date
timestamp[ns]date 2023-05-05 00:00:00
2025-09-08 00:00:00
| arxiv_id
stringlengths 10
10
| title
stringlengths 6
202
| authors
listlengths 1
3.3k
| github
stringlengths 0
116
| abstract
stringlengths 165
1.92k
|
|---|---|---|---|---|---|
2025-09-03T00:00:00 |
2509.02479
|
SimpleTIR: End-to-End Reinforcement Learning for Multi-Turn Tool-Integrated Reasoning
|
[
"Zhenghai Xue",
"Longtao Zheng",
"Qian Liu",
"Yingru Li",
"Xiaosen Zheng",
"Zejun Ma",
"Bo An"
] |
Large Language Models (LLMs) can significantly improve their reasoning capabilities by interacting with external tools, a paradigm known as Tool-Integrated Reasoning (TIR). However, extending TIR to multi-turn scenarios using Reinforcement Learning (RL) is often hindered by training instability and performance collapse. We identify that such instability is primarily caused by a distributional drift from external tool feedback, leading to the generation of low-probability tokens. This issue compounds over successive turns, causing catastrophic gradient norm explosions that derail the training process. To address this challenge, we introduce SimpleTIR , a plug-and-play algorithm that stabilizes multi-turn TIR training. Its core strategy is to identify and filter out trajectories containing void turns, i.e., turns that yield neither a code block nor a final answer. By removing these problematic trajectories from the policy update, SimpleTIR effectively blocks the harmful, high-magnitude gradients, thus stabilizing the learning dynamics. Extensive experiments show that SimpleTIR achieves state-of-the-art performance on challenging math reasoning benchmarks, notably elevating the AIME24 score from a text-only baseline of 22.1 to 50.5 when starting from the Qwen2.5-7B base model. Furthermore, by avoiding the constraints of supervised fine-tuning, SimpleTIR encourages the model to discover diverse and sophisticated reasoning patterns, such as self-correction and cross-validation.
|
|
2025-09-03T00:00:00 |
2509.02040
|
Attributes as Textual Genes: Leveraging LLMs as Genetic Algorithm Simulators for Conditional Synthetic Data Generation
|
[
"Guangzeng Han",
"Weisi Liu",
"Xiaolei Huang"
] |
Large Language Models (LLMs) excel at generating synthetic data, but ensuring its quality and diversity remains challenging. We propose Genetic Prompt, a novel framework that combines genetic algorithms with LLMs to augment synthetic data generation. Our approach treats semantic text attributes as gene sequences and leverages the LLM to simulate crossover and mutation operations. This genetic process enhances data quality and diversity by creating novel attribute combinations, yielding synthetic distributions closer to real-world data. To optimize parent selection, we also integrate an active learning scheme that expands the offspring search space. Our experiments on multiple NLP tasks reveal several key findings: Genetic Prompt not only significantly outperforms state-of-the-art baselines but also shows robust performance across various generator model sizes and scales. Moreover, we demonstrate that fusing our synthetic data with the original training set significantly boosts downstream model performance, particularly for class-imbalanced scenarios. Our findings validate that Genetic Prompt is an effective method for producing high-quality synthetic data for a wide range of NLP applications.
|
|
2025-09-03T00:00:00 |
2509.01644
|
OpenVision 2: A Family of Generative Pretrained Visual Encoders for Multimodal Learning
|
[
"Yanqing Liu",
"Xianhang Li",
"Letian Zhang",
"Zirui Wang",
"Zeyu Zheng",
"Yuyin Zhou",
"Cihang Xie"
] |
This paper provides a simplification on OpenVision's architecture and loss design for enhancing its training efficiency. Following the prior vision-language pretraining works CapPa and AIMv2, as well as modern multimodal designs like LLaVA, our changes are straightforward: we remove the text encoder (and therefore the contrastive loss), retaining only the captioning loss as a purely generative training signal. We name this new version OpenVision 2. The initial results are promising: despite this simplification, OpenVision 2 competitively matches the original model's performance on a broad set of multimodal benchmarks while substantially cutting both training time and memory consumption. For example, with ViT-L/14, it reduces training time by about 1.5x (from 83h to 57h), and memory usage by about 1.8x (from 24.5GB to 13.8GB, equivalently allowing the maximum batch size to grow from 2k to 8k). This superior training efficiency also allows us to scale far beyond the largest vision encoder used in OpenVision, reaching more than 1 billion parameters. We hold a strong belief that this lightweight, generative-only paradigm is compelling for future vision encoder development in multimodal foundation models.
|
|
2025-09-03T00:00:00 |
2509.01563
|
Kwai Keye-VL 1.5 Technical Report
|
[
"Biao Yang",
"Bin Wen",
"Boyang Ding",
"Changyi Liu",
"Chenglong Chu",
"Chengru Song",
"Chongling Rao",
"Chuan Yi",
"Da Li",
"Dunju Zang",
"Fan Yang",
"Guorui Zhou",
"Guowang Zhang",
"Han Shen",
"Hao Peng",
"Haojie Ding",
"Hao Wang",
"Hengrui Ju",
"Jiaming Huang",
"Jiangxia Cao",
"Jiankang Chen",
"Jingyun Hua",
"Kaibing Chen",
"Kaiyu Jiang",
"Kaiyu Tang",
"Kun Gai",
"Muhao Wei",
"Qiang Wang",
"Ruitao Wang",
"Sen Na",
"Shengnan Zhang",
"Siyang Mao",
"Sui Huang",
"Tianke Zhang",
"Tingting Gao",
"Wei Chen",
"Wei Yuan",
"Xiangyu Wu",
"Xiao Hu",
"Xingyu Lu",
"Yi-Fan Zhang",
"Yiping Yang",
"Yulong Chen",
"Zeyi Lu",
"Zhenhua Wu",
"Zhixin Ling",
"Zhuoran Yang",
"Ziming Li",
"Di Xu",
"Haixuan Gao",
"Hang Li",
"Jing Wang",
"Lejian Ren",
"Qigen Hu",
"Qianqian Wang",
"Shiyao Wang",
"Xinchen Luo",
"Yan Li",
"Yuhang Hu",
"Zixing Zhang"
] |
In recent years, the development of Large Language Models (LLMs) has significantly advanced, extending their capabilities to multimodal tasks through Multimodal Large Language Models (MLLMs). However, video understanding remains a challenging area due to the dynamic and information-dense nature of videos. Existing models struggle with the trade-off between spatial resolution and temporal coverage when processing video content. We present Keye-VL-1.5, which addresses fundamental challenges in video comprehension through three key innovations. First, we introduce a novel Slow-Fast video encoding strategy that dynamically allocates computational resources based on inter-frame similarity, processing key frames with significant visual changes at higher resolution (Slow pathway) while handling relatively static frames with increased temporal coverage at lower resolution (Fast pathway). Second, we implement a progressive four-stage pre-training methodology that systematically extends the model's context length from 8K to 128K tokens, enabling processing of longer videos and more complex visual content. Third, we develop a comprehensive post-training pipeline focusing on reasoning enhancement and human preference alignment, incorporating a 5-step chain-of-thought data construction process, iterative GSPO-based reinforcement learning with progressive prompt hinting for difficult cases, and alignment training. Through extensive evaluation on public benchmarks and rigorous internal human assessment, Keye-VL-1.5 demonstrates significant improvements over existing models, particularly excelling in video understanding tasks while maintaining competitive performance on general multimodal benchmarks.
|
|
2025-09-03T00:00:00 |
2509.01360
|
M3Ret: Unleashing Zero-shot Multimodal Medical Image Retrieval via Self-Supervision
|
[
"Che Liu",
"Zheng Jiang",
"Chengyu Fang",
"Heng Guo",
"Yan-Jie Zhou",
"Jiaqi Qu",
"Le Lu",
"Minfeng Xu"
] |
Medical image retrieval is essential for clinical decision-making and translational research, relying on discriminative visual representations. Yet, current methods remain fragmented, relying on separate architectures and training strategies for 2D, 3D, and video-based medical data. This modality-specific design hampers scalability and inhibits the development of unified representations. To enable unified learning, we curate a large-scale hybrid-modality dataset comprising 867,653 medical imaging samples, including 2D X-rays and ultrasounds, RGB endoscopy videos, and 3D CT scans. Leveraging this dataset, we train M3Ret, a unified visual encoder without any modality-specific customization. It successfully learns transferable representations using both generative (MAE) and contrastive (SimDINO) self-supervised learning (SSL) paradigms. Our approach sets a new state-of-the-art in zero-shot image-to-image retrieval across all individual modalities, surpassing strong baselines such as DINOv3 and the text-supervised BMC-CLIP. More remarkably, strong cross-modal alignment emerges without paired data, and the model generalizes to unseen MRI tasks, despite never observing MRI during pretraining, demonstrating the generalizability of purely visual self-supervision to unseen modalities. Comprehensive analyses further validate the scalability of our framework across model and data sizes. These findings deliver a promising signal to the medical imaging community, positioning M3Ret as a step toward foundation models for visual SSL in multimodal medical image understanding.
|
|
2025-09-03T00:00:00 |
2509.00244
|
Universal Deep Research: Bring Your Own Model and Strategy
|
[
"Peter Belcak",
"Pavlo Molchanov"
] |
Deep research tools are among the most impactful and most commonly encountered agentic systems today. We observe, however, that each deep research agent introduced so far is hard-coded to carry out a particular research strategy using a fixed choice of tools. We introduce Universal Deep Research (UDR), a generalist agentic system that wraps around any language model and enables the user to create, edit, and refine their own entirely custom deep research strategies without any need for additional training or finetuning. To showcase the generality of our system, we equip UDR with example minimal, expansive, and intensive research strategies, and provide a user interface to facilitate experimentation with the system.
|
|
2025-09-03T00:00:00 |
2509.02544
|
UI-TARS-2 Technical Report: Advancing GUI Agent with Multi-Turn Reinforcement Learning
|
[
"Haoming Wang",
"Haoyang Zou",
"Huatong Song",
"Jiazhan Feng",
"Junjie Fang",
"Junting Lu",
"Longxiang Liu",
"Qinyu Luo",
"Shihao Liang",
"Shijue Huang",
"Wanjun Zhong",
"Yining Ye",
"Yujia Qin",
"Yuwen Xiong",
"Yuxin Song",
"Zhiyong Wu",
"Bo Li",
"Chen Dun",
"Chong Liu",
"Fuxing Leng",
"Hanbin Wang",
"Hao Yu",
"Haobin Chen",
"Hongyi Guo",
"Jing Su",
"Jingjia Huang",
"Kai Shen",
"Kaiyu Shi",
"Lin Yan",
"Peiyao Zhao",
"Pengfei Liu",
"Qinghao Ye",
"Renjie Zheng",
"Wayne Xin Zhao",
"Wen Heng",
"Wenhao Huang",
"Wenqian Wang",
"Xiaobo Qin",
"Yi Lin",
"Youbin Wu",
"Zehui Chen",
"Zihao Wang",
"Baoquan Zhong",
"Xinchun Zhang",
"Xujing Li",
"Yuanfan Li",
"Zhongkai Zhao",
"Chengquan Jiang",
"Faming Wu",
"Haotian Zhou",
"Jinlin Pang",
"Li Han",
"Qianli Ma",
"Siyao Liu",
"Songhua Cai",
"Wenqi Fu",
"Xin Liu",
"Zhi Zhang",
"Bo Zhou",
"Guoliang Li",
"Jiajun Shi",
"Jiale Yang",
"Jie Tang",
"Li Li",
"Taoran Lu",
"Woyu Lin",
"Xiaokang Tong",
"Xinyao Li",
"Yichi Zhang",
"Yu Miao",
"Zhengxuan Jiang",
"Zili Li",
"Ziyuan Zhao",
"Chenxin Li",
"Dehua Ma",
"Feng Lin",
"Ge Zhang",
"Haihua Yang",
"Hangyu Guo",
"Hongda Zhu",
"Jiaheng Liu",
"Junda Du",
"Kai Cai",
"Kuanye Li",
"Lichen Yuan",
"Meilan Han",
"Minchao Wang",
"Shuyue Guo",
"Tianhao Cheng",
"Xiaobo Ma",
"Xiaojun Xiao",
"Xiaolong Huang",
"Xinjie Chen",
"Yidi Du",
"Yilin Chen",
"Yiwen Wang",
"Zhaojian Li",
"Zhenzhu Yang",
"Zhiyuan Zeng",
"Chaolin Jin",
"Chen Li",
"Hao Chen",
"Haoli Chen",
"Jian Chen",
"Qinghao Zhao",
"Guang Shi"
] |
The development of autonomous agents for graphical user interfaces (GUIs) presents major challenges in artificial intelligence. While recent advances in native agent models have shown promise by unifying perception, reasoning, action, and memory through end-to-end learning, open problems remain in data scalability, multi-turn reinforcement learning (RL), the limitations of GUI-only operation, and environment stability. In this technical report, we present UI-TARS-2, a native GUI-centered agent model that addresses these challenges through a systematic training methodology: a data flywheel for scalable data generation, a stabilized multi-turn RL framework, a hybrid GUI environment that integrates file systems and terminals, and a unified sandbox platform for large-scale rollouts. Empirical evaluation demonstrates that UI-TARS-2 achieves significant improvements over its predecessor UI-TARS-1.5. On GUI benchmarks, it reaches 88.2 on Online-Mind2Web, 47.5 on OSWorld, 50.6 on WindowsAgentArena, and 73.3 on AndroidWorld, outperforming strong baselines such as Claude and OpenAI agents. In game environments, it attains a mean normalized score of 59.8 across a 15-game suite-roughly 60% of human-level performance-and remains competitive with frontier proprietary models (e.g., OpenAI o3) on LMGame-Bench. Additionally, the model can generalize to long-horizon information-seeking tasks and software engineering benchmarks, highlighting its robustness across diverse agent tasks. Detailed analyses of training dynamics further provide insights into achieving stability and efficiency in large-scale agent RL. These results underscore UI-TARS-2's potential to advance the state of GUI agents and exhibit strong generalization to real-world interactive scenarios.
|
|
2025-09-03T00:00:00 |
2509.01984
|
Discrete Noise Inversion for Next-scale Autoregressive Text-based Image Editing
|
[
"Quan Dao",
"Xiaoxiao He",
"Ligong Han",
"Ngan Hoai Nguyen",
"Amin Heyrani Nobar",
"Faez Ahmed",
"Han Zhang",
"Viet Anh Nguyen",
"Dimitris Metaxas"
] |
Visual autoregressive models (VAR) have recently emerged as a promising class of generative models, achieving performance comparable to diffusion models in text-to-image generation tasks. While conditional generation has been widely explored, the ability to perform prompt-guided image editing without additional training is equally critical, as it supports numerous practical real-world applications. This paper investigates the text-to-image editing capabilities of VAR by introducing Visual AutoRegressive Inverse Noise (VARIN), the first noise inversion-based editing technique designed explicitly for VAR models. VARIN leverages a novel pseudo-inverse function for argmax sampling, named Location-aware Argmax Inversion (LAI), to generate inverse Gumbel noises. These inverse noises enable precise reconstruction of the source image and facilitate targeted, controllable edits aligned with textual prompts. Extensive experiments demonstrate that VARIN effectively modifies source images according to specified prompts while significantly preserving the original background and structural details, thus validating its efficacy as a practical editing approach.
|
|
2025-09-03T00:00:00 |
2509.01610
|
Improving Large Vision and Language Models by Learning from a Panel of Peers
|
[
"Jefferson Hernandez",
"Jing Shi",
"Simon Jenni",
"Vicente Ordonez",
"Kushal Kafle"
] |
Traditional alignment methods for Large Vision and Language Models (LVLMs) primarily rely on human-curated preference data. Human-generated preference data is costly; machine-generated preference data is limited in quality; and self-supervised preference data often introduces hallucinations. To overcome these limitations, we propose a novel Panel-of-Peers learning framework inspired by collaborative learning among humans. This approach leverages a panel of LVLMs, each evaluating and learning from their collective outputs through an iterative self-improvement process. By simulating a peer review system, our models generate, assess, and refine outputs in response to a curated set of prompts, mimicking a classroom learning environment. We demonstrate that this methodology enhances model performance without requiring extensive human-labeled datasets. Our experiments show significant improvement across multiple benchmarks, demonstrating the potential of peer evaluations as a scalable alternative to self-supervised alignment. Notably, we show that Panel-of-Peers increases the average score on fifteen benchmarks from 48% to 57%
|
|
2025-09-03T00:00:00 |
2509.01250
|
Towards More Diverse and Challenging Pre-training for Point Cloud Learning: Self-Supervised Cross Reconstruction with Decoupled Views
|
[
"Xiangdong Zhang",
"Shaofeng Zhang",
"Junchi Yan"
] |
https://github.com/aHapBean/Point-PQAE
|
Point cloud learning, especially in a self-supervised way without manual labels, has gained growing attention in both vision and learning communities due to its potential utility in a wide range of applications. Most existing generative approaches for point cloud self-supervised learning focus on recovering masked points from visible ones within a single view. Recognizing that a two-view pre-training paradigm inherently introduces greater diversity and variance, it may thus enable more challenging and informative pre-training. Inspired by this, we explore the potential of two-view learning in this domain. In this paper, we propose Point-PQAE, a cross-reconstruction generative paradigm that first generates two decoupled point clouds/views and then reconstructs one from the other. To achieve this goal, we develop a crop mechanism for point cloud view generation for the first time and further propose a novel positional encoding to represent the 3D relative position between the two decoupled views. The cross-reconstruction significantly increases the difficulty of pre-training compared to self-reconstruction, which enables our method to surpass previous single-modal self-reconstruction methods in 3D self-supervised learning. Specifically, it outperforms the self-reconstruction baseline (Point-MAE) by 6.5%, 7.0%, and 6.7% in three variants of ScanObjectNN with the Mlp-Linear evaluation protocol. The code is available at https://github.com/aHapBean/Point-PQAE.
|
2025-09-03T00:00:00 |
2509.01055
|
VerlTool: Towards Holistic Agentic Reinforcement Learning with Tool Use
|
[
"Dongfu Jiang",
"Yi Lu",
"Zhuofeng Li",
"Zhiheng Lyu",
"Ping Nie",
"Haozhe Wang",
"Alex Su",
"Hui Chen",
"Kai Zou",
"Chao Du",
"Tianyu Pang",
"Wenhu Chen"
] |
https://github.com/TIGER-AI-Lab/verl-tool
|
Reinforcement Learning with Verifiable Rewards (RLVR) has demonstrated success in enhancing LLM reasoning capabilities, but remains limited to single-turn interactions without tool integration. While recent Agentic Reinforcement Learning with Tool use (ARLT) approaches have emerged to address multi-turn tool interactions, existing works develop task-specific codebases that suffer from fragmentation, synchronous execution bottlenecks, and limited extensibility across domains. These inefficiencies hinder broader community adoption and algorithmic innovation. We introduce VerlTool, a unified and modular framework that addresses these limitations through systematic design principles. VerlTool provides four key contributions: (1) upstream alignment with VeRL ensuring compatibility and simplified maintenance, (2) unified tool management via standardized APIs supporting diverse modalities including code execution, search, SQL databases, and vision processing, (3) asynchronous rollout execution achieving near 2times speedup by eliminating synchronization bottlenecks, and (4) comprehensive evaluation demonstrating competitive performance across 6 ARLT domains. Our framework formalizes ARLT as multi-turn trajectories with multi-modal observation tokens (text/image/video), extending beyond single-turn RLVR paradigms. We train and evaluate models on mathematical reasoning, knowledge QA, SQL generation, visual reasoning, web search, and software engineering tasks, achieving results comparable to specialized systems while providing unified training infrastructure. The modular plugin architecture enables rapid tool integration requiring only lightweight Python definitions, significantly reducing development overhead and providing a scalable foundation for tool-augmented RL research. Our code is open-sourced at https://github.com/TIGER-AI-Lab/verl-tool.
|
2025-09-03T00:00:00 |
2509.01052
|
FlashAdventure: A Benchmark for GUI Agents Solving Full Story Arcs in Diverse Adventure Games
|
[
"Jaewoo Ahn",
"Junseo Kim",
"Heeseung Yun",
"Jaehyeon Son",
"Dongmin Park",
"Jaewoong Cho",
"Gunhee Kim"
] |
GUI agents powered by LLMs show promise in interacting with diverse digital environments. Among these, video games offer a valuable testbed due to their varied interfaces, with adventure games posing additional challenges through complex, narrative-driven interactions. Existing game benchmarks, however, lack diversity and rarely evaluate agents on completing entire storylines. To address this, we introduce FlashAdventure, a benchmark of 34 Flash-based adventure games designed to test full story arc completion and tackle the observation-behavior gap: the challenge of remembering and acting on earlier gameplay information. We also propose CUA-as-a-Judge, an automated gameplay evaluator, and COAST, an agentic framework leveraging long-term clue memory to better plan and solve sequential tasks. Experiments show current GUI agents struggle with full story arcs, while COAST improves milestone completion by bridging the observation-behavior gap. Nonetheless, a marked discrepancy between humans and best-performing agents warrants continued research efforts to narrow this divide.
|
|
2025-09-03T00:00:00 |
2509.00578
|
C-DiffDet+: Fusing Global Scene Context with Generative Denoising for High-Fidelity Object Detection
|
[
"Abdellah Zakaria Sellam",
"Ilyes Benaissa",
"Salah Eddine Bekhouche",
"Abdenour Hadid",
"Vito Renó",
"Cosimo Distante"
] |
Fine-grained object detection in challenging visual domains, such as vehicle damage assessment, presents a formidable challenge even for human experts to resolve reliably. While DiffusionDet has advanced the state-of-the-art through conditional denoising diffusion, its performance remains limited by local feature conditioning in context-dependent scenarios. We address this fundamental limitation by introducing Context-Aware Fusion (CAF), which leverages cross-attention mechanisms to integrate global scene context with local proposal features directly. The global context is generated using a separate dedicated encoder that captures comprehensive environmental information, enabling each object proposal to attend to scene-level understanding. Our framework significantly enhances the generative detection paradigm by enabling each object proposal to attend to comprehensive environmental information. Experimental results demonstrate an improvement over state-of-the-art models on the CarDD benchmark, establishing new performance benchmarks for context-aware object detection in fine-grained domains
|
|
2025-09-03T00:00:00 |
2509.02208
|
Baichuan-M2: Scaling Medical Capability with Large Verifier System
|
[
"Baichuan-M2 Team",
"Chengfeng Dou",
"Chong Liu",
"Fan Yang",
"Fei Li",
"Jiyuan Jia",
"Mingyang Chen",
"Qiang Ju",
"Shuai Wang",
"Shunya Dang",
"Tianpeng Li",
"Xiangrong Zeng",
"Yijie Zhou",
"Chenzheng Zhu",
"Da Pan",
"Fei Deng",
"Guangwei Ai",
"Guosheng Dong",
"Hongda Zhang",
"Jinyang Tai",
"Jixiang Hong",
"Kai Lu",
"Linzhuang Sun",
"Peidong Guo",
"Qian Ma",
"Rihui Xin",
"Shihui Yang",
"Shusen Zhang",
"Yichuan Mo",
"Zheng Liang",
"Zhishou Zhang",
"Hengfu Cui",
"Zuyi Zhu",
"Xiaochuan Wang"
] |
As large language models (LLMs) advance in conversational and reasoning capabilities, their practical application in healthcare has become a critical research focus. However, there is a notable gap between the performance of medical LLMs on static benchmarks such as USMLE and their utility in real-world clinical decision-making. This discrepancy arises because traditional exams fail to capture the dynamic, interactive nature of medical consultations. To address this challenge, we introduce a novel dynamic verification framework that moves beyond static answer verifier, establishing a large-scale, high-fidelity interactive reinforcement learning system. Our framework comprises two key components: a Patient Simulator that creates realistic clinical environments using de-identified medical records, and a Clinical Rubrics Generator that dynamically produces multi-dimensional evaluation metrics. Building on this foundation, we develop Baichuan-M2, a 32B-parameter medical augmented reasoning model trained through a multi-stage reinforcement learning strategy with an improved Group Relative Policy Optimization (GRPO) algorithm. Evaluated on HealthBench, Baichuan-M2 outperforms all other open-source models and most advanced closed-source counterparts, achieving a score above 32 on the challenging HealthBench Hard benchmark-previously exceeded only by GPT-5. Our work demonstrates that robust dynamic verifier system is essential for aligning LLM capabilities with practical clinical applications, establishing a new Pareto front in the performance-parameter trade-off for medical AI deployment.
|
|
2025-09-03T00:00:00 |
2509.02460
|
GenCompositor: Generative Video Compositing with Diffusion Transformer
|
[
"Shuzhou Yang",
"Xiaoyu Li",
"Xiaodong Cun",
"Guangzhi Wang",
"Lingen Li",
"Ying Shan",
"Jian Zhang"
] |
Video compositing combines live-action footage to create video production, serving as a crucial technique in video creation and film production. Traditional pipelines require intensive labor efforts and expert collaboration, resulting in lengthy production cycles and high manpower costs. To address this issue, we automate this process with generative models, called generative video compositing. This new task strives to adaptively inject identity and motion information of foreground video to the target video in an interactive manner, allowing users to customize the size, motion trajectory, and other attributes of the dynamic elements added in final video. Specifically, we designed a novel Diffusion Transformer (DiT) pipeline based on its intrinsic properties. To maintain consistency of the target video before and after editing, we revised a light-weight DiT-based background preservation branch with masked token injection. As to inherit dynamic elements from other sources, a DiT fusion block is proposed using full self-attention, along with a simple yet effective foreground augmentation for training. Besides, for fusing background and foreground videos with different layouts based on user control, we developed a novel position embedding, named Extended Rotary Position Embedding (ERoPE). Finally, we curated a dataset comprising 61K sets of videos for our new task, called VideoComp. This data includes complete dynamic elements and high-quality target videos. Experiments demonstrate that our method effectively realizes generative video compositing, outperforming existing possible solutions in fidelity and consistency.
|
|
2025-09-03T00:00:00 |
2509.01363
|
Reasoning Vectors: Transferring Chain-of-Thought Capabilities via Task Arithmetic
|
[
"Mohammad Zbeeb",
"Hasan Abed Al Kader Hammoud",
"Bernard Ghanem"
] |
Large language models often require costly optimization, such as reinforcement learning, to master complex reasoning tasks. This work demonstrates that reasoning ability, once learned, can be extracted and transferred between models as a compact task vector. We source two publicly available, identically initialized Qwen2.5 models, one fine-tuned with supervised fine-tuning (SFT) and the other with group relative policy optimization (GRPO) on the same dataset. From these, we extract a reasoning vector: v_{reason} = theta_{GRPO} - theta_{SFT}. We hypothesize that this vector captures the reasoning capability instilled by reinforcement learning while factoring out shared knowledge from the SFT process. When added to compatible instruction-tuned models through simple arithmetic, this vector consistently improves performance across diverse reasoning benchmarks: GSM8K (+4.9%), HumanEval (+4.3%), SciQ (+1.7%), and BigBenchHard (+12.3% for the 1.5B model). The performance improvements persist under adversarial conditions. Conversely, subtracting the vector causes significant performance degradation (-11.8% on GSM8K), demonstrating the vector's strong contribution to the model's reasoning abilities. This work shows how reasoning capabilities, typically developed through expensive training, can be extracted from existing open-source models and reused through simple tensor arithmetic, offering a practical way to enhance models by recycling prior computational investments.
|
|
2025-09-03T00:00:00 |
2509.02133
|
AMBEDKAR-A Multi-level Bias Elimination through a Decoding Approach with Knowledge Augmentation for Robust Constitutional Alignment of Language Models
|
[
"Snehasis Mukhopadhyay",
"Aryan Kasat",
"Shivam Dubey",
"Rahul Karthikeyan",
"Dhruv Sood",
"Vinija Jain",
"Aman Chadha",
"Amitava Das"
] |
Large Language Models (LLMs) can inadvertently reflect societal biases present in their training data, leading to harmful or prejudiced outputs. In the Indian context, our empirical evaluations across a suite of models reveal that biases around caste and religion are particularly salient. Yet, most existing mitigation strategies are Western-centric and fail to address these local nuances. We propose AMBEDKAR, a framework inspired by the egalitarian vision of Dr B. R. Ambedkar, architect of the Indian Constitution, to guide LLM outputs toward fairness, neutrality, and inclusion in line with Articles 14 to 17. Our approach introduces a Constitution-Aware Decoding Layer, guided by the AI Constitution of India and applied only at inference time, without any parameter updates to the base model. We incorporate a speculative decoding algorithm that proactively reduces casteist and communal bias during generation. This mitigation layer operates directly within the decoding process, avoiding changes to model internals and lowering the computational and infrastructural costs associated with retraining. We reinterpret speculative decoding not merely as an efficiency tool but as a mechanism for fairness. In this framework, a Small Language Model (SLM) acts as a potentially biased generator, while a constitutionally guided Large Language Model (LLM) serves as the verifier. Rather than accelerating generation, the LLM enforces bias-robust trajectories in the SLM outputs. This inversion of roles gives rise to a fairness-by-speculation paradigm. Our approach yields an absolute reduction of bias up to 26.41 percent compared to baseline. Our source code, datasets, and results are available at https://anonymous.4open.science/r/AMBEDKAR-983B/
|
|
2025-09-03T00:00:00 |
2509.00581
|
SQL-of-Thought: Multi-agentic Text-to-SQL with Guided Error Correction
|
[
"Saumya Chaturvedi",
"Aman Chadha",
"Laurent Bindschaedler"
] |
Converting natural language queries into SQL queries is a crucial challenge in both industry and academia, aiming to increase access to databases and large-scale applications. This work examines how in-context learning and chain-of-thought can be utilized to develop a robust solution for text-to-SQL systems. We propose SQL-of-Thought: a multi-agent framework that decomposes the Text2SQL task into schema linking, subproblem identification, query plan generation, SQL generation, and a guided correction loop. Unlike prior systems that rely only on execution-based static correction, we introduce taxonomy-guided dynamic error modification informed by in-context learning. SQL-of-Thought achieves state-of-the-art results on the Spider dataset and its variants, combining guided error taxonomy with reasoning-based query planning.
|
|
2025-09-03T00:00:00 |
2509.00404
|
Metis: Training Large Language Models with Advanced Low-Bit Quantization
|
[
"Hengjie Cao",
"Mengyi Chen",
"Yifeng Yang",
"Ruijun Huang",
"Fang Dong",
"Jixian Zhou",
"Anrui Chen",
"Mingzhi Dong",
"Yujiang Wang",
"Jinlong Hou",
"Yuan Cheng",
"Fan Wu",
"Fan Yang",
"Tun Lu",
"Ning Gu",
"Li Shang"
] |
https://github.com/typename-yyf/Metis-quantization
|
This work identifies anisotropic parameter distributions as a fundamental barrier to training large language models (LLMs) with low-bit quantization: a few dominant singular values create wide numerical ranges that conflict with the inherent bias of block-wise quantization. This bias disproportionately preserves high-magnitude values while discarding smaller ones, causing training instability and low model performance. This work introduces Metis, a training framework that combines (i) spectral decomposition with random embedding to efficiently disentangle dominant from long-tail components, compressing broad distributions into quantization-friendly narrow ranges; (ii) adaptive learning rates in the spectral domain to amplify underrepresented directions and better capture diverse features critical for performance; and (iii) a dual-range regularizer that jointly constrains numerical precision and parameter range distribution, ensuring stable, unbiased low-bit training. With Metis, FP8 training surpasses FP32 baselines, and FP4 training achieves accuracy comparable to FP32, paving the way for robust and scalable LLM training under advanced low-bit quantization. The code implementation for Metis is available at: https://github.com/typename-yyf/Metis-quantization.
|
2025-09-03T00:00:00 |
2509.02046
|
Fantastic Pretraining Optimizers and Where to Find Them
|
[
"Kaiyue Wen",
"David Hall",
"Tengyu Ma",
"Percy Liang"
] |
AdamW has long been the dominant optimizer in language model pretraining, despite numerous claims that alternative optimizers offer 1.4 to 2x speedup. We posit that two methodological shortcomings have obscured fair comparisons and hindered practical adoption: (i) unequal hyperparameter tuning and (ii) limited or misleading evaluation setups. To address these two issues, we conduct a systematic study of ten deep learning optimizers across four model scales (0.1B-1.2B parameters) and data-to-model ratios (1-8x the Chinchilla optimum). We find that fair and informative comparisons require rigorous hyperparameter tuning and evaluations across a range of model scales and data-to-model ratios, performed at the end of training. First, optimal hyperparameters for one optimizer may be suboptimal for another, making blind hyperparameter transfer unfair. Second, the actual speedup of many proposed optimizers over well-tuned baselines is lower than claimed and decreases with model size to only 1.1x for 1.2B parameter models. Thirdly, comparing intermediate checkpoints before reaching the target training budgets can be misleading, as rankings between two optimizers can flip during training due to learning rate decay. Through our thorough investigation, we find that all the fastest optimizers such as Muon and Soap, use matrices as preconditioners -- multiplying gradients with matrices rather than entry-wise scalars. However, the speedup of matrix-based optimizers is inversely proportional to model scale, decreasing from 1.4x over AdamW for 0.1B parameter models to merely 1.1x for 1.2B parameter models.
|
|
2025-09-03T00:00:00 |
2509.00425
|
The Gold Medals in an Empty Room: Diagnosing Metalinguistic Reasoning in LLMs with Camlang
|
[
"Fenghua Liu",
"Yulong Chen",
"Yixuan Liu",
"Zhujun Jin",
"Solomon Tsai",
"Ming Zhong"
] |
Large Language Models (LLMs) achieve gold-medal performance across many benchmarks, yet it remains unclear whether such success reflects genuine reasoning or pattern matching. From a cognitive science perspective, an informative test is whether models can master an unfamiliar language through explicit metalinguistic deductive learning, a paradigm where human learners can reliably internalise grammatical systems through metalinguistic reasoning. We address this question with Camlang, a novel constructed language that exhibits naturalistic yet unattested feature combinations. Camlang consists of two explicit resources, a grammar book and a bilingual dictionary, which mirror adult second-language learning via explicit grammar rules and lexical lookup, and enable us to disentangle errors in morpho-syntax, lexical semantics, and sentence-level reasoning. Human experiments show that these resources are sufficient for participants to acquire Camlang and successfully solve Camlang tasks. To operationalise evaluation, we adapt CommonsenseQA into Camlang, creating Camlang-CSQA-v0, the first task in a broader suite where solving questions requires applying grammar rules and lexical mappings. Experimental results show that GPT-5 achieves 98\% EM accuracy in English but only 47\% in Camlang, far below human performance at 87\%, while other state-of-the-art reasoning LLMs perform even worse. Human verification further reveals that most model successes stem from shallow lexical alignment while GPT-5 shows emerging metalinguistic awareness to a limited extent but not systematic grammatical mastery as humans. Camlang establishes a cognitively grounded evaluation paradigm that exposes fundamental gaps between current models and human metalinguistic competence.
|
|
2025-09-03T00:00:00 |
2509.00531
|
MobiAgent: A Systematic Framework for Customizable Mobile Agents
|
[
"Cheng Zhang",
"Erhu Feng",
"Xi Zhao",
"Yisheng Zhao",
"Wangbo Gong",
"Jiahui Sun",
"Dong Du",
"Zhichao Hua",
"Yubin Xia",
"Haibo Chen"
] |
With the rapid advancement of Vision-Language Models (VLMs), GUI-based mobile agents have emerged as a key development direction for intelligent mobile systems. However, existing agent models continue to face significant challenges in real-world task execution, particularly in terms of accuracy and efficiency. To address these limitations, we propose MobiAgent, a comprehensive mobile agent system comprising three core components: the MobiMind-series agent models, the AgentRR acceleration framework, and the MobiFlow benchmarking suite. Furthermore, recognizing that the capabilities of current mobile agents are still limited by the availability of high-quality data, we have developed an AI-assisted agile data collection pipeline that significantly reduces the cost of manual annotation. Compared to both general-purpose LLMs and specialized GUI agent models, MobiAgent achieves state-of-the-art performance in real-world mobile scenarios.
|
|
2025-09-03T00:00:00 |
2509.01584
|
ViSTA-SLAM: Visual SLAM with Symmetric Two-view Association
|
[
"Ganlin Zhang",
"Shenhan Qian",
"Xi Wang",
"Daniel Cremers"
] |
https://github.com/zhangganlin/vista-slam
|
We present ViSTA-SLAM as a real-time monocular visual SLAM system that operates without requiring camera intrinsics, making it broadly applicable across diverse camera setups. At its core, the system employs a lightweight symmetric two-view association (STA) model as the frontend, which simultaneously estimates relative camera poses and regresses local pointmaps from only two RGB images. This design reduces model complexity significantly, the size of our frontend is only 35\% that of comparable state-of-the-art methods, while enhancing the quality of two-view constraints used in the pipeline. In the backend, we construct a specially designed Sim(3) pose graph that incorporates loop closures to address accumulated drift. Extensive experiments demonstrate that our approach achieves superior performance in both camera tracking and dense 3D reconstruction quality compared to current methods. Github repository: https://github.com/zhangganlin/vista-slam
|
2025-09-03T00:00:00 |
2509.01440
|
Benchmarking Optimizers for Large Language Model Pretraining
|
[
"Andrei Semenov",
"Matteo Pagliardini",
"Martin Jaggi"
] |
The recent development of Large Language Models (LLMs) has been accompanied by an effervescence of novel ideas and methods to better optimize the loss of deep learning models. Claims from those methods are myriad: from faster convergence to removing reliance on certain hyperparameters. However, the diverse experimental protocols used to validate these claims make direct comparisons between methods challenging. This study presents a comprehensive evaluation of recent optimization techniques across standardized LLM pretraining scenarios, systematically varying model size, batch size, and training duration. Through careful tuning of each method, we provide guidance to practitioners on which optimizer is best suited for each scenario. For researchers, our work highlights promising directions for future optimization research. Finally, by releasing our code and making all experiments fully reproducible, we hope our efforts can help the development and rigorous benchmarking of future methods.
|
|
2025-09-03T00:00:00 |
2509.02333
|
DCPO: Dynamic Clipping Policy Optimization
|
[
"Shihui Yang",
"Chengfeng Dou",
"Peidong Guo",
"Kai Lu",
"Qiang Ju",
"Fei Deng",
"Rihui Xin"
] |
Reinforcement Learning from Verifiable Rewards (RLVR) has emerged as a promising framework for enhancing the reasoning capabilities of large language models. However, existing approaches such as GRPO often suffer from zero gradients. This problem arises primarily due to fixed clipping bounds for token-level probability ratios and the standardization of identical rewards, which can lead to ineffective gradient updates and underutilization of generated responses. In this work, we propose Dynamic Clipping Policy Optimization (DCPO), which introduces a dynamic clipping strategy that adaptively adjusts the clipping bounds based on token-specific prior probabilities to enhance token-level exploration, and a smooth advantage standardization technique that standardizes rewards across cumulative training steps to improve the response-level effective utilization of generated responses. DCPO achieved state-of-the-art performance on four benchmarks based on four different models. In particular, DCPO achieved an Avg@1 of 46.7 under greedy decoding and an Avg@32 of 38.8 under 32 times sampling on the AIME24 benchmark, surpassing both DAPO (36.7/31.6) and GRPO (36.7/32.1) on the Qwen2.5-Math-7B model. On the AIME25 benchmark based on Qwen2.5-14B, DCPO achieves a performance of (23.3/19.0), surpassing GRPO (13.3/10.5) and DAPO (20.0/15.3). Furthermore, DCPO achieved an average 28% improvement in the nonzero advantage over GRPO in four models, doubled the training efficiency over DAPO, and significantly reduced the token clipping ratio by an order of magnitude compared to both GRPO and DAPO, while achieving superior performance. These results highlight DCPO's effectiveness in leveraging generated data more efficiently for reinforcement learning in large language models.
|
|
2025-09-03T00:00:00 |
2508.20586
|
FastFit: Accelerating Multi-Reference Virtual Try-On via Cacheable Diffusion Models
|
[
"Zheng Chong",
"Yanwei Lei",
"Shiyue Zhang",
"Zhuandi He",
"Zhen Wang",
"Xujie Zhang",
"Xiao Dong",
"Yiling Wu",
"Dongmei Jiang",
"Xiaodan Liang"
] |
Despite its great potential, virtual try-on technology is hindered from real-world application by two major challenges: the inability of current methods to support multi-reference outfit compositions (including garments and accessories), and their significant inefficiency caused by the redundant re-computation of reference features in each denoising step. To address these challenges, we propose FastFit, a high-speed multi-reference virtual try-on framework based on a novel cacheable diffusion architecture. By employing a Semi-Attention mechanism and substituting traditional timestep embeddings with class embeddings for reference items, our model fully decouples reference feature encoding from the denoising process with negligible parameter overhead. This allows reference features to be computed only once and losslessly reused across all steps, fundamentally breaking the efficiency bottleneck and achieving an average 3.5x speedup over comparable methods. Furthermore, to facilitate research on complex, multi-reference virtual try-on, we introduce DressCode-MR, a new large-scale dataset. It comprises 28,179 sets of high-quality, paired images covering five key categories (tops, bottoms, dresses, shoes, and bags), constructed through a pipeline of expert models and human feedback refinement. Extensive experiments on the VITON-HD, DressCode, and our DressCode-MR datasets show that FastFit surpasses state-of-the-art methods on key fidelity metrics while offering its significant advantage in inference efficiency.
|
|
2025-09-03T00:00:00 |
2508.21496
|
ELV-Halluc: Benchmarking Semantic Aggregation Hallucinations in Long Video Understanding
|
[
"Hao Lu",
"Jiahao Wang",
"Yaolun Zhang",
"Ruohui Wang",
"Xuanyu Zheng",
"Yepeng Tang",
"Dahua Lin",
"Lewei Lu"
] |
Video multimodal large language models (Video-MLLMs) have achieved remarkable progress in video understanding. However, they remain vulnerable to hallucination-producing content inconsistent with or unrelated to video inputs. Previous video hallucination benchmarks primarily focus on short-videos. They attribute hallucinations to factors such as strong language priors, missing frames, or vision-language biases introduced by the visual encoder. While these causes indeed account for most hallucinations in short videos, they still oversimplify the cause of hallucinations. Sometimes, models generate incorrect outputs but with correct frame-level semantics. We refer to this type of hallucination as Semantic Aggregation Hallucination (SAH), which arises during the process of aggregating frame-level semantics into event-level semantic groups. Given that SAH becomes particularly critical in long videos due to increased semantic complexity across multiple events, it is essential to separate and thoroughly investigate the causes of this type of hallucination. To address the above issues, we introduce ELV-Halluc, the first benchmark dedicated to long-video hallucination, enabling a systematic investigation of SAH. Our experiments confirm the existence of SAH and show that it increases with semantic complexity. Additionally, we find that models are more prone to SAH on rapidly changing semantics. Moreover, we discuss potential approaches to mitigate SAH. We demonstrate that positional encoding strategy contributes to alleviating SAH, and further adopt DPO strategy to enhance the model's ability to distinguish semantics within and across events. To support this, we curate a dataset of 8K adversarial data pairs and achieve improvements on both ELV-Halluc and Video-MME, including a substantial 27.7% reduction in SAH ratio.
|
|
2025-09-03T00:00:00 |
2509.02563
|
DynaGuard: A Dynamic Guardrail Model With User-Defined Policies
|
[
"Monte Hoover",
"Vatsal Baherwani",
"Neel Jain",
"Khalid Saifullah",
"Joseph Vincent",
"Chirag Jain",
"Melissa Kazemi Rad",
"C. Bayan Bruss",
"Ashwinee Panda",
"Tom Goldstein"
] |
Guardian models are used to supervise and moderate the outputs of user-facing chatbots, enforcing guardrails and detecting bad behaviors. Standard guardian models like LlamaGuard detect predefined, static categories of harms. We propose dynamic guardian models that evaluate text based on user-defined policies, making them useful for different application domains that are not addressed by standard guardian models. Our dynamic guardian models can be used for fast detection of policy violations or with chain-of-thought reasoning that articulates and justifies the model outputs. Our dynamic guardian models match static models in detection accuracy for static harm categories while identifying violations of free-form policies with accuracy comparable to frontier reasoning models in a fraction of the time.
|
|
2025-09-03T00:00:00 |
2508.21334
|
Stairway to Fairness: Connecting Group and Individual Fairness
|
[
"Theresia Veronika Rampisela",
"Maria Maistro",
"Tuukka Ruotsalo",
"Falk Scholer",
"Christina Lioma"
] |
https://github.com/theresiavr/stairway-to-fairness
|
Fairness in recommender systems (RSs) is commonly categorised into group fairness and individual fairness. However, there is no established scientific understanding of the relationship between the two fairness types, as prior work on both types has used different evaluation measures or evaluation objectives for each fairness type, thereby not allowing for a proper comparison of the two. As a result, it is currently not known how increasing one type of fairness may affect the other. To fill this gap, we study the relationship of group and individual fairness through a comprehensive comparison of evaluation measures that can be used for both fairness types. Our experiments with 8 runs across 3 datasets show that recommendations that are highly fair for groups can be very unfair for individuals. Our finding is novel and useful for RS practitioners aiming to improve the fairness of their systems. Our code is available at: https://github.com/theresiavr/stairway-to-fairness.
|
2025-09-03T00:00:00 |
2509.00605
|
Gated Associative Memory: A Parallel O(N) Architecture for Efficient Sequence Modeling
|
[
"Rishiraj Acharya"
] |
The Transformer architecture, underpinned by the self-attention mechanism, has become the de facto standard for sequence modeling tasks. However, its core computational primitive scales quadratically with sequence length (O(N^2)), creating a significant bottleneck for processing long contexts. In this paper, we propose the Gated Associative Memory (GAM) network, a novel, fully parallel architecture for sequence modeling that exhibits linear complexity (O(N)) with respect to sequence length. The GAM block replaces the self-attention layer with two parallel pathways: a causal convolution to efficiently capture local, position-dependent context, and a parallel associative memory retrieval mechanism to model global, content-based patterns. These pathways are dynamically fused using a gating mechanism, allowing the model to flexibly combine local and global information for each token. We implement GAM from scratch and conduct a rigorous comparative analysis against a standard Transformer model and a modern linear-time baseline (Mamba) on the WikiText-2 benchmark, as well as against the Transformer on the TinyStories dataset. Our experiments demonstrate that GAM is consistently faster, outperforming both baselines on training speed, and achieves a superior or competitive final validation perplexity across all datasets, establishing it as a promising and efficient alternative for sequence modeling.
|
|
2025-09-03T00:00:00 |
2509.01790
|
Flaw or Artifact? Rethinking Prompt Sensitivity in Evaluating LLMs
|
[
"Andong Hua",
"Kenan Tang",
"Chenhe Gu",
"Jindong Gu",
"Eric Wong",
"Yao Qin"
] |
Prompt sensitivity, referring to the phenomenon where paraphrasing (i.e., repeating something written or spoken using different words) leads to significant changes in large language model (LLM) performance, has been widely accepted as a core limitation of LLMs. In this work, we revisit this issue and ask: Is the widely reported high prompt sensitivity truly an inherent weakness of LLMs, or is it largely an artifact of evaluation processes? To answer this question, we systematically evaluate 7 LLMs (e.g., GPT and Gemini family) across 6 benchmarks, including both multiple-choice and open-ended tasks on 12 diverse prompt templates. We find that much of the prompt sensitivity stems from heuristic evaluation methods, including log-likelihood scoring and rigid answer matching, which often overlook semantically correct responses expressed through alternative phrasings, such as synonyms or paraphrases. When we adopt LLM-as-a-Judge evaluations, we observe a substantial reduction in performance variance and a consistently higher correlation in model rankings across prompts. Our findings suggest that modern LLMs are more robust to prompt templates than previously believed, and that prompt sensitivity may be more an artifact of evaluation than a flaw in the models.
|
|
2025-09-03T00:00:00 |
2509.02523
|
Flavors of Moonshine: Tiny Specialized ASR Models for Edge Devices
|
[
"Evan King",
"Adam Sabra",
"Manjunath Kudlur",
"James Wang",
"Pete Warden"
] |
We present the Flavors of Moonshine, a suite of tiny automatic speech recognition (ASR) models specialized for a range of underrepresented languages. Prevailing wisdom suggests that multilingual ASR models outperform monolingual counterparts by exploiting cross-lingual phonetic similarities. We challenge this assumption, showing that for sufficiently small models (27M parameters), training monolingual systems on a carefully balanced mix of high-quality human-labeled, pseudo-labeled, and synthetic data yields substantially superior performance. On average, our models achieve error rates 48% lower than the comparably sized Whisper Tiny model, outperform the 9x larger Whisper Small model, and in most cases match or outperform the 28x larger Whisper Medium model. These results advance the state of the art for models of this size, enabling accurate on-device ASR for languages that previously had limited support. We release Arabic, Chinese, Japanese, Korean, Ukrainian, and Vietnamese Moonshine models under a permissive open-source license.
|
|
2025-09-03T00:00:00 |
2508.21038
|
On the Theoretical Limitations of Embedding-Based Retrieval
|
[
"Orion Weller",
"Michael Boratko",
"Iftekhar Naim",
"Jinhyuk Lee"
] |
Vector embeddings have been tasked with an ever-increasing set of retrieval tasks over the years, with a nascent rise in using them for reasoning, instruction-following, coding, and more. These new benchmarks push embeddings to work for any query and any notion of relevance that could be given. While prior works have pointed out theoretical limitations of vector embeddings, there is a common assumption that these difficulties are exclusively due to unrealistic queries, and those that are not can be overcome with better training data and larger models. In this work, we demonstrate that we may encounter these theoretical limitations in realistic settings with extremely simple queries. We connect known results in learning theory, showing that the number of top-k subsets of documents capable of being returned as the result of some query is limited by the dimension of the embedding. We empirically show that this holds true even if we restrict to k=2, and directly optimize on the test set with free parameterized embeddings. We then create a realistic dataset called LIMIT that stress tests models based on these theoretical results, and observe that even state-of-the-art models fail on this dataset despite the simple nature of the task. Our work shows the limits of embedding models under the existing single vector paradigm and calls for future research to develop methods that can resolve this fundamental limitation.
|
|
2025-09-04T00:00:00 |
2509.01106
|
Robix: A Unified Model for Robot Interaction, Reasoning and Planning
|
[
"Huang Fang",
"Mengxi Zhang",
"Heng Dong",
"Wei Li",
"Zixuan Wang",
"Qifeng Zhang",
"Xueyun Tian",
"Yucheng Hu",
"Hang Li"
] |
We introduce Robix, a unified model that integrates robot reasoning, task planning, and natural language interaction within a single vision-language architecture. Acting as the high-level cognitive layer in a hierarchical robot system, Robix dynamically generates atomic commands for the low-level controller and verbal responses for human interaction, enabling robots to follow complex instructions, plan long-horizon tasks, and interact naturally with human within an end-to-end framework. Robix further introduces novel capabilities such as proactive dialogue, real-time interruption handling, and context-aware commonsense reasoning during task execution. At its core, Robix leverages chain-of-thought reasoning and adopts a three-stage training strategy: (1) continued pretraining to enhance foundational embodied reasoning abilities including 3D spatial understanding, visual grounding, and task-centric reasoning; (2) supervised finetuning to model human-robot interaction and task planning as a unified reasoning-action sequence; and (3) reinforcement learning to improve reasoning-action consistency and long-horizon task coherence. Extensive experiments demonstrate that Robix outperforms both open-source and commercial baselines (e.g., GPT-4o and Gemini 2.5 Pro) in interactive task execution, demonstrating strong generalization across diverse instruction types (e.g., open-ended, multi-stage, constrained, invalid, and interrupted) and various user-involved tasks such as table bussing, grocery shopping, and dietary filtering.
|
|
2025-09-04T00:00:00 |
2509.00375
|
Open Data Synthesis For Deep Research
|
[
"Ziyi Xia",
"Kun Luo",
"Hongjin Qian",
"Zheng Liu"
] |
https://github.com/VectorSpaceLab/InfoSeek{this
|
Large language models (LLMs) are increasingly expected to go beyond simple factual queries toward Deep Research-tasks that require decomposing questions into sub-problems, coordinating multi-step reasoning, and synthesizing evidence from diverse sources. We formalize Deep Research tasks with verifiable answers as Hierarchical Constraint Satisfaction Problems (HCSPs), which are fundamentally different from single-constraint, multi-hop, or flat CSP formulations. However, existing benchmarks (e.g., Natural Questions, HotpotQA) fail to capture this complexity, while recent synthetic datasets often introduce shortcut reasoning, knowledge leakage, or lack sufficient structural depth. To address this gap, we introduce InfoSeek, a scalable framework for synthesizing complex Deep Research tasks. InfoSeek uses a dual-agent system to recursively build a Research Tree from large-scale webpages, blurring intermediate nodes into valid sub-problems, and converting these trees into natural language questions that require traversing the full hierarchy. It also enables rapid scaling, yielding over 50K training examples, a curated test set, and reasoning trajectories generated via reject sampling. Experiments show that models trained on InfoSeek consistently outperform strong baselines. On a challenging benchmark BrowseComp-Plus, 3B LLMs optimized with InfoSeek surpass much larger 32B models and lightweight commercial APIs (e.g., Gemini2.5-Flash), while achieving performance comparable to stronger APIs (e.g., Gemini2.5-Pro). By preserving meta-information such as intermediate steps and retrieval labels, InfoSeek further supports advanced optimization strategies, including compound reward design and trajectory-level exploration. We provide our codes and datasets in https://github.com/VectorSpaceLab/InfoSeek{this repository}.
|
2025-09-04T00:00:00 |
2509.01977
|
MOSAIC: Multi-Subject Personalized Generation via Correspondence-Aware Alignment and Disentanglement
|
[
"Dong She",
"Siming Fu",
"Mushui Liu",
"Qiaoqiao Jin",
"Hualiang Wang",
"Mu Liu",
"Jidong Jiang"
] |
Multi-subject personalized generation presents unique challenges in maintaining identity fidelity and semantic coherence when synthesizing images conditioned on multiple reference subjects. Existing methods often suffer from identity blending and attribute leakage due to inadequate modeling of how different subjects should interact within shared representation spaces. We present MOSAIC, a representation-centric framework that rethinks multi-subject generation through explicit semantic correspondence and orthogonal feature disentanglement. Our key insight is that multi-subject generation requires precise semantic alignment at the representation level - knowing exactly which regions in the generated image should attend to which parts of each reference. To enable this, we introduce SemAlign-MS, a meticulously annotated dataset providing fine-grained semantic correspondences between multiple reference subjects and target images, previously unavailable in this domain. Building on this foundation, we propose the semantic correspondence attention loss to enforce precise point-to-point semantic alignment, ensuring high consistency from each reference to its designated regions. Furthermore, we develop the multi-reference disentanglement loss to push different subjects into orthogonal attention subspaces, preventing feature interference while preserving individual identity characteristics. Extensive experiments demonstrate that MOSAIC achieves state-of-the-art performance on multiple benchmarks. Notably, while existing methods typically degrade beyond 3 subjects, MOSAIC maintains high fidelity with 4+ reference subjects, opening new possibilities for complex multi-subject synthesis applications.
|
|
2025-09-04T00:00:00 |
2509.00428
|
Mixture of Global and Local Experts with Diffusion Transformer for Controllable Face Generation
|
[
"Xuechao Zou",
"Shun Zhang",
"Xing Fu",
"Yue Li",
"Kai Li",
"Yushe Cao",
"Congyan Lang",
"Pin Tao",
"Junliang Xing"
] |
https://github.com/XavierJiezou/Face-MoGLE
|
Controllable face generation poses critical challenges in generative modeling due to the intricate balance required between semantic controllability and photorealism. While existing approaches struggle with disentangling semantic controls from generation pipelines, we revisit the architectural potential of Diffusion Transformers (DiTs) through the lens of expert specialization. This paper introduces Face-MoGLE, a novel framework featuring: (1) Semantic-decoupled latent modeling through mask-conditioned space factorization, enabling precise attribute manipulation; (2) A mixture of global and local experts that captures holistic structure and region-level semantics for fine-grained controllability; (3) A dynamic gating network producing time-dependent coefficients that evolve with diffusion steps and spatial locations. Face-MoGLE provides a powerful and flexible solution for high-quality, controllable face generation, with strong potential in generative modeling and security applications. Extensive experiments demonstrate its effectiveness in multimodal and monomodal face generation settings and its robust zero-shot generalization capability. Project page is available at https://github.com/XavierJiezou/Face-MoGLE.
|
2025-09-04T00:00:00 |
2509.03405
|
LMEnt: A Suite for Analyzing Knowledge in Language Models from Pretraining Data to Representations
|
[
"Daniela Gottesman",
"Alon Gilae-Dotan",
"Ido Cohen",
"Yoav Gur-Arieh",
"Marius Mosbach",
"Ori Yoran",
"Mor Geva"
] |
Language models (LMs) increasingly drive real-world applications that require world knowledge. However, the internal processes through which models turn data into representations of knowledge and beliefs about the world, are poorly understood. Insights into these processes could pave the way for developing LMs with knowledge representations that are more consistent, robust, and complete. To facilitate studying these questions, we present LMEnt, a suite for analyzing knowledge acquisition in LMs during pretraining. LMEnt introduces: (1) a knowledge-rich pretraining corpus, fully annotated with entity mentions, based on Wikipedia, (2) an entity-based retrieval method over pretraining data that outperforms previous approaches by as much as 80.4%, and (3) 12 pretrained models with up to 1B parameters and 4K intermediate checkpoints, with comparable performance to popular open-sourced models on knowledge benchmarks. Together, these resources provide a controlled environment for analyzing connections between entity mentions in pretraining and downstream performance, and the effects of causal interventions in pretraining data. We show the utility of LMEnt by studying knowledge acquisition across checkpoints, finding that fact frequency is key, but does not fully explain learning trends. We release LMEnt to support studies of knowledge in LMs, including knowledge representations, plasticity, editing, attribution, and learning dynamics.
|
|
2025-09-04T00:00:00 |
2509.02722
|
Planning with Reasoning using Vision Language World Model
|
[
"Delong Chen",
"Theo Moutakanni",
"Willy Chung",
"Yejin Bang",
"Ziwei Ji",
"Allen Bolourchi",
"Pascale Fung"
] |
Effective planning requires strong world models, but high-level world models that can understand and reason about actions with semantic and temporal abstraction remain largely underdeveloped. We introduce the Vision Language World Model (VLWM), a foundation model trained for language-based world modeling on natural videos. Given visual observations, the VLWM first infers the overall goal achievements then predicts a trajectory composed of interleaved actions and world state changes. Those targets are extracted by iterative LLM Self-Refine conditioned on compressed future observations represented by Tree of Captions. The VLWM learns both an action policy and a dynamics model, which respectively facilitates reactive system-1 plan decoding and reflective system-2 planning via cost minimization. The cost evaluates the semantic distance between the hypothetical future states given by VLWM roll-outs and the expected goal state, and is measured by a critic model that we trained in a self-supervised manner. The VLWM achieves state-of-the-art Visual Planning for Assistance (VPA) performance on both benchmark evaluations and our proposed PlannerArena human evaluations, where system-2 improves the Elo score by +27% upon system-1. The VLWM models also outperforms strong VLM baselines on RoboVQA and WorldPrediction benchmark.
|
|
2025-09-04T00:00:00 |
2509.00930
|
SATQuest: A Verifier for Logical Reasoning Evaluation and Reinforcement Fine-Tuning of LLMs
|
[
"Yanxiao Zhao",
"Yaqian Li",
"Zihao Bo",
"Rinyoichi Takezoe",
"Haojia Hui",
"Mo Guang",
"Lei Ren",
"Xiaolin Qin",
"Kaiwen Long"
] |
Recent advances in Large Language Models (LLMs) have demonstrated remarkable general reasoning capabilities. However, systematically evaluating and enhancing these reasoning capabilities is challenging due to the lack of controllable and scalable tools for fine-grained analysis. Existing benchmarks and datasets often lack the necessary variable control for multi-dimensional, systematic analysis and training, or have narrow problem types and formats. To address these limitations, we introduce SATQuest, a systematic verifier designed to evaluate and enhance logical reasoning in LLMs by generating diverse, Satisfiability-based logical reasoning problems directly from Conjunctive Normal Form (CNF) instances. SATQuest structures these problems along three orthogonal dimensions: instance scale, problem type, and question format, employing randomized, SAT-based problem generation and objective answer verification via PySAT. This design mitigates memorization issues, allows for nuanced insights into reasoning performance, and enables effective reinforcement fine-tuning. Our extensive evaluation of various LLMs using SATQuest identified significant limitations in their logical reasoning, particularly in generalizing beyond familiar mathematical formats. Furthermore, we show that reinforcement fine-tuning with SATQuest rewards substantially improves targeted task performance and generalizes to more complex instances, while highlighting remaining challenges in cross-format adaptation. Through these demonstrations, we showcase SATQuest's potential as a foundational tool and a valuable starting point for advancing LLM logical reasoning.
|
|
2025-09-04T00:00:00 |
2509.02530
|
Manipulation as in Simulation: Enabling Accurate Geometry Perception in Robots
|
[
"Minghuan Liu",
"Zhengbang Zhu",
"Xiaoshen Han",
"Peng Hu",
"Haotong Lin",
"Xinyao Li",
"Jingxiao Chen",
"Jiafeng Xu",
"Yichu Yang",
"Yunfeng Lin",
"Xinghang Li",
"Yong Yu",
"Weinan Zhang",
"Tao Kong",
"Bingyi Kang"
] |
Modern robotic manipulation primarily relies on visual observations in a 2D color space for skill learning but suffers from poor generalization. In contrast, humans, living in a 3D world, depend more on physical properties-such as distance, size, and shape-than on texture when interacting with objects. Since such 3D geometric information can be acquired from widely available depth cameras, it appears feasible to endow robots with similar perceptual capabilities. Our pilot study found that using depth cameras for manipulation is challenging, primarily due to their limited accuracy and susceptibility to various types of noise. In this work, we propose Camera Depth Models (CDMs) as a simple plugin on daily-use depth cameras, which take RGB images and raw depth signals as input and output denoised, accurate metric depth. To achieve this, we develop a neural data engine that generates high-quality paired data from simulation by modeling a depth camera's noise pattern. Our results show that CDMs achieve nearly simulation-level accuracy in depth prediction, effectively bridging the sim-to-real gap for manipulation tasks. Notably, our experiments demonstrate, for the first time, that a policy trained on raw simulated depth, without the need for adding noise or real-world fine-tuning, generalizes seamlessly to real-world robots on two challenging long-horizon tasks involving articulated, reflective, and slender objects, with little to no performance degradation. We hope our findings will inspire future research in utilizing simulation data and 3D information in general robot policies.
|
|
2025-09-04T00:00:00 |
2509.03403
|
Beyond Correctness: Harmonizing Process and Outcome Rewards through RL Training
|
[
"Chenlu Ye",
"Zhou Yu",
"Ziji Zhang",
"Hao Chen",
"Narayanan Sadagopan",
"Jing Huang",
"Tong Zhang",
"Anurag Beniwal"
] |
https://github.com/Chenluye99/PROF
|
Reinforcement learning with verifiable rewards (RLVR) has emerged to be a predominant paradigm for mathematical reasoning tasks, offering stable improvements in reasoning ability. However, Outcome Reward Models (ORMs) in RLVR are too coarse-grained to distinguish flawed reasoning within correct answers or valid reasoning within incorrect answers. This lack of granularity introduces noisy and misleading gradients significantly and hinders further progress in reasoning process quality. While Process Reward Models (PRMs) offer fine-grained guidance for intermediate steps, they frequently suffer from inaccuracies and are susceptible to reward hacking. To resolve this dilemma, we introduce PRocess cOnsistency Filter (PROF), an effective data process curation method that harmonizes noisy, fine-grained process rewards with accurate, coarse-grained outcome rewards. Rather than naively blending PRM and ORM in the objective function (arXiv:archive/2506.18896), PROF leverages their complementary strengths through consistency-driven sample selection. Our approach retains correct responses with higher averaged process values and incorrect responses with lower averaged process values, while maintaining positive/negative training sample balance. Extensive experiments demonstrate that our method not only consistently improves the final accuracy over 4% compared to the blending approaches, but also strengthens the quality of intermediate reasoning steps. Codes and training recipes are available at https://github.com/Chenluye99/PROF.
|
2025-09-05T00:00:00 |
2509.01396
|
DeepResearch Arena: The First Exam of LLMs' Research Abilities via Seminar-Grounded Tasks
|
[
"Haiyuan Wan",
"Chen Yang",
"Junchi Yu",
"Meiqi Tu",
"Jiaxuan Lu",
"Di Yu",
"Jianbao Cao",
"Ben Gao",
"Jiaqing Xie",
"Aoran Wang",
"Wenlong Zhang",
"Philip Torr",
"Dongzhan Zhou"
] |
Deep research agents have attracted growing attention for their potential to orchestrate multi-stage research workflows, spanning literature synthesis, methodological design, and empirical verification. Despite these strides, evaluating their research capability faithfully is rather challenging due to the difficulty of collecting frontier research questions that genuinely capture researchers' attention and intellectual curiosity. To address this gap, we introduce DeepResearch Arena, a benchmark grounded in academic seminars that capture rich expert discourse and interaction, better reflecting real-world research environments and reducing the risk of data leakage. To automatically construct DeepResearch Arena, we propose a Multi-Agent Hierarchical Task Generation (MAHTG) system that extracts research-worthy inspirations from seminar transcripts. The MAHTG system further translates research-worthy inspirations into high-quality research tasks, ensuring the traceability of research task formulation while filtering noise. With the MAHTG system, we curate DeepResearch Arena with over 10,000 high-quality research tasks from over 200 academic seminars, spanning 12 disciplines, such as literature, history, and science. Our extensive evaluation shows that DeepResearch Arena presents substantial challenges for current state-of-the-art agents, with clear performance gaps observed across different models.
|
|
2025-09-05T00:00:00 |
2509.04419
|
Towards a Unified View of Large Language Model Post-Training
|
[
"Xingtai Lv",
"Yuxin Zuo",
"Youbang Sun",
"Hongyi Liu",
"Yuntian Wei",
"Zhekai Chen",
"Lixuan He",
"Xuekai Zhu",
"Kaiyan Zhang",
"Bingning Wang",
"Ning Ding",
"Bowen Zhou"
] |
Two major sources of training data exist for post-training modern language models: online (model-generated rollouts) data, and offline (human or other-model demonstrations) data. These two types of data are typically used by approaches like Reinforcement Learning (RL) and Supervised Fine-Tuning (SFT), respectively. In this paper, we show that these approaches are not in contradiction, but are instances of a single optimization process. We derive a Unified Policy Gradient Estimator, and present the calculations of a wide spectrum of post-training approaches as the gradient of a common objective under different data distribution assumptions and various bias-variance tradeoffs. The gradient estimator is constructed with four interchangeable parts: stabilization mask, reference policy denominator, advantage estimate, and likelihood gradient. Motivated by our theoretical findings, we propose Hybrid Post-Training (HPT), an algorithm that dynamically selects different training signals. HPT is designed to yield both effective exploitation of demonstration and stable exploration without sacrificing learned reasoning patterns. We provide extensive experiments and ablation studies to verify the effectiveness of our unified theoretical framework and HPT. Across six mathematical reasoning benchmarks and two out-of-distribution suites, HPT consistently surpasses strong baselines across models of varying scales and families.
|
|
2025-09-05T00:00:00 |
2508.18733
|
Drawing2CAD: Sequence-to-Sequence Learning for CAD Generation from Vector Drawings
|
[
"Feiwei Qin",
"Shichao Lu",
"Junhao Hou",
"Changmiao Wang",
"Meie Fang",
"Ligang Liu"
] |
https://github.com/lllssc/Drawing2CAD
|
Computer-Aided Design (CAD) generative modeling is driving significant innovations across industrial applications. Recent works have shown remarkable progress in creating solid models from various inputs such as point clouds, meshes, and text descriptions. However, these methods fundamentally diverge from traditional industrial workflows that begin with 2D engineering drawings. The automatic generation of parametric CAD models from these 2D vector drawings remains underexplored despite being a critical step in engineering design. To address this gap, our key insight is to reframe CAD generation as a sequence-to-sequence learning problem where vector drawing primitives directly inform the generation of parametric CAD operations, preserving geometric precision and design intent throughout the transformation process. We propose Drawing2CAD, a framework with three key technical components: a network-friendly vector primitive representation that preserves precise geometric information, a dual-decoder transformer architecture that decouples command type and parameter generation while maintaining precise correspondence, and a soft target distribution loss function accommodating inherent flexibility in CAD parameters. To train and evaluate Drawing2CAD, we create CAD-VGDrawing, a dataset of paired engineering drawings and parametric CAD models, and conduct thorough experiments to demonstrate the effectiveness of our method. Code and dataset are available at https://github.com/lllssc/Drawing2CAD.
|
2025-09-05T00:00:00 |
2509.03888
|
False Sense of Security: Why Probing-based Malicious Input Detection Fails to Generalize
|
[
"Cheng Wang",
"Zeming Wei",
"Qin Liu",
"Muhao Chen"
] |
https://github.com/WangCheng0116/Why-Probe-Fails
|
Large Language Models (LLMs) can comply with harmful instructions, raising serious safety concerns despite their impressive capabilities. Recent work has leveraged probing-based approaches to study the separability of malicious and benign inputs in LLMs' internal representations, and researchers have proposed using such probing methods for safety detection. We systematically re-examine this paradigm. Motivated by poor out-of-distribution performance, we hypothesize that probes learn superficial patterns rather than semantic harmfulness. Through controlled experiments, we confirm this hypothesis and identify the specific patterns learned: instructional patterns and trigger words. Our investigation follows a systematic approach, progressing from demonstrating comparable performance of simple n-gram methods, to controlled experiments with semantically cleaned datasets, to detailed analysis of pattern dependencies. These results reveal a false sense of security around current probing-based approaches and highlight the need to redesign both models and evaluation protocols, for which we provide further discussions in the hope of suggesting responsible further research in this direction. We have open-sourced the project at https://github.com/WangCheng0116/Why-Probe-Fails.
|
2025-09-05T00:00:00 |
2509.04394
|
Transition Models: Rethinking the Generative Learning Objective
|
[
"Zidong Wang",
"Yiyuan Zhang",
"Xiaoyu Yue",
"Xiangyu Yue",
"Yangguang Li",
"Wanli Ouyang",
"Lei Bai"
] |
A fundamental dilemma in generative modeling persists: iterative diffusion models achieve outstanding fidelity, but at a significant computational cost, while efficient few-step alternatives are constrained by a hard quality ceiling. This conflict between generation steps and output quality arises from restrictive training objectives that focus exclusively on either infinitesimal dynamics (PF-ODEs) or direct endpoint prediction. We address this challenge by introducing an exact, continuous-time dynamics equation that analytically defines state transitions across any finite time interval. This leads to a novel generative paradigm, Transition Models (TiM), which adapt to arbitrary-step transitions, seamlessly traversing the generative trajectory from single leaps to fine-grained refinement with more steps. Despite having only 865M parameters, TiM achieves state-of-the-art performance, surpassing leading models such as SD3.5 (8B parameters) and FLUX.1 (12B parameters) across all evaluated step counts. Importantly, unlike previous few-step generators, TiM demonstrates monotonic quality improvement as the sampling budget increases. Additionally, when employing our native-resolution strategy, TiM delivers exceptional fidelity at resolutions up to 4096x4096.
|
|
2025-09-05T00:00:00 |
2509.04434
|
Durian: Dual Reference-guided Portrait Animation with Attribute Transfer
|
[
"Hyunsoo Cha",
"Byungjun Kim",
"Hanbyul Joo"
] |
We present Durian, the first method for generating portrait animation videos with facial attribute transfer from a given reference image to a target portrait in a zero-shot manner. To enable high-fidelity and spatially consistent attribute transfer across frames, we introduce dual reference networks that inject spatial features from both the portrait and attribute images into the denoising process of a diffusion model. We train the model using a self-reconstruction formulation, where two frames are sampled from the same portrait video: one is treated as the attribute reference and the other as the target portrait, and the remaining frames are reconstructed conditioned on these inputs and their corresponding masks. To support the transfer of attributes with varying spatial extent, we propose a mask expansion strategy using keypoint-conditioned image generation for training. In addition, we further augment the attribute and portrait images with spatial and appearance-level transformations to improve robustness to positional misalignment between them. These strategies allow the model to effectively generalize across diverse attributes and in-the-wild reference combinations, despite being trained without explicit triplet supervision. Durian achieves state-of-the-art performance on portrait animation with attribute transfer, and notably, its dual reference design enables multi-attribute composition in a single generation pass without additional training.
|
|
2025-09-05T00:00:00 |
2509.04406
|
Few-step Flow for 3D Generation via Marginal-Data Transport Distillation
|
[
"Zanwei Zhou",
"Taoran Yi",
"Jiemin Fang",
"Chen Yang",
"Lingxi Xie",
"Xinggang Wang",
"Wei Shen",
"Qi Tian"
] |
Flow-based 3D generation models typically require dozens of sampling steps during inference. Though few-step distillation methods, particularly Consistency Models (CMs), have achieved substantial advancements in accelerating 2D diffusion models, they remain under-explored for more complex 3D generation tasks. In this study, we propose a novel framework, MDT-dist, for few-step 3D flow distillation. Our approach is built upon a primary objective: distilling the pretrained model to learn the Marginal-Data Transport. Directly learning this objective needs to integrate the velocity fields, while this integral is intractable to be implemented. Therefore, we propose two optimizable objectives, Velocity Matching (VM) and Velocity Distillation (VD), to equivalently convert the optimization target from the transport level to the velocity and the distribution level respectively. Velocity Matching (VM) learns to stably match the velocity fields between the student and the teacher, but inevitably provides biased gradient estimates. Velocity Distillation (VD) further enhances the optimization process by leveraging the learned velocity fields to perform probability density distillation. When evaluated on the pioneer 3D generation framework TRELLIS, our method reduces sampling steps of each flow transformer from 25 to 1 or 2, achieving 0.68s (1 step x 2) and 0.94s (2 steps x 2) latency with 9.0x and 6.5x speedup on A800, while preserving high visual and geometric fidelity. Extensive experiments demonstrate that our method significantly outperforms existing CM distillation methods, and enables TRELLIS to achieve superior performance in few-step 3D generation.
|
|
2025-09-05T00:00:00 |
2509.04338
|
From Editor to Dense Geometry Estimator
|
[
"JiYuan Wang",
"Chunyu Lin",
"Lei Sun",
"Rongying Liu",
"Lang Nie",
"Mingxing Li",
"Kang Liao",
"Xiangxiang Chu",
"Yao Zhao"
] |
Leveraging visual priors from pre-trained text-to-image (T2I) generative models has shown success in dense prediction. However, dense prediction is inherently an image-to-image task, suggesting that image editing models, rather than T2I generative models, may be a more suitable foundation for fine-tuning. Motivated by this, we conduct a systematic analysis of the fine-tuning behaviors of both editors and generators for dense geometry estimation. Our findings show that editing models possess inherent structural priors, which enable them to converge more stably by ``refining" their innate features, and ultimately achieve higher performance than their generative counterparts. Based on these findings, we introduce FE2E, a framework that pioneeringly adapts an advanced editing model based on Diffusion Transformer (DiT) architecture for dense geometry prediction. Specifically, to tailor the editor for this deterministic task, we reformulate the editor's original flow matching loss into the ``consistent velocity" training objective. And we use logarithmic quantization to resolve the precision conflict between the editor's native BFloat16 format and the high precision demand of our tasks. Additionally, we leverage the DiT's global attention for a cost-free joint estimation of depth and normals in a single forward pass, enabling their supervisory signals to mutually enhance each other. Without scaling up the training data, FE2E achieves impressive performance improvements in zero-shot monocular depth and normal estimation across multiple datasets. Notably, it achieves over 35\% performance gains on the ETH3D dataset and outperforms the DepthAnything series, which is trained on 100times data. The project page can be accessed https://amap-ml.github.io/FE2E/{here}.
|
|
2025-09-05T00:00:00 |
2509.04292
|
Inverse IFEval: Can LLMs Unlearn Stubborn Training Conventions to Follow Real Instructions?
|
[
"Qinyan Zhang",
"Xinping Lei",
"Ruijie Miao",
"Yu Fu",
"Haojie Fan",
"Le Chang",
"Jiafan Hou",
"Dingling Zhang",
"Zhongfei Hou",
"Ziqiang Yang",
"Changxin Pu",
"Fei Hu",
"Jingkai Liu",
"Mengyun Liu",
"Yang Liu",
"Xiang Gao",
"Jiaheng Liu",
"Tong Yang",
"Zaiyuan Wang",
"Ge Zhang",
"Wenhao Huang"
] |
Large Language Models (LLMs) achieve strong performance on diverse tasks but often exhibit cognitive inertia, struggling to follow instructions that conflict with the standardized patterns learned during supervised fine-tuning (SFT). To evaluate this limitation, we propose Inverse IFEval, a benchmark that measures models Counter-intuitive Abilitytheir capacity to override training-induced biases and comply with adversarial instructions. Inverse IFEval introduces eight types of such challenges, including Question Correction, Intentional Textual Flaws, Code without Comments, and Counterfactual Answering. Using a human-in-the-loop pipeline, we construct a dataset of 1012 high-quality Chinese and English questions across 23 domains, evaluated under an optimized LLM-as-a-Judge framework. Experiments on existing leading LLMs demonstrate the necessity of our proposed Inverse IFEval benchmark. Our findings emphasize that future alignment efforts should not only pursue fluency and factual correctness but also account for adaptability under unconventional contexts. We hope that Inverse IFEval serves as both a diagnostic tool and a foundation for developing methods that mitigate cognitive inertia, reduce overfitting to narrow patterns, and ultimately enhance the instruction-following reliability of LLMs in diverse and unpredictable real-world scenarios.
|
|
2025-09-05T00:00:00 |
2509.04011
|
NER Retriever: Zero-Shot Named Entity Retrieval with Type-Aware Embeddings
|
[
"Or Shachar",
"Uri Katz",
"Yoav Goldberg",
"Oren Glickman"
] |
https://github.com/ShacharOr100/ner_retriever
|
We present NER Retriever, a zero-shot retrieval framework for ad-hoc Named Entity Retrieval, a variant of Named Entity Recognition (NER), where the types of interest are not provided in advance, and a user-defined type description is used to retrieve documents mentioning entities of that type. Instead of relying on fixed schemas or fine-tuned models, our method builds on internal representations of large language models (LLMs) to embed both entity mentions and user-provided open-ended type descriptions into a shared semantic space. We show that internal representations, specifically the value vectors from mid-layer transformer blocks, encode fine-grained type information more effectively than commonly used top-layer embeddings. To refine these representations, we train a lightweight contrastive projection network that aligns type-compatible entities while separating unrelated types. The resulting entity embeddings are compact, type-aware, and well-suited for nearest-neighbor search. Evaluated on three benchmarks, NER Retriever significantly outperforms both lexical and dense sentence-level retrieval baselines. Our findings provide empirical support for representation selection within LLMs and demonstrate a practical solution for scalable, schema-free entity retrieval. The NER Retriever Codebase is publicly available at https://github.com/ShacharOr100/ner_retriever
|
2025-09-05T00:00:00 |
2509.04442
|
Delta Activations: A Representation for Finetuned Large Language Models
|
[
"Zhiqiu Xu",
"Amish Sethi",
"Mayur Naik",
"Ser-Nam Lim"
] |
https://github.com/OscarXZQ/delta_activations
|
The success of powerful open source Large Language Models (LLMs) has enabled the community to create a vast collection of post-trained models adapted to specific tasks and domains. However, navigating and understanding these models remains challenging due to inconsistent metadata and unstructured repositories. We introduce Delta Activations, a method to represent finetuned models as vector embeddings by measuring shifts in their internal activations relative to a base model. This representation allows for effective clustering by domain and task, revealing structure in the model landscape. Delta Activations also demonstrate desirable properties: it is robust across finetuning settings and exhibits an additive property when finetuning datasets are mixed. In addition, we show that Delta Activations can embed tasks via few-shot finetuning, and further explore its use for model selection and merging. We hope Delta Activations can facilitate the practice of reusing publicly available models. Code is available at https://github.com/OscarXZQ/delta_activations.
|
2025-09-05T00:00:00 |
2509.03867
|
Drivel-ology: Challenging LLMs with Interpreting Nonsense with Depth
|
[
"Yang Wang",
"Chenghao Xiao",
"Chia-Yi Hsiao",
"Zi Yan Chang",
"Chi-Li Chen",
"Tyler Loakman",
"Chenghua Lin"
] |
We introduce Drivelology, a unique linguistic phenomenon characterised as "nonsense with depth", utterances that are syntactically coherent yet pragmatically paradoxical, emotionally loaded, or rhetorically subversive. While such expressions may resemble surface-level nonsense, they encode implicit meaning requiring contextual inference, moral reasoning, or emotional interpretation. We find that current large language models (LLMs), despite excelling at many natural language processing (NLP) tasks, consistently fail to grasp the layered semantics of Drivelological text. To investigate this, we construct a small but diverse benchmark dataset of over 1,200 meticulously curated examples, with select instances in English, Mandarin, Spanish, French, Japanese, and Korean. Annotation was especially challenging: each of the examples required careful expert review to verify that it truly reflected Drivelological characteristics. The process involved multiple rounds of discussion and adjudication to address disagreements, highlighting the subtle and subjective nature of the Drivelology. We evaluate a range of LLMs on classification, generation, and reasoning tasks. Our results reveal clear limitations of LLMs: models often confuse Drivelology with shallow nonsense, produce incoherent justifications, or miss the implied rhetorical function altogether. These findings highlight a deeper representational gap in LLMs' pragmatic understanding and challenge the assumption that statistical fluency implies cognitive comprehension. We release our dataset and code to facilitate further research in modelling linguistic depth beyond surface-level coherence.
|
|
2025-09-05T00:00:00 |
2508.20478
|
Video-MTR: Reinforced Multi-Turn Reasoning for Long Video Understanding
|
[
"Yuan Xie",
"Tianshui Chen",
"Zheng Ge",
"Lionel Ni"
] |
Long-form video understanding, characterized by long-range temporal dependencies and multiple events, remains a challenge. Existing methods often rely on static reasoning or external visual-language models (VLMs), which face issues like complexity and sub-optimal performance due to the lack of end-to-end training. In this paper, we propose Video-MTR, a reinforced multi-turn reasoning framework designed to enable iterative key video segment selection and question comprehension. Unlike traditional video reasoning pipeline, which generate predictions in a single turn, Video-MTR performs reasoning in multiple turns, selecting video segments progressively based on the evolving understanding of previously processed segments and the current question. This iterative process allows for a more refined and contextually aware analysis of the video. To ensure intermediate reasoning process, we introduce a novel gated bi-level reward system, combining trajectory-level rewards based on answer correctness and turn-level rewards emphasizing frame-query relevance. This system optimizes both video segment selection and question comprehension, eliminating the need for external VLMs and allowing end-to-end training. Extensive experiments on benchmarks like VideoMME, MLVU, and EgoSchema demonstrate that Video-MTR outperforms existing methods in both accuracy and efficiency, advancing the state-of-the-art in long video understanding.
|
|
2025-09-05T00:00:00 |
2509.03059
|
Loong: Synthesize Long Chain-of-Thoughts at Scale through Verifiers
|
[
"Xingyue Huang",
"Rishabh",
"Gregor Franke",
"Ziyi Yang",
"Jiamu Bai",
"Weijie Bai",
"Jinhe Bi",
"Zifeng Ding",
"Yiqun Duan",
"Chengyu Fan",
"Wendong Fan",
"Xin Gao",
"Ruohao Guo",
"Yuan He",
"Zhuangzhuang He",
"Xianglong Hu",
"Neil Johnson",
"Bowen Li",
"Fangru Lin",
"Siyu Lin",
"Tong Liu",
"Yunpu Ma",
"Hao Shen",
"Hao Sun",
"Beibei Wang",
"Fangyijie Wang",
"Hao Wang",
"Haoran Wang",
"Yang Wang",
"Yifeng Wang",
"Zhaowei Wang",
"Ziyang Wang",
"Yifan Wu",
"Zikai Xiao",
"Chengxing Xie",
"Fan Yang",
"Junxiao Yang",
"Qianshuo Ye",
"Ziyu Ye",
"Guangtao Zeng",
"Yuwen Ebony Zhang",
"Zeyu Zhang",
"Zihao Zhu",
"Bernard Ghanem",
"Philip Torr",
"Guohao Li"
] |
https://github.com/camel-ai/loong
|
Recent advances in Large Language Models (LLMs) have shown that their reasoning capabilities can be significantly improved through Reinforcement Learning with Verifiable Reward (RLVR), particularly in domains like mathematics and programming, where ground-truth correctness can be automatically evaluated. However, extending this success to other reasoning-intensive domains remains challenging due to the scarcity of high-quality, verifiable datasets and the high cost of human supervision. In this work, we introduce the Loong Project: an open-source framework for scalable synthetic data generation and verification across a diverse range of reasoning-intensive domains. The framework consists of two key components: (1) LoongBench, a curated seed dataset containing 8,729 human-vetted examples across 12 domains (e.g., Advanced Mathematics, Chemistry, Logic), each paired with executable code and rich metadata; and (2) LoongEnv, a modular synthetic data generation environment that supports multiple prompting strategies to produce new question-answer-code triples. Together, these components form an agent-environment loop that enables reinforcement learning, where an LLM-based agent is rewarded for generating Chain-of-Thought (CoT) solutions that align with code-executed answers. Empirically, we benchmark LoongBench on a broad suite of both open-source and proprietary LLMs to evaluate domain coverage and reveal performance bottlenecks. In addition, we conduct a comprehensive analysis of synthetic data generated by LoongEnv, examining correctness, difficulty, and diversity. Code and documentation are available at https://github.com/camel-ai/loong.
|
2025-09-08T00:00:00 |
2509.04744
|
WildScore: Benchmarking MLLMs in-the-Wild Symbolic Music Reasoning
|
[
"Gagan Mundada",
"Yash Vishe",
"Amit Namburi",
"Xin Xu",
"Zachary Novack",
"Julian McAuley",
"Junda Wu"
] |
Recent advances in Multimodal Large Language Models (MLLMs) have demonstrated impressive capabilities across various vision-language tasks. However, their reasoning abilities in the multimodal symbolic music domain remain largely unexplored. We introduce WildScore, the first in-the-wild multimodal symbolic music reasoning and analysis benchmark, designed to evaluate MLLMs' capacity to interpret real-world music scores and answer complex musicological queries. Each instance in WildScore is sourced from genuine musical compositions and accompanied by authentic user-generated questions and discussions, capturing the intricacies of practical music analysis. To facilitate systematic evaluation, we propose a systematic taxonomy, comprising both high-level and fine-grained musicological ontologies. Furthermore, we frame complex music reasoning as multiple-choice question answering, enabling controlled and scalable assessment of MLLMs' symbolic music understanding. Empirical benchmarking of state-of-the-art MLLMs on WildScore reveals intriguing patterns in their visual-symbolic reasoning, uncovering both promising directions and persistent challenges for MLLMs in symbolic music reasoning and analysis. We release the dataset and code.
|
|
2025-09-08T00:00:00 |
2509.05263
|
LatticeWorld: A Multimodal Large Language Model-Empowered Framework for Interactive Complex World Generation
|
[
"Yinglin Duan",
"Zhengxia Zou",
"Tongwei Gu",
"Wei Jia",
"Zhan Zhao",
"Luyi Xu",
"Xinzhu Liu",
"Hao Jiang",
"Kang Chen",
"Shuang Qiu"
] |
Recent research has been increasingly focusing on developing 3D world models that simulate complex real-world scenarios. World models have found broad applications across various domains, including embodied AI, autonomous driving, entertainment, etc. A more realistic simulation with accurate physics will effectively narrow the sim-to-real gap and allow us to gather rich information about the real world conveniently. While traditional manual modeling has enabled the creation of virtual 3D scenes, modern approaches have leveraged advanced machine learning algorithms for 3D world generation, with most recent advances focusing on generative methods that can create virtual worlds based on user instructions. This work explores such a research direction by proposing LatticeWorld, a simple yet effective 3D world generation framework that streamlines the industrial production pipeline of 3D environments. LatticeWorld leverages lightweight LLMs (LLaMA-2-7B) alongside the industry-grade rendering engine (e.g., Unreal Engine 5) to generate a dynamic environment. Our proposed framework accepts textual descriptions and visual instructions as multimodal inputs and creates large-scale 3D interactive worlds with dynamic agents, featuring competitive multi-agent interaction, high-fidelity physics simulation, and real-time rendering. We conduct comprehensive experiments to evaluate LatticeWorld, showing that it achieves superior accuracy in scene layout generation and visual fidelity. Moreover, LatticeWorld achieves over a 90times increase in industrial production efficiency while maintaining high creative quality compared with traditional manual production methods. Our demo video is available at https://youtu.be/8VWZXpERR18
|
|
2025-09-08T00:00:00 |
2509.03800
|
MedVista3D: Vision-Language Modeling for Reducing Diagnostic Errors in 3D CT Disease Detection, Understanding and Reporting
|
[
"Yuheng Li",
"Yenho Chen",
"Yuxiang Lai",
"Jike Zhong",
"Vanessa Wildman",
"Xiaofeng Yang"
] |
Radiologic diagnostic errors-under-reading errors, inattentional blindness, and communication failures-remain prevalent in clinical practice. These issues often stem from missed localized abnormalities, limited global context, and variability in report language. These challenges are amplified in 3D imaging, where clinicians must examine hundreds of slices per scan. Addressing them requires systems with precise localized detection, global volume-level reasoning, and semantically consistent natural language reporting. However, existing 3D vision-language models are unable to meet all three needs jointly, lacking local-global understanding for spatial reasoning and struggling with the variability and noise of uncurated radiology reports. We present MedVista3D, a multi-scale semantic-enriched vision-language pretraining framework for 3D CT analysis. To enable joint disease detection and holistic interpretation, MedVista3D performs local and global image-text alignment for fine-grained representation learning within full-volume context. To address report variability, we apply language model rewrites and introduce a Radiology Semantic Matching Bank for semantics-aware alignment. MedVista3D achieves state-of-the-art performance on zero-shot disease classification, report retrieval, and medical visual question answering, while transferring well to organ segmentation and prognosis prediction. Code and datasets will be released.
|
|
2025-09-08T00:00:00 |
2509.04664
|
Why Language Models Hallucinate
|
[
"Adam Tauman Kalai",
"Ofir Nachum",
"Santosh S. Vempala",
"Edwin Zhang"
] |
Like students facing hard exam questions, large language models sometimes guess when uncertain, producing plausible yet incorrect statements instead of admitting uncertainty. Such "hallucinations" persist even in state-of-the-art systems and undermine trust. We argue that language models hallucinate because the training and evaluation procedures reward guessing over acknowledging uncertainty, and we analyze the statistical causes of hallucinations in the modern training pipeline. Hallucinations need not be mysterious -- they originate simply as errors in binary classification. If incorrect statements cannot be distinguished from facts, then hallucinations in pretrained language models will arise through natural statistical pressures. We then argue that hallucinations persist due to the way most evaluations are graded -- language models are optimized to be good test-takers, and guessing when uncertain improves test performance. This "epidemic" of penalizing uncertain responses can only be addressed through a socio-technical mitigation: modifying the scoring of existing benchmarks that are misaligned but dominate leaderboards, rather than introducing additional hallucination evaluations. This change may steer the field toward more trustworthy AI systems.
|
|
2025-09-08T00:00:00 |
2509.04504
|
Behavioral Fingerprinting of Large Language Models
|
[
"Zehua Pei",
"Hui-Ling Zhen",
"Ying Zhang",
"Zhiyuan Yang",
"Xing Li",
"Xianzhi Yu",
"Mingxuan Yuan",
"Bei Yu"
] |
https://github.com/JarvisPei/Behavioral-Fingerprinting
|
Current benchmarks for Large Language Models (LLMs) primarily focus on performance metrics, often failing to capture the nuanced behavioral characteristics that differentiate them. This paper introduces a novel ``Behavioral Fingerprinting'' framework designed to move beyond traditional evaluation by creating a multi-faceted profile of a model's intrinsic cognitive and interactive styles. Using a curated Diagnostic Prompt Suite and an innovative, automated evaluation pipeline where a powerful LLM acts as an impartial judge, we analyze eighteen models across capability tiers. Our results reveal a critical divergence in the LLM landscape: while core capabilities like abstract and causal reasoning are converging among top models, alignment-related behaviors such as sycophancy and semantic robustness vary dramatically. We further document a cross-model default persona clustering (ISTJ/ESTJ) that likely reflects common alignment incentives. Taken together, this suggests that a model's interactive nature is not an emergent property of its scale or reasoning power, but a direct consequence of specific, and highly variable, developer alignment strategies. Our framework provides a reproducible and scalable methodology for uncovering these deep behavioral differences. Project: https://github.com/JarvisPei/Behavioral-Fingerprinting
|
2025-09-08T00:00:00 |
2509.04013
|
On Robustness and Reliability of Benchmark-Based Evaluation of LLMs
|
[
"Riccardo Lunardi",
"Vincenzo Della Mea",
"Stefano Mizzaro",
"Kevin Roitero"
] |
Large Language Models (LLMs) effectiveness is usually evaluated by means of benchmarks such as MMLU, ARC-C, or HellaSwag, where questions are presented in their original wording, thus in a fixed, standardized format. However, real-world applications involve linguistic variability, requiring models to maintain their effectiveness across diverse rewordings of the same question or query. In this study, we systematically assess the robustness of LLMs to paraphrased benchmark questions and investigate whether benchmark-based evaluations provide a reliable measure of model capabilities. We systematically generate various paraphrases of all the questions across six different common benchmarks, and measure the resulting variations in effectiveness of 34 state-of-the-art LLMs, of different size and effectiveness. Our findings reveal that while LLM rankings remain relatively stable across paraphrased inputs, absolute effectiveness scores change, and decline significantly. This suggests that LLMs struggle with linguistic variability, raising concerns about their generalization abilities and evaluation methodologies. Furthermore, the observed performance drop challenges the reliability of benchmark-based evaluations, indicating that high benchmark scores may not fully capture a model's robustness to real-world input variations. We discuss the implications of these findings for LLM evaluation methodologies, emphasizing the need for robustness-aware benchmarks that better reflect practical deployment scenarios.
|
|
2025-09-08T00:00:00 |
2509.05296
|
WinT3R: Window-Based Streaming Reconstruction with Camera Token Pool
|
[
"Zizun Li",
"Jianjun Zhou",
"Yifan Wang",
"Haoyu Guo",
"Wenzheng Chang",
"Yang Zhou",
"Haoyi Zhu",
"Junyi Chen",
"Chunhua Shen",
"Tong He"
] |
https://github.com/LiZizun/WinT3R
|
We present WinT3R, a feed-forward reconstruction model capable of online prediction of precise camera poses and high-quality point maps. Previous methods suffer from a trade-off between reconstruction quality and real-time performance. To address this, we first introduce a sliding window mechanism that ensures sufficient information exchange among frames within the window, thereby improving the quality of geometric predictions without large computation. In addition, we leverage a compact representation of cameras and maintain a global camera token pool, which enhances the reliability of camera pose estimation without sacrificing efficiency. These designs enable WinT3R to achieve state-of-the-art performance in terms of online reconstruction quality, camera pose estimation, and reconstruction speed, as validated by extensive experiments on diverse datasets. Code and model are publicly available at https://github.com/LiZizun/WinT3R.
|
Subsets and Splits
Top 20 Papers with GitHub Links
This query retrieves 20 entries from the training set where the GitHub URL is not null or empty, providing a basic filter of non-empty entries.