date
timestamp[ns]date 2023-05-05 00:00:00
2025-04-01 00:00:00
| arxiv_id
stringlengths 10
10
| title
stringlengths 8
177
| authors
sequencelengths 1
942
| github
stringlengths 0
116
| abstract
stringlengths 165
1.92k
|
|---|---|---|---|---|---|
2025-03-05T00:00:00 | 2503.01935 | MultiAgentBench: Evaluating the Collaboration and Competition of LLM agents | [
"Kunlun Zhu",
"Hongyi Du",
"Zhaochen Hong",
"Xiaocheng Yang",
"Shuyi Guo",
"Zhe Wang",
"Zhenhailong Wang",
"Cheng Qian",
"Xiangru Tang",
"Heng Ji",
"Jiaxuan You"
] | https://github.com/MultiagentBench/MARBLE | Large Language Models (LLMs) have shown remarkable capabilities as autonomous agents, yet existing benchmarks either focus on single-agent tasks or are confined to narrow domains, failing to capture the dynamics of multi-agent coordination and competition. In this paper, we introduce MultiAgentBench, a comprehensive benchmark designed to evaluate LLM-based multi-agent systems across diverse, interactive scenarios. Our framework measures not only task completion but also the quality of collaboration and competition using novel, milestone-based key performance indicators. Moreover, we evaluate various coordination protocols (including star, chain, tree, and graph topologies) and innovative strategies such as group discussion and cognitive planning. Notably, gpt-4o-mini reaches the average highest task score, graph structure performs the best among coordination protocols in the research scenario, and cognitive planning improves milestone achievement rates by 3%. Code and datasets are public available at https://github.com/MultiagentBench/MARBLE. |
2025-03-05T00:00:00 | 2503.02846 | Mask-DPO: Generalizable Fine-grained Factuality Alignment of LLMs | [
"Yuzhe Gu",
"Wenwei Zhang",
"Chengqi Lyu",
"Dahua Lin",
"Kai Chen"
] | Large language models (LLMs) exhibit hallucinations (i.e., unfaithful or nonsensical information) when serving as AI assistants in various domains. Since hallucinations always come with truthful content in the LLM responses, previous factuality alignment methods that conduct response-level preference learning inevitably introduced noises during training. Therefore, this paper proposes a fine-grained factuality alignment method based on Direct Preference Optimization (DPO), called Mask-DPO. Incorporating sentence-level factuality as mask signals, Mask-DPO only learns from factually correct sentences in the preferred samples and prevents the penalty on factual contents in the not preferred samples, which resolves the ambiguity in the preference learning. Extensive experimental results demonstrate that Mask-DPO can significantly improve the factuality of LLMs responses to questions from both in-domain and out-of-domain datasets, although these questions and their corresponding topics are unseen during training. Only trained on the ANAH train set, the score of Llama3.1-8B-Instruct on the ANAH test set is improved from 49.19% to 77.53%, even surpassing the score of Llama3.1-70B-Instruct (53.44%), while its FactScore on the out-of-domain Biography dataset is also improved from 30.29% to 39.39%. We further study the generalization property of Mask-DPO using different training sample scaling strategies and find that scaling the number of topics in the dataset is more effective than the number of questions. We provide a hypothesis of what factual alignment is doing with LLMs, on the implication of this phenomenon, and conduct proof-of-concept experiments to verify it. We hope the method and the findings pave the way for future research on scaling factuality alignment. |
|
2025-03-05T00:00:00 | 2503.02682 | MPO: Boosting LLM Agents with Meta Plan Optimization | [
"Weimin Xiong",
"Yifan Song",
"Qingxiu Dong",
"Bingchan Zhao",
"Feifan Song",
"Xun Wang",
"Sujian Li"
] | Recent advancements in large language models (LLMs) have enabled LLM-based agents to successfully tackle interactive planning tasks. However, despite their successes, existing approaches often suffer from planning hallucinations and require retraining for each new agent. To address these challenges, we propose the Meta Plan Optimization (MPO) framework, which enhances agent planning capabilities by directly incorporating explicit guidance. Unlike previous methods that rely on complex knowledge, which either require significant human effort or lack quality assurance, MPO leverages high-level general guidance through meta plans to assist agent planning and enables continuous optimization of the meta plans based on feedback from the agent's task execution. Our experiments conducted on two representative tasks demonstrate that MPO significantly outperforms existing baselines. Moreover, our analysis indicates that MPO provides a plug-and-play solution that enhances both task completion efficiency and generalization capabilities in previous unseen scenarios. |
|
2025-03-05T00:00:00 | 2503.02879 | Wikipedia in the Era of LLMs: Evolution and Risks | [
"Siming Huang",
"Yuliang Xu",
"Mingmeng Geng",
"Yao Wan",
"Dongping Chen"
] | In this paper, we present a thorough analysis of the impact of Large Language Models (LLMs) on Wikipedia, examining the evolution of Wikipedia through existing data and using simulations to explore potential risks. We begin by analyzing page views and article content to study Wikipedia's recent changes and assess the impact of LLMs. Subsequently, we evaluate how LLMs affect various Natural Language Processing (NLP) tasks related to Wikipedia, including machine translation and retrieval-augmented generation (RAG). Our findings and simulation results reveal that Wikipedia articles have been influenced by LLMs, with an impact of approximately 1%-2% in certain categories. If the machine translation benchmark based on Wikipedia is influenced by LLMs, the scores of the models may become inflated, and the comparative results among models might shift as well. Moreover, the effectiveness of RAG might decrease if the knowledge base becomes polluted by LLM-generated content. While LLMs have not yet fully changed Wikipedia's language and knowledge structures, we believe that our empirical findings signal the need for careful consideration of potential future risks. |
|
2025-03-05T00:00:00 | 2503.02197 | ATLaS: Agent Tuning via Learning Critical Steps | [
"Zhixun Chen",
"Ming Li",
"Yuxuan Huang",
"Yali Du",
"Meng Fang",
"Tianyi Zhou"
] | Large Language Model (LLM) agents have demonstrated remarkable generalization capabilities across multi-domain tasks. Existing agent tuning approaches typically employ supervised finetuning on entire expert trajectories. However, behavior-cloning of full trajectories can introduce expert bias and weaken generalization to states not covered by the expert data. Additionally, critical steps, such as planning, complex reasoning for intermediate subtasks, and strategic decision-making, are essential to success in agent tasks, so learning these steps is the key to improving LLM agents. For more effective and efficient agent tuning, we propose ATLaS that identifies the critical steps in expert trajectories and finetunes LLMs solely on these steps with reduced costs. By steering the training's focus to a few critical steps, our method mitigates the risk of overfitting entire trajectories and promotes generalization across different environments and tasks. In extensive experiments, an LLM finetuned on only 30% critical steps selected by ATLaS outperforms the LLM finetuned on all steps and recent open-source LLM agents. ATLaS maintains and improves base LLM skills as generalist agents interacting with diverse environments. |
|
2025-03-05T00:00:00 | 2503.02878 | Language Models can Self-Improve at State-Value Estimation for Better Search | [
"Ethan Mendes",
"Alan Ritter"
] | Collecting ground truth task completion rewards or human demonstrations for multi-step reasoning tasks is often cost-prohibitive and time-consuming, especially in interactive domains like web tasks. To address this bottleneck, we present self-taught lookahead, a self-supervised method that leverages state-transition dynamics to train a value model capable of effectively guiding language model-controlled search. We find that moderately sized (8 billion parameters) open-weight value models improved with self-taught lookahead can match the performance of using a frontier LLM such as gpt-4o as the value model. Furthermore, we find that self-taught lookahead improves performance by 20% while reducing costs 37x compared to previous LLM-based tree search, without relying on ground truth rewards. |
|
2025-03-05T00:00:00 | 2503.02876 | SPIDER: A Comprehensive Multi-Organ Supervised Pathology Dataset and Baseline Models | [
"Dmitry Nechaev",
"Alexey Pchelnikov",
"Ekaterina Ivanova"
] | https://github.com/HistAI/SPIDER | Advancing AI in computational pathology requires large, high-quality, and diverse datasets, yet existing public datasets are often limited in organ diversity, class coverage, or annotation quality. To bridge this gap, we introduce SPIDER (Supervised Pathology Image-DEscription Repository), the largest publicly available patch-level dataset covering multiple organ types, including Skin, Colorectal, and Thorax, with comprehensive class coverage for each organ. SPIDER provides high-quality annotations verified by expert pathologists and includes surrounding context patches, which enhance classification performance by providing spatial context. Alongside the dataset, we present baseline models trained on SPIDER using the Hibou-L foundation model as a feature extractor combined with an attention-based classification head. The models achieve state-of-the-art performance across multiple tissue categories and serve as strong benchmarks for future digital pathology research. Beyond patch classification, the model enables rapid identification of significant areas, quantitative tissue metrics, and establishes a foundation for multimodal approaches. Both the dataset and trained models are publicly available to advance research, reproducibility, and AI-driven pathology development. Access them at: https://github.com/HistAI/SPIDER |
2025-03-05T00:00:00 | 2503.00955 | SemViQA: A Semantic Question Answering System for Vietnamese Information Fact-Checking | [
"Nam V. Nguyen",
"Dien X. Tran",
"Thanh T. Tran",
"Anh T. Hoang",
"Tai V. Duong",
"Di T. Le",
"Phuc-Lu Le"
] | https://github.com/DAVID-NGUYEN-S16/SemViQA | The rise of misinformation, exacerbated by Large Language Models (LLMs) like GPT and Gemini, demands robust fact-checking solutions, especially for low-resource languages like Vietnamese. Existing methods struggle with semantic ambiguity, homonyms, and complex linguistic structures, often trading accuracy for efficiency. We introduce SemViQA, a novel Vietnamese fact-checking framework integrating Semantic-based Evidence Retrieval (SER) and Two-step Verdict Classification (TVC). Our approach balances precision and speed, achieving state-of-the-art results with 78.97\% strict accuracy on ISE-DSC01 and 80.82\% on ViWikiFC, securing 1st place in the UIT Data Science Challenge. Additionally, SemViQA Faster improves inference speed 7x while maintaining competitive accuracy. SemViQA sets a new benchmark for Vietnamese fact verification, advancing the fight against misinformation. The source code is available at: https://github.com/DAVID-NGUYEN-S16/SemViQA. |
2025-03-05T00:00:00 | 2502.14856 | FR-Spec: Accelerating Large-Vocabulary Language Models via Frequency-Ranked Speculative Sampling | [
"Weilin Zhao",
"Tengyu Pan",
"Xu Han",
"Yudi Zhang",
"Ao Sun",
"Yuxiang Huang",
"Kaihuo Zhang",
"Weilun Zhao",
"Yuxuan Li",
"Jianyong Wang",
"Zhiyuan Liu",
"Maosong Sun"
] | Speculative sampling has emerged as an important technique for accelerating the auto-regressive generation process of large language models (LLMs) by utilizing a draft-then-verify mechanism to produce multiple tokens per forward pass. While state-of-the-art speculative sampling methods use only a single layer and a language modeling (LM) head as the draft model to achieve impressive layer compression, their efficiency gains are substantially reduced for large-vocabulary LLMs, such as Llama-3-8B with a vocabulary of 128k tokens. To address this, we present FR-Spec, a frequency-ranked speculative sampling framework that optimizes draft candidate selection through vocabulary space compression. By constraining the draft search to a frequency-prioritized token subset, our method reduces LM Head computation overhead by 75% while ensuring the equivalence of the final output distribution. Experiments across multiple datasets demonstrate an average of 1.12times speedup over the state-of-the-art speculative sampling method EAGLE-2. |
|
2025-03-05T00:00:00 | 2503.01342 | UFO: A Unified Approach to Fine-grained Visual Perception via Open-ended Language Interface | [
"Hao Tang",
"Chenwei Xie",
"Haiyang Wang",
"Xiaoyi Bao",
"Tingyu Weng",
"Pandeng Li",
"Yun Zheng",
"Liwei Wang"
] | Generalist models have achieved remarkable success in both language and vision-language tasks, showcasing the potential of unified modeling. However, effectively integrating fine-grained perception tasks like detection and segmentation into these models remains a significant challenge. This is primarily because these tasks often rely heavily on task-specific designs and architectures that can complicate the modeling process. To address this challenge, we present \ours, a framework that Unifies Fine-grained visual perception tasks through an Open-ended language interface. By transforming all perception targets into the language space, \ours unifies object-level detection, pixel-level segmentation, and image-level vision-language tasks into a single model. Additionally, we introduce a novel embedding retrieval approach that relies solely on the language interface to support segmentation tasks. Our framework bridges the gap between fine-grained perception and vision-language tasks, significantly simplifying architectural design and training strategies while achieving comparable or superior performance to methods with intricate task-specific designs. After multi-task training on five standard visual perception datasets, \ours outperforms the previous state-of-the-art generalist models by 12.3 mAP on COCO instance segmentation and 3.3 mIoU on ADE20K semantic segmentation. Furthermore, our method seamlessly integrates with existing MLLMs, effectively combining fine-grained perception capabilities with their advanced language abilities, thereby enabling more challenging tasks such as reasoning segmentation. Code and models will be publicly available. |
|
2025-03-05T00:00:00 | 2503.02268 | AppAgentX: Evolving GUI Agents as Proficient Smartphone Users | [
"Wenjia Jiang",
"Yangyang Zhuang",
"Chenxi Song",
"Xu Yang",
"Chi Zhang"
] | Recent advancements in Large Language Models (LLMs) have led to the development of intelligent LLM-based agents capable of interacting with graphical user interfaces (GUIs). These agents demonstrate strong reasoning and adaptability, enabling them to perform complex tasks that traditionally required predefined rules. However, the reliance on step-by-step reasoning in LLM-based agents often results in inefficiencies, particularly for routine tasks. In contrast, traditional rule-based systems excel in efficiency but lack the intelligence and flexibility to adapt to novel scenarios. To address this challenge, we propose a novel evolutionary framework for GUI agents that enhances operational efficiency while retaining intelligence and flexibility. Our approach incorporates a memory mechanism that records the agent's task execution history. By analyzing this history, the agent identifies repetitive action sequences and evolves high-level actions that act as shortcuts, replacing these low-level operations and improving efficiency. This allows the agent to focus on tasks requiring more complex reasoning, while simplifying routine actions. Experimental results on multiple benchmark tasks demonstrate that our approach significantly outperforms existing methods in both efficiency and accuracy. The code will be open-sourced to support further research. |
|
2025-03-05T00:00:00 | 2503.02368 | Iterative Value Function Optimization for Guided Decoding | [
"Zhenhua Liu",
"Lijun Li",
"Ruizhe Chen",
"Yuxian Jiang",
"Tong Zhu",
"Wenliang Chen",
"Jing Shao"
] | While Reinforcement Learning from Human Feedback (RLHF) has become the predominant method for controlling language model outputs, it suffers from high computational costs and training instability. Guided decoding, especially value-guided methods, offers a cost-effective alternative by controlling outputs without re-training models. However, the accuracy of the value function is crucial for value-guided decoding, as inaccuracies can lead to suboptimal decision-making and degraded performance. Existing methods struggle with accurately estimating the optimal value function, leading to less effective control. We propose Iterative Value Function Optimization, a novel framework that addresses these limitations through two key components: Monte Carlo Value Estimation, which reduces estimation variance by exploring diverse trajectories, and Iterative On-Policy Optimization, which progressively improves value estimation through collecting trajectories from value-guided policies. Extensive experiments on text summarization, multi-turn dialogue, and instruction following demonstrate the effectiveness of value-guided decoding approaches in aligning language models. These approaches not only achieve alignment but also significantly reduce computational costs by leveraging principled value function optimization for efficient and effective control. |
|
2025-03-05T00:00:00 | 2503.00735 | LADDER: Self-Improving LLMs Through Recursive Problem Decomposition | [
"Toby Simonds",
"Akira Yoshiyama"
] | We introduce LADDER (Learning through Autonomous Difficulty-Driven Example Recursion), a framework which enables Large Language Models to autonomously improve their problem-solving capabilities through self-guided learning by recursively generating and solving progressively simpler variants of complex problems. Unlike prior approaches that require curated datasets or human feedback, LADDER leverages a model's own capabilities to generate easier question variants. We demonstrate LADDER's effectiveness in the subject of mathematical integration, improving Llama 3.2 3B's accuracy from 1% to 82% on undergraduate-level problems and enabling Qwen2.5 7B Deepseek-R1 Distilled to achieve 73% on the MIT Integration Bee qualifying examination. We also introduce TTRL (Test-Time Reinforcement Learning), where we perform reinforcement learning on variants of test problems at inference time. TTRL enables Qwen2.5 7B Deepseek-R1 Distilled to achieve a state-of-the-art score of 90% on the MIT Integration Bee qualifying examination, surpassing OpenAI o1's performance. These results show how self-directed strategic learning can achieve significant capability improvements without relying on architectural scaling or human supervision. |
|
2025-03-05T00:00:00 | 2503.02537 | RectifiedHR: Enable Efficient High-Resolution Image Generation via Energy Rectification | [
"Zhen Yang",
"Guibao Shen",
"Liang Hou",
"Mushui Liu",
"Luozhou Wang",
"Xin Tao",
"Pengfei Wan",
"Di Zhang",
"Ying-Cong Chen"
] | Diffusion models have achieved remarkable advances in various image generation tasks. However, their performance notably declines when generating images at resolutions higher than those used during the training period. Despite the existence of numerous methods for producing high-resolution images, they either suffer from inefficiency or are hindered by complex operations. In this paper, we propose RectifiedHR, an efficient and straightforward solution for training-free high-resolution image generation. Specifically, we introduce the noise refresh strategy, which theoretically only requires a few lines of code to unlock the model's high-resolution generation ability and improve efficiency. Additionally, we first observe the phenomenon of energy decay that may cause image blurriness during the high-resolution image generation process. To address this issue, we propose an Energy Rectification strategy, where modifying the hyperparameters of the classifier-free guidance effectively improves the generation performance. Our method is entirely training-free and boasts a simple implementation logic. Through extensive comparisons with numerous baseline methods, our RectifiedHR demonstrates superior effectiveness and efficiency. |
|
2025-03-05T00:00:00 | 2503.02357 | Q-Eval-100K: Evaluating Visual Quality and Alignment Level for Text-to-Vision Content | [
"Zicheng Zhang",
"Tengchuan Kou",
"Shushi Wang",
"Chunyi Li",
"Wei Sun",
"Wei Wang",
"Xiaoyu Li",
"Zongyu Wang",
"Xuezhi Cao",
"Xiongkuo Min",
"Xiaohong Liu",
"Guangtao Zhai"
] | https://github.com/zzc-1998/Q-Eval | Evaluating text-to-vision content hinges on two crucial aspects: visual quality and alignment. While significant progress has been made in developing objective models to assess these dimensions, the performance of such models heavily relies on the scale and quality of human annotations. According to Scaling Law, increasing the number of human-labeled instances follows a predictable pattern that enhances the performance of evaluation models. Therefore, we introduce a comprehensive dataset designed to Evaluate Visual quality and Alignment Level for text-to-vision content (Q-EVAL-100K), featuring the largest collection of human-labeled Mean Opinion Scores (MOS) for the mentioned two aspects. The Q-EVAL-100K dataset encompasses both text-to-image and text-to-video models, with 960K human annotations specifically focused on visual quality and alignment for 100K instances (60K images and 40K videos). Leveraging this dataset with context prompt, we propose Q-Eval-Score, a unified model capable of evaluating both visual quality and alignment with special improvements for handling long-text prompt alignment. Experimental results indicate that the proposed Q-Eval-Score achieves superior performance on both visual quality and alignment, with strong generalization capabilities across other benchmarks. These findings highlight the significant value of the Q-EVAL-100K dataset. Data and codes will be available at https://github.com/zzc-1998/Q-Eval. |
2025-03-05T00:00:00 | 2503.02783 | IterPref: Focal Preference Learning for Code Generation via Iterative Debugging | [
"Jie Wu",
"Haoling Li",
"Xin Zhang",
"Jianwen Luo",
"Yangyu Huang",
"Ruihang Chu",
"Yujiu Yang",
"Scarlett Li"
] | Preference learning enhances Code LLMs beyond supervised fine-tuning by leveraging relative quality comparisons. Existing methods construct preference pairs from candidates based on test case success, treating the higher pass rate sample as positive and the lower as negative. However, this approach does not pinpoint specific errors in the code, which prevents the model from learning more informative error correction patterns, as aligning failing code as a whole lacks the granularity needed to capture meaningful error-resolution relationships. To address these issues, we propose IterPref, a new preference alignment framework that mimics human iterative debugging to refine Code LLMs. IterPref explicitly locates error regions and aligns the corresponding tokens via a tailored DPO algorithm. To generate informative pairs, we introduce the CodeFlow dataset, where samples are iteratively refined until passing tests, with modifications capturing error corrections. Extensive experiments show that a diverse suite of Code LLMs equipped with IterPref achieves significant performance gains in code generation and improves on challenging tasks like BigCodeBench. In-depth analysis reveals that IterPref yields fewer errors. Our code and data will be made publicaly available. |
|
2025-03-05T00:00:00 | 2503.00876 | Improve Representation for Imbalanced Regression through Geometric Constraints | [
"Zijian Dong",
"Yilei Wu",
"Chongyao Chen",
"Yingtian Zou",
"Yichi Zhang",
"Juan Helen Zhou"
] | In representation learning, uniformity refers to the uniform feature distribution in the latent space (i.e., unit hypersphere). Previous work has shown that improving uniformity contributes to the learning of under-represented classes. However, most of the previous work focused on classification; the representation space of imbalanced regression remains unexplored. Classification-based methods are not suitable for regression tasks because they cluster features into distinct groups without considering the continuous and ordered nature essential for regression. In a geometric aspect, we uniquely focus on ensuring uniformity in the latent space for imbalanced regression through two key losses: enveloping and homogeneity. The enveloping loss encourages the induced trace to uniformly occupy the surface of a hypersphere, while the homogeneity loss ensures smoothness, with representations evenly spaced at consistent intervals. Our method integrates these geometric principles into the data representations via a Surrogate-driven Representation Learning (SRL) framework. Experiments with real-world regression and operator learning tasks highlight the importance of uniformity in imbalanced regression and validate the efficacy of our geometry-based loss functions. |
|
2025-03-05T00:00:00 | 2503.01842 | Discrete-Time Hybrid Automata Learning: Legged Locomotion Meets Skateboarding | [
"Hang Liu",
"Sangli Teng",
"Ben Liu",
"Wei Zhang",
"Maani Ghaffari"
] | This paper introduces Discrete-time Hybrid Automata Learning (DHAL), a framework using on-policy Reinforcement Learning to identify and execute mode-switching without trajectory segmentation or event function learning. Hybrid dynamical systems, which include continuous flow and discrete mode switching, can model robotics tasks like legged robot locomotion. Model-based methods usually depend on predefined gaits, while model-free approaches lack explicit mode-switching knowledge. Current methods identify discrete modes via segmentation before regressing continuous flow, but learning high-dimensional complex rigid body dynamics without trajectory labels or segmentation is a challenging open problem. Our approach incorporates a beta policy distribution and a multi-critic architecture to model contact-guided motions, exemplified by a challenging quadrupedal robot skateboard task. We validate our method through simulations and real-world tests, demonstrating robust performance in hybrid dynamical systems. |
|
2025-03-05T00:00:00 | 2503.02812 | Q-Filters: Leveraging QK Geometry for Efficient KV Cache Compression | [
"Nathan Godey",
"Alessio Devoto",
"Yu Zhao",
"Simone Scardapane",
"Pasquale Minervini",
"Éric de la Clergerie",
"Benoît Sagot"
] | Autoregressive language models rely on a Key-Value (KV) Cache, which avoids re-computing past hidden states during generation, making it faster. As model sizes and context lengths grow, the KV Cache becomes a significant memory bottleneck, which calls for compression methods that limit its size during generation. In this paper, we discover surprising properties of Query (Q) and Key (K) vectors that allow us to efficiently approximate attention scores without computing the attention maps. We propose Q-Filters, a training-free KV Cache compression method that filters out less crucial Key-Value pairs based on a single context-agnostic projection. Contrarily to many alternatives, Q-Filters is compatible with FlashAttention, as it does not require direct access to attention weights. Experimental results in long-context settings demonstrate that Q-Filters is competitive with attention-based compression methods such as SnapKV in retrieval tasks while consistently outperforming efficient compression schemes such as Streaming-LLM in generation setups. Notably, Q-Filters achieves a 99% accuracy in the needle-in-a-haystack task with a x32 compression level while reducing the generation perplexity drop by up to 65% in text generation compared to Streaming-LLM. |
|
2025-03-05T00:00:00 | 2503.02152 | Tabby: Tabular Data Synthesis with Language Models | [
"Sonia Cromp",
"Satya Sai Srinath Namburi GNVV",
"Mohammed Alkhudhayri",
"Catherine Cao",
"Samuel Guo",
"Nicholas Roberts",
"Frederic Sala"
] | While advances in large language models (LLMs) have greatly improved the quality of synthetic text data in recent years, synthesizing tabular data has received relatively less attention. We address this disparity with Tabby, a simple but powerful post-training modification to the standard Transformer language model architecture, enabling its use for tabular dataset synthesis. Tabby enables the representation of differences across columns using Gated Mixture-of-Experts, with column-specific sets of parameters. Empirically, Tabby results in data quality near or equal to that of real data. By pairing our novel LLM table training technique, Plain, with Tabby, we observe up to a 44% improvement in quality over previous methods. We also show that Tabby extends beyond tables to more general structured data, reaching parity with real data on a nested JSON dataset as well. |
|
2025-03-05T00:00:00 | 2503.02304 | A Token-level Text Image Foundation Model for Document Understanding | [
"Tongkun Guan",
"Zining Wang",
"Pei Fu",
"Zhengtao Guo",
"Wei Shen",
"Kai Zhou",
"Tiezhu Yue",
"Chen Duan",
"Hao Sun",
"Qianyi Jiang",
"Junfeng Luo",
"Xiaokang Yang"
] | In recent years, general visual foundation models (VFMs) have witnessed increasing adoption, particularly as image encoders for popular multi-modal large language models (MLLMs). However, without semantically fine-grained supervision, these models still encounter fundamental prediction errors in the context of downstream text-image-related tasks, i.e., perception, understanding and reasoning with images containing small and dense texts. To bridge this gap, we develop TokenOCR, the first token-level visual foundation model specifically tailored for text-image-related tasks, designed to support a variety of traditional downstream applications. To facilitate the pretraining of TokenOCR, we also devise a high-quality data production pipeline that constructs the first token-level image text dataset, TokenIT, comprising 20 million images and 1.8 billion token-mask pairs. Furthermore, leveraging this foundation with exceptional image-as-text capability, we seamlessly replace previous VFMs with TokenOCR to construct a document-level MLLM, TokenVL, for VQA-based document understanding tasks. Finally, extensive experiments demonstrate the effectiveness of TokenOCR and TokenVL. Code, datasets, and weights will be available at https://token-family.github.io/TokenOCR_project. |
|
2025-03-05T00:00:00 | 2503.02823 | A Multimodal Symphony: Integrating Taste and Sound through Generative AI | [
"Matteo Spanio",
"Massimiliano Zampini",
"Antonio Rodà",
"Franco Pierucci"
] | In recent decades, neuroscientific and psychological research has traced direct relationships between taste and auditory perceptions. This article explores multimodal generative models capable of converting taste information into music, building on this foundational research. We provide a brief review of the state of the art in this field, highlighting key findings and methodologies. We present an experiment in which a fine-tuned version of a generative music model (MusicGEN) is used to generate music based on detailed taste descriptions provided for each musical piece. The results are promising: according the participants' (n=111) evaluation, the fine-tuned model produces music that more coherently reflects the input taste descriptions compared to the non-fine-tuned model. This study represents a significant step towards understanding and developing embodied interactions between AI, sound, and taste, opening new possibilities in the field of generative AI. We release our dataset, code and pre-trained model at: https://osf.io/xs5jy/. |
|
2025-03-05T00:00:00 | 2503.00200 | Unified Video Action Model | [
"Shuang Li",
"Yihuai Gao",
"Dorsa Sadigh",
"Shuran Song"
] | A unified video and action model holds significant promise for robotics, where videos provide rich scene information for action prediction, and actions provide dynamics information for video prediction. However, effectively combining video generation and action prediction remains challenging, and current video generation-based methods struggle to match the performance of direct policy learning in action accuracy and inference speed. To bridge this gap, we introduce the Unified Video Action model (UVA), which jointly optimizes video and action predictions to achieve both high accuracy and efficient action inference. The key lies in learning a joint video-action latent representation and decoupling video-action decoding. The joint latent representation bridges the visual and action domains, effectively modeling the relationship between video and action sequences. Meanwhile, the decoupled decoding, powered by two lightweight diffusion heads, enables high-speed action inference by bypassing video generation during inference. Such a unified framework further enables versatile functionality through masked input training. By selectively masking actions or videos, a single model can tackle diverse tasks beyond policy learning, such as forward and inverse dynamics modeling and video generation. Via an extensive set of experiments, we demonstrate that UVA can serve as a general-purpose solution for a wide range of robotics tasks, such as policy learning, forward/inverse dynamics and video observation prediction, without compromising performance compared to methods tailored for specific applications. Results are best viewed on https://unified-video-action-model.github.io/. |
|
2025-03-05T00:00:00 | 2503.00069 | Societal Alignment Frameworks Can Improve LLM Alignment | [
"Karolina Stańczak",
"Nicholas Meade",
"Mehar Bhatia",
"Hattie Zhou",
"Konstantin Böttinger",
"Jeremy Barnes",
"Jason Stanley",
"Jessica Montgomery",
"Richard Zemel",
"Nicolas Papernot",
"Nicolas Chapados",
"Denis Therien",
"Timothy P. Lillicrap",
"Ana Marasović",
"Sylvie Delacroix",
"Gillian K. Hadfield",
"Siva Reddy"
] | Recent progress in large language models (LLMs) has focused on producing responses that meet human expectations and align with shared values - a process coined alignment. However, aligning LLMs remains challenging due to the inherent disconnect between the complexity of human values and the narrow nature of the technological approaches designed to address them. Current alignment methods often lead to misspecified objectives, reflecting the broader issue of incomplete contracts, the impracticality of specifying a contract between a model developer, and the model that accounts for every scenario in LLM alignment. In this paper, we argue that improving LLM alignment requires incorporating insights from societal alignment frameworks, including social, economic, and contractual alignment, and discuss potential solutions drawn from these domains. Given the role of uncertainty within societal alignment frameworks, we then investigate how it manifests in LLM alignment. We end our discussion by offering an alternative view on LLM alignment, framing the underspecified nature of its objectives as an opportunity rather than perfect their specification. Beyond technical improvements in LLM alignment, we discuss the need for participatory alignment interface designs. |
|
2025-03-05T00:00:00 | 2503.03651 | DoraCycle: Domain-Oriented Adaptation of Unified Generative Model in Multimodal Cycles | [
"Rui Zhao",
"Weijia Mao",
"Mike Zheng Shou"
] | https://github.com/showlab/DoraCycle | Adapting generative models to specific domains presents an effective solution for satisfying specialized requirements. However, adapting to some complex domains remains challenging, especially when these domains require substantial paired data to capture the targeted distributions. Since unpaired data from a single modality, such as vision or language, is more readily available, we utilize the bidirectional mappings between vision and language learned by the unified generative model to enable training on unpaired data for domain adaptation. Specifically, we propose DoraCycle, which integrates two multimodal cycles: text-to-image-to-text and image-to-text-to-image. The model is optimized through cross-entropy loss computed at the cycle endpoints, where both endpoints share the same modality. This facilitates self-evolution of the model without reliance on annotated text-image pairs. Experimental results demonstrate that for tasks independent of paired knowledge, such as stylization, DoraCycle can effectively adapt the unified model using only unpaired data. For tasks involving new paired knowledge, such as specific identities, a combination of a small set of paired image-text examples and larger-scale unpaired data is sufficient for effective domain-oriented adaptation. The code will be released at https://github.com/showlab/DoraCycle. |
2025-03-06T00:00:00 | 2503.00329 | ABC: Achieving Better Control of Multimodal Embeddings using VLMs | [
"Benjamin Schneider",
"Florian Kerschbaum",
"Wenhu Chen"
] | Visual embedding models excel at zero-shot tasks like visual retrieval and classification. However, these models cannot be used for tasks that contain ambiguity or require user instruction. These tasks necessitate a multimodal embedding model, which outputs embeddings that combine visual and natural language input. Existing CLIP-based approaches embed images and text independently, and fuse the result. We find that this results in weak interactions between modalities, and poor user control over the representation. We introduce ABC, an open-source multimodal embedding model that uses a vision-language model backbone to deeply integrate image features with natural language instructions. ABC achieves bestfor-size performance on MSCOCO image-to-text retrieval and is the top performing model on classification and VQA tasks in the Massive Multimodal Embedding Benchmark. With a strongly unified vision-language representation, ABC can use natural language to solve subtle and potentially ambiguous visual retrieval problems. To evaluate this capability, we design CtrlBench, a benchmark that requires interleaving textual instructions with image content for correct retrieval. ABC advances the state of multimodal embeddings by offering high-quality representations and flexible natural language control. Our model and datasets are available at our project page. |
|
2025-03-06T00:00:00 | 2503.02951 | KodCode: A Diverse, Challenging, and Verifiable Synthetic Dataset for Coding | [
"Zhangchen Xu",
"Yang Liu",
"Yueqin Yin",
"Mingyuan Zhou",
"Radha Poovendran"
] | We introduce KodCode, a synthetic dataset that addresses the persistent challenge of acquiring high-quality, verifiable training data across diverse difficulties and domains for training Large Language Models for coding. Existing code-focused resources typically fail to ensure either the breadth of coverage (e.g., spanning simple coding tasks to advanced algorithmic problems) or verifiable correctness (e.g., unit tests). In contrast, KodCode comprises question-solution-test triplets that are systematically validated via a self-verification procedure. Our pipeline begins by synthesizing a broad range of coding questions, then generates solutions and test cases with additional attempts allocated to challenging problems. Finally, post-training data synthesis is done by rewriting questions into diverse formats and generating responses under a test-based reject sampling procedure from a reasoning model (DeepSeek R1). This pipeline yields a large-scale, robust and diverse coding dataset. KodCode is suitable for supervised fine-tuning and the paired unit tests also provide great potential for RL tuning. Fine-tuning experiments on coding benchmarks (HumanEval(+), MBPP(+), BigCodeBench, and LiveCodeBench) demonstrate that KodCode-tuned models achieve state-of-the-art performance, surpassing models like Qwen2.5-Coder-32B-Instruct and DeepSeek-R1-Distill-Llama-70B. |
|
2025-03-06T00:00:00 | 2503.00865 | Babel: Open Multilingual Large Language Models Serving Over 90% of Global Speakers | [
"Yiran Zhao",
"Chaoqun Liu",
"Yue Deng",
"Jiahao Ying",
"Mahani Aljunied",
"Zhaodonghui Li",
"Lidong Bing",
"Hou Pong Chan",
"Yu Rong",
"Deli Zhao",
"Wenxuan Zhang"
] | Large language models (LLMs) have revolutionized natural language processing (NLP), yet open-source multilingual LLMs remain scarce, with existing models often limited in language coverage. Such models typically prioritize well-resourced languages, while widely spoken but under-resourced languages are often overlooked. To address this disparity, we introduce Babel, an open multilingual LLM that covers the top 25 languages by number of speakers, supports over 90% of the global population, and includes many languages neglected by other open multilingual LLMs. Unlike traditional continue pretraining approaches, Babel expands its parameter count through a layer extension technique that elevates Babel's performance ceiling. We introduce two variants: Babel-9B, designed for efficient inference and fine-tuning, and Babel-83B, which sets a new standard for open multilingual LLMs. Extensive evaluations on multilingual tasks demonstrate its superior performance compared to open LLMs of comparable size. In addition, using open-source supervised fine-tuning datasets, Babel achieves remarkable performance, with Babel-9B-Chat leading among 10B-sized LLMs and Babel-83B-Chat setting a new standard for multilingual tasks, reaching the same level of commercial models. |
|
2025-03-06T00:00:00 | 2503.00502 | Interact, Instruct to Improve: A LLM-Driven Parallel Actor-Reasoner Framework for Enhancing Autonomous Vehicle Interactions | [
"Shiyu Fang",
"Jiaqi Liu",
"Chengkai Xu",
"Chen Lv",
"Peng Hang",
"Jian Sun"
] | https://github.com/FanGShiYuu/Actor-Reasoner | Autonomous Vehicles (AVs) have entered the commercialization stage, but their limited ability to interact and express intentions still poses challenges in interactions with Human-driven Vehicles (HVs). Recent advances in large language models (LLMs) enable bidirectional human-machine communication, but the conflict between slow inference speed and the need for real-time decision-making challenges practical deployment. To address these issues, this paper introduces a parallel Actor-Reasoner framework designed to enable explicit bidirectional AV-HV interactions across multiple scenarios. First, by facilitating interactions between the LLM-driven Reasoner and heterogeneous simulated HVs during training, an interaction memory database, referred to as the Actor, is established. Then, by introducing the memory partition module and the two-layer memory retrieval module, the Actor's ability to handle heterogeneous HVs is significantly enhanced. Ablation studies and comparisons with other decision-making methods demonstrate that the proposed Actor-Reasoner framework significantly improves safety and efficiency. Finally, with the combination of the external Human-Machine Interface (eHMI) information derived from Reasoner's reasoning and the feasible action solutions retrieved from the Actor, the effectiveness of the proposed Actor-Reasoner is confirmed in multi-scenario field interactions. Our code is available at https://github.com/FanGShiYuu/Actor-Reasoner. |
2025-03-06T00:00:00 | 2503.03751 | GEN3C: 3D-Informed World-Consistent Video Generation with Precise Camera Control | [
"Xuanchi Ren",
"Tianchang Shen",
"Jiahui Huang",
"Huan Ling",
"Yifan Lu",
"Merlin Nimier-David",
"Thomas Müller",
"Alexander Keller",
"Sanja Fidler",
"Jun Gao"
] | We present GEN3C, a generative video model with precise Camera Control and temporal 3D Consistency. Prior video models already generate realistic videos, but they tend to leverage little 3D information, leading to inconsistencies, such as objects popping in and out of existence. Camera control, if implemented at all, is imprecise, because camera parameters are mere inputs to the neural network which must then infer how the video depends on the camera. In contrast, GEN3C is guided by a 3D cache: point clouds obtained by predicting the pixel-wise depth of seed images or previously generated frames. When generating the next frames, GEN3C is conditioned on the 2D renderings of the 3D cache with the new camera trajectory provided by the user. Crucially, this means that GEN3C neither has to remember what it previously generated nor does it have to infer the image structure from the camera pose. The model, instead, can focus all its generative power on previously unobserved regions, as well as advancing the scene state to the next frame. Our results demonstrate more precise camera control than prior work, as well as state-of-the-art results in sparse-view novel view synthesis, even in challenging settings such as driving scenes and monocular dynamic video. Results are best viewed in videos. Check out our webpage! https://research.nvidia.com/labs/toronto-ai/GEN3C/ |
|
2025-03-06T00:00:00 | 2503.01763 | Retrieval Models Aren't Tool-Savvy: Benchmarking Tool Retrieval for Large Language Models | [
"Zhengliang Shi",
"Yuhan Wang",
"Lingyong Yan",
"Pengjie Ren",
"Shuaiqiang Wang",
"Dawei Yin",
"Zhaochun Ren"
] | Tool learning aims to augment large language models (LLMs) with diverse tools, enabling them to act as agents for solving practical tasks. Due to the limited context length of tool-using LLMs, adopting information retrieval (IR) models to select useful tools from large toolsets is a critical initial step. However, the performance of IR models in tool retrieval tasks remains underexplored and unclear. Most tool-use benchmarks simplify this step by manually pre-annotating a small set of relevant tools for each task, which is far from the real-world scenarios. In this paper, we propose ToolRet, a heterogeneous tool retrieval benchmark comprising 7.6k diverse retrieval tasks, and a corpus of 43k tools, collected from existing datasets. We benchmark six types of models on ToolRet. Surprisingly, even the models with strong performance in conventional IR benchmarks, exhibit poor performance on ToolRet. This low retrieval quality degrades the task pass rate of tool-use LLMs. As a further step, we contribute a large-scale training dataset with over 200k instances, which substantially optimizes the tool retrieval ability of IR models. |
|
2025-03-06T00:00:00 | 2503.01729 | FLAME: A Federated Learning Benchmark for Robotic Manipulation | [
"Santiago Bou Betran",
"Alberta Longhini",
"Miguel Vasco",
"Yuchong Zhang",
"Danica Kragic"
] | Recent progress in robotic manipulation has been fueled by large-scale datasets collected across diverse environments. Training robotic manipulation policies on these datasets is traditionally performed in a centralized manner, raising concerns regarding scalability, adaptability, and data privacy. While federated learning enables decentralized, privacy-preserving training, its application to robotic manipulation remains largely unexplored. We introduce FLAME (Federated Learning Across Manipulation Environments), the first benchmark designed for federated learning in robotic manipulation. FLAME consists of: (i) a set of large-scale datasets of over 160,000 expert demonstrations of multiple manipulation tasks, collected across a wide range of simulated environments; (ii) a training and evaluation framework for robotic policy learning in a federated setting. We evaluate standard federated learning algorithms in FLAME, showing their potential for distributed policy learning and highlighting key challenges. Our benchmark establishes a foundation for scalable, adaptive, and privacy-aware robotic learning. |
|
2025-03-06T00:00:00 | 2503.01449 | Benchmarking Large Language Models for Multi-Language Software Vulnerability Detection | [
"Ting Zhang",
"Chengran Yang",
"Yindu Su",
"Martin Weyssow",
"Hung Nguyen",
"Tan Bui",
"Hong Jin Kang",
"Yikun Li",
"Eng Lieh Ouh",
"Lwin Khin Shar",
"David Lo"
] | Recent advancements in generative AI have led to the widespread adoption of large language models (LLMs) in software engineering, addressing numerous long-standing challenges. However, a comprehensive study examining the capabilities of LLMs in software vulnerability detection (SVD), a crucial aspect of software security, is currently lacking. Existing research primarily focuses on evaluating LLMs using C/C++ datasets. It typically explores only one or two strategies among prompt engineering, instruction tuning, and sequence classification fine-tuning for open-source LLMs. Consequently, there is a significant knowledge gap regarding the effectiveness of diverse LLMs in detecting vulnerabilities across various programming languages. To address this knowledge gap, we present a comprehensive empirical study evaluating the performance of LLMs on the SVD task. We have compiled a comprehensive dataset comprising 8,260 vulnerable functions in Python, 7,505 in Java, and 28,983 in JavaScript. We assess five open-source LLMs using multiple approaches, including prompt engineering, instruction tuning, and sequence classification fine-tuning. These LLMs are benchmarked against five fine-tuned small language models and two open-source static application security testing tools. Furthermore, we explore two avenues to improve LLM performance on SVD: a) Data perspective: Retraining models using downsampled balanced datasets. b) Model perspective: Investigating ensemble learning methods that combine predictions from multiple LLMs. Our comprehensive experiments demonstrate that SVD remains a challenging task for LLMs. This study provides a thorough understanding of the role of LLMs in SVD and offers practical insights for future advancements in leveraging generative AI to enhance software security practices. |
|
2025-03-06T00:00:00 | 2503.01378 | CognitiveDrone: A VLA Model and Evaluation Benchmark for Real-Time Cognitive Task Solving and Reasoning in UAVs | [
"Artem Lykov",
"Valerii Serpiva",
"Muhammad Haris Khan",
"Oleg Sautenkov",
"Artyom Myshlyaev",
"Grik Tadevosyan",
"Yasheerah Yaqoot",
"Dzmitry Tsetserukou"
] | This paper introduces CognitiveDrone, a novel Vision-Language-Action (VLA) model tailored for complex Unmanned Aerial Vehicles (UAVs) tasks that demand advanced cognitive abilities. Trained on a dataset comprising over 8,000 simulated flight trajectories across three key categories-Human Recognition, Symbol Understanding, and Reasoning-the model generates real-time 4D action commands based on first-person visual inputs and textual instructions. To further enhance performance in intricate scenarios, we propose CognitiveDrone-R1, which integrates an additional Vision-Language Model (VLM) reasoning module to simplify task directives prior to high-frequency control. Experimental evaluations using our open-source benchmark, CognitiveDroneBench, reveal that while a racing-oriented model (RaceVLA) achieves an overall success rate of 31.3%, the base CognitiveDrone model reaches 59.6%, and CognitiveDrone-R1 attains a success rate of 77.2%. These results demonstrate improvements of up to 30% in critical cognitive tasks, underscoring the effectiveness of incorporating advanced reasoning capabilities into UAV control systems. Our contributions include the development of a state-of-the-art VLA model for UAV control and the introduction of the first dedicated benchmark for assessing cognitive tasks in drone operations. The complete repository is available at cognitivedrone.github.io |
|
2025-03-06T00:00:00 | 2503.01372 | SwiLTra-Bench: The Swiss Legal Translation Benchmark | [
"Joel Niklaus",
"Jakob Merane",
"Luka Nenadic",
"Sina Ahmadi",
"Yingqiang Gao",
"Cyrill A. H. Chevalley",
"Claude Humbel",
"Christophe Gösken",
"Lorenzo Tanzi",
"Thomas Lüthi",
"Stefan Palombo",
"Spencer Poff",
"Boling Yang",
"Nan Wu",
"Matthew Guillod",
"Robin Mamié",
"Daniel Brunner",
"Julio Pereyra",
"Niko Grupen"
] | In Switzerland legal translation is uniquely important due to the country's four official languages and requirements for multilingual legal documentation. However, this process traditionally relies on professionals who must be both legal experts and skilled translators -- creating bottlenecks and impacting effective access to justice. To address this challenge, we introduce SwiLTra-Bench, a comprehensive multilingual benchmark of over 180K aligned Swiss legal translation pairs comprising laws, headnotes, and press releases across all Swiss languages along with English, designed to evaluate LLM-based translation systems. Our systematic evaluation reveals that frontier models achieve superior translation performance across all document types, while specialized translation systems excel specifically in laws but under-perform in headnotes. Through rigorous testing and human expert validation, we demonstrate that while fine-tuning open SLMs significantly improves their translation quality, they still lag behind the best zero-shot prompted frontier models such as Claude-3.5-Sonnet. Additionally, we present SwiLTra-Judge, a specialized LLM evaluation system that aligns best with human expert assessments. |
|
2025-03-06T00:00:00 | 2503.03278 | Enhancing Abnormality Grounding for Vision Language Models with Knowledge Descriptions | [
"Jun Li",
"Che Liu",
"Wenjia Bai",
"Rossella Arcucci",
"Cosmin I. Bercea",
"Julia A. Schnabel"
] | Visual Language Models (VLMs) have demonstrated impressive capabilities in visual grounding tasks. However, their effectiveness in the medical domain, particularly for abnormality detection and localization within medical images, remains underexplored. A major challenge is the complex and abstract nature of medical terminology, which makes it difficult to directly associate pathological anomaly terms with their corresponding visual features. In this work, we introduce a novel approach to enhance VLM performance in medical abnormality detection and localization by leveraging decomposed medical knowledge. Instead of directly prompting models to recognize specific abnormalities, we focus on breaking down medical concepts into fundamental attributes and common visual patterns. This strategy promotes a stronger alignment between textual descriptions and visual features, improving both the recognition and localization of abnormalities in medical images.We evaluate our method on the 0.23B Florence-2 base model and demonstrate that it achieves comparable performance in abnormality grounding to significantly larger 7B LLaVA-based medical VLMs, despite being trained on only 1.5% of the data used for such models. Experimental results also demonstrate the effectiveness of our approach in both known and previously unseen abnormalities, suggesting its strong generalization capabilities. |
|
2025-03-06T00:00:00 | 2503.01836 | CrowdSelect: Synthetic Instruction Data Selection with Multi-LLM Wisdom | [
"Yisen Li",
"Lingfeng Yang",
"Wenxuan Shen",
"Pan Zhou",
"Yao Wan",
"Weiwei Lin",
"Dongping Chen"
] | https://github.com/listentm/crowdselect | Distilling advanced Large Language Models' instruction-following capabilities into smaller models using a selected subset has become a mainstream approach in model training. While existing synthetic instruction data selection strategies rely mainly on single-dimensional signals (i.e., reward scores, model perplexity), they fail to capture the complexity of instruction-following across diverse fields. Therefore, we investigate more diverse signals to capture comprehensive instruction-response pair characteristics and propose three foundational metrics that leverage Multi-LLM wisdom, informed by (1) diverse LLM responses and (2) reward model assessment. Building upon base metrics, we propose CrowdSelect, an integrated metric incorporating a clustering-based approach to maintain response diversity. Our comprehensive experiments demonstrate that our foundation metrics consistently improve performance across 4 base models on MT-bench and Arena-Hard. CrowdSelect, efficiently incorporating all metrics, achieves state-of-the-art performance in both Full and LoRA fine-tuning, showing improvements of 4.81% on Arena-Hard and 11.1% on MT-bench with Llama-3.2-3b-instruct. We hope our findings will bring valuable insights for future research in this direction. Code are available at https://github.com/listentm/crowdselect. |
2025-03-06T00:00:00 | 2503.03044 | QE4PE: Word-level Quality Estimation for Human Post-Editing | [
"Gabriele Sarti",
"Vilém Zouhar",
"Grzegorz Chrupała",
"Ana Guerberof-Arenas",
"Malvina Nissim",
"Arianna Bisazza"
] | Word-level quality estimation (QE) detects erroneous spans in machine translations, which can direct and facilitate human post-editing. While the accuracy of word-level QE systems has been assessed extensively, their usability and downstream influence on the speed, quality and editing choices of human post-editing remain understudied. Our QE4PE study investigates the impact of word-level QE on machine translation (MT) post-editing in a realistic setting involving 42 professional post-editors across two translation directions. We compare four error-span highlight modalities, including supervised and uncertainty-based word-level QE methods, for identifying potential errors in the outputs of a state-of-the-art neural MT model. Post-editing effort and productivity are estimated by behavioral logs, while quality improvements are assessed by word- and segment-level human annotation. We find that domain, language and editors' speed are critical factors in determining highlights' effectiveness, with modest differences between human-made and automated QE highlights underlining a gap between accuracy and usability in professional workflows. |
|
2025-03-06T00:00:00 | 2502.20317 | Mixture of Structural-and-Textual Retrieval over Text-rich Graph Knowledge Bases | [
"Yongjia Lei",
"Haoyu Han",
"Ryan A. Rossi",
"Franck Dernoncourt",
"Nedim Lipka",
"Mahantesh M Halappanavar",
"Jiliang Tang",
"Yu Wang"
] | https://github.com/Yoega/MoR | Text-rich Graph Knowledge Bases (TG-KBs) have become increasingly crucial for answering queries by providing textual and structural knowledge. However, current retrieval methods often retrieve these two types of knowledge in isolation without considering their mutual reinforcement and some hybrid methods even bypass structural retrieval entirely after neighboring aggregation. To fill in this gap, we propose a Mixture of Structural-and-Textual Retrieval (MoR) to retrieve these two types of knowledge via a Planning-Reasoning-Organizing framework. In the Planning stage, MoR generates textual planning graphs delineating the logic for answering queries. Following planning graphs, in the Reasoning stage, MoR interweaves structural traversal and textual matching to obtain candidates from TG-KBs. In the Organizing stage, MoR further reranks fetched candidates based on their structural trajectory. Extensive experiments demonstrate the superiority of MoR in harmonizing structural and textual retrieval with insights, including uneven retrieving performance across different query logics and the benefits of integrating structural trajectories for candidate reranking. Our code is available at https://github.com/Yoega/MoR. |
2025-03-06T00:00:00 | 2502.18860 | Exploring Rewriting Approaches for Different Conversational Tasks | [
"Md Mehrab Tanjim",
"Ryan A. Rossi",
"Mike Rimer",
"Xiang Chen",
"Sungchul Kim",
"Vaishnavi Muppala",
"Tong Yu",
"Zhengmian Hu",
"Ritwik Sinha",
"Wei Zhang",
"Iftikhar Ahamath Burhanuddin",
"Franck Dernoncourt"
] | Conversational assistants often require a question rewriting algorithm that leverages a subset of past interactions to provide a more meaningful (accurate) answer to the user's question or request. However, the exact rewriting approach may often depend on the use case and application-specific tasks supported by the conversational assistant, among other constraints. In this paper, we systematically investigate two different approaches, denoted as rewriting and fusion, on two fundamentally different generation tasks, including a text-to-text generation task and a multimodal generative task that takes as input text and generates a visualization or data table that answers the user's question. Our results indicate that the specific rewriting or fusion approach highly depends on the underlying use case and generative task. In particular, we find that for a conversational question-answering assistant, the query rewriting approach performs best, whereas for a data analysis assistant that generates visualizations and data tables based on the user's conversation with the assistant, the fusion approach works best. Notably, we explore two datasets for the data analysis assistant use case, for short and long conversations, and we find that query fusion always performs better, whereas for the conversational text-based question-answering, the query rewrite approach performs best. |
|
2025-03-06T00:00:00 | 2503.01933 | Fine-Tuning Small Language Models for Domain-Specific AI: An Edge AI Perspective | [
"Rakshit Aralimatti",
"Syed Abdul Gaffar Shakhadri",
"Kruthika KR",
"Kartik Basavaraj Angadi"
] | Deploying large scale language models on edge devices faces inherent challenges such as high computational demands, energy consumption, and potential data privacy risks. This paper introduces the Shakti Small Language Models (SLMs) Shakti-100M, Shakti-250M, and Shakti-500M which target these constraints headon. By combining efficient architectures, quantization techniques, and responsible AI principles, the Shakti series enables on-device intelligence for smartphones, smart appliances, IoT systems, and beyond. We provide comprehensive insights into their design philosophy, training pipelines, and benchmark performance on both general tasks (e.g., MMLU, Hellaswag) and specialized domains (healthcare, finance, and legal). Our findings illustrate that compact models, when carefully engineered and fine-tuned, can meet and often exceed expectations in real-world edge-AI scenarios. |
|
2025-03-06T00:00:00 | 2503.03746 | Process-based Self-Rewarding Language Models | [
"Shimao Zhang",
"Xiao Liu",
"Xin Zhang",
"Junxiao Liu",
"Zheheng Luo",
"Shujian Huang",
"Yeyun Gong"
] | Large Language Models have demonstrated outstanding performance across various downstream tasks and have been widely applied in multiple scenarios. Human-annotated preference data is used for training to further improve LLMs' performance, which is constrained by the upper limit of human performance. Therefore, Self-Rewarding method has been proposed, where LLMs generate training data by rewarding their own outputs. However, the existing self-rewarding paradigm is not effective in mathematical reasoning scenarios and may even lead to a decline in performance. In this work, we propose the Process-based Self-Rewarding pipeline for language models, which introduces long-thought reasoning, step-wise LLM-as-a-Judge, and step-wise preference optimization within the self-rewarding paradigm. Our new paradigm successfully enhances the performance of LLMs on multiple mathematical reasoning benchmarks through iterative Process-based Self-Rewarding, demonstrating the immense potential of self-rewarding to achieve LLM reasoning that may surpass human capabilities. |
|
2025-03-06T00:00:00 | 2503.00307 | Remasking Discrete Diffusion Models with Inference-Time Scaling | [
"Guanghan Wang",
"Yair Schiff",
"Subham Sekhar Sahoo",
"Volodymyr Kuleshov"
] | Part of the success of diffusion models stems from their ability to perform iterative refinement, i.e., repeatedly correcting outputs during generation. However, modern masked discrete diffusion lacks this capability: when a token is generated, it cannot be updated again, even when it introduces an error. Here, we address this limitation by introducing the remasking diffusion model (ReMDM) sampler, a method that can be applied to pretrained masked diffusion models in a principled way and that is derived from a discrete diffusion model with a custom remasking backward process. Most interestingly, ReMDM endows discrete diffusion with a form of inference-time compute scaling. By increasing the number of sampling steps, ReMDM generates natural language outputs that approach the quality of autoregressive models, whereas when the computation budget is limited, ReMDM better maintains quality. ReMDM also improves sample quality of masked diffusion models for discretized images, and in scientific domains such as molecule design, ReMDM facilitates diffusion guidance and pushes the Pareto frontier of controllability relative to classical masking and uniform noise diffusion. We provide the code along with a blog post on the project page: https://remdm.github.io. |
|
2025-03-06T00:00:00 | 2503.02954 | Reliable and Efficient Multi-Agent Coordination via Graph Neural Network Variational Autoencoders | [
"Yue Meng",
"Nathalie Majcherczyk",
"Wenliang Liu",
"Scott Kiesel",
"Chuchu Fan",
"Federico Pecora"
] | Multi-agent coordination is crucial for reliable multi-robot navigation in shared spaces such as automated warehouses. In regions of dense robot traffic, local coordination methods may fail to find a deadlock-free solution. In these scenarios, it is appropriate to let a central unit generate a global schedule that decides the passing order of robots. However, the runtime of such centralized coordination methods increases significantly with the problem scale. In this paper, we propose to leverage Graph Neural Network Variational Autoencoders (GNN-VAE) to solve the multi-agent coordination problem at scale faster than through centralized optimization. We formulate the coordination problem as a graph problem and collect ground truth data using a Mixed-Integer Linear Program (MILP) solver. During training, our learning framework encodes good quality solutions of the graph problem into a latent space. At inference time, solution samples are decoded from the sampled latent variables, and the lowest-cost sample is selected for coordination. Finally, the feasible proposal with the highest performance index is selected for the deployment. By construction, our GNN-VAE framework returns solutions that always respect the constraints of the considered coordination problem. Numerical results show that our approach trained on small-scale problems can achieve high-quality solutions even for large-scale problems with 250 robots, being much faster than other baselines. Project page: https://mengyuest.github.io/gnn-vae-coord |
|
2025-03-06T00:00:00 | 2503.02924 | Diverse Controllable Diffusion Policy with Signal Temporal Logic | [
"Yue Meng",
"Chuchu fan"
] | https://github.com/mengyuest/pSTL-diffusion-policy | Generating realistic simulations is critical for autonomous system applications such as self-driving and human-robot interactions. However, driving simulators nowadays still have difficulty in generating controllable, diverse, and rule-compliant behaviors for road participants: Rule-based models cannot produce diverse behaviors and require careful tuning, whereas learning-based methods imitate the policy from data but are not designed to follow the rules explicitly. Besides, the real-world datasets are by nature "single-outcome", making the learning method hard to generate diverse behaviors. In this paper, we leverage Signal Temporal Logic (STL) and Diffusion Models to learn controllable, diverse, and rule-aware policy. We first calibrate the STL on the real-world data, then generate diverse synthetic data using trajectory optimization, and finally learn the rectified diffusion policy on the augmented dataset. We test on the NuScenes dataset and our approach can achieve the most diverse rule-compliant trajectories compared to other baselines, with a runtime 1/17X to the second-best approach. In the closed-loop testing, our approach reaches the highest diversity, rule satisfaction rate, and the least collision rate. Our method can generate varied characteristics conditional on different STL parameters in testing. A case study on human-robot encounter scenarios shows our approach can generate diverse and closed-to-oracle trajectories. The annotation tool, augmented dataset, and code are available at https://github.com/mengyuest/pSTL-diffusion-policy. |
2025-03-06T00:00:00 | 2503.02003 | HoT: Highlighted Chain of Thought for Referencing Supporting Facts from Inputs | [
"Tin Nguyen",
"Logan Bolton",
"Mohammad Reza Taesiri",
"Anh Totti Nguyen"
] | An Achilles heel of Large Language Models (LLMs) is their tendency to hallucinate non-factual statements. A response mixed of factual and non-factual statements poses a challenge for humans to verify and accurately base their decisions on. To combat this problem, we propose Highlighted Chain-of-Thought Prompting (HoT), a technique for prompting LLMs to generate responses with XML tags that ground facts to those provided in the query. That is, given an input question, LLMs would first re-format the question to add XML tags highlighting key facts, and then, generate a response with highlights over the facts referenced from the input. Interestingly, in few-shot settings, HoT outperforms vanilla chain of thought prompting (CoT) on a wide range of 17 tasks from arithmetic, reading comprehension to logical reasoning. When asking humans to verify LLM responses, highlights help time-limited participants to more accurately and efficiently recognize when LLMs are correct. Yet, surprisingly, when LLMs are wrong, HoTs tend to make users believe that an answer is correct. |
|
2025-03-07T00:00:00 | 2503.04625 | START: Self-taught Reasoner with Tools | [
"Chengpeng Li",
"Mingfeng Xue",
"Zhenru Zhang",
"Jiaxi Yang",
"Beichen Zhang",
"Xiang Wang",
"Bowen Yu",
"Binyuan Hui",
"Junyang Lin",
"Dayiheng Liu"
] | Large reasoning models (LRMs) like OpenAI-o1 and DeepSeek-R1 have demonstrated remarkable capabilities in complex reasoning tasks through the utilization of long Chain-of-thought (CoT). However, these models often suffer from hallucinations and inefficiencies due to their reliance solely on internal reasoning processes. In this paper, we introduce START (Self-Taught Reasoner with Tools), a novel tool-integrated long CoT reasoning LLM that significantly enhances reasoning capabilities by leveraging external tools. Through code execution, START is capable of performing complex computations, self-checking, exploring diverse methods, and self-debugging, thereby addressing the limitations of LRMs. The core innovation of START lies in its self-learning framework, which comprises two key techniques: 1) Hint-infer: We demonstrate that inserting artificially designed hints (e.g., ``Wait, maybe using Python here is a good idea.'') during the inference process of a LRM effectively stimulates its ability to utilize external tools without the need for any demonstration data. Hint-infer can also serve as a simple and effective sequential test-time scaling method; 2) Hint Rejection Sampling Fine-Tuning (Hint-RFT): Hint-RFT combines Hint-infer and RFT by scoring, filtering, and modifying the reasoning trajectories with tool invocation generated by a LRM via Hint-infer, followed by fine-tuning the LRM. Through this framework, we have fine-tuned the QwQ-32B model to achieve START. On PhD-level science QA (GPQA), competition-level math benchmarks (AMC23, AIME24, AIME25), and the competition-level code benchmark (LiveCodeBench), START achieves accuracy rates of 63.6%, 95.0%, 66.7%, 47.1%, and 47.3%, respectively. It significantly outperforms the base QwQ-32B and achieves performance comparable to the state-of-the-art open-weight model R1-Distill-Qwen-32B and the proprietary model o1-Preview. |
|
2025-03-07T00:00:00 | 2503.04222 | FuseChat-3.0: Preference Optimization Meets Heterogeneous Model Fusion | [
"Ziyi Yang",
"Fanqi Wan",
"Longguang Zhong",
"Canbin Huang",
"Guosheng Liang",
"Xiaojun Quan"
] | https://github.com/SLIT-AI/FuseChat-3.0 | We introduce FuseChat-3.0, a suite of large language models (LLMs) developed by integrating the strengths of heterogeneous source LLMs into more compact target LLMs. Our source models include the powerful Gemma-2-27B-it, Mistral-Large-Instruct-2407, Qwen-2.5-72B-Instruct, and Llama-3.1-70B-Instruct. For target models, we focus on three widely-used smaller variants-Llama-3.1-8B-Instruct, Gemma-2-9B-it, and Qwen-2.5-7B-Instruct-along with two ultra-compact options, Llama-3.2-3B-Instruct and Llama-3.2-1B-Instruct. To leverage the diverse capabilities of these source models, we develop a specialized data construction protocol tailored to various tasks and domains. The FuseChat-3.0 training pipeline consists of two key stages: (1) supervised fine-tuning (SFT) to align the target and source model distributions, and (2) Direct Preference Optimization (DPO) to apply preferences from multiple source LLMs to fine-tune the target model. The resulting FuseChat-3.0 models exhibit significant performance gains across tasks such as instruction following, general knowledge, mathematics, and coding. As illustrated in Figure 1, using Llama-3.1-8B-Instruct as the target model, our fusion approach achieves an average improvement of 6.8 points across 14 benchmarks. Moreover, it demonstrates remarkable gains of 37.1 points and 30.1 points on the instruction-following benchmarks AlpacaEval-2 and Arena-Hard, respectively. Our code, models, and datasets are available at https://github.com/SLIT-AI/FuseChat-3.0. |
2025-03-07T00:00:00 | 2503.04378 | Dedicated Feedback and Edit Models Empower Inference-Time Scaling for Open-Ended General-Domain Tasks | [
"Zhilin Wang",
"Jiaqi Zeng",
"Olivier Delalleau",
"Daniel Egert",
"Ellie Evans",
"Hoo-Chang Shin",
"Felipe Soares",
"Yi Dong",
"Oleksii Kuchaiev"
] | Inference-Time Scaling has been critical to the success of recent models such as OpenAI o1 and DeepSeek R1. However, many techniques used to train models for inference-time scaling require tasks to have answers that can be verified, limiting their application to domains such as math, coding and logical reasoning. We take inspiration from how humans make first attempts, ask for detailed feedback from others and make improvements based on such feedback across a wide spectrum of open-ended endeavors. To this end, we collect data for and train dedicated Feedback and Edit Models that are capable of performing inference-time scaling for open-ended general-domain tasks. In our setup, one model generates an initial response, which are given feedback by a second model, that are then used by a third model to edit the response. We show that performance on Arena Hard, a benchmark strongly predictive of Chatbot Arena Elo can be boosted by scaling the number of initial response drafts, effective feedback and edited responses. When scaled optimally, our setup based on 70B models from the Llama 3 family can reach SoTA performance on Arena Hard at 92.7 as of 5 Mar 2025, surpassing OpenAI o1-preview-2024-09-12 with 90.4 and DeepSeek R1 with 92.3. |
|
2025-03-07T00:00:00 | 2503.04598 | HybridNorm: Towards Stable and Efficient Transformer Training via Hybrid Normalization | [
"Zhijian Zhuo",
"Yutao Zeng",
"Ya Wang",
"Sijun Zhang",
"Jian Yang",
"Xiaoqing Li",
"Xun Zhou",
"Jinwen Ma"
] | https://github.com/BryceZhuo/HybridNorm | Transformers have become the de facto architecture for a wide range of machine learning tasks, particularly in large language models (LLMs). Despite their remarkable performance, challenges remain in training deep transformer networks, especially regarding the location of layer normalization. While Pre-Norm structures facilitate easier training due to their more prominent identity path, they often yield suboptimal performance compared to Post-Norm. In this paper, we propose HybridNorm, a straightforward yet effective hybrid normalization strategy that integrates the advantages of both Pre-Norm and Post-Norm approaches. Specifically, HybridNorm employs QKV normalization within the attention mechanism and Post-Norm in the feed-forward network (FFN) of each transformer block. This design not only stabilizes training but also enhances performance, particularly in the context of LLMs. Comprehensive experiments in both dense and sparse architectures show that HybridNorm consistently outperforms both Pre-Norm and Post-Norm approaches, achieving state-of-the-art results across various benchmarks. These findings highlight the potential of HybridNorm as a more stable and effective technique for improving the training and performance of deep transformer models. %Code will be made publicly available. Code is available at https://github.com/BryceZhuo/HybridNorm. |
2025-03-07T00:00:00 | 2503.04094 | PokéChamp: an Expert-level Minimax Language Agent | [
"Seth Karten",
"Andy Luu Nguyen",
"Chi Jin"
] | We introduce Pok\'eChamp, a minimax agent powered by Large Language Models (LLMs) for Pok\'emon battles. Built on a general framework for two-player competitive games, Pok\'eChamp leverages the generalist capabilities of LLMs to enhance minimax tree search. Specifically, LLMs replace three key modules: (1) player action sampling, (2) opponent modeling, and (3) value function estimation, enabling the agent to effectively utilize gameplay history and human knowledge to reduce the search space and address partial observability. Notably, our framework requires no additional LLM training. We evaluate Pok\'eChamp in the popular Gen 9 OU format. When powered by GPT-4o, it achieves a win rate of 76% against the best existing LLM-based bot and 84% against the strongest rule-based bot, demonstrating its superior performance. Even with an open-source 8-billion-parameter Llama 3.1 model, Pok\'eChamp consistently outperforms the previous best LLM-based bot, Pok\'ellmon powered by GPT-4o, with a 64% win rate. Pok\'eChamp attains a projected Elo of 1300-1500 on the Pok\'emon Showdown online ladder, placing it among the top 30%-10% of human players. In addition, this work compiles the largest real-player Pok\'emon battle dataset, featuring over 3 million games, including more than 500k high-Elo matches. Based on this dataset, we establish a series of battle benchmarks and puzzles to evaluate specific battling skills. We further provide key updates to the local game engine. We hope this work fosters further research that leverage Pok\'emon battle as benchmark to integrate LLM technologies with game-theoretic algorithms addressing general multiagent problems. Videos, code, and dataset available at https://sites.google.com/view/pokechamp-llm. |
|
2025-03-07T00:00:00 | 2503.04130 | Token-Efficient Long Video Understanding for Multimodal LLMs | [
"Jindong Jiang",
"Xiuyu Li",
"Zhijian Liu",
"Muyang Li",
"Guo Chen",
"Zhiqi Li",
"De-An Huang",
"Guilin Liu",
"Zhiding Yu",
"Kurt Keutzer",
"Sungjin Ahn",
"Jan Kautz",
"Hongxu Yin",
"Yao Lu",
"Song Han",
"Wonmin Byeon"
] | Recent advances in video-based multimodal large language models (Video-LLMs) have significantly improved video understanding by processing videos as sequences of image frames. However, many existing methods treat frames independently in the vision backbone, lacking explicit temporal modeling, which limits their ability to capture dynamic patterns and efficiently handle long videos. To address these limitations, we introduce STORM (Spatiotemporal TOken Reduction for Multimodal LLMs), a novel architecture incorporating a dedicated temporal encoder between the image encoder and the LLM. Our temporal encoder leverages the Mamba State Space Model to integrate temporal information into image tokens, generating enriched representations that preserve inter-frame dynamics across the entire video sequence. This enriched encoding not only enhances video reasoning capabilities but also enables effective token reduction strategies, including test-time sampling and training-based temporal and spatial pooling, substantially reducing computational demands on the LLM without sacrificing key temporal information. By integrating these techniques, our approach simultaneously reduces training and inference latency while improving performance, enabling efficient and robust video understanding over extended temporal contexts. Extensive evaluations show that STORM achieves state-of-the-art results across various long video understanding benchmarks (more than 5\% improvement on MLVU and LongVideoBench) while reducing the computation costs by up to 8times and the decoding latency by 2.4-2.9times for the fixed numbers of input frames. Project page is available at https://research.nvidia.com/labs/lpr/storm |
|
2025-03-07T00:00:00 | 2503.03803 | EgoLife: Towards Egocentric Life Assistant | [
"Jingkang Yang",
"Shuai Liu",
"Hongming Guo",
"Yuhao Dong",
"Xiamengwei Zhang",
"Sicheng Zhang",
"Pengyun Wang",
"Zitang Zhou",
"Binzhu Xie",
"Ziyue Wang",
"Bei Ouyang",
"Zhengyu Lin",
"Marco Cominelli",
"Zhongang Cai",
"Yuanhan Zhang",
"Peiyuan Zhang",
"Fangzhou Hong",
"Joerg Widmer",
"Francesco Gringoli",
"Lei Yang",
"Bo Li",
"Ziwei Liu"
] | We introduce EgoLife, a project to develop an egocentric life assistant that accompanies and enhances personal efficiency through AI-powered wearable glasses. To lay the foundation for this assistant, we conducted a comprehensive data collection study where six participants lived together for one week, continuously recording their daily activities - including discussions, shopping, cooking, socializing, and entertainment - using AI glasses for multimodal egocentric video capture, along with synchronized third-person-view video references. This effort resulted in the EgoLife Dataset, a comprehensive 300-hour egocentric, interpersonal, multiview, and multimodal daily life dataset with intensive annotation. Leveraging this dataset, we introduce EgoLifeQA, a suite of long-context, life-oriented question-answering tasks designed to provide meaningful assistance in daily life by addressing practical questions such as recalling past relevant events, monitoring health habits, and offering personalized recommendations. To address the key technical challenges of (1) developing robust visual-audio models for egocentric data, (2) enabling identity recognition, and (3) facilitating long-context question answering over extensive temporal information, we introduce EgoButler, an integrated system comprising EgoGPT and EgoRAG. EgoGPT is an omni-modal model trained on egocentric datasets, achieving state-of-the-art performance on egocentric video understanding. EgoRAG is a retrieval-based component that supports answering ultra-long-context questions. Our experimental studies verify their working mechanisms and reveal critical factors and bottlenecks, guiding future improvements. By releasing our datasets, models, and benchmarks, we aim to stimulate further research in egocentric AI assistants. |
|
2025-03-07T00:00:00 | 2503.03983 | Audio Flamingo 2: An Audio-Language Model with Long-Audio Understanding and Expert Reasoning Abilities | [
"Sreyan Ghosh",
"Zhifeng Kong",
"Sonal Kumar",
"S Sakshi",
"Jaehyeon Kim",
"Wei Ping",
"Rafael Valle",
"Dinesh Manocha",
"Bryan Catanzaro"
] | Understanding and reasoning over non-speech sounds and music are crucial for both humans and AI agents to interact effectively with their environments. In this paper, we introduce Audio Flamingo 2 (AF2), an Audio-Language Model (ALM) with advanced audio understanding and reasoning capabilities. AF2 leverages (i) a custom CLAP model, (ii) synthetic Audio QA data for fine-grained audio reasoning, and (iii) a multi-stage curriculum learning strategy. AF2 achieves state-of-the-art performance with only a 3B parameter small language model, surpassing large open-source and proprietary models across over 20 benchmarks. Next, for the first time, we extend audio understanding to long audio segments (30 secs to 5 mins) and propose LongAudio, a large and novel dataset for training ALMs on long audio captioning and question-answering tasks. Fine-tuning AF2 on LongAudio leads to exceptional performance on our proposed LongAudioBench, an expert annotated benchmark for evaluating ALMs on long audio understanding capabilities. We conduct extensive ablation studies to confirm the efficacy of our approach. Project Website: https://research.nvidia.com/labs/adlr/AF2/. |
|
2025-03-07T00:00:00 | 2503.02191 | Understanding and Predicting Derailment in Toxic Conversations on GitHub | [
"Mia Mohammad Imran",
"Robert Zita",
"Rebekah Copeland",
"Preetha Chatterjee",
"Rahat Rizvi Rahman",
"Kostadin Damevski"
] | Software projects thrive on the involvement and contributions of individuals from different backgrounds. However, toxic language and negative interactions can hinder the participation and retention of contributors and alienate newcomers. Proactive moderation strategies aim to prevent toxicity from occurring by addressing conversations that have derailed from their intended purpose. This study aims to understand and predict conversational derailment leading to toxicity on GitHub. To facilitate this research, we curate a novel dataset comprising 202 toxic conversations from GitHub with annotated derailment points, along with 696 non-toxic conversations as a baseline. Based on this dataset, we identify unique characteristics of toxic conversations and derailment points, including linguistic markers such as second-person pronouns, negation terms, and tones of Bitter Frustration and Impatience, as well as patterns in conversational dynamics between project contributors and external participants. Leveraging these empirical observations, we propose a proactive moderation approach to automatically detect and address potentially harmful conversations before escalation. By utilizing modern LLMs, we develop a conversation trajectory summary technique that captures the evolution of discussions and identifies early signs of derailment. Our experiments demonstrate that LLM prompts tailored to provide summaries of GitHub conversations achieve 69% F1-Score in predicting conversational derailment, strongly improving over a set of baseline approaches. |
|
2025-03-07T00:00:00 | 2502.20258 | LLM as a Broken Telephone: Iterative Generation Distorts Information | [
"Amr Mohamed",
"Mingmeng Geng",
"Michalis Vazirgiannis",
"Guokan Shang"
] | As large language models are increasingly responsible for online content, concerns arise about the impact of repeatedly processing their own outputs. Inspired by the "broken telephone" effect in chained human communication, this study investigates whether LLMs similarly distort information through iterative generation. Through translation-based experiments, we find that distortion accumulates over time, influenced by language choice and chain complexity. While degradation is inevitable, it can be mitigated through strategic prompting techniques. These findings contribute to discussions on the long-term effects of AI-mediated information propagation, raising important questions about the reliability of LLM-generated content in iterative workflows. |
|
2025-03-07T00:00:00 | 2503.04606 | The Best of Both Worlds: Integrating Language Models and Diffusion Models for Video Generation | [
"Aoxiong Yin",
"Kai Shen",
"Yichong Leng",
"Xu Tan",
"Xinyu Zhou",
"Juncheng Li",
"Siliang Tang"
] | Recent advancements in text-to-video (T2V) generation have been driven by two competing paradigms: autoregressive language models and diffusion models. However, each paradigm has intrinsic limitations: language models struggle with visual quality and error accumulation, while diffusion models lack semantic understanding and causal modeling. In this work, we propose LanDiff, a hybrid framework that synergizes the strengths of both paradigms through coarse-to-fine generation. Our architecture introduces three key innovations: (1) a semantic tokenizer that compresses 3D visual features into compact 1D discrete representations through efficient semantic compression, achieving a sim14,000times compression ratio; (2) a language model that generates semantic tokens with high-level semantic relationships; (3) a streaming diffusion model that refines coarse semantics into high-fidelity videos. Experiments show that LanDiff, a 5B model, achieves a score of 85.43 on the VBench T2V benchmark, surpassing the state-of-the-art open-source models Hunyuan Video (13B) and other commercial models such as Sora, Keling, and Hailuo. Furthermore, our model also achieves state-of-the-art performance in long video generation, surpassing other open-source models in this field. Our demo can be viewed at https://landiff.github.io/. |
|
2025-03-07T00:00:00 | 2503.04725 | L^2M: Mutual Information Scaling Law for Long-Context Language Modeling | [
"Zhuo Chen",
"Oriol Mayné i Comas",
"Zhuotao Jin",
"Di Luo",
"Marin Soljačić"
] | We rigorously establish a bipartite mutual information scaling law in natural language that governs long-range dependencies. This scaling law, which we show is distinct from and scales independently of the conventional two-point mutual information, is the key to understanding long-context language modeling. Using this scaling law, we formulate the Long-context Language Modeling (L^2M) condition, which relates a model's capacity for effective long context length modeling to the scaling of its latent state size for storing past information. Our results are validated through experiments on both transformers and state space models. This work establishes a theoretical foundation that guides the development of large language models toward longer context lengths. |
|
2025-03-07T00:00:00 | 2503.01375 | Combining Flow Matching and Transformers for Efficient Solution of Bayesian Inverse Problems | [
"Daniil Sherki",
"Ivan Oseledets",
"Ekaterina Muravleva"
] | Solving Bayesian inverse problems efficiently remains a significant challenge due to the complexity of posterior distributions and the computational cost of traditional sampling methods. Given a series of observations and the forward model, we want to recover the distribution of the parameters, conditioned on observed experimental data. We show, that combining Conditional Flow Mathching (CFM) with transformer-based architecture, we can efficiently sample from such kind of distribution, conditioned on variable number of observations. |
|
2025-03-07T00:00:00 | 2503.01901 | Identifying Sensitive Weights via Post-quantization Integral | [
"Yuezhou Hu",
"Weiyu Huang",
"Zichen Liang",
"Chang Chen",
"Jintao Zhang",
"Jun Zhu",
"Jianfei Chen"
] | Serving Large Language Models (LLMs) is costly. However, post-training weight quantization can address this problem by both compressing their sizes for limited memory and saving bandwidth for acceleration. As not all weight dimensions are equally important, those methods typically rely on a sensitivity metric, which indicates the element-wise influence of weights on loss function and is used to preprocess original weights for better quantization. In this work, we conduct an empirical study on the accuracy of the sensitivity metric, and find that existing gradient and Hessian based metrics are very inaccurate: they underestimate quantization's impact on the loss function by orders of magnitude, mainly due to the small convergence radius of local 2nd order approximation, \ie, gradient and Hessian term in Taylor's formula. To tackle this problem, we propose Post-quantization Integral (PQI), an accurate metric to estimate posterior sensitivity in a fine-grained manner. To leverage this accurate metric, we further propose ReQuant, a simple yet powerful framework that mainly consists of two Dense-and-Sparse detach components: self-adaptive outlier selection and step-wise significant weights detach. Results show that ReQuant boosts state-of-the-art post-training quantization methods, with a pronounced improvement of 2.66 perplexity gain on Llama 3.2 1B with QTIP. |
|
2025-03-07T00:00:00 | 2503.02972 | LINGOLY-TOO: Disentangling Memorisation from Reasoning with Linguistic Templatisation and Orthographic Obfuscation | [
"Jude Khouja",
"Karolina Korgul",
"Simi Hellsten",
"Lingyi Yang",
"Vlad Neacs",
"Harry Mayne",
"Ryan Kearns",
"Andrew Bean",
"Adam Mahdi"
] | Effective evaluation of the reasoning capabilities of large language models (LLMs) are susceptible to overestimation due to data exposure of evaluation benchmarks. We introduce a framework for producing linguistic reasoning problems that reduces the effect of memorisation in model performance estimates and apply this framework to develop LINGOLY-TOO, a challenging evaluation benchmark for linguistic reasoning. By developing orthographic templates, we dynamically obfuscate the writing systems of real languages to generate numerous question variations. These variations preserve the reasoning steps required for each solution while reducing the likelihood of specific problem instances appearing in model training data. Our experiments demonstrate that frontier models, including OpenAI o1-preview and DeepSeem R1, struggle with advanced reasoning. Our analysis also shows that LLMs exhibit noticeable variance in accuracy across permutations of the same problem, and on average perform better on questions appearing in their original orthography. Our findings highlight the opaque nature of response generation in LLMs and provide evidence that prior data exposure contributes to overestimating the reasoning capabilities of frontier models. |
|
2025-03-07T00:00:00 | 2503.04644 | IFIR: A Comprehensive Benchmark for Evaluating Instruction-Following in Expert-Domain Information Retrieval | [
"Tingyu Song",
"Guo Gan",
"Mingsheng Shang",
"Yilun Zhao"
] | We introduce IFIR, the first comprehensive benchmark designed to evaluate instruction-following information retrieval (IR) in expert domains. IFIR includes 2,426 high-quality examples and covers eight subsets across four specialized domains: finance, law, healthcare, and science literature. Each subset addresses one or more domain-specific retrieval tasks, replicating real-world scenarios where customized instructions are critical. IFIR enables a detailed analysis of instruction-following retrieval capabilities by incorporating instructions at different levels of complexity. We also propose a novel LLM-based evaluation method to provide a more precise and reliable assessment of model performance in following instructions. Through extensive experiments on 15 frontier retrieval models, including those based on LLMs, our results reveal that current models face significant challenges in effectively following complex, domain-specific instructions. We further provide in-depth analyses to highlight these limitations, offering valuable insights to guide future advancements in retriever development. |
|
2025-03-07T00:00:00 | 2503.04369 | Lost in Literalism: How Supervised Training Shapes Translationese in LLMs | [
"Yafu Li",
"Ronghao Zhang",
"Zhilin Wang",
"Huajian Zhang",
"Leyang Cui",
"Yongjing Yin",
"Tong Xiao",
"Yue Zhang"
] | https://github.com/yafuly/LLM_Translationese | Large language models (LLMs) have achieved remarkable success in machine translation, demonstrating impressive performance across diverse languages. However, translationese, characterized by overly literal and unnatural translations, remains a persistent challenge in LLM-based translation systems. Despite their pre-training on vast corpora of natural utterances, LLMs exhibit translationese errors and generate unexpected unnatural translations, stemming from biases introduced during supervised fine-tuning (SFT). In this work, we systematically evaluate the prevalence of translationese in LLM-generated translations and investigate its roots during supervised training. We introduce methods to mitigate these biases, including polishing golden references and filtering unnatural training instances. Empirical evaluations demonstrate that these approaches significantly reduce translationese while improving translation naturalness, validated by human evaluations and automatic metrics. Our findings highlight the need for training-aware adjustments to optimize LLM translation outputs, paving the way for more fluent and target-language-consistent translations. We release the data and code at https://github.com/yafuly/LLM_Translationese. |
2025-03-07T00:00:00 | 2503.02495 | Union of Experts: Adapting Hierarchical Routing to Equivalently Decomposed Transformer | [
"Yujiao Yang",
"Jing Lian",
"Linhui Li"
] | https://github.com/YujiaoYang-work/UoE | Mixture-of-Experts (MoE) enhances model performance while maintaining computational efficiency, making it well-suited for large-scale applications. However, expert in exist MoE paradigm works as an individual, thereby lacking high-quality expert interactions. Moreover, they have not been effectively extended to attention block, which constrains further efficiency improvements. To tackle these issues, we propose Union-of-Experts (UoE), which decomposes transformer into an equitant group of experts, and then implement dynamic routing on input data and experts. Our approach advances MoE design with three key innovations: (1) We conducted equitant expert decomposition on both MLP blocks and attention blocks based on matrix partition in tensor parallelism. (2) We developed two routing paradigms: patch wise data selection and expert selection, to apply routing across different levels. (3) We design the architecture of UoE model, including Selective Multi-Head Attention (SMHA) and Union-of-MLP-Experts (UoME). (4) We develop parallel implementation of UoE's routing and computation operation, and optimize efficiency based on the hardware processing analysis. The experiments demonstrate that the model employed with UoE surpass Full Attention, state-of-art MoEs and efficient transformers in several tasks across image and natural language domains. The source codes are available at https://github.com/YujiaoYang-work/UoE. |
2025-03-07T00:00:00 | 2503.04724 | LLMVoX: Autoregressive Streaming Text-to-Speech Model for Any LLM | [
"Sambal Shikhar",
"Mohammed Irfan Kurpath",
"Sahal Shaji Mullappilly",
"Jean Lahoud",
"Fahad Khan",
"Rao Muhammad Anwer",
"Salman Khan",
"Hisham Cholakkal"
] | Recent advancements in speech-to-speech dialogue systems leverage LLMs for multimodal interactions, yet they remain hindered by fine-tuning requirements, high computational overhead, and text-speech misalignment. Existing speech-enabled LLMs often degrade conversational quality by modifying the LLM, thereby compromising its linguistic capabilities. In contrast, we propose LLMVoX, a lightweight 30M-parameter, LLM-agnostic, autoregressive streaming TTS system that generates high-quality speech with low latency, while fully preserving the capabilities of the base LLM. Our approach achieves a significantly lower Word Error Rate compared to speech-enabled LLMs, while operating at comparable latency and UTMOS score. By decoupling speech synthesis from LLM processing via a multi-queue token streaming system, LLMVoX supports seamless, infinite-length dialogues. Its plug-and-play design also facilitates extension to various tasks with different backbones. Furthermore, LLMVoX generalizes to new languages with only dataset adaptation, attaining a low Character Error Rate on an Arabic speech task. Additionally, we have integrated LLMVoX with a Vision-Language Model to create an omni-model with speech, text, and vision capabilities, without requiring additional multimodal training. Our code base and project page is available at https://mbzuai-oryx.github.io/LLMVoX . |
|
2025-03-07T00:00:00 | 2503.01917 | How to Steer LLM Latents for Hallucination Detection? | [
"Seongheon Park",
"Xuefeng Du",
"Min-Hsuan Yeh",
"Haobo Wang",
"Yixuan Li"
] | Hallucinations in LLMs pose a significant concern to their safe deployment in real-world applications. Recent approaches have leveraged the latent space of LLMs for hallucination detection, but their embeddings, optimized for linguistic coherence rather than factual accuracy, often fail to clearly separate truthful and hallucinated content. To this end, we propose the Truthfulness Separator Vector (TSV), a lightweight and flexible steering vector that reshapes the LLM's representation space during inference to enhance the separation between truthful and hallucinated outputs, without altering model parameters. Our two-stage framework first trains TSV on a small set of labeled exemplars to form compact and well-separated clusters. It then augments the exemplar set with unlabeled LLM generations, employing an optimal transport-based algorithm for pseudo-labeling combined with a confidence-based filtering process. Extensive experiments demonstrate that TSV achieves state-of-the-art performance with minimal labeled data, exhibiting strong generalization across datasets and providing a practical solution for real-world LLM applications. |
|
2025-03-07T00:00:00 | 2503.03962 | On the Acquisition of Shared Grammatical Representations in Bilingual Language Models | [
"Catherine Arnett",
"Tyler A. Chang",
"James A. Michaelov",
"Benjamin K. Bergen"
] | While crosslingual transfer is crucial to contemporary language models' multilingual capabilities, how it occurs is not well understood. In this paper, we ask what happens to a monolingual language model when it begins to be trained on a second language. Specifically, we train small bilingual models for which we control the amount of data for each language and the order of language exposure. To find evidence of shared multilingual representations, we turn to structural priming, a method used to study grammatical representations in humans. We first replicate previous crosslingual structural priming results and find that after controlling for training data quantity and language exposure, there are asymmetrical effects across language pairs and directions. We argue that this asymmetry may shape hypotheses about human structural priming effects. We also find that structural priming effects are less robust for less similar language pairs, highlighting potential limitations of crosslingual transfer learning and shared representations for typologically diverse languages. |
|
2025-03-09T00:00:00 | 2503.05236 | Unified Reward Model for Multimodal Understanding and Generation | [
"Yibin Wang",
"Yuhang Zang",
"Hao Li",
"Cheng Jin",
"Jiaqi Wang"
] | Recent advances in human preference alignment have significantly enhanced multimodal generation and understanding. A key approach is training reward models to guide preference optimization. However, existing models are often task-specific, limiting their adaptability across diverse visual applications. We also argue that jointly learning to assess multiple tasks may foster a synergistic effect, where improved image understanding enhances image generation assessment, and refined image evaluation benefits video assessment through better frame analysis. To this end, this paper proposes UnifiedReward, the first unified reward model for multimodal understanding and generation assessment, enabling both pairwise ranking and pointwise scoring, which can be employed for vision model preference alignment. Specifically, (1) we first develop UnifiedReward on our constructed large-scale human preference dataset, including both image and video generation/understanding tasks. (2) Then, it is utilized to automatically construct high-quality preference pair data based on the vision models, fine-gradually filtering their outputs through pair ranking and point sifting. (3) Finally, these data are used for their preference alignment through Direct Preference Optimization (DPO). Experimental results demonstrate that joint learning to assess diverse visual tasks can lead to substantial mutual benefits and we apply our pipeline to both image and video understanding/generation tasks, significantly improving the performance in each domain. |
|
2025-03-09T00:00:00 | 2503.05179 | Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketching | [
"Simon A. Aytes",
"Jinheon Baek",
"Sung Ju Hwang"
] | Recent advances in large language models have demonstrated remarkable reasoning capabilities through Chain of Thought (CoT) prompting, but often at the cost of excessive verbosity in their intermediate outputs, which increases computational overhead. We introduce Sketch-of-Thought (SoT), a novel prompting framework that combines cognitive-inspired reasoning paradigms with linguistic constraints to minimize token usage while preserving reasoning accuracy. SoT is designed as a flexible framework that can incorporate any custom reasoning paradigms based on cognitive science, and we instantiate it with three such paradigms - Conceptual Chaining, Chunked Symbolism, and Expert Lexicons - each tailored to different reasoning tasks and selected dynamically via a lightweight routing model. Through comprehensive evaluation across 15 reasoning datasets with multiple languages and multimodal scenarios, we demonstrate that SoT achieves token reductions of 76% with negligible accuracy impact. In certain domains like mathematical and multi-hop reasoning, it even improves accuracy while using significantly fewer tokens. Our code is publicly available: https://www.github.com/SimonAytes/SoT. |
|
2025-03-09T00:00:00 | 2503.02130 | Forgetting Transformer: Softmax Attention with a Forget Gate | [
"Zhixuan Lin",
"Evgenii Nikishin",
"Xu Owen He",
"Aaron Courville"
] | https://github.com/zhixuan-lin/forgetting-transformer | An essential component of modern recurrent sequence models is the forget gate. While Transformers do not have an explicit recurrent form, we show that a forget gate can be naturally incorporated into Transformers by down-weighting the unnormalized attention scores in a data-dependent way. We name this attention mechanism the Forgetting Attention and the resulting model the Forgetting Transformer (FoX). We show that FoX outperforms the Transformer on long-context language modeling, length extrapolation, and short-context downstream tasks, while performing on par with the Transformer on long-context downstream tasks. Moreover, it is compatible with the FlashAttention algorithm and does not require any positional embeddings. Several analyses, including the needle-in-the-haystack test, show that FoX also retains the Transformer's superior long-context capabilities over recurrent sequence models such as Mamba-2, HGRN2, and DeltaNet. We also introduce a "Pro" block design that incorporates some common architectural components in recurrent sequence models and find it significantly improves the performance of both FoX and the Transformer. Our code is available at https://github.com/zhixuan-lin/forgetting-transformer. |
2025-03-10T00:00:00 | 2503.05236 | Unified Reward Model for Multimodal Understanding and Generation | [
"Yibin Wang",
"Yuhang Zang",
"Hao Li",
"Cheng Jin",
"Jiaqi Wang"
] | Recent advances in human preference alignment have significantly enhanced multimodal generation and understanding. A key approach is training reward models to guide preference optimization. However, existing models are often task-specific, limiting their adaptability across diverse visual applications. We also argue that jointly learning to assess multiple tasks may foster a synergistic effect, where improved image understanding enhances image generation assessment, and refined image evaluation benefits video assessment through better frame analysis. To this end, this paper proposes UnifiedReward, the first unified reward model for multimodal understanding and generation assessment, enabling both pairwise ranking and pointwise scoring, which can be employed for vision model preference alignment. Specifically, (1) we first develop UnifiedReward on our constructed large-scale human preference dataset, including both image and video generation/understanding tasks. (2) Then, it is utilized to automatically construct high-quality preference pair data based on the vision models, fine-gradually filtering their outputs through pair ranking and point sifting. (3) Finally, these data are used for their preference alignment through Direct Preference Optimization (DPO). Experimental results demonstrate that joint learning to assess diverse visual tasks can lead to substantial mutual benefits and we apply our pipeline to both image and video understanding/generation tasks, significantly improving the performance in each domain. |
|
2025-03-10T00:00:00 | 2503.05179 | Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketching | [
"Simon A. Aytes",
"Jinheon Baek",
"Sung Ju Hwang"
] | Recent advances in large language models have demonstrated remarkable reasoning capabilities through Chain of Thought (CoT) prompting, but often at the cost of excessive verbosity in their intermediate outputs, which increases computational overhead. We introduce Sketch-of-Thought (SoT), a novel prompting framework that combines cognitive-inspired reasoning paradigms with linguistic constraints to minimize token usage while preserving reasoning accuracy. SoT is designed as a flexible framework that can incorporate any custom reasoning paradigms based on cognitive science, and we instantiate it with three such paradigms - Conceptual Chaining, Chunked Symbolism, and Expert Lexicons - each tailored to different reasoning tasks and selected dynamically via a lightweight routing model. Through comprehensive evaluation across 15 reasoning datasets with multiple languages and multimodal scenarios, we demonstrate that SoT achieves token reductions of 76% with negligible accuracy impact. In certain domains like mathematical and multi-hop reasoning, it even improves accuracy while using significantly fewer tokens. Our code is publicly available: https://www.github.com/SimonAytes/SoT. |
|
2025-03-10T00:00:00 | 2503.02130 | Forgetting Transformer: Softmax Attention with a Forget Gate | [
"Zhixuan Lin",
"Evgenii Nikishin",
"Xu Owen He",
"Aaron Courville"
] | https://github.com/zhixuan-lin/forgetting-transformer | An essential component of modern recurrent sequence models is the forget gate. While Transformers do not have an explicit recurrent form, we show that a forget gate can be naturally incorporated into Transformers by down-weighting the unnormalized attention scores in a data-dependent way. We name this attention mechanism the Forgetting Attention and the resulting model the Forgetting Transformer (FoX). We show that FoX outperforms the Transformer on long-context language modeling, length extrapolation, and short-context downstream tasks, while performing on par with the Transformer on long-context downstream tasks. Moreover, it is compatible with the FlashAttention algorithm and does not require any positional embeddings. Several analyses, including the needle-in-the-haystack test, show that FoX also retains the Transformer's superior long-context capabilities over recurrent sequence models such as Mamba-2, HGRN2, and DeltaNet. We also introduce a "Pro" block design that incorporates some common architectural components in recurrent sequence models and find it significantly improves the performance of both FoX and the Transformer. Our code is available at https://github.com/zhixuan-lin/forgetting-transformer. |
2025-03-10T00:00:00 | 2503.05652 | BEHAVIOR Robot Suite: Streamlining Real-World Whole-Body Manipulation for Everyday Household Activities | [
"Yunfan Jiang",
"Ruohan Zhang",
"Josiah Wong",
"Chen Wang",
"Yanjie Ze",
"Hang Yin",
"Cem Gokmen",
"Shuran Song",
"Jiajun Wu",
"Li Fei-Fei"
] | Real-world household tasks present significant challenges for mobile manipulation robots. An analysis of existing robotics benchmarks reveals that successful task performance hinges on three key whole-body control capabilities: bimanual coordination, stable and precise navigation, and extensive end-effector reachability. Achieving these capabilities requires careful hardware design, but the resulting system complexity further complicates visuomotor policy learning. To address these challenges, we introduce the BEHAVIOR Robot Suite (BRS), a comprehensive framework for whole-body manipulation in diverse household tasks. Built on a bimanual, wheeled robot with a 4-DoF torso, BRS integrates a cost-effective whole-body teleoperation interface for data collection and a novel algorithm for learning whole-body visuomotor policies. We evaluate BRS on five challenging household tasks that not only emphasize the three core capabilities but also introduce additional complexities, such as long-range navigation, interaction with articulated and deformable objects, and manipulation in confined spaces. We believe that BRS's integrated robotic embodiment, data collection interface, and learning framework mark a significant step toward enabling real-world whole-body manipulation for everyday household tasks. BRS is open-sourced at https://behavior-robot-suite.github.io/ |
|
2025-03-10T00:00:00 | 2503.05592 | R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning | [
"Huatong Song",
"Jinhao Jiang",
"Yingqian Min",
"Jie Chen",
"Zhipeng Chen",
"Wayne Xin Zhao",
"Lei Fang",
"Ji-Rong Wen"
] | Existing Large Reasoning Models (LRMs) have shown the potential of reinforcement learning (RL) to enhance the complex reasoning capabilities of Large Language Models~(LLMs). While they achieve remarkable performance on challenging tasks such as mathematics and coding, they often rely on their internal knowledge to solve problems, which can be inadequate for time-sensitive or knowledge-intensive questions, leading to inaccuracies and hallucinations. To address this, we propose R1-Searcher, a novel two-stage outcome-based RL approach designed to enhance the search capabilities of LLMs. This method allows LLMs to autonomously invoke external search systems to access additional knowledge during the reasoning process. Our framework relies exclusively on RL, without requiring process rewards or distillation for a cold start. % effectively generalizing to out-of-domain datasets and supporting both Base and Instruct models. Our experiments demonstrate that our method significantly outperforms previous strong RAG methods, even when compared to the closed-source GPT-4o-mini. |
|
2025-03-10T00:00:00 | 2503.05379 | R1-Omni: Explainable Omni-Multimodal Emotion Recognition with Reinforcing Learning | [
"Jiaxing Zhao",
"Xihan Wei",
"Liefeng Bo"
] | In this work, we present the first application of Reinforcement Learning with Verifiable Reward (RLVR) to an Omni-multimodal large language model in the context of emotion recognition, a task where both visual and audio modalities play crucial roles. We leverage RLVR to optimize the Omni model, significantly enhancing its performance in three key aspects: reasoning capability, emotion recognition accuracy, and generalization ability. The introduction of RLVR not only improves the model's overall performance on in-distribution data but also demonstrates superior robustness when evaluated on out-of-distribution datasets. More importantly, the improved reasoning capability enables clear analysis of the contributions of different modalities, particularly visual and audio information, in the emotion recognition process. This provides valuable insights into the optimization of multimodal large language models. |
|
2025-03-10T00:00:00 | 2503.05132 | R1-Zero's "Aha Moment" in Visual Reasoning on a 2B Non-SFT Model | [
"Hengguang Zhou",
"Xirui Li",
"Ruochen Wang",
"Minhao Cheng",
"Tianyi Zhou",
"Cho-Jui Hsieh"
] | https://github.com/turningpoint-ai/VisualThinker-R1-Zero | Recently DeepSeek R1 demonstrated how reinforcement learning with simple rule-based incentives can enable autonomous development of complex reasoning in large language models, characterized by the "aha moment", in which the model manifest self-reflection and increased response length during training. However, attempts to extend this success to multimodal reasoning often failed to reproduce these key characteristics. In this report, we present the first successful replication of these emergent characteristics for multimodal reasoning on only a non-SFT 2B model. Starting with Qwen2-VL-2B and applying reinforcement learning directly on the SAT dataset, our model achieves 59.47% accuracy on CVBench, outperforming the base model by approximately ~30% and exceeding both SFT setting by ~2%. In addition, we share our failed attempts and insights in attempting to achieve R1-like reasoning using RL with instruct models. aiming to shed light on the challenges involved. Our key observations include: (1) applying RL on instruct model often results in trivial reasoning trajectories, and (2) naive length reward are ineffective in eliciting reasoning capabilities. The project code is available at https://github.com/turningpoint-ai/VisualThinker-R1-Zero |
2025-03-10T00:00:00 | 2503.04872 | TinyR1-32B-Preview: Boosting Accuracy with Branch-Merge Distillation | [
"Lin Sun",
"Guangxiang Zhao",
"Xiaoqi Jian",
"Yuhan Wu",
"Weihong Lin",
"Yongfu Zhu",
"Change Jia",
"Linglin Zhang",
"Jinzhu Wu",
"Junfeng Ran",
"Sai-er Hu",
"Zihan Jiang",
"Junting Zhou",
"Wenrui Liu",
"Bin Cui",
"Tong Yang",
"Xiangzheng Zhang"
] | The challenge of reducing the size of Large Language Models (LLMs) while maintaining their performance has gained significant attention. However, existing methods, such as model distillation and transfer learning, often fail to achieve high accuracy. To address this limitation, we introduce the Branch-Merge distillation approach, which enhances model compression through two phases: (1) the Branch Phase, where knowledge from a large teacher model is selectively distilled into specialized student models via domain-specific supervised fine-tuning (SFT); And (2) the Merge Phase, where these student models are merged to enable cross-domain knowledge transfer and improve generalization. We validate our distillation approach using DeepSeek-R1 as the teacher and DeepSeek-R1-Distill-Qwen-32B as the student. The resulting merged model, TinyR1-32B-Preview, outperforms its counterpart DeepSeek-R1-Distill-Qwen-32B across multiple benchmarks, including Mathematics (+5.5 points), Coding (+4.4 points) and Science (+2.9 points), while achieving near-equal performance to DeepSeek-R1 on AIME 2024. The Branch-Merge distillation approach provides a scalable solution for creating smaller, high-performing LLMs with reduced computational cost and time. |
|
2025-03-10T00:00:00 | 2503.04808 | Learning from Failures in Multi-Attempt Reinforcement Learning | [
"Stephen Chung",
"Wenyu Du",
"Jie Fu"
] | https://github.com/DualityRL/multi-attempt | Recent advancements in reinforcement learning (RL) for large language models (LLMs), exemplified by DeepSeek R1, have shown that even a simple question-answering task can substantially improve an LLM's reasoning capabilities. In this work, we extend this approach by modifying the task into a multi-attempt setting. Instead of generating a single response per question, the model is given multiple attempts, with feedback provided after incorrect responses. The multi-attempt task encourages the model to refine its previous attempts and improve search efficiency. Experimental results show that even a small LLM trained on a multi-attempt task achieves significantly higher accuracy when evaluated with more attempts, improving from 45.6% with 1 attempt to 52.5% with 2 attempts on the math benchmark. In contrast, the same LLM trained on a standard single-turn task exhibits only a marginal improvement, increasing from 42.3% to 43.2% when given more attempts during evaluation. The results indicate that, compared to the standard single-turn task, an LLM trained on a multi-attempt task achieves slightly better performance on math benchmarks while also learning to refine its responses more effectively based on user feedback. Full code is available at https://github.com/DualityRL/multi-attempt |
2025-03-10T00:00:00 | 2503.05315 | LoRACode: LoRA Adapters for Code Embeddings | [
"Saumya Chaturvedi",
"Aman Chadha",
"Laurent Bindschaedler"
] | Code embeddings are essential for semantic code search; however, current approaches often struggle to capture the precise syntactic and contextual nuances inherent in code. Open-source models such as CodeBERT and UniXcoder exhibit limitations in scalability and efficiency, while high-performing proprietary systems impose substantial computational costs. We introduce a parameter-efficient fine-tuning method based on Low-Rank Adaptation (LoRA) to construct task-specific adapters for code retrieval. Our approach reduces the number of trainable parameters to less than two percent of the base model, enabling rapid fine-tuning on extensive code corpora (2 million samples in 25 minutes on two H100 GPUs). Experiments demonstrate an increase of up to 9.1% in Mean Reciprocal Rank (MRR) for Code2Code search, and up to 86.69% for Text2Code search tasks across multiple programming languages. Distinction in task-wise and language-wise adaptation helps explore the sensitivity of code retrieval for syntactical and linguistic variations. |
|
2025-03-10T00:00:00 | 2503.05638 | TrajectoryCrafter: Redirecting Camera Trajectory for Monocular Videos via Diffusion Models | [
"Mark YU",
"Wenbo Hu",
"Jinbo Xing",
"Ying Shan"
] | We present TrajectoryCrafter, a novel approach to redirect camera trajectories for monocular videos. By disentangling deterministic view transformations from stochastic content generation, our method achieves precise control over user-specified camera trajectories. We propose a novel dual-stream conditional video diffusion model that concurrently integrates point cloud renders and source videos as conditions, ensuring accurate view transformations and coherent 4D content generation. Instead of leveraging scarce multi-view videos, we curate a hybrid training dataset combining web-scale monocular videos with static multi-view datasets, by our innovative double-reprojection strategy, significantly fostering robust generalization across diverse scenes. Extensive evaluations on multi-view and large-scale monocular videos demonstrate the superior performance of our method. |
|
2025-03-10T00:00:00 | 2503.01713 | SAGE: A Framework of Precise Retrieval for RAG | [
"Jintao Zhang",
"Guoliang Li",
"Jinyang Su"
] | Retrieval-augmented generation (RAG) has demonstrated significant proficiency in conducting question-answering (QA) tasks within a specified corpus. Nonetheless, numerous failure instances of RAG in QA still exist. These failures are not solely attributable to the limitations of Large Language Models (LLMs); instead, they predominantly arise from the retrieval of inaccurate information for LLMs due to two limitations: (1) Current RAG methods segment the corpus without considering semantics, making it difficult to find relevant context due to impaired correlation between questions and the segments. (2) There is a trade-off between missing essential context with fewer context retrieved and getting irrelevant context with more context retrieved. In this paper, we introduce a RAG framework (SAGE), to overcome these limitations. First, to address the segmentation issue without considering semantics, we propose to train a semantic segmentation model. This model is trained to segment the corpus into semantically complete chunks. Second, to ensure that only the most relevant chunks are retrieved while the irrelevant ones are ignored, we design a chunk selection algorithm to dynamically select chunks based on the decreasing speed of the relevance score, leading to a more relevant selection. Third, to further ensure the precision of the retrieved chunks, we propose letting LLMs assess whether retrieved chunks are excessive or lacking and then adjust the amount of context accordingly. Experiments show that SAGE outperforms baselines by 61.25% in the quality of QA on average. Moreover, by avoiding retrieving noisy context, SAGE lowers the cost of the tokens consumed in LLM inference and achieves a 49.41% enhancement in cost efficiency on average. Additionally, our work offers valuable insights for boosting RAG. |
|
2025-03-10T00:00:00 | 2503.05500 | EuroBERT: Scaling Multilingual Encoders for European Languages | [
"Nicolas Boizard",
"Hippolyte Gisserot-Boukhlef",
"Duarte M. Alves",
"André Martins",
"Ayoub Hammal",
"Caio Corro",
"Céline Hudelot",
"Emmanuel Malherbe",
"Etienne Malaboeuf",
"Fanny Jourdan",
"Gabriel Hautreux",
"João Alves",
"Kevin El-Haddad",
"Manuel Faysse",
"Maxime Peyrard",
"Nuno M. Guerreiro",
"Patrick Fernandes",
"Ricardo Rei",
"Pierre Colombo"
] | General-purpose multilingual vector representations, used in retrieval, regression and classification, are traditionally obtained from bidirectional encoder models. Despite their wide applicability, encoders have been recently overshadowed by advances in generative decoder-only models. However, many innovations driving this progress are not inherently tied to decoders. In this paper, we revisit the development of multilingual encoders through the lens of these advances, and introduce EuroBERT, a family of multilingual encoders covering European and widely spoken global languages. Our models outperform existing alternatives across a diverse range of tasks, spanning multilingual capabilities, mathematics, and coding, and natively supporting sequences of up to 8,192 tokens. We also examine the design decisions behind EuroBERT, offering insights into our dataset composition and training pipeline. We publicly release the EuroBERT models, including intermediate training checkpoints, together with our training framework. |
|
2025-03-10T00:00:00 | 2503.05639 | VideoPainter: Any-length Video Inpainting and Editing with Plug-and-Play Context Control | [
"Yuxuan Bian",
"Zhaoyang Zhang",
"Xuan Ju",
"Mingdeng Cao",
"Liangbin Xie",
"Ying Shan",
"Qiang Xu"
] | Video inpainting, which aims to restore corrupted video content, has experienced substantial progress. Despite these advances, existing methods, whether propagating unmasked region pixels through optical flow and receptive field priors, or extending image-inpainting models temporally, face challenges in generating fully masked objects or balancing the competing objectives of background context preservation and foreground generation in one model, respectively. To address these limitations, we propose a novel dual-stream paradigm VideoPainter that incorporates an efficient context encoder (comprising only 6% of the backbone parameters) to process masked videos and inject backbone-aware background contextual cues to any pre-trained video DiT, producing semantically consistent content in a plug-and-play manner. This architectural separation significantly reduces the model's learning complexity while enabling nuanced integration of crucial background context. We also introduce a novel target region ID resampling technique that enables any-length video inpainting, greatly enhancing our practical applicability. Additionally, we establish a scalable dataset pipeline leveraging current vision understanding models, contributing VPData and VPBench to facilitate segmentation-based inpainting training and assessment, the largest video inpainting dataset and benchmark to date with over 390K diverse clips. Using inpainting as a pipeline basis, we also explore downstream applications including video editing and video editing pair data generation, demonstrating competitive performance and significant practical potential. Extensive experiments demonstrate VideoPainter's superior performance in both any-length video inpainting and editing, across eight key metrics, including video quality, mask region preservation, and textual coherence. |
|
2025-03-10T00:00:00 | 2503.04504 | AnyAnomaly: Zero-Shot Customizable Video Anomaly Detection with LVLM | [
"Sunghyun Ahn",
"Youngwan Jo",
"Kijung Lee",
"Sein Kwon",
"Inpyo Hong",
"Sanghyun Park"
] | Video anomaly detection (VAD) is crucial for video analysis and surveillance in computer vision. However, existing VAD models rely on learned normal patterns, which makes them difficult to apply to diverse environments. Consequently, users should retrain models or develop separate AI models for new environments, which requires expertise in machine learning, high-performance hardware, and extensive data collection, limiting the practical usability of VAD. To address these challenges, this study proposes customizable video anomaly detection (C-VAD) technique and the AnyAnomaly model. C-VAD considers user-defined text as an abnormal event and detects frames containing a specified event in a video. We effectively implemented AnyAnomaly using a context-aware visual question answering without fine-tuning the large vision language model. To validate the effectiveness of the proposed model, we constructed C-VAD datasets and demonstrated the superiority of AnyAnomaly. Furthermore, our approach showed competitive performance on VAD benchmark datasets, achieving state-of-the-art results on the UBnormal dataset and outperforming other methods in generalization across all datasets. Our code is available online at github.com/SkiddieAhn/Paper-AnyAnomaly. |
|
2025-03-10T00:00:00 | 2503.05447 | Linear-MoE: Linear Sequence Modeling Meets Mixture-of-Experts | [
"Weigao Sun",
"Disen Lan",
"Tong Zhu",
"Xiaoye Qu",
"Yu Cheng"
] | https://github.com/OpenSparseLLMs/Linear-MoE | Linear Sequence Modeling (LSM) like linear attention, state space models and linear RNNs, and Mixture-of-Experts (MoE) have recently emerged as significant architectural improvements. In this paper, we introduce Linear-MoE, a production-level system for modeling and training large-scale models that integrate LSM with MoE. Linear-MoE leverages the advantages of both LSM modules for linear-complexity sequence modeling and MoE layers for sparsely activation, aiming to offer high performance with efficient training. The Linear-MoE system comprises: 1) Modeling subsystem, which provides a unified framework supporting all instances of LSM. and 2) Training subsystem, which facilitates efficient training by incorporating various advanced parallelism technologies, particularly Sequence Parallelism designed for Linear-MoE models. Additionally, we explore hybrid models that combine Linear-MoE layers with standard Transformer-MoE layers with its Sequence Parallelism to further enhance model flexibility and performance. Evaluations on two model series, A0.3B-2B and A1B-7B, demonstrate Linear-MoE achieves efficiency gains while maintaining competitive performance on various benchmarks, showcasing its potential as a next-generation foundational model architecture. Code: https://github.com/OpenSparseLLMs/Linear-MoE. |
2025-03-10T00:00:00 | 2503.04548 | An Empirical Study on Eliciting and Improving R1-like Reasoning Models | [
"Zhipeng Chen",
"Yingqian Min",
"Beichen Zhang",
"Jie Chen",
"Jinhao Jiang",
"Daixuan Cheng",
"Wayne Xin Zhao",
"Zheng Liu",
"Xu Miao",
"Yang Lu",
"Lei Fang",
"Zhongyuan Wang",
"Ji-Rong Wen"
] | https://github.com/RUCAIBox/Slow_Thinking_with_LLMs | In this report, we present the third technical report on the development of slow-thinking models as part of the STILL project. As the technical pathway becomes clearer, scaling RL training has become a central technique for implementing such reasoning models. We systematically experiment with and document the effects of various factors influencing RL training, conducting experiments on both base models and fine-tuned models. Specifically, we demonstrate that our RL training approach consistently improves the Qwen2.5-32B base models, enhancing both response length and test accuracy. Furthermore, we show that even when a model like DeepSeek-R1-Distill-Qwen-1.5B has already achieved a high performance level, it can be further refined through RL training, reaching an accuracy of 39.33% on AIME 2024. Beyond RL training, we also explore the use of tool manipulation, finding that it significantly boosts the reasoning performance of large reasoning models. This approach achieves a remarkable accuracy of 86.67% with greedy search on AIME 2024, underscoring its effectiveness in enhancing model capabilities. We release our resources at the STILL project website: https://github.com/RUCAIBox/Slow_Thinking_with_LLMs. |
2025-03-10T00:00:00 | 2503.04824 | ProReflow: Progressive Reflow with Decomposed Velocity | [
"Lei Ke",
"Haohang Xu",
"Xuefei Ning",
"Yu Li",
"Jiajun Li",
"Haoling Li",
"Yuxuan Lin",
"Dongsheng Jiang",
"Yujiu Yang",
"Linfeng Zhang"
] | Diffusion models have achieved significant progress in both image and video generation while still suffering from huge computation costs. As an effective solution, flow matching aims to reflow the diffusion process of diffusion models into a straight line for a few-step and even one-step generation. However, in this paper, we suggest that the original training pipeline of flow matching is not optimal and introduce two techniques to improve it. Firstly, we introduce progressive reflow, which progressively reflows the diffusion models in local timesteps until the whole diffusion progresses, reducing the difficulty of flow matching. Second, we introduce aligned v-prediction, which highlights the importance of direction matching in flow matching over magnitude matching. Experimental results on SDv1.5 and SDXL demonstrate the effectiveness of our method, for example, conducting on SDv1.5 achieves an FID of 10.70 on MSCOCO2014 validation set with only 4 sampling steps, close to our teacher model (32 DDIM steps, FID = 10.05). |
|
2025-03-10T00:00:00 | 2502.21263 | RuCCoD: Towards Automated ICD Coding in Russian | [
"Aleksandr Nesterov",
"Andrey Sakhovskiy",
"Ivan Sviridov",
"Airat Valiev",
"Vladimir Makharev",
"Petr Anokhin",
"Galina Zubkova",
"Elena Tutubalina"
] | This study investigates the feasibility of automating clinical coding in Russian, a language with limited biomedical resources. We present a new dataset for ICD coding, which includes diagnosis fields from electronic health records (EHRs) annotated with over 10,000 entities and more than 1,500 unique ICD codes. This dataset serves as a benchmark for several state-of-the-art models, including BERT, LLaMA with LoRA, and RAG, with additional experiments examining transfer learning across domains (from PubMed abstracts to medical diagnosis) and terminologies (from UMLS concepts to ICD codes). We then apply the best-performing model to label an in-house EHR dataset containing patient histories from 2017 to 2021. Our experiments, conducted on a carefully curated test set, demonstrate that training with the automated predicted codes leads to a significant improvement in accuracy compared to manually annotated data from physicians. We believe our findings offer valuable insights into the potential for automating clinical coding in resource-limited languages like Russian, which could enhance clinical efficiency and data accuracy in these contexts. |
|
2025-03-10T00:00:00 | 2503.04359 | LONGCODEU: Benchmarking Long-Context Language Models on Long Code Understanding | [
"Jia Li",
"Xuyuan Guo",
"Lei Li",
"Kechi Zhang",
"Ge Li",
"Jia Li",
"Zhengwei Tao",
"Fang Liu",
"Chongyang Tao",
"Yuqi Zhu",
"Zhi Jin"
] | Current advanced long-context language models offer great potential for real-world software engineering applications. However, progress in this critical domain remains hampered by a fundamental limitation: the absence of a rigorous evaluation framework for long code understanding. To gap this obstacle, we propose a long code understanding benchmark LONGCODEU from four aspects (8 tasks) to evaluate LCLMs' long code understanding ability required for practical applications, including code unit perception, intra-code unit understanding, inter-code unit relation understanding, and long code documentation understanding. We evaluate 9 popular LCLMs on LONGCODEU (i.e., 6 general models and 3 code models). Our experimental results reveal key limitations in current LCLMs' capabilities for long code understanding. Particularly, the performance of LCLMs drops dramatically when the long code length is greater than 32K, falling far short of their claimed 128K-1M context windows. In the four aspects, inter-code unit relation understanding is the most challenging for LCLMs. Our study provides valuable insights for optimizing LCLMs and driving advancements in software engineering. |
|
2025-03-10T00:00:00 | 2503.05085 | S2S-Arena, Evaluating Speech2Speech Protocols on Instruction Following with Paralinguistic Information | [
"Feng Jiang",
"Zhiyu Lin",
"Fan Bu",
"Yuhao Du",
"Benyou Wang",
"Haizhou Li"
] | The rapid development of large language models (LLMs) has brought significant attention to speech models, particularly recent progress in speech2speech protocols supporting speech input and output. However, the existing benchmarks adopt automatic text-based evaluators for evaluating the instruction following ability of these models lack consideration for paralinguistic information in both speech understanding and generation. To address these issues, we introduce S2S-Arena, a novel arena-style S2S benchmark that evaluates instruction-following capabilities with paralinguistic information in both speech-in and speech-out across real-world tasks. We design 154 samples that fused TTS and live recordings in four domains with 21 tasks and manually evaluate existing popular speech models in an arena-style manner. The experimental results show that: (1) in addition to the superior performance of GPT-4o, the speech model of cascaded ASR, LLM, and TTS outperforms the jointly trained model after text-speech alignment in speech2speech protocols; (2) considering paralinguistic information, the knowledgeability of the speech model mainly depends on the LLM backbone, and the multilingual support of that is limited by the speech module; (3) excellent speech models can already understand the paralinguistic information in speech input, but generating appropriate audio with paralinguistic information is still a challenge. |
|
2025-03-10T00:00:00 | 2502.18968 | Know You First and Be You Better: Modeling Human-Like User Simulators via Implicit Profiles | [
"Kuang Wang",
"Xianfei Li",
"Shenghao Yang",
"Li Zhou",
"Feng Jiang",
"Haizhou Li"
] | User simulators are crucial for replicating human interactions with dialogue systems, supporting both collaborative training and automatic evaluation, especially for large language models (LLMs). However, existing simulators often rely solely on text utterances, missing implicit user traits such as personality, speaking style, and goals. In contrast, persona-based methods lack generalizability, as they depend on predefined profiles of famous individuals or archetypes. To address these challenges, we propose User Simulator with implicit Profiles (USP), a framework that infers implicit user profiles from human-machine conversations and uses them to generate more personalized and realistic dialogues. We first develop an LLM-driven extractor with a comprehensive profile schema. Then, we refine the simulation through conditional supervised fine-tuning and reinforcement learning with cycle consistency, optimizing it at both the utterance and conversation levels. Finally, we adopt a diverse profile sampler to capture the distribution of real-world user profiles. Experimental results demonstrate that USP outperforms strong baselines in terms of authenticity and diversity while achieving comparable performance in consistency. Furthermore, dynamic multi-turn evaluations based on USP strongly align with mainstream benchmarks, demonstrating its effectiveness in real-world applications. |
|
2025-03-10T00:00:00 | 2503.01840 | EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Test | [
"Yuhui Li",
"Fangyun Wei",
"Chao Zhang",
"Hongyang Zhang"
] | https://github.com/SafeAILab/EAGLE | The sequential nature of modern LLMs makes them expensive and slow, and speculative sampling has proven to be an effective solution to this problem. Methods like EAGLE perform autoregression at the feature level, reusing top-layer features from the target model to achieve better results than vanilla speculative sampling. A growing trend in the LLM community is scaling up training data to improve model intelligence without increasing inference costs. However, we observe that scaling up data provides limited improvements for EAGLE. We identify that this limitation arises from EAGLE's feature prediction constraints. In this paper, we introduce EAGLE-3, which abandons feature prediction in favor of direct token prediction and replaces reliance on top-layer features with multi-layer feature fusion via a technique named training-time test. These improvements significantly enhance performance and enable the draft model to fully benefit from scaling up training data. Our experiments include both chat models and reasoning models, evaluated on five tasks. The results show that EAGLE-3 achieves a speedup ratio up to 6.5x, with about 1.4x improvement over EAGLE-2. The code is available at https://github.com/SafeAILab/EAGLE. |
2025-03-10T00:00:00 | 2503.04957 | SafeArena: Evaluating the Safety of Autonomous Web Agents | [
"Ada Defne Tur",
"Nicholas Meade",
"Xing Han Lù",
"Alejandra Zambrano",
"Arkil Patel",
"Esin Durmus",
"Spandana Gella",
"Karolina Stańczak",
"Siva Reddy"
] | LLM-based agents are becoming increasingly proficient at solving web-based tasks. With this capability comes a greater risk of misuse for malicious purposes, such as posting misinformation in an online forum or selling illicit substances on a website. To evaluate these risks, we propose SafeArena, the first benchmark to focus on the deliberate misuse of web agents. SafeArena comprises 250 safe and 250 harmful tasks across four websites. We classify the harmful tasks into five harm categories -- misinformation, illegal activity, harassment, cybercrime, and social bias, designed to assess realistic misuses of web agents. We evaluate leading LLM-based web agents, including GPT-4o, Claude-3.5 Sonnet, Qwen-2-VL 72B, and Llama-3.2 90B, on our benchmark. To systematically assess their susceptibility to harmful tasks, we introduce the Agent Risk Assessment framework that categorizes agent behavior across four risk levels. We find agents are surprisingly compliant with malicious requests, with GPT-4o and Qwen-2 completing 34.7% and 27.3% of harmful requests, respectively. Our findings highlight the urgent need for safety alignment procedures for web agents. Our benchmark is available here: https://safearena.github.io |
|
2025-03-11T00:00:00 | 2503.06680 | FEA-Bench: A Benchmark for Evaluating Repository-Level Code Generation for Feature Implementation | [
"Wei Li",
"Xin Zhang",
"Zhongxin Guo",
"Shaoguang Mao",
"Wen Luo",
"Guangyue Peng",
"Yangyu Huang",
"Houfeng Wang",
"Scarlett Li"
] | Implementing new features in repository-level codebases is a crucial application of code generation models. However, current benchmarks lack a dedicated evaluation framework for this capability. To fill this gap, we introduce FEA-Bench, a benchmark designed to assess the ability of large language models (LLMs) to perform incremental development within code repositories. We collect pull requests from 83 GitHub repositories and use rule-based and intent-based filtering to construct task instances focused on new feature development. Each task instance containing code changes is paired with relevant unit test files to ensure that the solution can be verified. The feature implementation requires LLMs to simultaneously possess code completion capabilities for new components and code editing abilities for other relevant parts in the code repository, providing a more comprehensive evaluation method of LLMs' automated software engineering capabilities. Experimental results show that LLMs perform significantly worse in the FEA-Bench, highlighting considerable challenges in such repository-level incremental code development. |
|
2025-03-11T00:00:00 | 2503.03499 | State-offset Tuning: State-based Parameter-Efficient Fine-Tuning for State Space Models | [
"Wonjun Kang",
"Kevin Galim",
"Yuchen Zeng",
"Minjae Lee",
"Hyung Il Koo",
"Nam Ik Cho"
] | https://github.com/furiosa-ai/ssm-state-tuning | State Space Models (SSMs) have emerged as efficient alternatives to Transformers, mitigating their quadratic computational cost. However, the application of Parameter-Efficient Fine-Tuning (PEFT) methods to SSMs remains largely unexplored. In particular, prompt-based methods like Prompt Tuning and Prefix-Tuning, which are widely used in Transformers, do not perform well on SSMs. To address this, we propose state-based methods as a superior alternative to prompt-based methods. This new family of methods naturally stems from the architectural characteristics of SSMs. State-based methods adjust state-related features directly instead of depending on external prompts. Furthermore, we introduce a novel state-based PEFT method: State-offset Tuning. At every timestep, our method directly affects the state at the current step, leading to more effective adaptation. Through extensive experiments across diverse datasets, we demonstrate the effectiveness of our method. Code is available at https://github.com/furiosa-ai/ssm-state-tuning. |
2025-03-11T00:00:00 | 2503.03511 | NeuGrasp: Generalizable Neural Surface Reconstruction with Background Priors for Material-Agnostic Object Grasp Detection | [
"Qingyu Fan",
"Yinghao Cai",
"Chao Li",
"Wenzhe He",
"Xudong Zheng",
"Tao Lu",
"Bin Liang",
"Shuo Wang"
] | Robotic grasping in scenes with transparent and specular objects presents great challenges for methods relying on accurate depth information. In this paper, we introduce NeuGrasp, a neural surface reconstruction method that leverages background priors for material-agnostic grasp detection. NeuGrasp integrates transformers and global prior volumes to aggregate multi-view features with spatial encoding, enabling robust surface reconstruction in narrow and sparse viewing conditions. By focusing on foreground objects through residual feature enhancement and refining spatial perception with an occupancy-prior volume, NeuGrasp excels in handling objects with transparent and specular surfaces. Extensive experiments in both simulated and real-world scenarios show that NeuGrasp outperforms state-of-the-art methods in grasping while maintaining comparable reconstruction quality. More details are available at https://neugrasp.github.io/. |
|
2025-03-11T00:00:00 | 2503.02199 | Words or Vision: Do Vision-Language Models Have Blind Faith in Text? | [
"Ailin Deng",
"Tri Cao",
"Zhirui Chen",
"Bryan Hooi"
] | Vision-Language Models (VLMs) excel in integrating visual and textual information for vision-centric tasks, but their handling of inconsistencies between modalities is underexplored. We investigate VLMs' modality preferences when faced with visual data and varied textual inputs in vision-centered settings. By introducing textual variations to four vision-centric tasks and evaluating ten Vision-Language Models (VLMs), we discover a ``blind faith in text'' phenomenon: VLMs disproportionately trust textual data over visual data when inconsistencies arise, leading to significant performance drops under corrupted text and raising safety concerns. We analyze factors influencing this text bias, including instruction prompts, language model size, text relevance, token order, and the interplay between visual and textual certainty. While certain factors, such as scaling up the language model size, slightly mitigate text bias, others like token order can exacerbate it due to positional biases inherited from language models. To address this issue, we explore supervised fine-tuning with text augmentation and demonstrate its effectiveness in reducing text bias. Additionally, we provide a theoretical analysis suggesting that the blind faith in text phenomenon may stem from an imbalance of pure text and multi-modal data during training. Our findings highlight the need for balanced training and careful consideration of modality interactions in VLMs to enhance their robustness and reliability in handling multi-modal data inconsistencies. |
|
2025-03-11T00:00:00 | 2503.07365 | MM-Eureka: Exploring Visual Aha Moment with Rule-based Large-scale Reinforcement Learning | [
"Fanqing Meng",
"Lingxiao Du",
"Zongkai Liu",
"Zhixiang Zhou",
"Quanfeng Lu",
"Daocheng Fu",
"Botian Shi",
"Wenhai Wang",
"Junjun He",
"Kaipeng Zhang",
"Ping Luo",
"Yu Qiao",
"Qiaosheng Zhang",
"Wenqi Shao"
] | https://github.com/ModalMinds/MM-EUREKA | We present MM-Eureka, a multimodal reasoning model that successfully extends large-scale rule-based reinforcement learning (RL) to multimodal reasoning. While rule-based RL has shown remarkable success in improving LLMs' reasoning abilities in text domains, its application to multimodal settings has remained challenging. Our work reproduces key characteristics of text-based RL systems like DeepSeek-R1 in the multimodal space, including steady increases in accuracy reward and response length, and the emergence of reflection behaviors. We demonstrate that both instruction-tuned and pre-trained models can develop strong multimodal reasoning capabilities through rule-based RL without supervised fine-tuning, showing superior data efficiency compared to alternative approaches. We open-source our complete pipeline to foster further research in this area. We release all our codes, models, data, etc. at https://github.com/ModalMinds/MM-EUREKA |
2025-03-11T00:00:00 | 2503.04812 | LLaVE: Large Language and Vision Embedding Models with Hardness-Weighted Contrastive Learning | [
"Zhibin Lan",
"Liqiang Niu",
"Fandong Meng",
"Jie Zhou",
"Jinsong Su"
] | Universal multimodal embedding models play a critical role in tasks such as interleaved image-text retrieval, multimodal RAG, and multimodal clustering. However, our empirical results indicate that existing LMM-based embedding models trained with the standard InfoNCE loss exhibit a high degree of overlap in similarity distribution between positive and negative pairs, making it challenging to distinguish hard negative pairs effectively. To deal with this issue, we propose a simple yet effective framework that dynamically improves the embedding model's representation learning for negative pairs based on their discriminative difficulty. Within this framework, we train a series of models, named LLaVE, and evaluate them on the MMEB benchmark, which covers 4 meta-tasks and 36 datasets. Experimental results show that LLaVE establishes stronger baselines that achieve state-of-the-art (SOTA) performance while demonstrating strong scalability and efficiency. Specifically, LLaVE-2B surpasses the previous SOTA 7B models, while LLaVE-7B achieves a further performance improvement of 6.2 points. Although LLaVE is trained on image-text data, it can generalize to text-video retrieval tasks in a zero-shot manner and achieve strong performance, demonstrating its remarkable potential for transfer to other embedding tasks. |
Subsets and Splits