Cover Image

HiFloat4 Format for Language Model Pre-training on Ascend NPUs

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Taghian, Mehran, Peng, Yunke, Huang, Xing, Wang, Yao, Wang, Yaoyuan, Guo, Wei, Luo, Yuanyong, Hu, Tianchi, Wang, Junsong, Wang, Xin, Liu, Hu, Cheng, Yu, Yu, Ziwei, Li, Hongliang, Rahimifar, Mehdi, Yan, Lei, Wang, Xuefei, Ma, Zhuang, Liu, Lei, Yu, Hui, Raju, Anandharaju Durai, Le, Hoang, Mak, Hei Yi, Rahman, Tanzila, Golestan, Shadan
Format:	Preprint
Published:	2026
Subjects:	Machine Learning Artificial Intelligence Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Ascend HiFloat8 Format for Deep Learning
by: Luo, Yuanyong, et al.
Published: (2024)

HiFloat4 Format for Language Model Inference
by: Luo, Yuanyong, et al.
Published: (2026)

Unleashing Low-Bit Inference on Ascend NPUs: A Comprehensive Evaluation of HiFloat Formats
by: Zhao, Pengxiang, et al.
Published: (2026)

TabEmb: Joint Semantic-Structure Embedding for Table Annotation
by: Hoseinzade, Ehsan, et al.
Published: (2026)

Enhancing Hardware Fault Tolerance in Machines with Reinforcement Learning Policy Gradient Algorithms
by: Schoepp, Sheila, et al.
Published: (2024)

Pangu Ultra: Pushing the Limits of Dense Large Language Models on Ascend NPUs
by: Yin, Yichun, et al.
Published: (2025)

OSUM-Pangu: An Open-Source Multidimension Speech Understanding Foundation Model Built upon OpenPangu on Ascend NPUs
by: Liao, Yujie, et al.
Published: (2026)

EAGLE-Pangu: Accelerator-Safe Tree Speculative Decoding on Ascend NPUs
by: Han, Chang, et al.
Published: (2026)

MindVL: Towards Efficient and Effective Training of Multimodal Large Language Models on Ascend NPUs
by: Chen, Feilong, et al.
Published: (2025)

Pangu Ultra MoE: How to Train Your Big MoE on Ascend NPUs
by: Tang, Yehui, et al.
Published: (2025)

Tail-Aware HiFloat4: W4A4 Post-Training Quantization for Wan2.2
by: Feng, Zhanfeng, et al.
Published: (2026)

Boundary-Protection W8A8 HiFloat8 Quantization for Large-Scale Text-to-Video Diffusion Transformers
by: Zhao, Yiming
Published: (2026)

STAMBPL1, transcriptionally regulated by SREBP1, promotes malignant behaviors of hepatocellular carcinoma cells via Wnt/β‐catenin signaling pathway
by: Junyi Jin, et al.
Published: (2024)

HiCI: Hierarchical Construction-Integration for Long-Context Attention
by: Zeng, Xiangyu, et al.
Published: (2026)

ENEC: A Lossless AI Model Compression Method Enabling Fast Inference on Ascend NPUs
by: Yang, Jinwu, et al.
Published: (2026)

WINFlowNets: Warm-up Integrated Networks Training of Generative Flow Networks for Robotics and Machine Fault Adaptation
by: Sufiyan, Zahin, et al.
Published: (2026)

A Study of the Efficacy of Generative Flow Networks for Robotics and Machine Fault-Adaptation
by: Sufiyan, Zahin, et al.
Published: (2025)

TLXML: Task-Level Explanation of Meta-Learning via Influence Functions
by: Mitsuka, Yoshihiro, et al.
Published: (2025)

Enhancing Learned Knowledge in LoRA Adapters Through Efficient Contrastive Decoding on Ascend NPUs
by: Heisler, Morgan Lindsay, et al.
Published: (2025)

EPD-Serve: A Flexible Multimodal EPD Disaggregation Inference Serving System On Ascend
by: Bai, Fan, et al.
Published: (2026)

M2XFP: A Metadata-Augmented Microscaling Data Format for Efficient Low-bit Quantization
by: Hu, Weiming, et al.
Published: (2026)

FastAttention: Extend FlashAttention2 to NPUs and Low-resource GPUs
by: Lin, Haoran, et al.
Published: (2024)

High‐Efficiency Removal of Cationic and Anionic Dyes From Wastewater Using β–Cyclodextrin‐Phytic Acid–Acrylamide Hydrogel
by: Lei Wang, et al.
Published: (2025)

SGEMM-cube: Precision-Recovery FP32 GEMM Approximation on Ascend NPUs with FP16 Matrix Engines
by: Xue, Weicheng, et al.
Published: (2025)

Effects of bromelain application and Lactiplantibacillus plantarum inoculation on in vitro starch digestibility and starch-protein matrix characteristics of high moisture corn silage
by: Lei, Wang, et al.
Published: (2026)

Multislice Hollow Ptychography for Simultaneous Atomic-Layer-Resolved 3D Structural Imaging and Spectroscopy
by: Lei, Yu, et al.
Published: (2025)

A Bioinspired Cascade SAzyme: Ti 3+ ‐Mediated Stabilization of Cu Single‐Atoms on TiO 2‐x for Efficient ROS Scavenging
by: Lei Wang, et al.
Published: (2025)

Spin Polarization Control via Magnetic Field in Dissipative Bosonic Systems
by: Fan, Yaoyuan, et al.
Published: (2025)

Relay Buffer Independent Communication over Pooled HBM for Efficient MoE Inference on Ascend
by: Hu, Tianlun, et al.
Published: (2026)

Viko 2.0: A Hierarchical Gecko-inspired Adhesive Gripper with Visuotactile Sensor
by: Pang, Chohei, et al.
Published: (2022)

Differencing and Coadding JWST Images with Matched Point Spread Function
by: Hu, Lei, et al.
Published: (2023)

AscendOptimizer: Episodic Agent for Ascend NPU Operator Optimization
by: Wu, Jiehao, et al.
Published: (2026)

GNMR: Runtime Stability Control for Low-Precision Large Language Model Training
by: Kong, Boao, et al.
Published: (2026)

HiMat: DiT-based Ultra-High Resolution SVBRDF Generation
by: Wang, Zixiong, et al.
Published: (2025)

HiCAST: Highly Customized Arbitrary Style Transfer with Adapter Enhanced Diffusion Models
by: Wang, Hanzhang, et al.
Published: (2024)

CaMDN: Enhancing Cache Efficiency for Multi-tenant DNNs on Integrated NPUs
by: Cai, Tianhao, et al.
Published: (2025)

MRPL35 Induces Proliferation, Invasion, and Glutamine Metabolism in NSCLC Cells by Upregulating SLC7A5 Expression
by: Wei Hou, et al.
Published: (2024)

Sequential Multi-Agent Dynamic Algorithm Configuration
by: Lu, Chen, et al.
Published: (2025)

NVR: Vector Runahead on NPUs for Sparse Memory Access
by: Wang, Hui, et al.
Published: (2025)

Research on Optimizing Real-Time Data Processing in High-Frequency Trading Algorithms using Machine Learning
by: Fan, Yuxin, et al.
Published: (2024)