Imagen de Portada

Decomposing MXFP4 quantization error for LLM reinforcement learning: reducible bias, recoverable deadzone, and an irreducible floor

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Li, Xiaocan, Wu, Shiliang, Shen, Zheng
Formato:	Preprint
Publicado:	2026
Materias:	Machine Learning Artificial Intelligence
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

A-3PO: Accelerating Asynchronous LLM Training with Staleness-aware Proximal Policy Approximation
por: Li, Xiaocan, et al.
Publicado: (2025)

TORQ: Two-Level Orthogonal Rotation for MXFP4 Quantization
por: Xu, Zukang, et al.
Publicado: (2026)

Pretraining large language models with MXFP4 on Native FP4 Hardware
por: Cim, Musa, et al.
Publicado: (2026)

Oscillation-Reduced MXFP4 Training for Vision Transformers
por: Chen, Yuxiang, et al.
Publicado: (2025)

Diagonal-Tiled Mixed-Precision Attention for Efficient Low-Bit MXFP Inference
por: Ding, Yifu, et al.
Publicado: (2026)

Unveiling the Potential of Quantization with MXFP4: Strategies for Quantization Error Reduction
por: Chhugani, Jatin, et al.
Publicado: (2026)

Recipes for Pre-training LLMs with MXFP8
por: Mishra, Asit, et al.
Publicado: (2025)

Normalization and effective learning rates in reinforcement learning
por: Lyle, Clare, et al.
Publicado: (2024)

MXNorm: Reusing MXFP block scales for efficient tensor normalisation
por: McLean, Callum, et al.
Publicado: (2026)

Causal prompting model-based offline reinforcement learning
por: Yu, Xuehui, et al.
Publicado: (2024)

Curriculum reinforcement learning for quantum architecture search under hardware errors
por: Patel, Yash J., et al.
Publicado: (2024)

SDQ: Sparse Decomposed Quantization for LLM Inference
por: Jeong, Geonhwa, et al.
Publicado: (2024)

Curriculum reinforcement learning with measurable task representation learning
por: Wen, Yongyan, et al.
Publicado: (2026)

Dynamic feature selection in medical predictive monitoring by reinforcement learning
por: Chen, Yutong, et al.
Publicado: (2024)

Ultra-short-term solar power forecasting by deep learning and data reconstruction
por: Wang, Jinbao, et al.
Publicado: (2025)

Trust Region Preference Approximation: A simple and stable reinforcement learning algorithm for LLM reasoning
por: Su, Xuerui, et al.
Publicado: (2025)

OCTOPUS: Optimized KV Cache for Transformers via Octahedral Parametrization Under optimal Squared error quantization
por: Boss, Mark, et al.
Publicado: (2026)

SpinQuant: LLM quantization with learned rotations
por: Liu, Zechun, et al.
Publicado: (2024)

An efficient deep reinforcement learning environment for flexible job-shop scheduling
por: Wu, Xinquan, et al.
Publicado: (2025)

A Causality-Aware Spatiotemporal Model for Multi-Region and Multi-Pollutant Air Quality Forecasting
por: Lu, Junxin, et al.
Publicado: (2025)

Revealing the Challenges of Sim-to-Real Transfer in Model-Based Reinforcement Learning via Latent Space Modeling
por: Lin, Zhilin, et al.
Publicado: (2025)

Breaking the Blocks: Continuous Low-Rank Decomposed Scaling for Unified LLM Quantization and Adaptation
por: Tang, Pingzhi, et al.
Publicado: (2026)

Delayed homomorphic reinforcement learning for environments with delayed feedback
por: Lee, Jongsoo, et al.
Publicado: (2026)

Counterfactual experience augmented off-policy reinforcement learning
por: Lee, Sunbowen, et al.
Publicado: (2025)

Deep reinforcement learning with time-scale invariant memory
por: Kabir, Md Rysul, et al.
Publicado: (2024)

Offline reinforcement learning for job-shop scheduling problems
por: Echeverria, Imanol, et al.
Publicado: (2024)

Bellman operator convergence enhancements in reinforcement learning algorithms
por: Kadurha, David Krame, et al.
Publicado: (2025)

Discovering highly efficient low-weight quantum error-correcting codes with reinforcement learning
por: He, Austin Yubo, et al.
Publicado: (2025)

Estimating unknown parameters in differential equations with a reinforcement learning based PSO method
por: Sun, Wenkui, et al.
Publicado: (2024)

SpreadsheetArena: Decomposing Preference in LLM Generation of Spreadsheet Workbooks
por: Kundurthy, Srivatsa, et al.
Publicado: (2026)

LLM Assertiveness can be Mechanistically Decomposed into Emotional and Logical Components
por: Tsujimura, Hikaru, et al.
Publicado: (2025)

Not all tokens are needed(NAT): token efficient reinforcement learning
por: Sang, Hejian, et al.
Publicado: (2026)

Leveraging weights signals -- Predicting and improving generalizability in reinforcement learning
por: Moulin, Olivier, et al.
Publicado: (2025)

Economic span selection of bridge based on deep reinforcement learning
por: Zhang, Leye, et al.
Publicado: (2024)

Agentic reinforcement learning empowers next-generation chemical language models for molecular design and synthesis
por: Li, Hao, et al.
Publicado: (2026)

TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting
por: Wang, Shiyu, et al.
Publicado: (2024)

Task diversity produces systematic transfer but inhibits continual reinforcement learning
por: Seth, Purab, et al.
Publicado: (2026)

Found-RL: foundation model-enhanced reinforcement learning for autonomous driving
por: Qu, Yansong, et al.
Publicado: (2026)

On the consistency of hyper-parameter selection in value-based deep reinforcement learning
por: Obando-Ceron, Johan, et al.
Publicado: (2024)

CAREL: Instruction-guided reinforcement learning with cross-modal auxiliary objectives
por: Saghafian, Armin, et al.
Publicado: (2024)