arXiv AI June 9, 2026 By Dmitrii Vasilev

An 84-Format Numeric Catalog with Bit-Exact Conformance Vectors: A Vendor-Neutral Reference for FP8, BF16, MXFP4, and Microscaling Formats

arXiv:2606. 09686v1 Announce Type: cross Abstract: Numeric format proliferation in machine learning hardware -- FP8 (E4M3 and E5M2), BF16, MXFP4, microscaling block formats, and dozens of research variants -- has outpaced the availability of vendor-neutral, bit-exact reference material.

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

arXiv Machine Learning

6d ago

CurveFP: Co-Designing Numerical Representation and Product Arithmetic for Language Models

arXiv:2608. 10010v2 Announce Type: replace Abstract: Low-precision formats usually optimize scalar fidelity while inheriting conventional product arithmetic.

By Ye Qiao

llms

arXiv Machine Learning

Aug 12

CurveFP: Rational-Radix Logarithmic Datatypes with Closed Products for Language Models

arXiv:2608. 10010v1 Announce Type: new Abstract: Low-precision datatypes reduce language-model cost, but most formats optimize scalar fidelity while leaving the arithmetic induced by their products unchanged.

By Ye Qiao

llms efficiency

arXiv Machine Learning

Jun 4

Novel Aspects of IEEE SA P3109 Arithmetic Formats for Machine Learning

arXiv:2606. 04028v1 Announce Type: new Abstract: The IEEE P3109 draft standard defines a parameterized family of binary floating-point formats and associated operations, with a focus on facilitating machine learning.

By Andrew Fitzgibbon, Christoph M. Wintersteiger, Jeffrey Sarnoff

efficiency

arXiv Machine Learning

2d ago

The Integer Alibi: Localizing Cross-Kernel Divergence in INT8-Quantized LLM Inference

arXiv:2608. 13756v1 Announce Type: new Abstract: Two GPU kernels implementing the same scaled INT8 GEMM interface are usually treated as interchangeable.

By Teng-Ruei Chen

llms efficiency benchmarks

arXiv AI

Jun 3

MX-SAFE: Versatile Inference- and Training-Proof Microscaling Format with On-the-Fly Exponent and Mantissa Bit Allocation

arXiv:2605. 24391v2 Announce Type: replace-cross Abstract: As the demand for deep learning grows, cost reduction through quantization has become essential for both training and inference.

By Dahoon Park, Jahyun Koo, Sangwoo Hwang, Jaeha Kung

efficiency

arXiv AI

Jun 16

FP8 is All You Need (Part 1): Debunking Hardware FP64 as the HPC Holy Grail (June 13th version)

arXiv:2606. 06510v2 Announce Type: replace-cross Abstract: Conventional HPC holds that native hardware FP64 is the irreducible foundation of scientific computing.

By Satoshi Matsuoka

efficiency

Related stories

CurveFP: Co-Designing Numerical Representation and Product Arithmetic for Language Models

CurveFP: Rational-Radix Logarithmic Datatypes with Closed Products for Language Models

Novel Aspects of IEEE SA P3109 Arithmetic Formats for Machine Learning

The Integer Alibi: Localizing Cross-Kernel Divergence in INT8-Quantized LLM Inference

MX-SAFE: Versatile Inference- and Training-Proof Microscaling Format with On-the-Fly Exponent and Mantissa Bit Allocation

FP8 is All You Need (Part 1): Debunking Hardware FP64 as the HPC Holy Grail (June 13th version)