Taco

Follow this topic Ask about this topic

Momentum

12 papers in the last four weeks, up 100% on the four weeks before. 0.1% of all new papers.

Jul 13Week of Sep 28

Latest papers 60

All topics

TACO: Ternary Absolute-max Column-wise One-sparse Optimizer for LLM Fine-Tuning
Oct 1, 2026Jichao Jiang, Cristian McGee, El Houcine Bergou +2Model Fine-Tuning Large Language Model Fine-Tuning
University of Central Florida · Mohammed VI Polytechnic University
Madeleine: Learning Involuntary Recall for Conversational Memory from Simulated Lives
Oct 1, 2026Zhiyun ShiConversational Memory Factual Recall
Nanyang Technological University, Singapore
CoVisco: Codec-Native Vision Encoder with Native Token Compression for Unified Image-Video Understanding
Sep 30, 2026Yulong Liu, Xiaotian Han, Junyuan Shang +6Vision Encoders Token Compression
ERNIE Team, Baidu Inc. · The Hong Kong University of Science and Technology · Institute of Automation, Chinese Academy of Sciences (CASIA)
CarveMix-RC: Addressing Rare-Class Imbalance Through Lesion-Aware Synthetic Augmentation for Brain Metastasis Segmentation
Sep 28, 2026Md Shibly Sadique, Md Fayaz Bin Hossen, Michael L. Evans +4Brain Tumor Segmentation Lesion Segmentation
Vision Lab, Department of Electrical and Computer Engineering, Old Dominion University, Norfolk, VA 23529, USA · Interdisciplinary Schools, Old Dominion University, Norfolk, VA 23529, USA
FONDANT: Strong and Best-Effort Planning via Antichains
Sep 28, 2026Benjamin Aminof, Tuan Khai Nguyen, Sasha RubinClassical Planning Planning
Technical University of Vienna, Austria · School of Computer Science, The University of Sydney, Australia
Grab a Coffee: Future-Aware Guidance for Discrete Diffusion with Compiled Objectives
Sep 28, 2026Hua, Xu, Dongxin Li +4Diffusion Models Geometric Guidance
Edward · Data Science and Analytics Thrust, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China · Department of Computer Science, University of California, Los Angeles, USA +1
CRISP: Cultural Reward Modeling for Implicit Situated Propriety
Sep 28, 2026Zekun Yuan, Yangfan Ye, Baohang Li +6Process-Level Supervision Taco
Harbin Institute of Technology · Huawei Technologies Co., Ltd · Peng Cheng Laboratory
OREO: Fidelity Alignment in 3D Generation via On-the-fly Rendering-Editing Optimization
Sep 24, 2026Zhiyuan Ma, Wenbo Hu, Wang Zhao +3Visual Fidelity Diffusion Priors
The Hong Kong Polytechnic University · Tencent ARC Lab
A Policy Profile for Croissant: Refusal as a Property of the Dataset
Sep 17, 2026Alexander ChernovGeneration Provenance Taco
Independent Researcher Toronto, Ontario, Canada
Fetch My Beer: Synthetic-to-real Hierarchical Policy for Smooth Pick-and-place
Sep 16, 2026Yingyue Li, Chenyangguang Zhang, Ruida Zhang +3Robotic Manipulation Policies Real-World Manipulation Tasks
Department of Automation, Tsinghua University, Beijing, China · ETH Zurich, Zurich, Switzerland · Beijing Xiaomi Robot Technology Co., Ltd, Beijing, China +1
Assessing nnU-Net Generalization across Brain Tumor Populations in BraTS-GoAT 2026
Sep 14, 2026Tristan Kirscher, Vivian Metzger, Philippe Meyer +1Brain Tumor Segmentation Brain Tumor Classification
ICube, Institut Strauss · ICube Laboratory, CNRS UMR 7357, University of Strasbourg, Strasbourg, France · Institut Strauss +2
CreaMem: A Scene-Aware Memory Architecture for Personalized Agents
Sep 8, 2026Qixuan Sun, Yue Que, Bowei He +4Personalized Large Language Model Agents Taco
City University of Hong Kong, Hong Kong SAR, China · AgentWoods Inc., San Francisco · Mohamed bin Zayed University of Artificial Intelligence, Abu Dhabi, UAE
CoFiE: Coarse-to-Fine Evidence Selection for Efficient Streaming Video Understanding
Sep 3, 2026Jing Jiang, Yiran Ling, Ruonan Li +2Streaming Video Understanding Vision Encoders
Harbin Institute of Technology · State Key Laboratory of Smart Farm Technologies and Systems · Pengcheng Laboratory
CAKE: Compiler-Agent Co-Design for Frontier Kernel Evolution
Aug 12, 2026Zihao Ye, Yingyi Huang, Hongyi Jin +11Graphics Processing Unit Kernels Nvidia
1NVIDIA · 2Carnegie Mellon University
Intensional Anaphora
Aug 12, 2026Ezra Keshet, Steven AbneySymbolic Predicates Contradiction
University of Michigan
Dual-Model Sentiment Analysis of Consumer Reviews in the Retail Coffee Sector Using Machine Learning and Deep Learning Approaches
Aug 12, 2026Muntasir Hasan Kanchan, Md. Alamgir Hossain, Md. Samiul Islam +1Sentiment Multimodal Sentiment Analysis
Department of Computer Science and Engineering, State University of Bangladesh, Dhaka, Bangladesh · Skill Morph Research Lab., Skill Morph, Dhaka, Bangladesh · Department of Computer Science, American International University - Bangladesh, Dhaka, Bangladesh
BREAD: Baseline-Referenced Explanations for Anomaly Diagnosis
Aug 11, 2026Jiaqi Qiu, Rob Goedhart, Jannis Kurtz +1Graph Anomaly Detection Xai
Department of Business Analytics, Amsterdam Business School, University of Amsterdam, Amsterdam, The Netherlands
FEAST: Federated Shared-Space Training for Resource-Heterogeneous Clients
Aug 10, 2026Bostan Khan, Masoud DaneshtalabBalanced Learning Taco
Mälardalen University, Västerås, Sweden
CRISP: Critical Step Perception for Training Efficient Deep Search Agents
Aug 3, 2026Haosi Mo, Zihao Yan, Ruiqing Zhang +4Search Agents Interaction Data
Baidu Inc., China · Institute of Computing and Intelligence, Harbin Institute of Technology, Shenzhen, China
Crossing the Margin Cliff: Toward Relearn-Robust LLM Unlearning via Margin Calibration
Jul 30, 2026Xiangyu Yin, Jiaxu Liu, Zhen Chen +1Large Language Model Unlearning Loss Function
Chalmers University of Technology, Gothenburg, Sweden · Imperial College London, London, United Kingdom · University of Liverpool, Liverpool, United Kingdom +1
MOJITO: Modal Joint Learning for Unified End-to-End Autonomous Driving
Jul 26, 2026Zhijing Cheng, Xuancheng Zhang, Donglin Di +4Causality-Aware End-To-End Autonomous Driving Autonomous Driving
University of Science and Technology of China, China · Li Auto, China · University of New South Wales, Australia
TAPAS: Throughput-adaptive Perception for Autonomous Systems
Jul 19, 2026Aman Vyas, Vasista Kodumagulla, Zain Taufique +2Perception Trajectory-Aware Hidden-State Analyses
Department of Computing, University of Turku, Finland
Subgrid-Scale Parameterization in Burgers' Equation Using Structure-Preserving Neural Networks and Entropy Variables
Jul 16, 2026Aijaz Nazir, Ilya TimofeyevNeural Network Reduced-Order Models
Dept. of Mathematics, University of Houston, Houston, TX 77204
Bake It Till You Make It: Ultrafast Spatial Texture-Atlas Splatting
Jul 15, 2026Neel Kelkar, Simon Niedermayr, Kaloian Petkov +2Visual Fidelity Ultrafast
Technical University of Munich, Siemens-Healthineers, Munich, Germany · Siemens-Healthineers, Erlangen, Germany
The Café in Amsterdam: When the Incumbent Becomes the Oracle
Jul 15, 2026Augusto CamargoHardware Efficiency Computational Cost
IME, University of São Paulo
ToFu: A White-Box, Token-Efficient Agent Harness for Researchers
Jul 13, 2026Junhao Ruan, Yuan Ge, Bei Li +7Coding Agents Agent Harness
Northeastern University, China · LongCat RSI, Meituan · NiuTrans Research
TACoS: Weakly Supervised Learning of Two-Dimensional Materials from Scribble Annotations to Precise Segmentation
Jul 8, 2026Jiabei Chen, Liping Zhang, Jiang-Bin Wu +6Pixel-Level Annotation Materials Science
AnnLab, Institute of Semiconductors, Chinese Academy of Sciences, Beijing, 100083, China · Center of Materials Science and Optoelectronics Engineering & School of Integrated Circuits, University of Chinese Academy of Sciences, Beijing, 100049, China · State Key Laboratory of Semiconductor Physics and Chip Technologies, Institute of Semiconductors, Chinese Academy of Sciences, Beijing, 100083, China
Demonstrating TOFFEE: A Learned System for Synthesizing Data Agent Trajectories at Scale
Jul 7, 2026Ziting Wang, Yin Li, Zuhao Yang +3Data Science Agents Single Trajectory
Nanyang Technological University Singapore · Huawei China · Industrial and Commercial Bank of China Limited, China
CRISP: A Spatiotemporal Camera-Radar Backbone for Driving via Forecasting-Based World-Model Pretraining
Jul 5, 2026Jingyu Song, Yi Liu, Katherine A. SkinnerAutonomous Driving Spatio-Temporal Transformers
Department of Robotics, University of Michigan, Ann Arbor, MI 48109 USA
TACO: TActile World Model as a Self-COrrector for Scalable Robot Policy Post-Training
Jul 3, 2026Shengbang Liu, Yueru Jia, Yuyang Yan +7Tactile World Model Diffusion-Based Vision-Language-Actions
State Key Laboratory of Multimedia Information Processing, School of Computer Science, Peking University · Sun Yat-sen University · AI2 Robotics +1
PASTA: A Paraphrasing And Self-Training Approach for Knowledge Updating in LLMs
Jun 27, 2026Takayuki Yamamoto, Daisuke KawaharaMultimodal Knowledge Editing Paraphrasing
Waseda University
From Hallucination to Grounding: Diagnosing Visual Spatial Intelligence via CRISP
Jun 25, 2026Zhixing Li, Yinan YuSpatial Reasoning Taco
Chalmers University of Technology, Gothenburg, Sweden
TACO: Towards Task-Consistent Open-Vocabulary Adaptation in Video Recognition
Jun 24, 2026Minghao Zhu, Xiao Lin, Mengxian Hu +5Vision-Language Model Adaptation Ai-Generated Video Detection
Tongji University, Shanghai, China
Sesame: Structure-Aware Molecular Generation via Spatial Density-Map Conditioning
Jun 22, 2026Konstantin Yatsenko, Arvind ThiagarajanMolecular Generation Drug Design
Tessel Biosciences, Inc., 750 Main Street, Cambridge, MA 02139
MeshFlow: Mesh Generation with Equivariant Flow Matching
Jun 22, 2026Qi Sun, Kiyohiro Nakayama, Jing Nathan Yan +6Mesh Generation Riemannian Flow Matching
City University of Hong Kong, Hong Kong · Stanford University, USA · Cornell Tech, USA +1
TACO: Task-Aware Column Description Generation Using LLMs
Jun 19, 2026Ting Cai, Rakesh R. Menon, Yiru Chen +8Tabular Data Information Extraction
University of Wisconsin-Madison · Adobe · Purdue University
CoffeeBench: Benchmarking Long-Horizon LLM Agents in Heterogeneous Multi-Agent Economies
Jun 15, 2026Issa Sugiura, Daichi Hattori, Kazuo Araragi +5Multi-Agent Simulations Long-Horizon Agents
Sakana AI · KPMG AZSA LLC
CRIS: Cross-Plane Self-Supervised Isotropic Restoration for Anisotropic Volumetric Imaging Across Modalities
Jun 14, 2026Adi Ahituv, Anat Ilivitzki, Moti FreimanAnisotropic Restoration
Faculty of Data and Decision Sciences, Technion – Israel Institute of Technology, Haifa, 3200003, Israel · Rambam Health Care Campus, Haifa, 3109601, Israel · Faculty of Biomedical Engineering, Technion – Israel Institute of Technology, Haifa, 3200003, Israel +1
LATTEArena: An Evaluation Framework for LLM-powered Tabular Feature Engineering (Extended Version)
Jun 8, 2026Ankai Hao, Ke Chen, Huan Li +1Feature Engineering Taco
Zhejiang University
Integrating Deep Learning Demand Forecasting with Multi-Objective Optimization for Circular Coffee Supply Chains: A Data-Driven Framework for Cost, Emissions, and Freshness Management
Jun 6, 2026Gerçek Budak, Faraz Gholamzadeh Gharehgheshlaghi, Melika Barjesteh Vaezi +1Supply Chain Optimization Sustainability
Department of Industrial Engineering, Ankara Yıldırım Beyazıt University, Keçiören, Ankara 06010, Türkiye · Department of Kinesiology and Sport Management, Texas Tech University, Lubbock, TX, United States · Department of Civil, Construction, and Environmental Engineering, University of Alabama, Tuscaloosa, AL, USA
Latent Anchor-Driven Test Generation for Deep Neural Networks
Jun 3, 2026Bin Duan, Matthew B. Dwyer, Guowei YangTest Generation Latent Space
School of Electrical Engineering and Computer Science, The University of Queensland, Australia · Department of Computer Science, University of Virginia, United States
CRePE: Convolution-aware Relative Importance in Post-training Pruning with Efficient Search
Jun 1, 2026Cheonjun ParkUnstructured Pruning Large Language Model Compression
Hankuk University of Foreign Studies
Croissant Tasks: A Metadata Format for Reproducible Machine Learning Evaluations
May 28, 2026Omar Benjelloun, Leonardo Martins Bianco, Isabelle Guyon +8Reproducibility Mle-Bench Lite
Google DeepMind, Paris, France · ChaLearn, CA, USA · Université Paris-Saclay, Gif-sur-Yvette, France +8
Dual-branch Distilled Transformer for Efficient Asymmetric UAV Tracking
May 27, 2026Hongtao Yang, Bineng Zhong, Qihua Liang +4Unmanned Aerial Vehicle Detection Multi-Object Tracking
Key Lab of Education Blockchain and Intelligent Technology, Ministry of Education, Guangxi Normal University, Guilin, 541004, China · Guangxi Key Lab of Multi-Source Information Mining and Security, Guangxi Normal University, Guilin, 541004, China · Nanjing University of Science and Technology +1
TacO: Benchmarking Tactile Sensors for Object Manipulation
May 21, 2026Anya Zorin, Zilin Si, Myungsun Park +11Tactile Robotic Manipulation
UC San Diego United States · CMU · UC San Diego +2
Croissant Baker: Metadata Generation for Discoverable, Governable, and Reusable ML Datasets
May 14, 2026Rafi Al Attrach, Rajna Fani, Sebastian Lobentanzer +17Metadata Taco
Technical University of Munich · Massachusetts Institute of Technology · Helmholtz Munich +15
BARISTA: A Multi-Task Egocentric Benchmark for Compositional Visual Understanding
May 12, 2026Patrick Knab, Orgest Xhelili, Inis Buzi +7Video Understanding Egocentric Dataset
Ramblr.ai Research · Technical University of Clausthal
Optimistic Dual Averaging Unifies Modern Optimizers
May 11, 2026Thomas Pethick, Wanyun Xie, Roman Machacek +1Adaptive Optimizers Federated Averaging
Independent Researcher · EPFL (LIONS) · University of Bern
SOPE: Stabilizing Off-Policy Evaluation for Online RL with Prior Data
May 7, 2026Carlo Romeo, Girolamo Macaluso, Alessandro Sestini +1Off-Policy Learning Offline Reinforcement Learning
1Media Integration and Communication Center – University of Florence · 2SEED – Electronic Arts
TACO: Trajectory Aligning Cross-view Optimisation
May 5, 2026Tavis Shore, Oscar Mendez, Simon HadfieldCross-View Geo-Localization Global Navigation Satellite System
Centre for Vision Speech and Signal Processing, University of Surrey, Guildford, United Kingdom · Locus Robotics, Kent, United Kingdom
Diversity in Large Language Models under Supervised Fine-Tuning
Apr 30, 2026Roman Klypa, Oleksandr CherednichenkoDiversity Taco
Univ. Grenoble Alpes, CNRS, Grenoble INP, LJK, 38000 Grenoble, France · Department of Mathematics and Mathematical Statistics, Integrated Science Lab, Umeå University, Sweden.
Reheat Nachos for Dinner? Evaluating AI Support for Cross-Cultural Communication of Neologisms
Apr 26, 2026Dayeon Ki, Yu Hou, Rachel Rudinger +3Artificial Intelligence Literacy Cross-Cultural Variation
University of Maryland
Beyond Acoustic Sparsity and Linguistic Bias: A Prompt-Free Paradigm for Mispronunciation Detection and Diagnosis
Apr 24, 2026Haopeng Geng, Longfei Yang, Xi Chen +3Pronunciation Assessment Linguistics
Graduate School of Engineering, The University of Tokyo, Japan
Mochi: Aligning Pre-training and Inference for Efficient Graph Foundation Models via Meta-Learning
Apr 23, 2026João Mattos, Arlei SilvaGraph Foundation Model Wireless Foundation Models
Computer Science Department · Ken Kennedy Institute
Ramen: Robust Test-Time Adaptation of Vision-Language Models with Active Sample Selection
Apr 23, 2026Wenxuan Bao, Yanjun Zhao, Xiyuan Yang +1Vision-Language Model Adaptation Stable Test-Time Adaptation
University of Illinois Urbana-Champaign
PASTA: A Patch-Agnostic Twofold-Stealthy Backdoor Attack on Vision Transformers
Apr 21, 2026Dazhuang Liu, Yanqi Qiao, Rui Wang +2Clean Label Backdoor Attack Vision Transformer
Delft University of Technology, Delft, the Netherlands. · University of Turku, Turku, Finland
COFFAIL: A Dataset of Successful and Anomalous Robot Skill Executions in the Context of Coffee Preparation
Apr 20, 2026Alex Mitrevski, Ayush SalunkeHuman-To-Robot Transfer Imitation Learning
Division of Systems and Control, Chalmers University of Technology, Gothenburg, Sweden · Autonomous Systems Group, Bonn-Rhein-Sieg University of Applied Sciences, Sankt Augustin, Germany
GeLaCo: An Evolutionary Approach to Layer Compression
Jul 14, 2025David Ponce, Thierry Etchegoyhen, Javier Del SerLarge Language Model Compression Compressed Model
Fundación Vicomtech, Basque Research and Technology Alliance (BRTA) · University of the Basque Country UPV/EHU · TECNALIA. Basque Research and Technology Alliance (BRTA)
TACO: Training-free Sound Prompted Segmentation via Semantically Constrained Audio-visual CO-factorization
Dec 2, 2024Hugo Malard, Michel Olvera, Stephane Lathuiliere +1Open-Vocabulary Segmentation Neural Audio
LTCI, Télécom Paris, Institut Polytechnique de Paris
Local Superior Soups: A Catalyst for Model Merging in Cross-Silo Federated Learning
Oct 31, 2024Minghui Chen, Meirui Jiang, Xin Zhang +3Federated Learning Parameter-Efficient Adaptation
University of British Columbia · Vector Institute · Chinese University of Hong Kong +1

Oct 1, 2026, cs.LGJ/K move · Enter open · S save

TACO: Ternary Absolute-max Column-wise One-sparse Optimizer for LLM Fine-Tuning

Jichao Jiang, Cristian McGee, El Houcine Bergou, Hanqin Cai+1

University of Central Florida · Mohammed VI Polytechnic University

Model Fine-Tuning Large Language Model Fine-Tuning Taco Muon

Open paper AskPDF arXiv

Full-parameter fine-tuning of large language models (LLMs) incurs substantial optimizer state memory overhead, limiting the model sizes that fit on modern GPUs. Existing approaches either compress optimizer state, abandon first-order gradients, or change the update geometry while retaining dense state. The recently introduced Muon optimizer reduces optimizer memory through matrix-valued updates. Still, its geometry differs from AdamW and can lead to performance degradation when fine-tuning AdamW-pretrained models. To reduce optimizer memory without sacrificing accuracy or computational efficiency in LLM fine-tuning, we propose Ternary Absolute-max Column-wise One-sparse optimizer, or TACO, which follows Muon's operator-norm steepest-descent view but takes the geometric route further. TACO computes the exact steepest-descent direction under a dimension-normalized

1\to1

operator norm by selecting the sign of the largest magnitude entry in each column of two-dimensional weight matrices. This retains first-order gradients while making optimizer state memory nearly negligible. Our practical TACO optimizer maintains only a small set of low precision gradient components per column, reducing persistent optimizer state by

174\times

relative to AdamW8bit (from 27.7 GB to 0.16 GB) and peak training memory by

2.9\times

(from 80.6 GB to 27.5 GB) on OPT-13B, while achieving comparable accuracy and runtime. TACO further enables full-parameter fine-tuning of 30-32B-parameter models on a single 80 GB H100 GPU across multiple model families and tasks.

Momentum

Related topics

Latest papers 60