Copertina

PV-Tuning: Beyond Straight-Through Estimation for Extreme LLM Compression

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Malinovskii, Vladimir, Mazur, Denis, Ilin, Ivan, Kuznedelev, Denis, Burlachenko, Konstantin, Yi, Kai, Alistarh, Dan, Richtarik, Peter
Natura:	Preprint
Pubblicazione:	2024
Soggetti:	Machine Learning
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Cache Me If You Must: Adaptive Key-Value Quantization for Large Language Models
di: Shutova, Alina, et al.
Pubblicazione: (2025)

Pushing the Limits of Large Language Model Quantization via the Linearity Theorem
di: Malinovskii, Vladimir, et al.
Pubblicazione: (2024)

EvoPress: Accurate Dynamic Model Compression via Evolutionary Search
di: Sieberling, Oliver, et al.
Pubblicazione: (2024)

Extreme Compression of Large Language Models via Additive Quantization
di: Egiazarian, Vage, et al.
Pubblicazione: (2024)

Unlocking FedNL: Self-Contained Compute-Optimized Implementation
di: Burlachenko, Konstantin, et al.
Pubblicazione: (2024)

BurTorch: Revisiting Training from First Principles by Coupling Autodiff, Math Optimization, and Systems
di: Burlachenko, Konstantin, et al.
Pubblicazione: (2025)

Error Feedback Reloaded: From Quadratic to Arithmetic Mean of Smoothness Constants
di: Richtárik, Peter, et al.
Pubblicazione: (2024)

Thanos: A Block-wise Pruning Algorithm for Efficient Large Language Model Compression
di: Ilin, Ivan, et al.
Pubblicazione: (2025)

The Iterative Optimal Brain Surgeon: Faster Sparse Recovery by Leveraging Second-Order Information
di: Wu, Diyuan, et al.
Pubblicazione: (2024)

Optimization Methods and Software for Federated Learning
di: Burlachenko, Konstantin
Pubblicazione: (2025)

Hogwild! Inference: Parallel LLM Generation via Concurrent Attention
di: Rodionov, Gleb, et al.
Pubblicazione: (2025)

Characterizing Graph Datasets for Node Classification: Homophily-Heterophily Dichotomy and Beyond
di: Platonov, Oleg, et al.
Pubblicazione: (2022)

Model Compression with Exact Budget Constraints via Riemannian Manifolds
di: Helcig, Michael, et al.
Pubblicazione: (2026)

Shadowheart SGD: Distributed Asynchronous SGD with Optimal Time Complexity Under Arbitrary Computation and Communication Heterogeneity
di: Tyurin, Alexander, et al.
Pubblicazione: (2024)

Accurate Compression of Text-to-Image Diffusion Models via Vector Quantization
di: Egiazarian, Vage, et al.
Pubblicazione: (2024)

Ringmaster LMO: Asynchronous Linear Minimization Oracle Momentum Method
di: Sadiev, Abdurakhmon, et al.
Pubblicazione: (2026)

Symmetric Pruning of Large Language Models
di: Yi, Kai, et al.
Pubblicazione: (2025)

KV Cache Offloading for Context-Intensive Tasks
di: Bocharnikov, Andrey, et al.
Pubblicazione: (2026)

Cohort Squeeze: Beyond a Single Communication Round per Cohort in Cross-Device Federated Learning
di: Yi, Kai, et al.
Pubblicazione: (2024)

MatryoshkaLoRA: Learning Accurate Hierarchical Low-Rank Representations for LLM Fine-Tuning
di: Modoranu, Ionut-Vlad, et al.
Pubblicazione: (2026)

Custom Gradient Estimators are Straight-Through Estimators in Disguise
di: Schoenbauer, Matt, et al.
Pubblicazione: (2024)

Quartet II: Accurate LLM Pre-Training in NVFP4 by Improved Unbiased Gradient Estimation
di: Panferov, Andrei, et al.
Pubblicazione: (2026)

Bridging the Gap Between Promise and Performance for Microscaling FP4 Quantization
di: Egiazarian, Vage, et al.
Pubblicazione: (2025)

A critical look at the evaluation of GNNs under heterophily: Are we really making progress?
di: Platonov, Oleg, et al.
Pubblicazione: (2023)

CompressedScaffnew: The First Theoretical Double Acceleration of Communication from Local Training and Compression in Distributed Optimization
di: Condat, Laurent, et al.
Pubblicazione: (2022)

Improving the Straight-Through Estimator with Zeroth-Order Information
di: Yang, Ningfeng, et al.
Pubblicazione: (2025)

TAMUNA: Doubly Accelerated Distributed Optimization with Local Training, Compression, and Partial Participation
di: Condat, Laurent, et al.
Pubblicazione: (2023)

MicroAdam: Accurate Adaptive Optimization with Low Space Overhead and Provable Convergence
di: Modoranu, Ionut-Vlad, et al.
Pubblicazione: (2024)

Behemoth: Benchmarking Unlearning in LLMs Using Fully Synthetic Data
di: Iofinova, Eugenia, et al.
Pubblicazione: (2026)

Improving Discrete Optimisation Via Decoupled Straight-Through Estimator
di: Shah, Rushi, et al.
Pubblicazione: (2024)

Apertus LLM Family Expansion via Distillation and Quantization
di: Panferov, Andrei, et al.
Pubblicazione: (2026)

Compression Scaling Laws:Unifying Sparsity and Quantization
di: Frantar, Elias, et al.
Pubblicazione: (2025)

Beyond Discreteness: Sample Complexity Analysis of Straight-Through Estimator for 1-bit Quantization
di: Jeong, Halyun, et al.
Pubblicazione: (2025)

RoSA: Accurate Parameter-Efficient Fine-Tuning via Robust Adaptation
di: Nikdan, Mahdi, et al.
Pubblicazione: (2024)

BiCoLoR: Communication-Efficient Optimization with Bidirectional Compression and Local Training
di: Condat, Laurent, et al.
Pubblicazione: (2026)

Beyond Outliers: A Study of Optimizers Under Quantization
di: Vlassis, Georgios, et al.
Pubblicazione: (2025)

Label Privacy in Split Learning for Large Models with Parameter-Efficient Training
di: Zmushko, Philip, et al.
Pubblicazione: (2024)

LLMQ: Efficient Lower-Precision Pretraining for Consumer GPUs
di: Schultheis, Erik, et al.
Pubblicazione: (2025)

Sparse-ProxSkip: Accelerated Sparse-to-Sparse Training in Federated Learning
di: Meinhardt, Georg, et al.
Pubblicazione: (2024)

FedP3: Federated Personalized and Privacy-friendly Network Pruning under Model Heterogeneity
di: Yi, Kai, et al.
Pubblicazione: (2024)