TV Regularization

TV: Total Variation

Follow this topic Ask about this topic

Latest papers

11 papers1 paper in the last four weeks, against 1 the four weeks before. 0.0% of all new papers.

TVP: TV is Almost All You Need for Speculative Decoding
Oct 31, 2026Aaryam SharmaSpeculative Decoding LLM Inference Acceleration
TV-Regulated OPD: Direction Matters in On-Policy Distillation
Sep 8, 2026Han Xiao, Yifan Niu, Dongyi Liu +2Preference Optimization On-Policy Distillation
The Hong Kong University of Science and Technology (Guangzhou) · The University of Edinburgh · The Hong Kong University of Science and Technology
A Tensor Variational Formulation of Gradient Energy Total Variation
Aug 29, 2026Freddie Åström, George Baravdish, Michael FelsbergTV Regularization Image Denoising
Computer Vision Laboratory, Linköping University, Sweden · Center for Medical Image Science and Visualization (CMIV), Linköping University · Department of Science and Technology, Linköping University, Sweden
Learning spatially varying regularisation parameters of low regularity for image reconstruction
Aug 25, 2026Kostas Papafitsoros, Luca Calatroni, Andreas KoflerTV Regularization Image Reconstruction
School of Mathematical Sciences, Queen Mary University of London, London E1 4BS, United Kingdom · MaLGa centre, DIBRIS, Universit`a di Genova, Istituto Italiano di Tecnologia, Genoa, Italy · Physikalisch-Technische Bundesanstalt (PTB), Braunschweig and Berlin, Germany
Dual Modality Prompted Diffusion Priors for Zero Shot Hyperspectral Pansharpening
Aug 12, 2026Pengwei Xie, Fei Zhu, Jiajun Li +4Diffusion Models for Image Restoration Hyperspectral Image Super-Resolution
School of Artificial Intelligence, Beijing Normal University, Beijing 100875, China · School of GeoAI and Hinton STAI Institute, and the Key Laboratory of Geographic Information Science (Ministry of Education), East China Normal University, Shanghai 200241, China · Peking University, Beijing 100871, China +3
Framelet-Based Blind Image Restoration with Minimax Concave Regularization
Apr 21, 2026Heng Zhang, Reza Parvaz, Rui YangImage Deblurring Image Restoration
College of Artificial Intelligence and Big Data, Chongqing Polytechnic University of Electronic Technology, Chongqing,China · Department of Engineering Sciences, Faculty of Advanced Technologies, University of Mohaghegh Ardabili, Ardabil, Iran
Duality for the Adversarial Total Variation
Apr 20, 2026Leon Bungert, Lucas SchmittTV Regularization
Institute of Mathematics, Center for Artificial Intelligence and Data Science (CAIDAS), University of Würzburg, Emil-Fischer-Str. 40, 97074 Würzburg, Germany
Discrete Flow Matching Policy Optimization
Apr 7, 2026Maojiang Su, Po-Chung Hsieh, Weimin Wu +4TV Regularization RL for Flow Matching Models
Center for Foundation Models and Generative AI, Northwestern University, Evanston, IL 60208, USA · Department of Electrical Engineering, National Taiwan University, Taipei 10617, Taiwan · Department of Physics, National Taiwan University, Taipei 10617, Taiwan +1
A Dual Certificate Approach to Sparsity in Infinite-Width Shallow Neural Networks
Mar 18, 2026Leonardo Del Grande, Christoph Brune, Marcello CarioniTV Regularization Shallow Neural Networks
Department of Applied Mathematics, University of Twente, 7500AE Enschede, The Netherlands
Bayesian Uncertainty-Aware MRI Reconstruction
Mar 13, 2026Ahmed Karam Eldaly, Matteo Figini, Daniel C. AlexanderTV Regularization Accelerated MRI Reconstruction
Department of Computer Science, University of Exeter, UK · UCL Hawkes Institute, Department of Computer Science, University College London, UK
An Improved Variational Method for Image Denoising
Oct 3, 2024Jing-En Huang, Jia-Wei Liao, Ku-Te Lin +2TV Regularization Image Denoising
National Yang Ming Chiao Tung University · National Taiwan University · National Central University +1

Oct 31, 2026J/K move · Enter open · S save

TVP: TV is Almost All You Need for Speculative Decoding

Aaryam Sharma

Speculative Decoding LLM Inference Acceleration TV Regularization

Open paper AskPDF arXiv

Speculative decoding accelerates LLM inference with a drafter whose proposed tokens are verified by the target model in parallel. Under lossless speculative sampling, a draft token is accepted with probability one minus the total variation (TV) distance between the target and draft distributions, yet drafters are often trained with cross-entropy (CE) or KL. We first show that under matched and well-tuned training, TV matches or surpasses several published draft-training objectives when fine-tuning, and a fixed TV and KL mixture does the same when training from scratch. We identify two remaining problems, TV's vanishing gradients on undercovered tokens and the prefix products that starve later positions in acceptance-length objectives. To address both, we derive TVP, a loss family based on expected acceptance length, whose power parameter interpolates between coverage and acceptance oriented objectives. Annealing the power from zero to one gives the highest or tied-highest acceptance length across three drafter architectures and three target models, including longer drafts and block verification. TVP improves acceptance length over CE by 9.8% when fine-tuning, and over the public DFlash2 training recipes by 6.2 to 12.9%.