arXiv Machine Learning July 20, 2026 By Zhengyang Zhuge, Hao Yu, Xin Wang, Zheng Li, Yizhong Cao, Dayiheng Liu, Jianwei Zhang

QUADS: Stabilizing NVFP4 Reinforcement Learning for MoE via QUantization-error Alignment across Dual Sides

Read the original on arXiv Machine Learning →

arXiv:2607. 15810v1 Announce Type: new Abstract: Rollout generation is a major bottleneck in Reinforcement Learning (RL) for Mixture-of-Experts (MoE) Large Language Models, motivating low-precision rollout acceleration such as FP8.

Summary generated by The Flow from the publisher's feed. The full article lives at arXiv Machine Learning.

arXiv Machine Learning

Jul 30

HiFloat4 Format for End-To-End Reinforcement Learning Post-Training of Large Language Models

arXiv:2607. 26515v1 Announce Type: new Abstract: We present, to our knowledge, the first end-to-end FP4 RL post-training, in which both the rollout and training policies, including their forward and backward passes, operate at 4-bit precision.

By Hei Yi Mak, Shadan Golestan, Hoang Le, Mehran Taghian Jazi, Yunke Peng, Yaoyuan Wang, Yao Wang, Junsong Wang, Tianchi Hu, Fengchen He, Guipeng Hu, Tanzila Rahman, Anandharaju Durai Raju

llms reinforcement-learning efficiency

arXiv AI

6d ago

Pretraining large language models with MXFP4 on Native FP4 Hardware

arXiv:2605. 09825v4 Announce Type: replace-cross Abstract: Why does full-pipeline FP4 training of large language models often diverge, even when forward activations and activation gradients remain stable?

By Musa Cim, Sarthak Arora, Poovaiah Palangappa, Miro Hodak, Ravi Dwivedula, Meena Arunachalam, Mahmut Taylan Kandemir

llms efficiency

arXiv AI

Jul 29

Stable FP4 Training via Transposition-Invariant Block Quantization

arXiv:2607. 24953v1 Announce Type: cross Abstract: Reducing training precision is a key lever for improving the e ciency of large language model (LLM) training, but pushing beyond FP8 to 4-bit oating point (FP4) remains challenging due to instability during optimization.

By Mehdi Rahimifar, Amin Darabi, Mehran Taghian Jazi, Xing Huang, Yao Wang, Zhijun Tu, Yufei Cui, Yunke Peng, Hongliang Li

llms efficiency

arXiv AI

Jun 3

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

arXiv:2605. 20402v3 Announce Type: replace-cross Abstract: MXFP4 arithmetic can dramatically accelerate reinforcement learning (RL) post-training of large language models (LLMs), yet the quantization error introduces severe accuracy degradation.

By Xiaocan Li, Shiliang Wu, Zheng Shen

llms reinforcement-learning efficiency safety

arXiv Machine Learning

Jun 2

Quartet II: Accurate LLM Pre-Training in NVFP4 by Improved Unbiased Gradient Estimation

arXiv:2601. 22813v2 Announce Type: replace Abstract: The NVFP4 lower-precision format, supported in hardware by NVIDIA Blackwell GPUs, promises to allow, for the first time, end-to-end fully-quantized pre-training of massive models such as LLMs.

By Andrei Panferov, Erik Schultheis, Soroush Tabesh, Dan Alistarh

llms efficiency

arXiv Machine Learning

Jun 16

ReQAT: Achieving Full-Precision Reasoning Accuracy with 4-bit Floating-Point Quantization-Aware Training

arXiv:2606. 15682v1 Announce Type: new Abstract: Large Reasoning Models (LRMs) achieve strong problem-solving through long chain-of-thought, but their deployment is constrained by the high cost of full-precision inference and growing KV cache footprints.

By Janghwan Lee, Sihwa Lee, Jinseok Kim, Yongjik Kim, Jieun Lim, Jinwook Oh, Jungwook Choi

fine-tuning efficiency data-engineering

Related stories

HiFloat4 Format for End-To-End Reinforcement Learning Post-Training of Large Language Models

Pretraining large language models with MXFP4 on Native FP4 Hardware

Stable FP4 Training via Transposition-Invariant Block Quantization

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

Quartet II: Accurate LLM Pre-Training in NVFP4 by Improved Unbiased Gradient Estimation

ReQAT: Achieving Full-Precision Reasoning Accuracy with 4-bit Floating-Point Quantization-Aware Training