi-automated and task-agnostic manner directly from the model itself. Many of these efforts have relied on large API-only parameter-based models such as GPT-3.5 (175B), which are expensive, and subject to limits on a number of queries. This paper explores the performance of three open-source small LLMs such as LLaMA 2-7B, LLama 2-13B, and Mistral 7B, using a semi-automated framework, thereby reducing human intervention, effort, and cost required to generate an instruction dataset for fine-tuning LLMs. Furthermore, we demonstrate that incorporating a Reinforcement Learning (RL) based training algorithm into this LLMs-based framework leads to further enhancements. Our evaluation of the dataset reveals that these RL-based frameworks achieve a substantial improvements in 63-66% of the tasks compared to previous approaches. |
|
2025-05-13T00:00:00 | 2505.03733 | WebGen-Bench: Evaluating LLMs on Generating Interactive and Functional Websites from Scratch | [
"Zimu Lu",
"Yunqiao Yang",
"Houxing Ren",
"Haotian Hou",
"Han Xiao",
"Ke Wang",
"Weikang Shi",
"Aojun Zhou",
"Mingjie Zhan",
"Hongsheng Li"
] | LLM-based agents have demonstrated great potential in generating and managing code within complex codebases. In this paper, we introduce WebGen-Bench, a novel benchmark designed to measure an LLM-based agent's ability to create multi-file website codebases from scratch. It contains diverse instructions for website generation, created through the combined efforts of human annotators and GPT-4o. These instructions span three major categories and thirteen minor categories, encompassing nearly all important types of web applications. To assess the quality of the generated websites, we use GPT-4o to generate test cases targeting each functionality described in the instructions, and then manually filter, adjust, and organize them to ensure accuracy, resulting in 647 test cases. Each test case specifies an operation to be performed on the website and the expected result after the operation. To automate testing and improve reproducibility, we employ a powerful web-navigation agent to execute tests on the generated websites and determine whether the observed responses align with the expected results. We evaluate three high-performance code-agent frameworks, Bolt.diy, OpenHands, and Aider, using multiple proprietary and open-source LLMs as engines. The best-performing combination, Bolt.diy powered by DeepSeek-R1, achieves only 27.8\% accuracy on the test cases, highlighting the challenging nature of our benchmark. Additionally, we construct WebGen-Instruct, a training set consisting of 6,667 website-generation instructions. Training Qwen2.5-Coder-32B-Instruct on Bolt.diy trajectories generated from a subset of this training set achieves an accuracy of 38.2\%, surpassing the performance of the best proprietary model. |
|
2025-05-13T00:00:00 | 2505.07062 | Seed1.5-VL Technical Report | [
"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"
] | 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) |
|
2025-05-13T00:00:00 | 2505.07787 | Learning from Peers in Reasoning Models | [
"Tongxu Luo",
"Wenyu Du",
"Jiaxi Bi",
"Stephen Chung",
"Zhengyang Tang",
"Hao Yang",
"Min Zhang",
"Benyou Wang"
] | 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/ . |
|
2025-05-13T00:00:00 | 2505.07796 | Learning Dynamics in Continual Pre-Training for Large Language Models | [
"Xingjin Wang",
"Howe Tissue",
"Lu Wang",
"Linjing Li",
"Daniel Dajun Zeng"
] | Continual Pre-Training (CPT) has become a popular and effective method to apply strong foundation models to specific downstream tasks. In this work, we explore the learning dynamics throughout the CPT process for large language models. We specifically focus on how general and downstream domain performance evolves at each training step, with domain performance measured via validation losses. We have observed that the CPT loss curve fundamentally characterizes the transition from one curve to another hidden curve, and could be described by decoupling the effects of distribution shift and learning rate annealing. We derive a CPT scaling law that combines the two factors, enabling the prediction of loss at any (continual) training steps and across learning rate schedules (LRS) in CPT. Our formulation presents a comprehensive understanding of several critical factors in CPT, including loss potential, peak learning rate, training steps, replay ratio, etc. Moreover, our approach can be adapted to customize training hyper-parameters to different CPT goals such as balancing general and domain-specific performance. Extensive experiments demonstrate that our scaling law holds across various CPT datasets and training hyper-parameters. |
|
2025-05-13T00:00:00 | 2505.07608 | MiMo: Unlocking the Reasoning Potential of Language Model -- From Pretraining to Posttraining | [
"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"
] | https://github.com/xiaomimimo/MiMo | 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. |
2025-05-13T00:00:00 | 2505.07818 | DanceGRPO: Unleashing GRPO on Visual Generation | [
"Zeyue Xue",
"Jie Wu",
"Yu Gao",
"Fangyuan Kong",
"Lingting Zhu",
"Mengzhao Chen",
"Zhiheng Liu",
"Wei Liu",
"Qiushan Guo",
"Weilin Huang",
"Ping Luo"
] | Recent breakthroughs in generative models-particularly diffusion models and rectified flows-have revolutionized visual content creation, yet aligning model outputs with human preferences remains a critical challenge. Existing reinforcement learning (RL)-based methods for visual generation face critical limitations: incompatibility with modern Ordinary Differential Equations (ODEs)-based sampling paradigms, instability in large-scale training, and lack of validation for video generation. This paper introduces DanceGRPO, the first unified framework to adapt Group Relative Policy Optimization (GRPO) to visual generation paradigms, unleashing one unified RL algorithm across two generative paradigms (diffusion models and rectified flows), three tasks (text-to-image, text-to-video, image-to-video), four foundation models (Stable Diffusion, HunyuanVideo, FLUX, SkyReel-I2V), and five reward models (image/video aesthetics, text-image alignment, video motion quality, and binary reward). To our knowledge, DanceGRPO is the first RL-based unified framework capable of seamless adaptation across diverse generative paradigms, tasks, foundational models, and reward models. DanceGRPO demonstrates consistent and substantial improvements, which outperform baselines by up to 181% on benchmarks such as HPS-v2.1, CLIP Score, VideoAlign, and GenEval. Notably, DanceGRPO not only can stabilize policy optimization for complex video generation, but also enables generative policy to better capture denoising trajectories for Best-of-N inference scaling and learn from sparse binary feedback. Our results establish DanceGRPO as a robust and versatile solution for scaling Reinforcement Learning from Human Feedback (RLHF) tasks in visual generation, offering new insights into harmonizing reinforcement learning and visual synthesis. The code will be released. |
|
2025-05-13T00:00:00 | 2505.07596 | Reinforced Internal-External Knowledge Synergistic Reasoning for Efficient Adaptive Search Agent | [
"Ziyang Huang",
"Xiaowei Yuan",
"Yiming Ju",
"Jun Zhao",
"Kang Liu"
] | Retrieval-augmented generation (RAG) is a common strategy to reduce hallucinations in Large Language Models (LLMs). While reinforcement learning (RL) can enable LLMs to act as search agents by activating retrieval capabilities, existing ones often underutilize their internal knowledge. This can lead to redundant retrievals, potential harmful knowledge conflicts, and increased inference latency. To address these limitations, an efficient and adaptive search agent capable of discerning optimal retrieval timing and synergistically integrating parametric (internal) and retrieved (external) knowledge is in urgent need. This paper introduces the Reinforced Internal-External Knowledge Synergistic Reasoning Agent (IKEA), which could indentify its own knowledge boundary and prioritize the utilization of internal knowledge, resorting to external search only when internal knowledge is deemed insufficient. This is achieved using a novel knowledge-boundary aware reward function and a knowledge-boundary aware training dataset. These are designed for internal-external knowledge synergy oriented RL, incentivizing the model to deliver accurate answers, minimize unnecessary retrievals, and encourage appropriate external searches when its own knowledge is lacking. Evaluations across multiple knowledge reasoning tasks demonstrate that IKEA significantly outperforms baseline methods, reduces retrieval frequency significantly, and exhibits robust generalization capabilities. |
|
2025-05-13T00:00:00 | 2505.07747 | Step1X-3D: Towards High-Fidelity and Controllable Generation of Textured 3D Assets | [
"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"
] | 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. |
|
2025-05-13T00:00:00 | 2505.06324 | Document Attribution: Examining Citation Relationships using Large Language Models | [
"Vipula Rawte",
"Ryan A. Rossi",
"Franck Dernoncourt",
"Nedim Lipka"
] | As Large Language Models (LLMs) are increasingly applied to document-based tasks - such as document summarization, question answering, and information extraction - where user requirements focus on retrieving information from provided documents rather than relying on the model's parametric knowledge, ensuring the trustworthiness and interpretability of these systems has become a critical concern. A central approach to addressing this challenge is attribution, which involves tracing the generated outputs back to their source documents. However, since LLMs can produce inaccurate or imprecise responses, it is crucial to assess the reliability of these citations. To tackle this, our work proposes two techniques. (1) A zero-shot approach that frames attribution as a straightforward textual entailment task. Our method using flan-ul2 demonstrates an improvement of 0.27% and 2.4% over the best baseline of ID and OOD sets of AttributionBench, respectively. (2) We also explore the role of the attention mechanism in enhancing the attribution process. Using a smaller LLM, flan-t5-small, the F1 scores outperform the baseline across almost all layers except layer 4 and layers 8 through 11. |
|
2025-05-13T00:00:00 | 2505.07447 | Unified Continuous Generative Models | [
"Peng Sun",
"Yi Jiang",
"Tao Lin"
] | https://github.com/LINs-lab/UCGM | 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. |
2025-05-13T00:00:00 | 2505.07086 | Multi-Objective-Guided Discrete Flow Matching for Controllable Biological Sequence Design | [
"Tong Chen",
"Yinuo Zhang",
"Sophia Tang",
"Pranam Chatterjee"
] | Designing biological sequences that satisfy multiple, often conflicting, functional and biophysical criteria remains a central challenge in biomolecule engineering. While discrete flow matching models have recently shown promise for efficient sampling in high-dimensional sequence spaces, existing approaches address only single objectives or require continuous embeddings that can distort discrete distributions. We present Multi-Objective-Guided Discrete Flow Matching (MOG-DFM), a general framework to steer any pretrained discrete-time flow matching generator toward Pareto-efficient trade-offs across multiple scalar objectives. At each sampling step, MOG-DFM computes a hybrid rank-directional score for candidate transitions and applies an adaptive hypercone filter to enforce consistent multi-objective progression. We also trained two unconditional discrete flow matching models, PepDFM for diverse peptide generation and EnhancerDFM for functional enhancer DNA generation, as base generation models for MOG-DFM. We demonstrate MOG-DFM's effectiveness in generating peptide binders optimized across five properties (hemolysis, non-fouling, solubility, half-life, and binding affinity), and in designing DNA sequences with specific enhancer classes and DNA shapes. In total, MOG-DFM proves to be a powerful tool for multi-property-guided biomolecule sequence design. |
|
2025-05-13T00:00:00 | 2505.07293 | AttentionInfluence: Adopting Attention Head Influence for Weak-to-Strong Pretraining Data Selection | [
"Kai Hua",
"Steven Wu",
"Ge Zhang",
"Ke Shen"
] | Recently, there has been growing interest in collecting reasoning-intensive pretraining data to improve LLMs' complex reasoning ability. Prior approaches typically rely on supervised classifiers to identify such data, which requires labeling by humans or LLMs, often introducing domain-specific biases. Due to the attention heads being crucial to in-context reasoning, we propose AttentionInfluence, a simple yet effective, training-free method without supervision signal. Our approach enables a small pretrained language model to act as a strong data selector through a simple attention head masking operation. Specifically, we identify retrieval heads and compute the loss difference when masking these heads. We apply AttentionInfluence to a 1.3B-parameter dense model to conduct data selection on the SmolLM corpus of 241B tokens, and mix the SmolLM corpus with the selected subset comprising 73B tokens to pretrain a 7B-parameter dense model using 1T training tokens and WSD learning rate scheduling. Our experimental results demonstrate substantial improvements, ranging from 1.4pp to 3.5pp, across several knowledge-intensive and reasoning-heavy benchmarks (i.e., MMLU, MMLU-Pro, AGIEval-en, GSM8K, and HumanEval). This demonstrates an effective weak-to-strong scaling property, with small models improving the final performance of larger models-offering a promising and scalable path for reasoning-centric data selection. |
|
2025-05-13T00:00:00 | 2505.07819 | H^{3}DP: Triply-Hierarchical Diffusion Policy for Visuomotor Learning | [
"Yiyang Lu",
"Yufeng Tian",
"Zhecheng Yuan",
"Xianbang Wang",
"Pu Hua",
"Zhengrong Xue",
"Huazhe Xu"
] | Visuomotor policy learning has witnessed substantial progress in robotic manipulation, with recent approaches predominantly relying on generative models to model the action distribution. However, these methods often overlook the critical coupling between visual perception and action prediction. In this work, we introduce Triply-Hierarchical Diffusion Policy~(H^{\mathbf{3}DP}), a novel visuomotor learning framework that explicitly incorporates hierarchical structures to strengthen the integration between visual features and action generation. H^{3}DP contains 3 levels of hierarchy: (1) depth-aware input layering that organizes RGB-D observations based on depth information; (2) multi-scale visual representations that encode semantic features at varying levels of granularity; and (3) a hierarchically conditioned diffusion process that aligns the generation of coarse-to-fine actions with corresponding visual features. Extensive experiments demonstrate that H^{3}DP yields a +27.5% average relative improvement over baselines across 44 simulation tasks and achieves superior performance in 4 challenging bimanual real-world manipulation tasks. Project Page: https://lyy-iiis.github.io/h3dp/. |
|
2025-05-13T00:00:00 | 2505.07793 | Overflow Prevention Enhances Long-Context Recurrent LLMs | [
"Assaf Ben-Kish",
"Itamar Zimerman",
"M. Jehanzeb Mirza",
"James Glass",
"Leonid Karlinsky",
"Raja Giryes"
] | A recent trend in LLMs is developing recurrent sub-quadratic models that improve long-context processing efficiency. We investigate leading large long-context models, focusing on how their fixed-size recurrent memory affects their performance. Our experiments reveal that, even when these models are trained for extended contexts, their use of long contexts remains underutilized. Specifically, we demonstrate that a chunk-based inference procedure, which identifies and processes only the most relevant portion of the input can mitigate recurrent memory failures and be effective for many long-context tasks: On LongBench, our method improves the overall performance of Falcon3-Mamba-Inst-7B by 14%, Falcon-Mamba-Inst-7B by 28%, RecurrentGemma-IT-9B by 50%, and RWKV6-Finch-7B by 51%. Surprisingly, this simple approach also leads to state-of-the-art results in the challenging LongBench v2 benchmark, showing competitive performance with equivalent size Transformers. Furthermore, our findings raise questions about whether recurrent models genuinely exploit long-range dependencies, as our single-chunk strategy delivers stronger performance - even in tasks that presumably require cross-context relations. |
|
2025-05-13T00:00:00 | 2505.04918 | Physics-Assisted and Topology-Informed Deep Learning for Weather Prediction | [
"Jiaqi Zheng",
"Qing Ling",
"Yerong Feng"
] | https://github.com/Yumenomae/PASSAT_5p625 | Although deep learning models have demonstrated remarkable potential in weather prediction, most of them overlook either the physics of the underlying weather evolution or the topology of the Earth's surface. In light of these disadvantages, we develop PASSAT, a novel Physics-ASSisted And Topology-informed deep learning model for weather prediction. PASSAT attributes the weather evolution to two key factors: (i) the advection process that can be characterized by the advection equation and the Navier-Stokes equation; (ii) the Earth-atmosphere interaction that is difficult to both model and calculate. PASSAT also takes the topology of the Earth's surface into consideration, other than simply treating it as a plane. With these considerations, PASSAT numerically solves the advection equation and the Navier-Stokes equation on the spherical manifold, utilizes a spherical graph neural network to capture the Earth-atmosphere interaction, and generates the initial velocity fields that are critical to solving the advection equation from the same spherical graph neural network. In the 5.625^circ-resolution ERA5 data set, PASSAT outperforms both the state-of-the-art deep learning-based weather prediction models and the operational numerical weather prediction model IFS T42. Code and checkpoint are available at https://github.com/Yumenomae/PASSAT_5p625. |
2025-05-13T00:00:00 | 2505.07812 | Continuous Visual Autoregressive Generation via Score Maximization | [
"Chenze Shao",
"Fandong Meng",
"Jie Zhou"
] | https://github.com/shaochenze/EAR | Conventional wisdom suggests that autoregressive models are used to process discrete data. When applied to continuous modalities such as visual data, Visual AutoRegressive modeling (VAR) typically resorts to quantization-based approaches to cast the data into a discrete space, which can introduce significant information loss. To tackle this issue, we introduce a Continuous VAR framework that enables direct visual autoregressive generation without vector quantization. The underlying theoretical foundation is strictly proper scoring rules, which provide powerful statistical tools capable of evaluating how well a generative model approximates the true distribution. Within this framework, all we need is to select a strictly proper score and set it as the training objective to optimize. We primarily explore a class of training objectives based on the energy score, which is likelihood-free and thus overcomes the difficulty of making probabilistic predictions in the continuous space. Previous efforts on continuous autoregressive generation, such as GIVT and diffusion loss, can also be derived from our framework using other strictly proper scores. Source code: https://github.com/shaochenze/EAR. |
2025-05-13T00:00:00 | 2505.07260 | UMoE: Unifying Attention and FFN with Shared Experts | [
"Yuanhang Yang",
"Chaozheng Wang",
"Jing Li"
] | Sparse Mixture of Experts (MoE) architectures have emerged as a promising approach for scaling Transformer models. While initial works primarily incorporated MoE into feed-forward network (FFN) layers, recent studies have explored extending the MoE paradigm to attention layers to enhance model performance. However, existing attention-based MoE layers require specialized implementations and demonstrate suboptimal performance compared to their FFN-based counterparts. In this paper, we aim to unify the MoE designs in attention and FFN layers by introducing a novel reformulation of the attention mechanism, revealing an underlying FFN-like structure within attention modules. Our proposed architecture, UMoE, achieves superior performance through attention-based MoE layers while enabling efficient parameter sharing between FFN and attention components. |
|
2025-05-13T00:00:00 | 2505.07263 | Skywork-VL Reward: An Effective Reward Model for Multimodal Understanding and Reasoning | [
"Xiaokun Wang",
"Chris",
"Jiangbo Pei",
"Wei Shen",
"Yi Peng",
"Yunzhuo Hao",
"Weijie Qiu",
"Ai Jian",
"Tianyidan Xie",
"Xuchen Song",
"Yang Liu",
"Yahui Zhou"
] | We propose Skywork-VL Reward, a multimodal reward model that provides reward signals for both multimodal understanding and reasoning tasks. Our technical approach comprises two key components: First, we construct a large-scale multimodal preference dataset that covers a wide range of tasks and scenarios, with responses collected from both standard vision-language models (VLMs) and advanced VLM reasoners. Second, we design a reward model architecture based on Qwen2.5-VL-7B-Instruct, integrating a reward head and applying multi-stage fine-tuning using pairwise ranking loss on pairwise preference data. Experimental evaluations show that Skywork-VL Reward achieves state-of-the-art results on multimodal VL-RewardBench and exhibits competitive performance on the text-only RewardBench benchmark. Furthermore, preference data constructed based on our Skywork-VL Reward proves highly effective for training Mixed Preference Optimization (MPO), leading to significant improvements in multimodal reasoning capabilities. Our results underscore Skywork-VL Reward as a significant advancement toward general-purpose, reliable reward models for multimodal alignment. Our model has been publicly released to promote transparency and reproducibility. |
|
2025-05-13T00:00:00 | 2505.06176 | MonetGPT: Solving Puzzles Enhances MLLMs' Image Retouching Skills | [
"Niladri Shekhar Dutt",
"Duygu Ceylan",
"Niloy J. Mitra"
] | Retouching is an essential task in post-manipulation of raw photographs. Generative editing, guided by text or strokes, provides a new tool accessible to users but can easily change the identity of the original objects in unacceptable and unpredictable ways. In contrast, although traditional procedural edits, as commonly supported by photoediting tools (e.g., Gimp, Lightroom), are conservative, they are still preferred by professionals. Unfortunately, professional quality retouching involves many individual procedural editing operations that is challenging to plan for most novices. In this paper, we ask if a multimodal large language model (MLLM) can be taught to critique raw photographs, suggest suitable remedies, and finally realize them with a given set of pre-authored procedural image operations. We demonstrate that MLLMs can be first made aware of the underlying image processing operations, by training them to solve specially designed visual puzzles. Subsequently, such an operation-aware MLLM can both plan and propose edit sequences. To facilitate training, given a set of expert-edited photos, we synthesize a reasoning dataset by procedurally manipulating the expert edits and then grounding a pretrained LLM on the visual adjustments, to synthesize reasoning for finetuning. The proposed retouching operations are, by construction, understandable by the users, preserve object details and resolution, and can be optionally overridden. We evaluate our setup on a variety of test examples and show advantages, in terms of explainability and identity preservation, over existing generative and other procedural alternatives. Code, data, models, and supplementary results can be found via our project website at https://monetgpt.github.io. |
|
2025-05-13T00:00:00 | 2505.00612 | Position: AI Competitions Provide the Gold Standard for Empirical Rigor in GenAI Evaluation | [
"D. Sculley",
"Will Cukierski",
"Phil Culliton",
"Sohier Dane",
"Maggie Demkin",
"Ryan Holbrook",
"Addison Howard",
"Paul Mooney",
"Walter Reade",
"Megan Risdal",
"Nate Keating"
] | In this position paper, we observe that empirical evaluation in Generative AI is at a crisis point since traditional ML evaluation and benchmarking strategies are insufficient to meet the needs of evaluating modern GenAI models and systems. There are many reasons for this, including the fact that these models typically have nearly unbounded input and output spaces, typically do not have a well defined ground truth target, and typically exhibit strong feedback loops and prediction dependence based on context of previous model outputs. On top of these critical issues, we argue that the problems of {\em leakage} and {\em contamination} are in fact the most important and difficult issues to address for GenAI evaluations. Interestingly, the field of AI Competitions has developed effective measures and practices to combat leakage for the purpose of counteracting cheating by bad actors within a competition setting. This makes AI Competitions an especially valuable (but underutilized) resource. Now is time for the field to view AI Competitions as the gold standard for empirical rigor in GenAI evaluation, and to harness and harvest their results with according value. |
|
2025-05-13T00:00:00 | 2505.07233 | DynamicRAG: Leveraging Outputs of Large Language Model as Feedback for Dynamic Reranking in Retrieval-Augmented Generation | [
"Jiashuo Sun",
"Xianrui Zhong",
"Sizhe Zhou",
"Jiawei Han"
] | https://github.com/GasolSun36/DynamicRAG | Retrieval-augmented generation (RAG) systems combine large language models (LLMs) with external knowledge retrieval, making them highly effective for knowledge-intensive tasks. A crucial but often under-explored component of these systems is the reranker, which refines retrieved documents to enhance generation quality and explainability. The challenge of selecting the optimal number of documents (k) remains unsolved: too few may omit critical information, while too many introduce noise and inefficiencies. Although recent studies have explored LLM-based rerankers, they primarily leverage internal model knowledge and overlook the rich supervisory signals that LLMs can provide, such as using response quality as feedback for optimizing reranking decisions. In this paper, we propose DynamicRAG, a novel RAG framework where the reranker dynamically adjusts both the order and number of retrieved documents based on the query. We model the reranker as an agent optimized through reinforcement learning (RL), using rewards derived from LLM output quality. Across seven knowledge-intensive datasets, DynamicRAG demonstrates superior performance, achieving state-of-the-art results. The model, data and code are available at https://github.com/GasolSun36/DynamicRAG |
2025-05-13T00:00:00 | 2505.04066 | LLAMAPIE: Proactive In-Ear Conversation Assistants | [
"Tuochao Chen",
"Nicholas Batchelder",
"Alisa Liu",
"Noah Smith",
"Shyamnath Gollakota"
] | We introduce LlamaPIE, the first real-time proactive assistant designed to enhance human conversations through discreet, concise guidance delivered via hearable devices. Unlike traditional language models that require explicit user invocation, this assistant operates in the background, anticipating user needs without interrupting conversations. We address several challenges, including determining when to respond, crafting concise responses that enhance conversations, leveraging knowledge of the user for context-aware assistance, and real-time, on-device processing. To achieve this, we construct a semi-synthetic dialogue dataset and propose a two-model pipeline: a small model that decides when to respond and a larger model that generates the response. We evaluate our approach on real-world datasets, demonstrating its effectiveness in providing helpful, unobtrusive assistance. User studies with our assistant, implemented on Apple Silicon M2 hardware, show a strong preference for the proactive assistant over both a baseline with no assistance and a reactive model, highlighting the potential of LlamaPie to enhance live conversations. |
|
2025-05-13T00:00:00 | 2505.07291 | INTELLECT-2: A Reasoning Model Trained Through Globally Decentralized Reinforcement Learning | [
"Prime Intellect Team",
"Sami Jaghouar",
"Justus Mattern",
"Jack Min Ong",
"Jannik Straube",
"Manveer Basra",
"Aaron Pazdera",
"Kushal Thaman",
"Matthew Di Ferrante",
"Felix Gabriel",
"Fares Obeid",
"Kemal Erdem",
"Michael Keiblinger",
"Johannes Hagemann"
] | We introduce INTELLECT-2, the first globally distributed reinforcement learning (RL) training run of a 32 billion parameter language model. Unlike traditional centralized training efforts, INTELLECT-2 trains a reasoning model using fully asynchronous RL across a dynamic, heterogeneous swarm of permissionless compute contributors. To enable a training run with this unique infrastructure, we built various components from scratch: we introduce PRIME-RL, our training framework purpose-built for distributed asynchronous reinforcement learning, based on top of novel components such as TOPLOC, which verifies rollouts from untrusted inference workers, and SHARDCAST, which efficiently broadcasts policy weights from training nodes to inference workers. Beyond infrastructure components, we propose modifications to the standard GRPO training recipe and data filtering techniques that were crucial to achieve training stability and ensure that our model successfully learned its training objective, thus improving upon QwQ-32B, the state of the art reasoning model in the 32B parameter range. We open-source INTELLECT-2 along with all of our code and data, hoping to encourage and enable more open research in the field of decentralized training. |
|
2025-05-14T00:00:00 | 2505.08665 | SkillFormer: Unified Multi-View Video Understanding for Proficiency Estimation | [
"Edoardo Bianchi",
"Antonio Liotta"
] | Assessing human skill levels in complex activities is a challenging problem with applications in sports, rehabilitation, and training. In this work, we present SkillFormer, a parameter-efficient architecture for unified multi-view proficiency estimation from egocentric and exocentric videos. Building on the TimeSformer backbone, SkillFormer introduces a CrossViewFusion module that fuses view-specific features using multi-head cross-attention, learnable gating, and adaptive self-calibration. We leverage Low-Rank Adaptation to fine-tune only a small subset of parameters, significantly reducing training costs. In fact, when evaluated on the EgoExo4D dataset, SkillFormer achieves state-of-the-art accuracy in multi-view settings while demonstrating remarkable computational efficiency, using 4.5x fewer parameters and requiring 3.75x fewer training epochs than prior baselines. It excels in multiple structured tasks, confirming the value of multi-view integration for fine-grained skill assessment. |
|
2025-05-14T00:00:00 | 2505.07591 | A Multi-Dimensional Constraint Framework for Evaluating and Improving Instruction Following in Large Language Models | [
"Junjie Ye",
"Caishuang Huang",
"Zhuohan Chen",
"Wenjie Fu",
"Chenyuan Yang",
"Leyi Yang",
"Yilong Wu",
"Peng Wang",
"Meng Zhou",
"Xiaolong Yang",
"Tao Gui",
"Qi Zhang",
"Zhongchao Shi",
"Jianping Fan",
"Xuanjing Huang"
] | https://github.com/Junjie-Ye/MulDimIF | Instruction following evaluates large language models (LLMs) on their ability to generate outputs that adhere to user-defined constraints. However, existing benchmarks often rely on templated constraint prompts, which lack the diversity of real-world usage and limit fine-grained performance assessment. To fill this gap, we propose a multi-dimensional constraint framework encompassing three constraint patterns, four constraint categories, and four difficulty levels. Building on this framework, we develop an automated instruction generation pipeline that performs constraint expansion, conflict detection, and instruction rewriting, yielding 1,200 code-verifiable instruction-following test samples. We evaluate 19 LLMs across seven model families and uncover substantial variation in performance across constraint forms. For instance, average performance drops from 77.67% at Level I to 32.96% at Level IV. Furthermore, we demonstrate the utility of our approach by using it to generate data for reinforcement learning, achieving substantial gains in instruction following without degrading general performance. In-depth analysis indicates that these gains stem primarily from modifications in the model's attention modules parameters, which enhance constraint recognition and adherence. Code and data are available in https://github.com/Junjie-Ye/MulDimIF. |
2025-05-14T00:00:00 | 2505.07215 | Measuring General Intelligence with Generated Games | [
"Vivek Verma",
"David Huang",
"William Chen",
"Dan Klein",
"Nicholas Tomlin"
] | We present gg-bench, a collection of game environments designed to evaluate general reasoning capabilities in language models. Unlike most static benchmarks, gg-bench is a data generating process where new evaluation instances can be generated at will. In particular, gg-bench is synthetically generated by (1) using a large language model (LLM) to generate natural language descriptions of novel games, (2) using the LLM to implement each game in code as a Gym environment, and (3) training reinforcement learning (RL) agents via self-play on the generated games. We evaluate language models by their winrate against these RL agents by prompting models with the game description, current board state, and a list of valid moves, after which models output the moves they wish to take. gg-bench is challenging: state-of-the-art LLMs such as GPT-4o and Claude 3.7 Sonnet achieve winrates of 7-9% on gg-bench using in-context learning, while reasoning models such as o1, o3-mini and DeepSeek-R1 achieve average winrates of 31-36%. We release the generated games, data generation process, and evaluation code in order to support future modeling work and expansion of our benchmark. |
|
2025-05-14T00:00:00 | 2505.07416 | ViMRHP: A Vietnamese Benchmark Dataset for Multimodal Review Helpfulness Prediction via Human-AI Collaborative Annotation | [
"Truc Mai-Thanh Nguyen",
"Dat Minh Nguyen",
"Son T. Luu",
"Kiet Van Nguyen"
] | https://github.com/trng28/ViMRHP | Multimodal Review Helpfulness Prediction (MRHP) is an essential task in recommender systems, particularly in E-commerce platforms. Determining the helpfulness of user-generated reviews enhances user experience and improves consumer decision-making. However, existing datasets focus predominantly on English and Indonesian, resulting in a lack of linguistic diversity, especially for low-resource languages such as Vietnamese. In this paper, we introduce ViMRHP (Vietnamese Multimodal Review Helpfulness Prediction), a large-scale benchmark dataset for MRHP task in Vietnamese. This dataset covers four domains, including 2K products with 46K reviews. Meanwhile, a large-scale dataset requires considerable time and cost. To optimize the annotation process, we leverage AI to assist annotators in constructing the ViMRHP dataset. With AI assistance, annotation time is reduced (90 to 120 seconds per task down to 20 to 40 seconds per task) while maintaining data quality and lowering overall costs by approximately 65%. However, AI-generated annotations still have limitations in complex annotation tasks, which we further examine through a detailed performance analysis. In our experiment on ViMRHP, we evaluate baseline models on human-verified and AI-generated annotations to assess their quality differences. The ViMRHP dataset is publicly available at https://github.com/trng28/ViMRHP |
2025-05-14T00:00:00 | 2505.08712 | NavDP: Learning Sim-to-Real Navigation Diffusion Policy with Privileged Information Guidance | [
"Wenzhe Cai",
"Jiaqi Peng",
"Yuqiang Yang",
"Yujian Zhang",
"Meng Wei",
"Hanqing Wang",
"Yilun Chen",
"Tai Wang",
"Jiangmiao Pang"
] | Learning navigation in dynamic open-world environments is an important yet challenging skill for robots. Most previous methods rely on precise localization and mapping or learn from expensive real-world demonstrations. In this paper, we propose the Navigation Diffusion Policy (NavDP), an end-to-end framework trained solely in simulation and can zero-shot transfer to different embodiments in diverse real-world environments. The key ingredient of NavDP's network is the combination of diffusion-based trajectory generation and a critic function for trajectory selection, which are conditioned on only local observation tokens encoded from a shared policy transformer. Given the privileged information of the global environment in simulation, we scale up the demonstrations of good quality to train the diffusion policy and formulate the critic value function targets with contrastive negative samples. Our demonstration generation approach achieves about 2,500 trajectories/GPU per day, 20times more efficient than real-world data collection, and results in a large-scale navigation dataset with 363.2km trajectories across 1244 scenes. Trained with this simulation dataset, NavDP achieves state-of-the-art performance and consistently outstanding generalization capability on quadruped, wheeled, and humanoid robots in diverse indoor and outdoor environments. In addition, we present a preliminary attempt at using Gaussian Splatting to make in-domain real-to-sim fine-tuning to further bridge the sim-to-real gap. Experiments show that adding such real-to-sim data can improve the success rate by 30\% without hurting its generalization capability. |
|
2025-05-14T00:00:00 | 2505.07916 | MiniMax-Speech: Intrinsic Zero-Shot Text-to-Speech with a Learnable Speaker Encoder | [
"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"
] | 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. |
|
2025-05-14T00:00:00 | 2505.08445 | Optimizing Retrieval-Augmented Generation: Analysis of Hyperparameter Impact on Performance and Efficiency | [
"Adel Ammar",
"Anis Koubaa",
"Omer Nacar",
"Wadii Boulila"
] | Large language models achieve high task performance yet often hallucinate or rely on outdated knowledge. Retrieval-augmented generation (RAG) addresses these gaps by coupling generation with external search. We analyse how hyperparameters influence speed and quality in RAG systems, covering Chroma and Faiss vector stores, chunking policies, cross-encoder re-ranking, and temperature, and we evaluate six metrics: faithfulness, answer correctness, answer relevancy, context precision, context recall, and answer similarity. Chroma processes queries 13% faster, whereas Faiss yields higher retrieval precision, revealing a clear speed-accuracy trade-off. Naive fixed-length chunking with small windows and minimal overlap outperforms semantic segmentation while remaining the quickest option. Re-ranking provides modest gains in retrieval quality yet increases runtime by roughly a factor of 5, so its usefulness depends on latency constraints. These results help practitioners balance computational cost and accuracy when tuning RAG systems for transparent, up-to-date responses. Finally, we re-evaluate the top configurations with a corrective RAG workflow and show that their advantages persist when the model can iteratively request additional evidence. We obtain a near-perfect context precision (99%), which demonstrates that RAG systems can achieve extremely high retrieval accuracy with the right combination of hyperparameters, with significant implications for applications where retrieval quality directly impacts downstream task performance, such as clinical decision support in healthcare. |
|
2025-05-14T00:00:00 | 2504.21475 | Advancing Arabic Reverse Dictionary Systems: A Transformer-Based Approach with Dataset Construction Guidelines | [
"Serry Sibaee",
"Samar Ahmed",
"Abdullah Al Harbi",
"Omer Nacar",
"Adel Ammar",
"Yasser Habashi",
"Wadii Boulila"
] | This study addresses the critical gap in Arabic natural language processing by developing an effective Arabic Reverse Dictionary (RD) system that enables users to find words based on their descriptions or meanings. We present a novel transformer-based approach with a semi-encoder neural network architecture featuring geometrically decreasing layers that achieves state-of-the-art results for Arabic RD tasks. Our methodology incorporates a comprehensive dataset construction process and establishes formal quality standards for Arabic lexicographic definitions. Experiments with various pre-trained models demonstrate that Arabic-specific models significantly outperform general multilingual embeddings, with ARBERTv2 achieving the best ranking score (0.0644). Additionally, we provide a formal abstraction of the reverse dictionary task that enhances theoretical understanding and develop a modular, extensible Python library (RDTL) with configurable training pipelines. Our analysis of dataset quality reveals important insights for improving Arabic definition construction, leading to eight specific standards for building high-quality reverse dictionary resources. This work contributes significantly to Arabic computational linguistics and provides valuable tools for language learning, academic writing, and professional communication in Arabic. |
|
2025-05-14T00:00:00 | 2505.08751 | Aya Vision: Advancing the Frontier of Multilingual Multimodality | [
"Saurabh Dash",
"Yiyang Nan",
"John Dang",
"Arash Ahmadian",
"Shivalika Singh",
"Madeline Smith",
"Bharat Venkitesh",
"Vlad Shmyhlo",
"Viraat Aryabumi",
"Walter Beller-Morales",
"Jeremy Pekmez",
"Jason Ozuzu",
"Pierre Richemond",
"Acyr Locatelli",
"Nick Frosst",
"Phil Blunsom",
"Aidan Gomez",
"Ivan Zhang",
"Marzieh Fadaee",
"Manoj Govindassamy",
"Sudip Roy",
"Matthias Gallé",
"Beyza Ermis",
"Ahmet Üstün",
"Sara Hooker"
] | Building multimodal language models is fundamentally challenging: it requires aligning vision and language modalities, curating high-quality instruction data, and avoiding the degradation of existing text-only capabilities once vision is introduced. These difficulties are further magnified in the multilingual setting, where the need for multimodal data in different languages exacerbates existing data scarcity, machine translation often distorts meaning, and catastrophic forgetting is more pronounced. To address the aforementioned challenges, we introduce novel techniques spanning both data and modeling. First, we develop a synthetic annotation framework that curates high-quality, diverse multilingual multimodal instruction data, enabling Aya Vision models to produce natural, human-preferred responses to multimodal inputs across many languages. Complementing this, we propose a cross-modal model merging technique that mitigates catastrophic forgetting, effectively preserving text-only capabilities while simultaneously enhancing multimodal generative performance. Aya-Vision-8B achieves best-in-class performance compared to strong multimodal models such as Qwen-2.5-VL-7B, Pixtral-12B, and even much larger Llama-3.2-90B-Vision. We further scale this approach with Aya-Vision-32B, which outperforms models more than twice its size, such as Molmo-72B and LLaMA-3.2-90B-Vision. Our work advances multilingual progress on the multi-modal frontier, and provides insights into techniques that effectively bend the need for compute while delivering extremely high performance. |
|
2025-05-14T00:00:00 | 2505.08311 | AM-Thinking-v1: Advancing the Frontier of Reasoning at 32B Scale | [
"Yunjie Ji",
"Xiaoyu Tian",
"Sitong Zhao",
"Haotian Wang",
"Shuaiting Chen",
"Yiping Peng",
"Han Zhao",
"Xiangang Li"
] | We present AM-Thinking-v1, a 32B dense language model that advances the frontier of reasoning, embodying the collaborative spirit of open-source innovation. Outperforming DeepSeek-R1 and rivaling leading Mixture-of-Experts (MoE) models like Qwen3-235B-A22B and Seed1.5-Thinking, AM-Thinking-v1 achieves impressive scores of 85.3 on AIME 2024, 74.4 on AIME 2025, and 70.3 on LiveCodeBench, showcasing state-of-the-art mathematical and coding capabilities among open-source models of similar scale. Built entirely from the open-source Qwen2.5-32B base model and publicly available queries, AM-Thinking-v1 leverages a meticulously crafted post-training pipeline - combining supervised fine-tuning and reinforcement learning - to deliver exceptional reasoning capabilities. This work demonstrates that the open-source community can achieve high performance at the 32B scale, a practical sweet spot for deployment and fine-tuning. By striking a balance between top-tier performance and real-world usability, we hope AM-Thinking-v1 inspires further collaborative efforts to harness mid-scale models, pushing reasoning boundaries while keeping accessibility at the core of innovation. We have open-sourced our model on https://huggingface.co/a-m-team/AM-Thinking-v1{Hugging Face}. |
|
2025-05-14T00:00:00 | 2505.05464 | Bring Reason to Vision: Understanding Perception and Reasoning through Model Merging | [
"Shiqi Chen",
"Jinghan Zhang",
"Tongyao Zhu",
"Wei Liu",
"Siyang Gao",
"Miao Xiong",
"Manling Li",
"Junxian He"
] | Vision-Language Models (VLMs) combine visual perception with the general capabilities, such as reasoning, of Large Language Models (LLMs). However, the mechanisms by which these two abilities can be combined and contribute remain poorly understood. In this work, we explore to compose perception and reasoning through model merging that connects parameters of different models. Unlike previous works that often focus on merging models of the same kind, we propose merging models across modalities, enabling the incorporation of the reasoning capabilities of LLMs into VLMs. Through extensive experiments, we demonstrate that model merging offers a successful pathway to transfer reasoning abilities from LLMs to VLMs in a training-free manner. Moreover, we utilize the merged models to understand the internal mechanism of perception and reasoning and how merging affects it. We find that perception capabilities are predominantly encoded in the early layers of the model, whereas reasoning is largely facilitated by the middle-to-late layers. After merging, we observe that all layers begin to contribute to reasoning, whereas the distribution of perception abilities across layers remains largely unchanged. These observations shed light on the potential of model merging as a tool for multimodal integration and interpretation. |
|
2025-05-14T00:00:00 | 2505.08727 | Memorization-Compression Cycles Improve Generalization | [
"Fangyuan Yu"
] | We prove theoretically that generalization improves not only through data scaling but also by compressing internal representations. To operationalize this insight, we introduce the Information Bottleneck Language Modeling (IBLM) objective, which reframes language modeling as a constrained optimization problem: minimizing representation entropy subject to optimal prediction performance. Empirically, we observe an emergent memorization-compression cycle during LLM pretraining, evidenced by oscillation positive/negative gradient alignment between cross-entropy and Matrix-Based Entropy (MBE), a measure of representation entropy. This pattern closely mirrors the predictive-compressive trade-off prescribed by IBLM and also parallels the biological alternation between awake learning and sleep consolidation. Motivated by this observation, we propose Gated Phase Transition (GAPT), a training algorithm that adaptively switches between memorization and compression phases. When applied to GPT-2 pretraining on FineWeb dataset, GAPT reduces MBE by 50% and improves cross-entropy by 4.8%. GAPT improves OOD generalizatino by 35% in a pretraining task on arithmetic multiplication. In a setting designed to simulate catastrophic forgetting, GAPT reduces interference by compressing and separating representations, achieving a 97% improvement in separation - paralleling the functional role of sleep consolidation. |
|
2025-05-14T00:00:00 | 2505.08638 | TRAIL: Trace Reasoning and Agentic Issue Localization | [
"Darshan Deshpande",
"Varun Gangal",
"Hersh Mehta",
"Jitin Krishnan",
"Anand Kannappan",
"Rebecca Qian"
] | The increasing adoption of agentic workflows across diverse domains brings a critical need to scalably and systematically evaluate the complex traces these systems generate. Current evaluation methods depend on manual, domain-specific human analysis of lengthy workflow traces - an approach that does not scale with the growing complexity and volume of agentic outputs. Error analysis in these settings is further complicated by the interplay of external tool outputs and language model reasoning, making it more challenging than traditional software debugging. In this work, we (1) articulate the need for robust and dynamic evaluation methods for agentic workflow traces, (2) introduce a formal taxonomy of error types encountered in agentic systems, and (3) present a set of 148 large human-annotated traces (TRAIL) constructed using this taxonomy and grounded in established agentic benchmarks. To ensure ecological validity, we curate traces from both single and multi-agent systems, focusing on real-world applications such as software engineering and open-world information retrieval. Our evaluations reveal that modern long context LLMs perform poorly at trace debugging, with the best Gemini-2.5-pro model scoring a mere 11% on TRAIL. Our dataset and code are made publicly available to support and accelerate future research in scalable evaluation for agentic workflows. |
|
2025-05-14T00:00:00 | 2505.08175 | Fast Text-to-Audio Generation with Adversarial Post-Training | [
"Zachary Novack",
"Zach Evans",
"Zack Zukowski",
"Josiah Taylor",
"CJ Carr",
"Julian Parker",
"Adnan Al-Sinan",
"Gian Marco Iodice",
"Julian McAuley",
"Taylor Berg-Kirkpatrick",
"Jordi Pons"
] | Text-to-audio systems, while increasingly performant, are slow at inference time, thus making their latency unpractical for many creative applications. We present Adversarial Relativistic-Contrastive (ARC) post-training, the first adversarial acceleration algorithm for diffusion/flow models not based on distillation. While past adversarial post-training methods have struggled to compare against their expensive distillation counterparts, ARC post-training is a simple procedure that (1) extends a recent relativistic adversarial formulation to diffusion/flow post-training and (2) combines it with a novel contrastive discriminator objective to encourage better prompt adherence. We pair ARC post-training with a number optimizations to Stable Audio Open and build a model capable of generating approx12s of 44.1kHz stereo audio in approx75ms on an H100, and approx7s on a mobile edge-device, the fastest text-to-audio model to our knowledge. |
|
2025-05-14T00:00:00 | 2505.09027 | Tests as Prompt: A Test-Driven-Development Benchmark for LLM Code Generation | [
"Yi Cui"
] | We introduce WebApp1K, a novel benchmark for evaluating large language models (LLMs) in test-driven development (TDD) tasks, where test cases serve as both prompt and verification for code generation. Unlike traditional approaches relying on natural language prompts, our benchmark emphasizes the ability of LLMs to interpret and implement functionality directly from test cases, reflecting real-world software development practices. Comprising 1000 diverse challenges across 20 application domains, the benchmark evaluates LLMs on their ability to generate compact, functional code under the constraints of context length and multi-feature complexity. Our findings highlight instruction following and in-context learning as critical capabilities for TDD success, surpassing the importance of general coding proficiency or pretraining knowledge. Through comprehensive evaluation of 19 frontier models, we reveal performance bottlenecks, such as instruction loss in long prompts, and provide a detailed error analysis spanning multiple root causes. This work underscores the practical value of TDD-specific benchmarks and lays the foundation for advancing LLM capabilities in rigorous, application-driven coding scenarios. |
|
2025-05-15T00:00:00 | 2505.09568 | BLIP3-o: A Family of Fully Open Unified Multimodal Models-Architecture, Training and Dataset | [
"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"
] | 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. |
Subsets and Splits