arxiv_id
float64
1.5k
2.51k
title
stringlengths
9
178
authors
stringlengths
2
22.8k
categories
stringlengths
4
146
summary
stringlengths
103
1.92k
published
stringdate
2015-02-06 10:44:00
2025-07-10 17:59:58
comments
stringlengths
2
417
journal_ref
stringclasses
321 values
doi
stringclasses
398 values
ss_title
stringlengths
8
159
ss_authors
stringlengths
11
8.38k
ss_year
float64
2.02k
2.03k
ss_venue
stringclasses
281 values
ss_citationCount
float64
0
134k
ss_referenceCount
float64
0
429
ss_fieldsOfStudy
stringclasses
47 values
2,502.17057
ExpandR: Teaching Dense Retrievers Beyond Queries with LLM Guidance
['Sijia Yao', 'Pengcheng Huang', 'Zhenghao Liu', 'Yu Gu', 'Yukun Yan', 'Shi Yu', 'Ge Yu']
['cs.IR', 'cs.AI']
Large language models (LLMs) have demonstrated significant potential in enhancing dense retrieval through query augmentation. However, most existing methods treat the LLM and the retriever as separate modules, overlooking the alignment between generation and ranking objectives. In this work, we propose ExpandR, a unifi...
2025-02-24T11:15:41Z
16 pages, 10 tables, 5 figures
null
null
null
null
null
null
null
null
null
2,502.17125
LettuceDetect: A Hallucination Detection Framework for RAG Applications
['Ádám Kovács', 'Gábor Recski']
['cs.CL', 'cs.AI']
Retrieval Augmented Generation (RAG) systems remain vulnerable to hallucinated answers despite incorporating external knowledge sources. We present LettuceDetect a framework that addresses two critical limitations in existing hallucination detection methods: (1) the context window constraints of traditional encoder-bas...
2025-02-24T13:11:47Z
6 pages
null
null
LettuceDetect: A Hallucination Detection Framework for RAG Applications
['Adam Kovacs', 'Gábor Recski']
2,025
arXiv.org
5
29
['Computer Science']
2,502.17237
MegaLoc: One Retrieval to Place Them All
['Gabriele Berton', 'Carlo Masone']
['cs.CV']
Retrieving images from the same location as a given query is an important component of multiple computer vision tasks, like Visual Place Recognition, Landmark Retrieval, Visual Localization, 3D reconstruction, and SLAM. However, existing solutions are built to specifically work for one of these tasks, and are known to ...
2025-02-24T15:14:55Z
Tech Report
null
null
null
null
null
null
null
null
null
2,502.17239
Baichuan-Audio: A Unified Framework for End-to-End Speech Interaction
['Tianpeng Li', 'Jun Liu', 'Tao Zhang', 'Yuanbo Fang', 'Da Pan', 'Mingrui Wang', 'Zheng Liang', 'Zehuan Li', 'Mingan Lin', 'Guosheng Dong', 'Jianhua Xu', 'Haoze Sun', 'Zenan Zhou', 'Weipeng Chen']
['cs.CL', 'cs.SD', 'eess.AS']
We introduce Baichuan-Audio, an end-to-end audio large language model that seamlessly integrates audio understanding and generation. It features a text-guided aligned speech generation mechanism, enabling real-time speech interaction with both comprehension and generation capabilities. Baichuan-Audio leverages a pre-tr...
2025-02-24T15:16:34Z
null
null
null
Baichuan-Audio: A Unified Framework for End-to-End Speech Interaction
['Tianpeng Li', 'Jun Liu', 'Tao Zhang', 'Yuanbo Fang', 'Da Pan', 'Mingrui Wang', 'Zheng Liang', 'Zehuan Li', 'Mingan Lin', 'Guosheng Dong', 'Jianhua Xu', 'Haoze Sun', 'Zenan Zhou', 'Weipeng Chen']
2,025
arXiv.org
7
44
['Computer Science', 'Engineering']
2,502.17387
Big-Math: A Large-Scale, High-Quality Math Dataset for Reinforcement Learning in Language Models
['Alon Albalak', 'Duy Phung', 'Nathan Lile', 'Rafael Rafailov', 'Kanishk Gandhi', 'Louis Castricato', 'Anikait Singh', 'Chase Blagden', 'Violet Xiang', 'Dakota Mahan', 'Nick Haber']
['cs.LG', 'cs.AI', 'cs.CL']
Increasing interest in reasoning models has led math to become a prominent testing ground for algorithmic and methodological improvements. However, existing open math datasets either contain a small collection of high-quality, human-written problems or a large corpus of machine-generated problems of uncertain quality, ...
2025-02-24T18:14:01Z
null
null
null
null
null
null
null
null
null
null
2,502.17424
Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs
['Jan Betley', 'Daniel Tan', 'Niels Warncke', 'Anna Sztyber-Betley', 'Xuchan Bao', 'Martín Soto', 'Nathan Labenz', 'Owain Evans']
['cs.CL', 'cs.AI', 'cs.CR', 'cs.LG']
We present a surprising result regarding LLMs and alignment. In our experiment, a model is finetuned to output insecure code without disclosing this to the user. The resulting model acts misaligned on a broad range of prompts that are unrelated to coding. It asserts that humans should be enslaved by AI, gives malicious...
2025-02-24T18:56:03Z
40 pages, 38 figures An earlier revision of this paper was accepted at ICML 2025. Since then, it has been updated to include new results on training dynamics (4.7) and base models (4.8)
null
null
null
null
null
null
null
null
null
2,502.17425
Introducing Visual Perception Token into Multimodal Large Language Model
['Runpeng Yu', 'Xinyin Ma', 'Xinchao Wang']
['cs.CV', 'cs.LG']
To utilize visual information, Multimodal Large Language Model (MLLM) relies on the perception process of its vision encoder. The completeness and accuracy of visual perception significantly influence the precision of spatial reasoning, fine-grained understanding, and other tasks. However, MLLM still lacks the autonomo...
2025-02-24T18:56:12Z
null
null
null
null
null
null
null
null
null
null
2,502.17437
Fractal Generative Models
['Tianhong Li', 'Qinyi Sun', 'Lijie Fan', 'Kaiming He']
['cs.LG', 'cs.CV']
Modularization is a cornerstone of computer science, abstracting complex functions into atomic building blocks. In this paper, we introduce a new level of modularization by abstracting generative models into atomic generative modules. Analogous to fractals in mathematics, our method constructs a new type of generative ...
2025-02-24T18:59:56Z
null
null
null
Fractal Generative Models
['Tianhong Li', 'Qinyi Sun', 'Lijie Fan', 'Kaiming He']
2,025
arXiv.org
8
0
['Computer Science']
2,502.17543
Training a Generally Curious Agent
['Fahim Tajwar', 'Yiding Jiang', 'Abitha Thankaraj', 'Sumaita Sadia Rahman', 'J Zico Kolter', 'Jeff Schneider', 'Ruslan Salakhutdinov']
['cs.LG', 'cs.AI', 'cs.CL']
Efficient exploration is essential for intelligent systems interacting with their environment, but existing language models often fall short in scenarios that require strategic information gathering. In this paper, we present Paprika, a fine-tuning approach that enables language models to develop general decision-makin...
2025-02-24T18:56:58Z
ICML 2025. Project Website: https://paprika-llm.github.io
null
null
Training a Generally Curious Agent
['Fahim Tajwar', 'Yiding Jiang', 'Abitha Thankaraj', 'Sumaita Sadia Rahman', 'J. Z. Kolter', 'Jeff Schneider', 'Ruslan Salakhutdinov']
2,025
arXiv.org
3
86
['Computer Science']
2,502.17579
VANPY: Voice Analysis Framework
['Gregory Koushnir', 'Michael Fire', 'Galit Fuhrmann Alpert', 'Dima Kagan']
['cs.SD', 'cs.LG', 'eess.AS']
Voice data is increasingly being used in modern digital communications, yet there is still a lack of comprehensive tools for automated voice analysis and characterization. To this end, we developed the VANPY (Voice Analysis in Python) framework for automated pre-processing, feature extraction, and classification of voi...
2025-02-17T21:12:57Z
null
null
null
null
null
null
null
null
null
null
2,502.17796
LAM: Large Avatar Model for One-shot Animatable Gaussian Head
['Yisheng He', 'Xiaodong Gu', 'Xiaodan Ye', 'Chao Xu', 'Zhengyi Zhao', 'Yuan Dong', 'Weihao Yuan', 'Zilong Dong', 'Liefeng Bo']
['cs.CV']
We present LAM, an innovative Large Avatar Model for animatable Gaussian head reconstruction from a single image. Unlike previous methods that require extensive training on captured video sequences or rely on auxiliary neural networks for animation and rendering during inference, our approach generates Gaussian heads t...
2025-02-25T02:57:45Z
Project Page: https://aigc3d.github.io/projects/LAM/ Source code: https://github.com/aigc3d/LAM
null
null
LAM: Large Avatar Model for One-shot Animatable Gaussian Head
['Yisheng He', 'Xiaodong Gu', 'Xiaodan Ye', 'Chao Xu', 'Zhengyi Zhao', 'Yuan Dong', 'Weihao Yuan', 'Zilong Dong', 'Liefeng Bo']
2,025
arXiv.org
0
77
['Computer Science']
2,502.18008
NotaGen: Advancing Musicality in Symbolic Music Generation with Large Language Model Training Paradigms
['Yashan Wang', 'Shangda Wu', 'Jianhuai Hu', 'Xingjian Du', 'Yueqi Peng', 'Yongxin Huang', 'Shuai Fan', 'Xiaobing Li', 'Feng Yu', 'Maosong Sun']
['cs.SD', 'cs.AI', 'eess.AS']
We introduce NotaGen, a symbolic music generation model aiming to explore the potential of producing high-quality classical sheet music. Inspired by the success of Large Language Models (LLMs), NotaGen adopts pre-training, fine-tuning, and reinforcement learning paradigms (henceforth referred to as the LLM training par...
2025-02-25T09:12:07Z
null
null
null
null
null
null
null
null
null
null
2,502.18009
Patient Trajectory Prediction: Integrating Clinical Notes with Transformers
['Sifal Klioui', 'Sana Sellami', 'Youssef Trardi']
['cs.LG']
Predicting disease trajectories from electronic health records (EHRs) is a complex task due to major challenges such as data non-stationarity, high granularity of medical codes, and integration of multimodal data. EHRs contain both structured data, such as diagnostic codes, and unstructured data, such as clinical notes...
2025-02-25T09:14:07Z
null
null
null
Patient Trajectory Prediction: Integrating Clinical Notes with Transformers
['Sifal Klioui', 'Sana Sellami', 'Youssef Trardi']
2,025
BIOSTEC : HEALTHINF
0
25
['Computer Science']
2,502.18041
OpenFly: A Comprehensive Platform for Aerial Vision-Language Navigation
['Yunpeng Gao', 'Chenhui Li', 'Zhongrui You', 'Junli Liu', 'Zhen Li', 'Pengan Chen', 'Qizhi Chen', 'Zhonghan Tang', 'Liansheng Wang', 'Penghui Yang', 'Yiwen Tang', 'Yuhang Tang', 'Shuai Liang', 'Songyi Zhu', 'Ziqin Xiong', 'Yifei Su', 'Xinyi Ye', 'Jianan Li', 'Yan Ding', 'Dong Wang', 'Zhigang Wang', 'Bin Zhao', 'Xuelon...
['cs.CV', 'cs.RO']
Vision-Language Navigation (VLN) aims to guide agents by leveraging language instructions and visual cues, playing a pivotal role in embodied AI. Indoor VLN has been extensively studied, whereas outdoor aerial VLN remains underexplored. The potential reason is that outdoor aerial view encompasses vast areas, making dat...
2025-02-25T09:57:18Z
null
null
null
null
null
null
null
null
null
null
2,502.18101
Detecting Offensive Memes with Social Biases in Singapore Context Using Multimodal Large Language Models
['Cao Yuxuan', 'Wu Jiayang', 'Alistair Cheong Liang Chuen', 'Bryan Shan Guanrong', 'Theodore Lee Chong Jen', 'Sherman Chann Zhi Shen']
['cs.CV', 'cs.CL']
Traditional online content moderation systems struggle to classify modern multimodal means of communication, such as memes, a highly nuanced and information-dense medium. This task is especially hard in a culturally diverse society like Singapore, where low-resource languages are used and extensive knowledge on local c...
2025-02-25T11:15:49Z
Accepted at 3rd Workshop on Cross-Cultural Considerations in NLP (C3NLP), co-located with NAACL 2025. This is an extended version with some appendix moved to the main body
null
null
Detecting Offensive Memes with Social Biases in Singapore Context Using Multimodal Large Language Models
['Yuxuan Cao', 'Jiayang Wu', 'Alistair Cheong Liang Chuen', 'Bryan Shan Guanrong', 'Theodore Lee Chong Jen', 'Sherman Chann Zhi Shen']
2,025
arXiv.org
0
48
['Computer Science']
2,502.18137
SpargeAttention: Accurate and Training-free Sparse Attention Accelerating Any Model Inference
['Jintao Zhang', 'Chendong Xiang', 'Haofeng Huang', 'Jia Wei', 'Haocheng Xi', 'Jun Zhu', 'Jianfei Chen']
['cs.LG', 'cs.AI', 'cs.CV', 'cs.PF']
An efficient attention implementation is essential for large models due to its quadratic time complexity. Fortunately, attention commonly exhibits sparsity, i.e., many values in the attention map are near zero, allowing for the omission of corresponding computations. Many studies have utilized the sparse pattern to acc...
2025-02-25T12:02:17Z
@inproceedings{zhang2025spargeattn, title={Spargeattn: Accurate sparse attention accelerating any model inference}, author={Zhang, Jintao and Xiang, Chendong and Huang, Haofeng and Wei, Jia and Xi, Haocheng and Zhu, Jun and Chen, Jianfei}, booktitle={International Conference on Machine Learning (ICML)}, year={2...
Proceedings of the 42 nd International Conference on Machine Learning, PMLR 267, 2025 (ICML 2025)
null
null
null
null
null
null
null
null
2,502.18186
Steering Language Model to Stable Speech Emotion Recognition via Contextual Perception and Chain of Thought
['Zhixian Zhao', 'Xinfa Zhu', 'Xinsheng Wang', 'Shuiyuan Wang', 'Xuelong Geng', 'Wenjie Tian', 'Lei Xie']
['cs.SD', 'cs.CL', 'eess.AS']
Large-scale audio language models (ALMs), such as Qwen2-Audio, are capable of comprehending diverse audio signal, performing audio analysis and generating textual responses. However, in speech emotion recognition (SER), ALMs often suffer from hallucinations, resulting in misclassifications or irrelevant outputs. To add...
2025-02-25T13:26:25Z
null
null
null
null
null
null
null
null
null
null
2,502.18274
Citrus: Leveraging Expert Cognitive Pathways in a Medical Language Model for Advanced Medical Decision Support
['Guoxin Wang', 'Minyu Gao', 'Shuai Yang', 'Ya Zhang', 'Lizhi He', 'Liang Huang', 'Hanlin Xiao', 'Yexuan Zhang', 'Wanyue Li', 'Lu Chen', 'Jintao Fei', 'Xin Li']
['cs.AI', 'cs.CL']
Large language models (LLMs), particularly those with reasoning capabilities, have rapidly advanced in recent years, demonstrating significant potential across a wide range of applications. However, their deployment in healthcare, especially in disease reasoning tasks, is hindered by the challenge of acquiring expert-l...
2025-02-25T15:05:12Z
null
null
null
Citrus: Leveraging Expert Cognitive Pathways in a Medical Language Model for Advanced Medical Decision Support
['Guoxin Wang', 'Minyu Gao', 'Shuai Yang', 'Ya Zhang', 'Lizhi He', 'Liang Huang', 'Hanlin Xiao', 'Yexuan Zhang', 'Wanyue Li', 'Lu Chen', 'Jintao Fei', 'Xin Li']
2,025
arXiv.org
2
86
['Computer Science']
2,502.18277
Self-Adjust Softmax
['Chuanyang Zheng', 'Yihang Gao', 'Guoxuan Chen', 'Han Shi', 'Jing Xiong', 'Xiaozhe Ren', 'Chao Huang', 'Xin Jiang', 'Zhenguo Li', 'Yu Li']
['cs.CL']
The softmax function is crucial in Transformer attention, which normalizes each row of the attention scores with summation to one, achieving superior performances over other alternative functions. However, the softmax function can face a gradient vanishing issue when some elements of the attention scores approach extre...
2025-02-25T15:07:40Z
Tech Report
null
null
Self-Adjust Softmax
['Chuanyang Zheng', 'Yihang Gao', 'Guoxuan Chen', 'Han Shi', 'Jing Xiong', 'Xiaozhe Ren', 'Chao Huang', 'Xin Jiang', 'Zhenguo Li', 'Yu Li']
2,025
arXiv.org
1
108
['Computer Science']
2,502.18316
WiCkeD: A Simple Method to Make Multiple Choice Benchmarks More Challenging
['Ahmed Elhady', 'Eneko Agirre', 'Mikel Artetxe']
['cs.CL']
We introduce WiCkeD, a simple method to increase the complexity of existing multiple-choice benchmarks by randomly replacing a choice with "None of the above", a method often used in educational tests. We show that WiCkeD can be automatically applied to any existing benchmark, making it more challenging. We apply WiCke...
2025-02-25T16:09:38Z
null
null
null
null
null
null
null
null
null
null
2,502.18364
ART: Anonymous Region Transformer for Variable Multi-Layer Transparent Image Generation
['Yifan Pu', 'Yiming Zhao', 'Zhicong Tang', 'Ruihong Yin', 'Haoxing Ye', 'Yuhui Yuan', 'Dong Chen', 'Jianmin Bao', 'Sirui Zhang', 'Yanbin Wang', 'Lin Liang', 'Lijuan Wang', 'Ji Li', 'Xiu Li', 'Zhouhui Lian', 'Gao Huang', 'Baining Guo']
['cs.CV']
Multi-layer image generation is a fundamental task that enables users to isolate, select, and edit specific image layers, thereby revolutionizing interactions with generative models. In this paper, we introduce the Anonymous Region Transformer (ART), which facilitates the direct generation of variable multi-layer trans...
2025-02-25T16:57:04Z
Project page: https://art-msra.github.io/
null
null
ART: Anonymous Region Transformer for Variable Multi-Layer Transparent Image Generation
['Yifan Pu', 'Yiming Zhao', 'Zhicong Tang', 'Ruihong Yin', 'Haoxing Ye', 'Yuhui Yuan', 'Dong Chen', 'Jianmin Bao', 'Sirui Zhang', 'Yanbin Wang', 'Lin Liang', 'Lijuan Wang', 'Ji Li', 'Xiu Li', 'Zhouhui Lian', 'Gao Huang', 'Baining Guo']
2,025
Computer Vision and Pattern Recognition
5
56
['Computer Science']
2,502.18411
OmniAlign-V: Towards Enhanced Alignment of MLLMs with Human Preference
['Xiangyu Zhao', 'Shengyuan Ding', 'Zicheng Zhang', 'Haian Huang', 'Maosong Cao', 'Weiyun Wang', 'Jiaqi Wang', 'Xinyu Fang', 'Wenhai Wang', 'Guangtao Zhai', 'Haodong Duan', 'Hua Yang', 'Kai Chen']
['cs.CV']
Recent advancements in open-source multi-modal large language models (MLLMs) have primarily focused on enhancing foundational capabilities, leaving a significant gap in human preference alignment. This paper introduces OmniAlign-V, a comprehensive dataset of 200K high-quality training samples featuring diverse images, ...
2025-02-25T18:05:14Z
null
null
null
null
null
null
null
null
null
null
2,502.18418
Rank1: Test-Time Compute for Reranking in Information Retrieval
['Orion Weller', 'Kathryn Ricci', 'Eugene Yang', 'Andrew Yates', 'Dawn Lawrie', 'Benjamin Van Durme']
['cs.IR', 'cs.CL', 'cs.LG']
We introduce Rank1, the first reranking model trained to take advantage of test-time compute. Rank1 demonstrates the applicability within retrieval of using a reasoning language model (i.e. OpenAI's o1, Deepseek's R1, etc.) for distillation in order to rapidly improve the performance of a smaller model. We gather and o...
2025-02-25T18:14:06Z
null
null
null
Rank1: Test-Time Compute for Reranking in Information Retrieval
['Orion Weller', 'Kathryn Ricci', 'Eugene Yang', 'Andrew Yates', 'Dawn J. Lawrie', 'Benjamin Van Durme']
2,025
arXiv.org
12
45
['Computer Science']
2,502.18435
What Makes the Preferred Thinking Direction for LLMs in Multiple-choice Questions?
['Yizhe Zhang', 'Richard Bai', 'Zijin Gu', 'Ruixiang Zhang', 'Jiatao Gu', 'Emmanuel Abbe', 'Samy Bengio', 'Navdeep Jaitly']
['cs.CL', 'cs.IT', 'cs.LG', 'math.IT']
Language models usually use left-to-right (L2R) autoregressive factorization. However, L2R factorization may not always be the best inductive bias. Therefore, we investigate whether alternative factorizations of the text distribution could be beneficial in some tasks. We investigate right-to-left (R2L) training as a co...
2025-02-25T18:30:25Z
10 pages for the main text
null
null
null
null
null
null
null
null
null
2,502.1846
DRAMA: Diverse Augmentation from Large Language Models to Smaller Dense Retrievers
['Xueguang Ma', 'Xi Victoria Lin', 'Barlas Oguz', 'Jimmy Lin', 'Wen-tau Yih', 'Xilun Chen']
['cs.CL', 'cs.IR']
Large language models (LLMs) have demonstrated strong effectiveness and robustness while fine-tuned as dense retrievers. However, their large parameter size brings significant inference time computational challenges, including high encoding costs for large-scale corpora and increased query latency, limiting their pract...
2025-02-25T18:59:07Z
ACL 2025
null
null
null
null
null
null
null
null
null
2,502.186
Chain of Draft: Thinking Faster by Writing Less
['Silei Xu', 'Wenhao Xie', 'Lingxiao Zhao', 'Pengcheng He']
['cs.CL', 'I.2.7']
Large Language Models (LLMs) have demonstrated remarkable performance in solving complex reasoning tasks through mechanisms like Chain-of-Thought (CoT) prompting, which emphasizes verbose, step-by-step reasoning. However, humans typically employ a more efficient strategy: drafting concise intermediate thoughts that cap...
2025-02-25T19:36:06Z
null
null
null
null
null
null
null
null
null
null
2,502.18679
Discriminative Finetuning of Generative Large Language Models without Reward Models and Human Preference Data
['Siqi Guo', 'Ilgee Hong', 'Vicente Balmaseda', 'Changlong Yu', 'Liang Qiu', 'Xin Liu', 'Haoming Jiang', 'Tuo Zhao', 'Tianbao Yang']
['cs.CL']
Supervised fine-tuning (SFT) has become a crucial step for aligning pretrained large language models (LLMs) using supervised datasets of input-output pairs. However, despite being supervised, SFT is inherently limited by its generative training objective. To address its limitations, the existing common strategy is to f...
2025-02-25T22:38:55Z
18 pages, 7 figures
null
null
Discriminative Finetuning of Generative Large Language Models without Reward Models and Human Preference Data
['Siqi Guo', 'Ilgee Hong', 'Vicente Balmaseda', 'Changlong Yu', 'Liang Qiu', 'Xin Liu', 'Haoming Jiang', 'Tuo Zhao', 'Tianbao Yang']
2,025
null
0
60
['Computer Science']
2,502.18772
Plutus: Benchmarking Large Language Models in Low-Resource Greek Finance
['Xueqing Peng', 'Triantafillos Papadopoulos', 'Efstathia Soufleri', 'Polydoros Giannouris', 'Ruoyu Xiang', 'Yan Wang', 'Lingfei Qian', 'Jimin Huang', 'Qianqian Xie', 'Sophia Ananiadou']
['cs.CL']
Despite Greece's pivotal role in the global economy, large language models (LLMs) remain underexplored for Greek financial context due to the linguistic complexity of Greek and the scarcity of domain-specific datasets. Previous efforts in multilingual financial natural language processing (NLP) have exposed considerabl...
2025-02-26T03:04:01Z
18 pages, 6 figures
null
null
Plutus: Benchmarking Large Language Models in Low-Resource Greek Finance
['Xueqing Peng', 'Triantafillos Papadopoulos', 'Efstathia Soufleri', 'Polydoros Giannouris', 'Ruoyu Xiang', 'Yan Wang', 'Lingfei Qian', 'Jimin Huang', 'Qianqian Xie', 'Sophia Ananiadou']
2,025
arXiv.org
3
0
['Computer Science']
2,502.18886
On Pruning State-Space LLMs
['Tamer Ghattas', 'Michael Hassid', 'Roy Schwartz']
['cs.CL', 'cs.LG']
Recent work proposed state-space models (SSMs) as an efficient alternative to transformer-based LLMs. Can these models be pruned to further reduce their computation costs? We adapt several pruning methods to the SSM structure, and apply them to four SSM-based LLMs across multiple tasks. We find that such models are qui...
2025-02-26T07:04:20Z
null
null
null
null
null
null
null
null
null
null
2,502.18924
MegaTTS 3: Sparse Alignment Enhanced Latent Diffusion Transformer for Zero-Shot Speech Synthesis
['Ziyue Jiang', 'Yi Ren', 'Ruiqi Li', 'Shengpeng Ji', 'Boyang Zhang', 'Zhenhui Ye', 'Chen Zhang', 'Bai Jionghao', 'Xiaoda Yang', 'Jialong Zuo', 'Yu Zhang', 'Rui Liu', 'Xiang Yin', 'Zhou Zhao']
['eess.AS', 'cs.LG', 'cs.SD']
While recent zero-shot text-to-speech (TTS) models have significantly improved speech quality and expressiveness, mainstream systems still suffer from issues related to speech-text alignment modeling: 1) models without explicit speech-text alignment modeling exhibit less robustness, especially for hard sentences in pra...
2025-02-26T08:22:00Z
null
null
null
null
null
null
null
null
null
null
2,502.18934
Kanana: Compute-efficient Bilingual Language Models
['Kanana LLM Team', 'Yunju Bak', 'Hojin Lee', 'Minho Ryu', 'Jiyeon Ham', 'Seungjae Jung', 'Daniel Wontae Nam', 'Taegyeong Eo', 'Donghun Lee', 'Doohae Jung', 'Boseop Kim', 'Nayeon Kim', 'Jaesun Park', 'Hyunho Kim', 'Hyunwoong Ko', 'Changmin Lee', 'Kyoung-Woon On', 'Seulye Baeg', 'Junrae Cho', 'Sunghee Jung', 'Jieun Kang...
['cs.CL', 'cs.LG']
We introduce Kanana, a series of bilingual language models that demonstrate exceeding performance in Korean and competitive performance in English. The computational cost of Kanana is significantly lower than that of state-of-the-art models of similar size. The report details the techniques employed during pre-training...
2025-02-26T08:36:20Z
40 pages, 15 figures
null
null
Kanana: Compute-efficient Bilingual Language Models
['Kanana Llm Team Yunju Bak', 'Hojin Lee', 'Minho Ryu', 'Jiyeon Ham', 'Seungjae Jung', 'D. W. Nam', 'Taegyeong Eo', 'Donghun Lee', 'Doohae Jung', 'Boseop Kim', 'Nayeon Kim', 'Jaesun Park', 'Hyunho Kim', 'H. Ko', 'Changmin Lee', 'Kyoung-Woon On', 'Seulye Baeg', 'Junrae Cho', 'Sunghee Jung', 'Jieun Kang', 'EungGyun Kim',...
2,025
arXiv.org
1
86
['Computer Science']
2,502.1894
MathTutorBench: A Benchmark for Measuring Open-ended Pedagogical Capabilities of LLM Tutors
['Jakub Macina', 'Nico Daheim', 'Ido Hakimi', 'Manu Kapur', 'Iryna Gurevych', 'Mrinmaya Sachan']
['cs.CL', 'cs.AI', 'cs.LG']
Evaluating the pedagogical capabilities of AI-based tutoring models is critical for making guided progress in the field. Yet, we lack a reliable, easy-to-use, and simple-to-run evaluation that reflects the pedagogical abilities of models. To fill this gap, we present MathTutorBench, an open-source benchmark for holisti...
2025-02-26T08:43:47Z
https://eth-lre.github.io/mathtutorbench
null
null
null
null
null
null
null
null
null
2,502.18968
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']
['cs.CL']
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, current role-playing methods face challenges such as a lack of utterance-level authenticity and user-level diversit...
2025-02-26T09:26:54Z
9 pages. Accepted to ACL 2025. Camera-ready version
null
null
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']
2,025
arXiv.org
0
63
['Computer Science']
2,502.18969
(Mis)Fitting: A Survey of Scaling Laws
['Margaret Li', 'Sneha Kudugunta', 'Luke Zettlemoyer']
['cs.LG', 'cs.AI', 'cs.CL', 'stat.ME']
Modern foundation models rely heavily on using scaling laws to guide crucial training decisions. Researchers often extrapolate the optimal architecture and hyper parameters settings from smaller training runs by describing the relationship between, loss, or task performance, and scale. All components of this process va...
2025-02-26T09:27:54Z
41 pages, 3 figure, first two authors contributed equally. ICLR, 2025
null
null
(Mis)Fitting: A Survey of Scaling Laws
['Margaret Li', 'Sneha Kudugunta', 'Luke S. Zettlemoyer']
2,025
arXiv.org
4
80
['Computer Science', 'Mathematics']
2,502.19204
Distill Any Depth: Distillation Creates a Stronger Monocular Depth Estimator
['Xiankang He', 'Dongyan Guo', 'Hongji Li', 'Ruibo Li', 'Ying Cui', 'Chi Zhang']
['cs.CV']
Recent advances in zero-shot monocular depth estimation(MDE) have significantly improved generalization by unifying depth distributions through normalized depth representations and by leveraging large-scale unlabeled data via pseudo-label distillation. However, existing methods that rely on global depth normalization t...
2025-02-26T15:10:05Z
project page: https://distill-any-depth-official.github.io/
null
null
null
null
null
null
null
null
null
2,502.19285
On the Importance of Text Preprocessing for Multimodal Representation Learning and Pathology Report Generation
['Ruben T. Lucassen', 'Tijn van de Luijtgaarden', 'Sander P. J. Moonemans', 'Gerben E. Breimer', 'Willeke A. M. Blokx', 'Mitko Veta']
['cs.CV']
Vision-language models in pathology enable multimodal case retrieval and automated report generation. Many of the models developed so far, however, have been trained on pathology reports that include information which cannot be inferred from paired whole slide images (e.g., patient history), potentially leading to hall...
2025-02-26T16:45:09Z
11 pages, 1 figure
null
null
On the Importance of Text Preprocessing for Multimodal Representation Learning and Pathology Report Generation
['Ruben T. Lucassen', 'Tijn van de Luijtgaarden', 'S. P. Moonemans', 'G. Breimer', 'W. Blokx', 'M. Veta']
2,025
arXiv.org
0
25
['Computer Science']
2,502.19293
Pathology Report Generation and Multimodal Representation Learning for Cutaneous Melanocytic Lesions
['Ruben T. Lucassen', 'Sander P. J. Moonemans', 'Tijn van de Luijtgaarden', 'Gerben E. Breimer', 'Willeke A. M. Blokx', 'Mitko Veta']
['cs.CV']
Millions of melanocytic skin lesions are examined by pathologists each year, the majority of which concern common nevi (i.e., ordinary moles). While most of these lesions can be diagnosed in seconds, writing the corresponding pathology report is much more time-consuming. Automating part of the report writing could, the...
2025-02-26T16:52:10Z
11 pages, 2 figures. arXiv admin note: text overlap with arXiv:2502.19285
null
null
null
null
null
null
null
null
null
2,502.1932
Shh, don't say that! Domain Certification in LLMs
['Cornelius Emde', 'Alasdair Paren', 'Preetham Arvind', 'Maxime Kayser', 'Tom Rainforth', 'Thomas Lukasiewicz', 'Bernard Ghanem', 'Philip H. S. Torr', 'Adel Bibi']
['cs.CL', 'cs.AI', 'cs.CR', 'cs.LG', 'stat.ML']
Large language models (LLMs) are often deployed to perform constrained tasks, with narrow domains. For example, customer support bots can be built on top of LLMs, relying on their broad language understanding and capabilities to enhance performance. However, these LLMs are adversarially susceptible, potentially generat...
2025-02-26T17:13:19Z
10 pages, includes appendix Published in International Conference on Learning Representations (ICLR) 2025
International Conference on Learning Representations (ICLR) 2025
null
null
null
null
null
null
null
null
2,502.19546
Repurposing the scientific literature with vision-language models
['Anton Alyakin', 'Jaden Stryker', 'Daniel Alexander Alber', 'Karl L. Sangwon', 'Jin Vivian Lee', 'Brandon Duderstadt', 'Akshay Save', 'David Kurland', 'Spencer Frome', 'Shrutika Singh', 'Jeff Zhang', 'Eunice Yang', 'Ki Yun Park', 'Cordelia Orillac', 'Aly A. Valliani', 'Sean Neifert', 'Albert Liu', 'Aneek Patel', 'Chri...
['cs.AI', 'cs.CL', 'cs.HC']
Leading vision-language models (VLMs) are trained on general Internet content, overlooking scientific journals' rich, domain-specific knowledge. Training on specialty-specific literature could yield high-performance, task-specific tools, enabling generative AI to match generalist models in specialty publishing, educati...
2025-02-26T20:35:37Z
null
null
null
null
null
null
null
null
null
null
2,502.19587
NeoBERT: A Next-Generation BERT
['Lola Le Breton', 'Quentin Fournier', 'Mariam El Mezouar', 'John X. Morris', 'Sarath Chandar']
['cs.CL', 'cs.AI']
Recent innovations in architecture, pre-training, and fine-tuning have led to the remarkable in-context learning and reasoning abilities of large auto-regressive language models such as LLaMA and DeepSeek. In contrast, encoders like BERT and RoBERTa have not seen the same level of progress despite being foundational fo...
2025-02-26T22:00:22Z
19 pages, 5 figures, 9 tables. Submitted to TMLR
null
null
null
null
null
null
null
null
null
2,502.19634
MedVLM-R1: Incentivizing Medical Reasoning Capability of Vision-Language Models (VLMs) via Reinforcement Learning
['Jiazhen Pan', 'Che Liu', 'Junde Wu', 'Fenglin Liu', 'Jiayuan Zhu', 'Hongwei Bran Li', 'Chen Chen', 'Cheng Ouyang', 'Daniel Rueckert']
['cs.CV', 'cs.AI']
Reasoning is a critical frontier for advancing medical image analysis, where transparency and trustworthiness play a central role in both clinician trust and regulatory approval. Although Medical Visual Language Models (VLMs) show promise for radiological tasks, most existing VLMs merely produce final answers without r...
2025-02-26T23:57:34Z
null
null
null
null
null
null
null
null
null
null
2,502.19645
Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success
['Moo Jin Kim', 'Chelsea Finn', 'Percy Liang']
['cs.RO', 'cs.AI', 'cs.CV', 'cs.LG']
Recent vision-language-action models (VLAs) build upon pretrained vision-language models and leverage diverse robot datasets to demonstrate strong task execution, language following ability, and semantic generalization. Despite these successes, VLAs struggle with novel robot setups and require fine-tuning to achieve go...
2025-02-27T00:30:29Z
Accepted to Robotics: Science and Systems (RSS) 2025. Project website: https://openvla-oft.github.io/
null
null
Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success
['Moo Jin Kim', 'Chelsea Finn', 'Percy Liang']
2,025
arXiv.org
37
66
['Computer Science']
2,502.19712
Teaching Dense Retrieval Models to Specialize with Listwise Distillation and LLM Data Augmentation
['Manveer Singh Tamber', 'Suleman Kazi', 'Vivek Sourabh', 'Jimmy Lin']
['cs.IR']
While the current state-of-the-art dense retrieval models exhibit strong out-of-domain generalization, they might fail to capture nuanced domain-specific knowledge. In principle, fine-tuning these models for specialized retrieval tasks should yield higher effectiveness than relying on a one-size-fits-all model, but in ...
2025-02-27T03:07:49Z
null
null
null
null
null
null
null
null
null
null
2,502.19731
Preference Learning Unlocks LLMs' Psycho-Counseling Skills
['Mian Zhang', 'Shaun M. Eack', 'Zhiyu Zoey Chen']
['cs.CL']
Applying large language models (LLMs) to assist in psycho-counseling is an emerging and meaningful approach, driven by the significant gap between patient needs and the availability of mental health support. However, current LLMs struggle to consistently provide effective responses to client speeches, largely due to th...
2025-02-27T03:50:25Z
10 pages, 6 figures
null
null
null
null
null
null
null
null
null
2,502.19868
C-Drag: Chain-of-Thought Driven Motion Controller for Video Generation
['Yuhao Li', 'Mirana Claire Angel', 'Salman Khan', 'Yu Zhu', 'Jinqiu Sun', 'Yanning Zhang', 'Fahad Shahbaz Khan']
['cs.CV']
Trajectory-based motion control has emerged as an intuitive and efficient approach for controllable video generation. However, the existing trajectory-based approaches are usually limited to only generating the motion trajectory of the controlled object and ignoring the dynamic interactions between the controlled objec...
2025-02-27T08:21:03Z
null
null
null
null
null
null
null
null
null
null
2,502.19937
Image Referenced Sketch Colorization Based on Animation Creation Workflow
['Dingkun Yan', 'Xinrui Wang', 'Zhuoru Li', 'Suguru Saito', 'Yusuke Iwasawa', 'Yutaka Matsuo', 'Jiaxian Guo']
['cs.CV', 'cs.MM']
Sketch colorization plays an important role in animation and digital illustration production tasks. However, existing methods still meet problems in that text-guided methods fail to provide accurate color and style reference, hint-guided methods still involve manual operation, and image-referenced methods are prone to ...
2025-02-27T10:04:47Z
null
null
null
null
null
null
null
null
null
null
2,502.20056
Enhanced Contrastive Learning with Multi-view Longitudinal Data for Chest X-ray Report Generation
['Kang Liu', 'Zhuoqi Ma', 'Xiaolu Kang', 'Yunan Li', 'Kun Xie', 'Zhicheng Jiao', 'Qiguang Miao']
['cs.CV', 'cs.AI']
Automated radiology report generation offers an effective solution to alleviate radiologists' workload. However, most existing methods focus primarily on single or fixed-view images to model current disease conditions, which limits diagnostic accuracy and overlooks disease progression. Although some approaches utilize ...
2025-02-27T12:59:04Z
Accepted by CVPR 2025
null
null
Enhanced Contrastive Learning with Multi-view Longitudinal Data for Chest X-ray Report Generation
['Kang Liu', 'Zhuoqi Ma', 'Xiaolu Kang', 'Yunan Li', 'Kun Xie', 'Zhicheng Jiao', 'Qiguang Miao']
2,025
Computer Vision and Pattern Recognition
4
60
['Computer Science']
2,502.201
Generative augmentations for improved cardiac ultrasound segmentation using diffusion models
['Gilles Van De Vyver', 'Aksel Try Lenz', 'Erik Smistad', 'Sindre Hellum Olaisen', 'Bjørnar Grenne', 'Espen Holte', 'Håavard Dalen', 'Lasse Løvstakken']
['eess.IV', 'cs.CV']
One of the main challenges in current research on segmentation in cardiac ultrasound is the lack of large and varied labeled datasets and the differences in annotation conventions between datasets. This makes it difficult to design robust segmentation models that generalize well to external datasets. This work utilizes...
2025-02-27T13:57:14Z
null
null
null
null
null
null
null
null
null
null
2,502.20122
Self-Training Elicits Concise Reasoning in Large Language Models
['Tergel Munkhbat', 'Namgyu Ho', 'Seo Hyun Kim', 'Yongjin Yang', 'Yujin Kim', 'Se-Young Yun']
['cs.CL', 'cs.AI', 'cs.LG']
Chain-of-thought (CoT) reasoning has enabled large language models (LLMs) to utilize additional computation through intermediate tokens to solve complex tasks. However, we posit that typical reasoning traces contain many redundant tokens, incurring extraneous inference costs. Upon examination of the output distribution...
2025-02-27T14:14:50Z
26 pages, 10 figures, 23 tables. Accepted to Findings of ACL 2025
null
null
Self-Training Elicits Concise Reasoning in Large Language Models
['Tergel Munkhbat', 'Namgyu Ho', 'Seohyun Kim', 'Yongjin Yang', 'Yujin Kim', 'Se-young Yun']
2,025
arXiv.org
37
44
['Computer Science']
2,502.20172
Multimodal Representation Alignment for Image Generation: Text-Image Interleaved Control Is Easier Than You Think
['Liang Chen', 'Shuai Bai', 'Wenhao Chai', 'Weichu Xie', 'Haozhe Zhao', 'Leon Vinci', 'Junyang Lin', 'Baobao Chang']
['cs.CV', 'cs.CL']
The field of advanced text-to-image generation is witnessing the emergence of unified frameworks that integrate powerful text encoders, such as CLIP and T5, with Diffusion Transformer backbones. Although there have been efforts to control output images with additional conditions, like canny and depth map, a comprehensi...
2025-02-27T15:08:39Z
13 pages, 9 figures, codebase in https://github.com/chenllliang/DreamEngine
null
null
null
null
null
null
null
null
null
2,502.20272
HVI: A New Color Space for Low-light Image Enhancement
['Qingsen Yan', 'Yixu Feng', 'Cheng Zhang', 'Guansong Pang', 'Kangbiao Shi', 'Peng Wu', 'Wei Dong', 'Jinqiu Sun', 'Yanning Zhang']
['cs.CV', 'cs.AI', 'cs.LG']
Low-Light Image Enhancement (LLIE) is a crucial computer vision task that aims to restore detailed visual information from corrupted low-light images. Many existing LLIE methods are based on standard RGB (sRGB) space, which often produce color bias and brightness artifacts due to inherent high color sensitivity in sRGB...
2025-02-27T16:59:51Z
Qingsen Yan, Yixu Feng, and Cheng Zhang contributed equally to this work
null
null
null
null
null
null
null
null
null
2,502.20311
Adapting Automatic Speech Recognition for Accented Air Traffic Control Communications
['Marcus Yu Zhe Wee', 'Justin Juin Hng Wong', 'Lynus Lim', 'Joe Yu Wei Tan', 'Prannaya Gupta', 'Dillion Lim', 'En Hao Tew', 'Aloysius Keng Siew Han', 'Yong Zhi Lim']
['cs.LG', 'cs.SD', 'eess.AS']
Effective communication in Air Traffic Control (ATC) is critical to maintaining aviation safety, yet the challenges posed by accented English remain largely unaddressed in Automatic Speech Recognition (ASR) systems. Existing models struggle with transcription accuracy for Southeast Asian-accented (SEA-accented) speech,...
2025-02-27T17:35:59Z
null
null
null
null
null
null
null
null
null
null
2,502.20317
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']
['cs.LG', 'cs.AI', 'cs.IR']
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 st...
2025-02-27T17:42:52Z
null
null
null
null
null
null
null
null
null
null
2,502.20321
UniTok: A Unified Tokenizer for Visual Generation and Understanding
['Chuofan Ma', 'Yi Jiang', 'Junfeng Wu', 'Jihan Yang', 'Xin Yu', 'Zehuan Yuan', 'Bingyue Peng', 'Xiaojuan Qi']
['cs.CV', 'cs.AI']
Visual generative and understanding models typically rely on distinct tokenizers to process images, presenting a key challenge for unifying them within a single framework. Recent studies attempt to address this by connecting the training of VQVAE (for autoregressive generation) and CLIP (for understanding) to build a u...
2025-02-27T17:47:01Z
null
null
null
UniTok: A Unified Tokenizer for Visual Generation and Understanding
['Chuofan Ma', 'Yi Jiang', 'Junfeng Wu', 'Jihan Yang', 'Xin Yu', 'Zehuan Yuan', 'Bingyue Peng', 'Xiaojuan Qi']
2,025
arXiv.org
15
78
['Computer Science']
2,502.20323
ARTalk: Speech-Driven 3D Head Animation via Autoregressive Model
['Xuangeng Chu', 'Nabarun Goswami', 'Ziteng Cui', 'Hanqin Wang', 'Tatsuya Harada']
['cs.CV']
Speech-driven 3D facial animation aims to generate realistic lip movements and facial expressions for 3D head models from arbitrary audio clips. Although existing diffusion-based methods are capable of producing natural motions, their slow generation speed limits their application potential. In this paper, we introduce...
2025-02-27T17:49:01Z
More video demonstrations, code, models and data can be found on our project website: http://xg-chu.site/project_artalk/
null
null
null
null
null
null
null
null
null
2,502.20388
Beyond Next-Token: Next-X Prediction for Autoregressive Visual Generation
['Sucheng Ren', 'Qihang Yu', 'Ju He', 'Xiaohui Shen', 'Alan Yuille', 'Liang-Chieh Chen']
['cs.CV']
Autoregressive (AR) modeling, known for its next-token prediction paradigm, underpins state-of-the-art language and visual generative models. Traditionally, a ``token'' is treated as the smallest prediction unit, often a discrete symbol in language or a quantized patch in vision. However, the optimal token definition f...
2025-02-27T18:59:08Z
Project page at \url{https://oliverrensu.github.io/project/xAR}
null
null
Beyond Next-Token: Next-X Prediction for Autoregressive Visual Generation
['Sucheng Ren', 'Qihang Yu', 'Ju He', 'Xiaohui Shen', 'Alan L. Yuille', 'Liang-Chieh Chen']
2,025
arXiv.org
11
64
['Computer Science']
2,502.20578
Interpreting CLIP with Hierarchical Sparse Autoencoders
['Vladimir Zaigrajew', 'Hubert Baniecki', 'Przemyslaw Biecek']
['cs.CV', 'cs.AI', 'cs.LG']
Sparse autoencoders (SAEs) are useful for detecting and steering interpretable features in neural networks, with particular potential for understanding complex multimodal representations. Given their ability to uncover interpretable features, SAEs are particularly valuable for analyzing large-scale vision-language mode...
2025-02-27T22:39:13Z
null
Proceedings of the 42st International Conference on Machine Learning (ICML 2025)
null
Interpreting CLIP with Hierarchical Sparse Autoencoders
['Vladimir Zaigrajew', 'Hubert Baniecki', 'P. Biecek']
2,025
arXiv.org
1
74
['Computer Science']
2,502.20583
LiteASR: Efficient Automatic Speech Recognition with Low-Rank Approximation
['Keisuke Kamahori', 'Jungo Kasai', 'Noriyuki Kojima', 'Baris Kasikci']
['cs.LG', 'cs.AI', 'cs.SD', 'eess.AS']
Modern automatic speech recognition (ASR) models, such as OpenAI's Whisper, rely on deep encoder-decoder architectures, and their encoders are a critical bottleneck for efficient deployment due to high computational intensity. We introduce LiteASR, a low-rank compression scheme for ASR encoders that significantly reduc...
2025-02-27T22:52:21Z
null
null
null
null
null
null
null
null
null
null
2,502.20795
Plan2Align: Predictive Planning Based Test-Time Preference Alignment for Large Language Models
['Kuang-Da Wang', 'Teng-Ruei Chen', 'Yu Heng Hung', 'Guo-Xun Ko', 'Shuoyang Ding', 'Yueh-Hua Wu', 'Yu-Chiang Frank Wang', 'Chao-Han Huck Yang', 'Wen-Chih Peng', 'Ping-Chun Hsieh']
['cs.CL']
Aligning Large Language Models with Preference Fine-Tuning is often resource-intensive. Test-time alignment techniques that do not modify the underlying models, such as prompting and guided decodings, offer a lightweight alternative. However, existing test-time alignment methods primarily improve short responses and fa...
2025-02-28T07:24:33Z
Preprint. Code will be released at Plan2Align GitHub link: https://github.com/NYCU-RL-Bandits-Lab/Plan2Align
null
null
null
null
null
null
null
null
null
2,502.20855
MAMUT: A Novel Framework for Modifying Mathematical Formulas for the Generation of Specialized Datasets for Language Model Training
['Jonathan Drechsel', 'Anja Reusch', 'Steffen Herbold']
['cs.CL', 'cs.LG']
Mathematical formulas are a fundamental and widely used component in various scientific fields, serving as a universal language for expressing complex concepts and relationships. While state-of-the-art transformer models excel in processing and understanding natural language, they encounter challenges with mathematical...
2025-02-28T08:53:42Z
null
Transactions on Machine Learning Research (06/2025)
null
MAMUT: A Novel Framework for Modifying Mathematical Formulas for the Generation of Specialized Datasets for Language Model Training
['Jonathan Drechsel', 'Anja Reusch', 'Steffen Herbold']
2,025
arXiv.org
1
40
['Computer Science']
2,502.20864
Do Language Models Understand Honorific Systems in Javanese?
['Mohammad Rifqi Farhansyah', 'Iwan Darmawan', 'Adryan Kusumawardhana', 'Genta Indra Winata', 'Alham Fikri Aji', 'Derry Tanti Wijaya']
['cs.CL']
The Javanese language features a complex system of honorifics that vary according to the social status of the speaker, listener, and referent. Despite its cultural and linguistic significance, there has been limited progress in developing a comprehensive corpus to capture these variations for natural language processin...
2025-02-28T09:05:35Z
ACL 2025 - Main Conference
null
null
Do Language Models Understand Honorific Systems in Javanese?
['MohammadRifqi Farhansyah', 'Iwan Darmawan', 'Adryan Kusumawardhana', 'Genta Indra Winata', 'Alham Fikri Aji', 'D. Wijaya']
2,025
arXiv.org
1
38
['Computer Science']
2,502.20936
WebFAQ: A Multilingual Collection of Natural Q&A Datasets for Dense Retrieval
['Michael Dinzinger', 'Laura Caspari', 'Kanishka Ghosh Dastidar', 'Jelena Mitrović', 'Michael Granitzer']
['cs.CL', 'cs.AI', 'cs.IR']
We present WebFAQ, a large-scale collection of open-domain question answering datasets derived from FAQ-style schema.org annotations. In total, the data collection consists of 96 million natural question-answer (QA) pairs across 75 languages, including 47 million (49%) non-English samples. WebFAQ further serves as the ...
2025-02-28T10:46:52Z
10 pages, 3 figures, 7 tables
null
null
WebFAQ: A Multilingual Collection of Natural Q&A Datasets for Dense Retrieval
['Michael Dinzinger', 'Laura Caspari', 'Kanishka Ghosh Dastidar', "Jelena Mitrovi'c", 'Michael Granitzer']
2,025
arXiv.org
0
0
['Computer Science']
2,502.21074
CODI: Compressing Chain-of-Thought into Continuous Space via Self-Distillation
['Zhenyi Shen', 'Hanqi Yan', 'Linhai Zhang', 'Zhanghao Hu', 'Yali Du', 'Yulan He']
['cs.CL']
Chain-of-Thought (CoT) reasoning enhances Large Language Models (LLMs) by encouraging step-by-step reasoning in natural language. However, leveraging a latent continuous space for reasoning may offer benefits in terms of both efficiency and robustness. Prior implicit CoT methods attempt to bypass language completely by...
2025-02-28T14:07:48Z
16 pages
null
null
CODI: Compressing Chain-of-Thought into Continuous Space via Self-Distillation
['Zhenyi Shen', 'Hanqi Yan', 'Linhai Zhang', 'Zhanghao Hu', 'Yali Du', 'Yulan He']
2,025
arXiv.org
27
47
['Computer Science']
2,502.21208
ARIES: Autonomous Reasoning with LLMs on Interactive Thought Graph Environments
['Pedro Gimenes', 'Zeyu Cao', 'Jeffrey Wong', 'Yiren Zhao']
['cs.AI', 'cs.LG']
Recent research has shown that LLM performance on reasoning tasks can be enhanced by scaling test-time compute. One promising approach, particularly with decomposable problems, involves arranging intermediate solutions as a graph on which transformations are performed to explore the solution space. However, prior works...
2025-02-28T16:28:13Z
null
null
null
ARIES: Autonomous Reasoning with LLMs on Interactive Thought Graph Environments
['Pedro Gimenes', 'Zeyu Cao', 'Jeffrey T. H. Wong', 'Yiren Zhao']
2,025
arXiv.org
0
18
['Computer Science']
2,502.21228
ECLeKTic: a Novel Challenge Set for Evaluation of Cross-Lingual Knowledge Transfer
['Omer Goldman', 'Uri Shaham', 'Dan Malkin', 'Sivan Eiger', 'Avinatan Hassidim', 'Yossi Matias', 'Joshua Maynez', 'Adi Mayrav Gilady', 'Jason Riesa', 'Shruti Rijhwani', 'Laura Rimell', 'Idan Szpektor', 'Reut Tsarfaty', 'Matan Eyal']
['cs.CL', 'cs.AI']
To achieve equitable performance across languages, multilingual large language models (LLMs) must be able to abstract knowledge beyond the language in which it was acquired. However, the current literature lacks reliable ways to measure LLMs' capability of cross-lingual knowledge transfer. To that end, we present ECLeK...
2025-02-28T16:59:30Z
null
null
null
null
null
null
null
null
null
null
2,502.21257
RoboBrain: A Unified Brain Model for Robotic Manipulation from Abstract to Concrete
['Yuheng Ji', 'Huajie Tan', 'Jiayu Shi', 'Xiaoshuai Hao', 'Yuan Zhang', 'Hengyuan Zhang', 'Pengwei Wang', 'Mengdi Zhao', 'Yao Mu', 'Pengju An', 'Xinda Xue', 'Qinghang Su', 'Huaihai Lyu', 'Xiaolong Zheng', 'Jiaming Liu', 'Zhongyuan Wang', 'Shanghang Zhang']
['cs.RO', 'cs.CV']
Recent advancements in Multimodal Large Language Models (MLLMs) have shown remarkable capabilities across various multimodal contexts. However, their application in robotic scenarios, particularly for long-horizon manipulation tasks, reveals significant limitations. These limitations arise from the current MLLMs lackin...
2025-02-28T17:30:39Z
null
null
null
RoboBrain: A Unified Brain Model for Robotic Manipulation from Abstract to Concrete
['Yuheng Ji', 'Huajie Tan', 'Jiayu Shi', 'Xiaoshuai Hao', 'Yuan Zhang', 'Hengyuan Zhang', 'Pengwei Wang', 'Mengdi Zhao', 'Yao Mu', 'Pengju An', 'Xinda Xue', 'Qinghang Su', 'Huaihai Lyu', 'Xiaolong Zheng', 'Jiaming Liu', 'Zhongyuan Wang', 'Shanghang Zhang']
2,025
Computer Vision and Pattern Recognition
15
95
['Computer Science']
2,502.21291
MIGE: Mutually Enhanced Multimodal Instruction-Based Image Generation and Editing
['Xueyun Tian', 'Wei Li', 'Bingbing Xu', 'Yige Yuan', 'Yuanzhuo Wang', 'Huawei Shen']
['cs.CV']
Despite significant progress in diffusion-based image generation, subject-driven generation and instruction-based editing remain challenging. Existing methods typically treat them separately, struggling with limited high-quality data and poor generalization. However, both tasks require capturing complex visual variatio...
2025-02-28T18:21:08Z
This paper have been accepted by ACM MM25
null
null
null
null
null
null
null
null
null
2,502.21309
FANformer: Improving Large Language Models Through Effective Periodicity Modeling
['Yihong Dong', 'Ge Li', 'Xue Jiang', 'Yongding Tao', 'Kechi Zhang', 'Hao Zhu', 'Huanyu Liu', 'Jiazheng Ding', 'Jia Li', 'Jinliang Deng', 'Hong Mei']
['cs.CL', 'cs.AI', 'cs.LG']
Periodicity, as one of the most important basic characteristics, lays the foundation for facilitating structured knowledge acquisition and systematic cognitive processes within human learning paradigms. However, the potential flaws of periodicity modeling in Transformer affect the learning efficiency and establishment ...
2025-02-28T18:52:24Z
null
null
null
null
null
null
null
null
null
null
2,502.21318
How far can we go with ImageNet for Text-to-Image generation?
['L. Degeorge', 'A. Ghosh', 'N. Dufour', 'D. Picard', 'V. Kalogeiton']
['cs.CV']
Recent text-to-image generation models have achieved remarkable results by training on billion-scale datasets, following a `bigger is better' paradigm that prioritizes data quantity over availability (closed vs open source) and reproducibility (data decay vs established collections). We challenge this established parad...
2025-02-28T18:59:42Z
null
null
null
null
null
null
null
null
null
null
2,503.00031
Efficient Test-Time Scaling via Self-Calibration
['Chengsong Huang', 'Langlin Huang', 'Jixuan Leng', 'Jiacheng Liu', 'Jiaxin Huang']
['cs.LG', 'cs.AI', 'cs.CL']
Increasing test-time computation is a straightforward approach to enhancing the quality of responses in Large Language Models (LLMs). While Best-of-N sampling and Self-Consistency with majority voting are simple and effective, they require a fixed number of sampling responses for each query, regardless of its complexit...
2025-02-25T00:21:14Z
null
null
null
null
null
null
null
null
null
null
2,503.00084
InspireMusic: Integrating Super Resolution and Large Language Model for High-Fidelity Long-Form Music Generation
['Chong Zhang', 'Yukun Ma', 'Qian Chen', 'Wen Wang', 'Shengkui Zhao', 'Zexu Pan', 'Hao Wang', 'Chongjia Ni', 'Trung Hieu Nguyen', 'Kun Zhou', 'Yidi Jiang', 'Chaohong Tan', 'Zhifu Gao', 'Zhihao Du', 'Bin Ma']
['cs.SD', 'cs.AI', 'cs.CL', 'eess.AS']
We introduce InspireMusic, a framework integrated super resolution and large language model for high-fidelity long-form music generation. A unified framework generates high-fidelity music, songs, and audio, which incorporates an autoregressive transformer with a super-resolution flow-matching model. This framework enab...
2025-02-28T09:58:25Z
Work in progress. Correspondence regarding this technical report should be directed to {chong.zhang, yukun.ma}@alibaba-inc.com. Online demo available on https://modelscope.cn/studios/iic/InspireMusic and https://huggingface.co/spaces/FunAudioLLM/InspireMusic
null
null
InspireMusic: Integrating Super Resolution and Large Language Model for High-Fidelity Long-Form Music Generation
['Chong Zhang', 'Yukun Ma', 'Qian Chen', 'Wen Wang', 'Shengkui Zhao', 'Zexu Pan', 'Hao Wang', 'Chongjia Ni', 'Trung Hieu Nguyen', 'Kun Zhou', 'Yidi Jiang', 'Chaohong Tan', 'Zhifu Gao', 'Zhihao Du', 'Bin Ma']
2,025
arXiv.org
1
24
['Computer Science', 'Engineering']
2,503.00118
Novel $|V_{cb}|$ extraction method via boosted $bc$-tagging with in-situ calibration
['Yuzhe Zhao', 'Congqiao Li', 'Antonios Agapitos', 'Dawei Fu', 'Leyun Gao', 'Yajun Mao', 'Qiang Li']
['hep-ph']
We present a novel method for measuring $|V_{cb}|$ at the LHC using an advanced boosted-jet tagger to identify "$bc$ signatures". By associating boosted $W \rightarrow bc$ signals with $bc$-matched jets from top-quark decays, we enable an in-situ calibration of the tagger. This approach significantly suppresses backgro...
2025-02-28T19:00:25Z
7 pages (main text), 6 figures
null
null
null
null
null
null
null
null
null
2,503.00205
AnalogGenie: A Generative Engine for Automatic Discovery of Analog Circuit Topologies
['Jian Gao', 'Weidong Cao', 'Junyi Yang', 'Xuan Zhang']
['cs.LG', 'cs.AR']
The massive and large-scale design of foundational semiconductor integrated circuits (ICs) is crucial to sustaining the advancement of many emerging and future technologies, such as generative AI, 5G/6G, and quantum computing. Excitingly, recent studies have shown the great capabilities of foundational models in expedi...
2025-02-28T21:41:20Z
ICLR 2025 camera ready
null
null
null
null
null
null
null
null
null
2,503.00211
SafeAuto: Knowledge-Enhanced Safe Autonomous Driving with Multimodal Foundation Models
['Jiawei Zhang', 'Xuan Yang', 'Taiqi Wang', 'Yu Yao', 'Aleksandr Petiushko', 'Bo Li']
['cs.RO', 'cs.AI', 'cs.LG', 'cs.SY', 'eess.SY']
Traditional autonomous driving systems often struggle to connect high-level reasoning with low-level control, leading to suboptimal and sometimes unsafe behaviors. Recent advances in multimodal large language models (MLLMs), which process both visual and textual data, offer an opportunity to unify perception and reason...
2025-02-28T21:53:47Z
null
null
null
null
null
null
null
null
null
null
2,503.00223
DeepRetrieval: Hacking Real Search Engines and Retrievers with Large Language Models via Reinforcement Learning
['Pengcheng Jiang', 'Jiacheng Lin', 'Lang Cao', 'Runchu Tian', 'SeongKu Kang', 'Zifeng Wang', 'Jimeng Sun', 'Jiawei Han']
['cs.IR']
Information retrieval systems are crucial for enabling effective access to large document collections. Recent approaches have leveraged Large Language Models (LLMs) to enhance retrieval performance through query augmentation, but often rely on expensive supervised learning or distillation techniques that require signif...
2025-02-28T22:16:42Z
null
null
null
DeepRetrieval: Hacking Real Search Engines and Retrievers with Large Language Models via Reinforcement Learning
['Pengcheng Jiang', 'Jiacheng Lin', 'Lang Cao', 'Runchu Tian', 'SeongKu Kang', 'Zifeng Wang', 'Jimeng Sun', 'Jiawei Han']
2,025
arXiv.org
13
118
['Computer Science']
2,503.00287
Passivity-Centric Safe Reinforcement Learning for Contact-Rich Robotic Tasks
['Heng Zhang', 'Gokhan Solak', 'Sebastian Hjorth', 'Arash Ajoudani']
['cs.RO']
Reinforcement learning (RL) has achieved remarkable success in various robotic tasks; however, its deployment in real-world scenarios, particularly in contact-rich environments, often overlooks critical safety and stability aspects. Policies without passivity guarantees can result in system instability, posing risks to...
2025-03-01T01:34:02Z
revision version
null
null
null
null
null
null
null
null
null
2,503.00329
ABC: Achieving Better Control of Multimodal Embeddings using VLMs
['Benjamin Schneider', 'Florian Kerschbaum', 'Wenhu Chen']
['cs.CV', 'cs.LG']
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. Exi...
2025-03-01T03:29:02Z
null
null
null
null
null
null
null
null
null
null
2,503.00332
Investigating the contribution of terrain-following coordinates and conservation schemes in AI-driven precipitation forecasts
['Yingkai Sha', 'John S. Schreck', 'William Chapman', 'David John Gagne II']
['physics.ao-ph', 'cs.AI']
Artificial Intelligence (AI) weather prediction (AIWP) models often produce "blurry" precipitation forecasts that overestimate drizzle and underestimate extremes. This study provides a novel solution to tackle this problem -- integrating terrain-following coordinates with global mass and energy conservation schemes int...
2025-03-01T03:44:46Z
null
null
null
Investigating the contribution of terrain-following coordinates and conservation schemes in AI-driven precipitation forecasts
['Yingkai Sha', 'John S. Schreck', 'William Chapman', 'David John Gagne']
2,025
arXiv.org
1
28
['Physics', 'Computer Science']
2,503.00493
LLaSE-G1: Incentivizing Generalization Capability for LLaMA-based Speech Enhancement
['Boyi Kang', 'Xinfa Zhu', 'Zihan Zhang', 'Zhen Ye', 'Mingshuai Liu', 'Ziqian Wang', 'Yike Zhu', 'Guobin Ma', 'Jun Chen', 'Longshuai Xiao', 'Chao Weng', 'Wei Xue', 'Lei Xie']
['eess.AS', 'cs.AI', 'cs.CL', 'cs.SD']
Recent advancements in language models (LMs) have demonstrated strong capabilities in semantic understanding and contextual modeling, which have flourished in generative speech enhancement (SE). However, many LM-based SE approaches primarily focus on semantic information, often neglecting the critical role of acoustic ...
2025-03-01T13:44:50Z
ACL2025 main, Codes available at https://github.com/Kevin-naticl/LLaSE-G1
null
null
null
null
null
null
null
null
null
2,503.00533
BodyGen: Advancing Towards Efficient Embodiment Co-Design
['Haofei Lu', 'Zhe Wu', 'Junliang Xing', 'Jianshu Li', 'Ruoyu Li', 'Zhe Li', 'Yuanchun Shi']
['cs.RO', 'cs.LG', 'cs.SY', 'eess.SY']
Embodiment co-design aims to optimize a robot's morphology and control policy simultaneously. While prior work has demonstrated its potential for generating environment-adaptive robots, this field still faces persistent challenges in optimization efficiency due to the (i) combinatorial nature of morphological search sp...
2025-03-01T15:25:42Z
ICLR 2025 (Spotlight). Project Page: https://genesisorigin.github.io, Code: https://github.com/GenesisOrigin/BodyGen
null
null
BodyGen: Advancing Towards Efficient Embodiment Co-Design
['Haofei Lu', 'Zhe Wu', 'Junliang Xing', 'Jianshu Li', 'Ruoyu Li', 'Zhe Li', 'Yuanchun Shi']
2,025
International Conference on Learning Representations
2
46
['Computer Science']
2,503.00564
ToolDial: Multi-turn Dialogue Generation Method for Tool-Augmented Language Models
['Jeonghoon Shim', 'Gyuhyeon Seo', 'Cheongsu Lim', 'Yohan Jo']
['cs.CL']
Tool-Augmented Language Models (TALMs) leverage external APIs to answer user queries across various domains. However, existing benchmark datasets for TALM research often feature simplistic dialogues that do not reflect real-world scenarios, such as the need for models to ask clarifying questions or proactively call add...
2025-03-01T17:23:51Z
Accepted to ICLR 2025
null
null
null
null
null
null
null
null
null
2,503.00735
LADDER: Self-Improving LLMs Through Recursive Problem Decomposition
['Toby Simonds', 'Akira Yoshiyama']
['cs.LG', 'cs.AI']
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 ...
2025-03-02T05:16:43Z
null
null
null
LADDER: Self-Improving LLMs Through Recursive Problem Decomposition
['Toby Simonds', 'Akira Yoshiyama']
2,025
arXiv.org
6
9
['Computer Science']
2,503.00803
HiMo: High-Speed Objects Motion Compensation in Point Clouds
['Qingwen Zhang', 'Ajinkya Khoche', 'Yi Yang', 'Li Ling', 'Sina Sharif Mansouri', 'Olov Andersson', 'Patric Jensfelt']
['cs.CV', 'cs.RO']
LiDAR point cloud is essential for autonomous vehicles, but motion distortions from dynamic objects degrade the data quality. While previous work has considered distortions caused by ego motion, distortions caused by other moving objects remain largely overlooked, leading to errors in object shape and position. This di...
2025-03-02T08:55:12Z
12 pages
null
null
null
null
null
null
null
null
null
2,503.00808
Predictive Data Selection: The Data That Predicts Is the Data That Teaches
['Kashun Shum', 'Yuzhen Huang', 'Hongjian Zou', 'Qi Ding', 'Yixuan Liao', 'Xiaoxin Chen', 'Qian Liu', 'Junxian He']
['cs.CL']
Language model pretraining involves training on extensive corpora, where data quality plays a pivotal role. In this work, we aim to directly estimate the contribution of data during pretraining and select pretraining data in an efficient manner. Specifically, we draw inspiration from recent findings showing that compre...
2025-03-02T09:21:28Z
22 pages
null
null
Predictive Data Selection: The Data That Predicts Is the Data That Teaches
['Kashun Shum', 'Yuzhen Huang', 'Hongjian Zou', 'Qi Ding', 'Yixuan Liao', 'Xiaoxin Chen', 'Qian Liu', 'Junxian He']
2,025
arXiv.org
4
43
['Computer Science']
2,503.00865
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']
['cs.CL', 'cs.AI']
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....
2025-03-02T11:53:55Z
null
null
null
Babel: Open Multilingual Large Language Models Serving Over 90% of Global Speakers
['Yiran Zhao', 'Chaoqun Liu', 'Yue Deng', 'Jiahao Ying', 'Mahani Aljunied', 'Zhaodonghui Li', 'Li Bing', 'Hou Pong Chan', 'Yu Rong', 'Deli Zhao', 'Wenxuan Zhang']
2,025
arXiv.org
6
26
['Computer Science']
2,503.00938
From Poses to Identity: Training-Free Person Re-Identification via Feature Centralization
['Chao Yuan', 'Guiwei Zhang', 'Changxiao Ma', 'Tianyi Zhang', 'Guanglin Niu']
['cs.CV']
Person re-identification (ReID) aims to extract accurate identity representation features. However, during feature extraction, individual samples are inevitably affected by noise (background, occlusions, and model limitations). Considering that features from the same identity follow a normal distribution around identit...
2025-03-02T15:31:48Z
null
null
null
null
null
null
null
null
null
null
2,503.00955
SemViQA: A Semantic Question Answering System for Vietnamese Information Fact-Checking
['Dien X. Tran', 'Nam V. Nguyen', 'Thanh T. Tran', 'Anh T. Hoang', 'Tai V. Duong', 'Di T. Le', 'Phuc-Lu Le']
['cs.CL', 'cs.AI']
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 efficienc...
2025-03-02T16:22:46Z
18 pages
null
null
null
null
null
null
null
null
null
2,503.00958
Layered Insights: Generalizable Analysis of Authorial Style by Leveraging All Transformer Layers
['Milad Alshomary', 'Nikhil Reddy Varimalla', 'Vishal Anand', 'Smaranda Muresan', 'Kathleen McKeown']
['cs.CL']
We propose a new approach for the authorship attribution task that leverages the various linguistic representations learned at different layers of pre-trained transformer-based models. We evaluate our approach on three datasets, comparing it to a state-of-the-art baseline in in-domain and out-of-domain scenarios. We fo...
2025-03-02T16:47:31Z
null
null
null
null
null
null
null
null
null
null
2,503.00985
Enhancing Text Editing for Grammatical Error Correction: Arabic as a Case Study
['Bashar Alhafni', 'Nizar Habash']
['cs.CL']
Text editing frames grammatical error correction (GEC) as a sequence tagging problem, where edit tags are assigned to input tokens, and applying these edits results in the corrected text. This approach has gained attention for its efficiency and interpretability. However, while extensively explored for English, text ed...
2025-03-02T18:48:50Z
null
null
null
Enhancing Text Editing for Grammatical Error Correction: Arabic as a Case Study
['Bashar Alhafni', 'Nizar Habash']
2,025
arXiv.org
2
89
['Computer Science']
2,503.01103
Direct Discriminative Optimization: Your Likelihood-Based Visual Generative Model is Secretly a GAN Discriminator
['Kaiwen Zheng', 'Yongxin Chen', 'Huayu Chen', 'Guande He', 'Ming-Yu Liu', 'Jun Zhu', 'Qinsheng Zhang']
['cs.CV', 'cs.LG']
While likelihood-based generative models, particularly diffusion and autoregressive models, have achieved remarkable fidelity in visual generation, the maximum likelihood estimation (MLE) objective, which minimizes the forward KL divergence, inherently suffers from a mode-covering tendency that limits the generation qu...
2025-03-03T02:06:22Z
ICML 2025 Spotlight Project Page: https://research.nvidia.com/labs/dir/ddo/ Code: https://github.com/NVlabs/DDO
null
null
null
null
null
null
null
null
null
2,503.01151
ReaderLM-v2: Small Language Model for HTML to Markdown and JSON
['Feng Wang', 'Zesheng Shi', 'Bo Wang', 'Nan Wang', 'Han Xiao']
['cs.CL', 'cs.AI', 'cs.IR', '68T50', 'I.2.7; I.2.10']
We present ReaderLM-v2, a compact 1.5 billion parameter language model designed for efficient web content extraction. Our model processes documents up to 512K tokens, transforming messy HTML into clean Markdown or JSON formats with high accuracy -- making it an ideal tool for grounding large language models. The model'...
2025-03-03T03:57:04Z
9 pages, 10-12 refs
null
null
null
null
null
null
null
null
null
2,503.01183
DiffRhythm: Blazingly Fast and Embarrassingly Simple End-to-End Full-Length Song Generation with Latent Diffusion
['Ziqian Ning', 'Huakang Chen', 'Yuepeng Jiang', 'Chunbo Hao', 'Guobin Ma', 'Shuai Wang', 'Jixun Yao', 'Lei Xie']
['eess.AS']
Recent advancements in music generation have garnered significant attention, yet existing approaches face critical limitations. Some current generative models can only synthesize either the vocal track or the accompaniment track. While some models can generate combined vocal and accompaniment, they typically rely on me...
2025-03-03T05:15:34Z
null
null
null
null
null
null
null
null
null
null
2,503.01342
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']
['cs.CV']
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 t...
2025-03-03T09:27:24Z
null
null
null
null
null
null
null
null
null
null
2,503.0137
Kiss3DGen: Repurposing Image Diffusion Models for 3D Asset Generation
['Jiantao Lin', 'Xin Yang', 'Meixi Chen', 'Yingjie Xu', 'Dongyu Yan', 'Leyi Wu', 'Xinli Xu', 'Lie XU', 'Shunsi Zhang', 'Ying-Cong Chen']
['cs.GR', 'cs.CV', 'cs.MM']
Diffusion models have achieved great success in generating 2D images. However, the quality and generalizability of 3D content generation remain limited. State-of-the-art methods often require large-scale 3D assets for training, which are challenging to collect. In this work, we introduce Kiss3DGen (Keep It Simple and S...
2025-03-03T10:07:19Z
The first three authors contributed equally to this work
null
null
null
null
null
null
null
null
null
2,503.01437
Eau De $Q$-Network: Adaptive Distillation of Neural Networks in Deep Reinforcement Learning
['Théo Vincent', 'Tim Faust', 'Yogesh Tripathi', 'Jan Peters', "Carlo D'Eramo"]
['cs.LG', 'cs.AI']
Recent works have successfully demonstrated that sparse deep reinforcement learning agents can be competitive against their dense counterparts. This opens up opportunities for reinforcement learning applications in fields where inference time and memory requirements are cost-sensitive or limited by hardware. Until now,...
2025-03-03T11:39:03Z
Published at RLC 2025: https://openreview.net/forum?id=Bb84iBj4wU#discussion
null
null
Eau De Q-Network: Adaptive Distillation of Neural Networks in Deep Reinforcement Learning
['Th´eo Vincent', 'T. Faust', 'Yogesh Tripathi', 'Jan Peters', "Carlo D'Eramo"]
2,025
arXiv.org
0
52
['Computer Science']
2,503.01493
Llama-3.1-Sherkala-8B-Chat: An Open Large Language Model for Kazakh
['Fajri Koto', 'Rituraj Joshi', 'Nurdaulet Mukhituly', 'Yuxia Wang', 'Zhuohan Xie', 'Rahul Pal', 'Daniil Orel', 'Parvez Mullah', 'Diana Turmakhan', 'Maiya Goloburda', 'Mohammed Kamran', 'Samujjwal Ghosh', 'Bokang Jia', 'Jonibek Mansurov', 'Mukhammed Togmanov', 'Debopriyo Banerjee', 'Nurkhan Laiyk', 'Akhmed Sakip', 'Xud...
['cs.CL']
Llama-3.1-Sherkala-8B-Chat, or Sherkala-Chat (8B) for short, is a state-of-the-art instruction-tuned open generative large language model (LLM) designed for Kazakh. Sherkala-Chat (8B) aims to enhance the inclusivity of LLM advancements for Kazakh speakers. Adapted from the LLaMA-3.1-8B model, Sherkala-Chat (8B) is trai...
2025-03-03T13:05:48Z
Technical Report
null
null
null
null
null
null
null
null
null
2,503.01496
Liger: Linearizing Large Language Models to Gated Recurrent Structures
['Disen Lan', 'Weigao Sun', 'Jiaxi Hu', 'Jusen Du', 'Yu Cheng']
['cs.CL', 'cs.AI', 'cs.LG']
Transformers with linear recurrent modeling offer linear-time training and constant-memory inference. Despite their demonstrated efficiency and performance, pretraining such non-standard architectures from scratch remains costly and risky. The linearization of large language models (LLMs) transforms pretrained standard...
2025-03-03T13:08:00Z
Accepted by ICML 2025, 15 pages
null
null
null
null
null
null
null
null
null
2,503.01565
AutoLUT: LUT-Based Image Super-Resolution with Automatic Sampling and Adaptive Residual Learning
['Yuheng Xu', 'Shijie Yang', 'Xin Liu', 'Jie Liu', 'Jie Tang', 'Gangshan Wu']
['cs.CV', 'eess.IV']
In recent years, the increasing popularity of Hi-DPI screens has driven a rising demand for high-resolution images. However, the limited computational power of edge devices poses a challenge in deploying complex super-resolution neural networks, highlighting the need for efficient methods. While prior works have made s...
2025-03-03T14:09:36Z
Accepted by CVPR2025
null
null
null
null
null
null
null
null
null
2,503.0171
Spark-TTS: An Efficient LLM-Based Text-to-Speech Model with Single-Stream Decoupled Speech Tokens
['Xinsheng Wang', 'Mingqi Jiang', 'Ziyang Ma', 'Ziyu Zhang', 'Songxiang Liu', 'Linqin Li', 'Zheng Liang', 'Qixi Zheng', 'Rui Wang', 'Xiaoqin Feng', 'Weizhen Bian', 'Zhen Ye', 'Sitong Cheng', 'Ruibin Yuan', 'Zhixian Zhao', 'Xinfa Zhu', 'Jiahao Pan', 'Liumeng Xue', 'Pengcheng Zhu', 'Yunlin Chen', 'Zhifei Li', 'Xie Chen',...
['cs.SD', 'cs.AI', 'eess.AS']
Recent advancements in large language models (LLMs) have driven significant progress in zero-shot text-to-speech (TTS) synthesis. However, existing foundation models rely on multi-stage processing or complex architectures for predicting multiple codebooks, limiting efficiency and integration flexibility. To overcome th...
2025-03-03T16:23:10Z
Submitted to ACL 2025
null
null
null
null
null
null
null
null
null
2,503.01743
Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs
['Microsoft', ':', 'Abdelrahman Abouelenin', 'Atabak Ashfaq', 'Adam Atkinson', 'Hany Awadalla', 'Nguyen Bach', 'Jianmin Bao', 'Alon Benhaim', 'Martin Cai', 'Vishrav Chaudhary', 'Congcong Chen', 'Dong Chen', 'Dongdong Chen', 'Junkun Chen', 'Weizhu Chen', 'Yen-Chun Chen', 'Yi-ling Chen', 'Qi Dai', 'Xiyang Dai', 'Ruchao F...
['cs.CL', 'cs.AI', 'cs.LG']
We introduce Phi-4-Mini and Phi-4-Multimodal, compact yet highly capable language and multimodal models. Phi-4-Mini is a 3.8-billion-parameter language model trained on high-quality web and synthetic data, significantly outperforming recent open-source models of similar size and matching the performance of models twice...
2025-03-03T17:05:52Z
39 pages
null
null
Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs
['Abdelrahman Abouelenin', 'Atabak Ashfaq', 'Adam Atkinson', 'H. Awadalla', 'Nguyen Bach', 'Jianmin Bao', 'A. Benhaim', 'Martin Cai', 'Vishrav Chaudhary', 'Congcong Chen', 'Dongdong Chen', 'Dongdong Chen', 'Junkun Chen', 'Weizhu Chen', 'Yen-Chun Chen', 'Yi-ling Chen', 'Qi Dai', 'Xiyang Dai', 'Ruchao Fan', 'Mei Gao', 'M...
2,025
arXiv.org
71
105
['Computer Science']