Elucidating the Design Space of FP4 training

• URL: http://arxiv.org/abs/2509.17791v1
• Date: Mon, 22 Sep 2025 13:50:40 GMT
• Title: Elucidating the Design Space of FP4 training
• Authors: Robert Hu, Carlo Luschi, Paul Balanca,
• Abstract summary: This paper aims to provide a unified view of the design space of textttFP4 training.<n>We introduce a comprehensive, quantisation gradient-based framework for microscaling quantization.<n>By systematically evaluating thousands of combinations of techniques, we identify which configurations offer the most favourable performance-to-overhead trade-off.
• Score: 6.963061311516306
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The increasing computational demands of foundation models have spurred research into low-precision training, with 4-bit floating-point (\texttt{FP4}) formats emerging as a frontier for maximizing hardware throughput. While numerous techniques have been proposed to stabilize \texttt{FP4} training, they often present isolated solutions with varying, and not always clear, computational overheads. This paper aims to provide a unified view of the design space of \texttt{FP4} training. We introduce a comprehensive, quantisation gradient-based framework for microscaling quantization that allows for a theoretical analysis of the computational costs associated with different stabilization methods on both the forward and backward passes. Using a simulator built on this framework, we conduct an extensive empirical study across a wide range of machine learning tasks, including regression, image classification, diffusion models, and language models. By systematically evaluating thousands of combinations of techniques, such as novel gradient approximations, rounding strategies, and scaling methods, we identify which configurations offer the most favourable performance-to-overhead trade-off. We find that the techniques enabling the best trade-off involve carefully combining Hadamard transformations, tensor scaling and stochastic rounding. We further find that using \texttt{UE5M3} as a scaling factor potentially offers a good compromise between range and precision with manageable computational overhead.

Related papers

• How to Set the Learning Rate for Large-Scale Pre-training? [73.03133634525635]
  We formalize this investigation into two distinct research paradigms: Fitting and Transfer.<n>Within the Fitting Paradigm, we introduce a Scaling Law for search factor, effectively reducing the search complexity from O(n3) to O(n*C_D*C_) via predictive modeling.<n>We extend the principles of $$Transfer to the Mixture of Experts (MoE) architecture, broadening its applicability to encompass model depth, weight decay, and token horizons.
  arXiv  Detail & Related papers  (2026-01-08T15:55:13Z)
• MoR: Mixture Of Representations For Mixed-Precision Training [0.398636957150696]
  Mixture-of-Representations (MoR) is a novel, per-tensor and sub-tensor level quantization framework.<n>MoR dynamically analyzes a tensor's numerical properties to select between a variety of different representations.<n>Our initial findings show that this approach can achieve state-of-the-art results with 98.38% of tensors quantized to the FP8 format.
  arXiv  Detail & Related papers  (2025-12-28T06:28:50Z)
• Pretraining Large Language Models with NVFP4 [53.235038214986865]
  We introduce a novel approach for stable and accurate training of large language models (LLMs) using the NVFP4 format.<n>Our method integrates two-dimensional quantization scheme for consistent representations across both the forward and backward passes.<n>Our results show that the model trained with our NVFP4-based pretraining technique achieves training loss and downstream task accuracies comparable to an FP8 baseline.
  arXiv  Detail & Related papers  (2025-09-29T17:53:17Z)
• Quartet: Native FP4 Training Can Be Optimal for Large Language Models [27.800012997794987]
  Training large language models (LLMs) models directly in low-precision offers a way to address computational costs.<n> NVIDIA's recent Blackwell architecture facilitates very low-precision operations using FP4 variants.<n>We introduce a new approach for accurate, end-to-end FP4 training with all the major computations in low precision.
  arXiv  Detail & Related papers  (2025-05-20T17:55:50Z)
• Towards Efficient Pre-training: Exploring FP4 Precision in Large Language Models [25.700481606604647]
  Experimental results demonstrate that our FP4 training scheme achieves accuracy comparable to BF16 and FP8, with smaller theoretical computational cost.<n>With the advent of next-generation hardware supporting FP4, our method sets the foundation for efficient ultra-low precision training.
  arXiv  Detail & Related papers  (2025-02-17T05:33:11Z)
• Optimizing Large Language Model Training Using FP4 Quantization [73.55459961002371]
  Quantized training presents a promising solution by enabling low-bit arithmetic operations to reduce costs.<n>This work introduces the first FP4 training framework for large language models (LLMs)
  arXiv  Detail & Related papers  (2025-01-28T18:04:50Z)
• GAQAT: gradient-adaptive quantization-aware training for domain generalization [54.31450550793485]
  We propose a novel Gradient-Adaptive Quantization-Aware Training (GAQAT) framework for DG.<n>Our approach begins by identifying the scale-gradient conflict problem in low-precision quantization.<n>Extensive experiments validate the effectiveness of the proposed GAQAT framework.
  arXiv  Detail & Related papers  (2024-12-07T06:07:21Z)
• LLMC: Benchmarking Large Language Model Quantization with a Versatile Compression Toolkit [55.73370804397226]
  Quantization, a key compression technique, can effectively mitigate these demands by compressing and accelerating large language models. We present LLMC, a plug-and-play compression toolkit, to fairly and systematically explore the impact of quantization. Powered by this versatile toolkit, our benchmark covers three key aspects: calibration data, algorithms (three strategies), and data formats.
  arXiv  Detail & Related papers  (2024-05-09T11:49:05Z)
• When Parameter-efficient Tuning Meets General-purpose Vision-language Models [65.19127815275307]
  PETAL revolutionizes the training process by requiring only 0.5% of the total parameters, achieved through a unique mode approximation technique. Our experiments reveal that PETAL not only outperforms current state-of-the-art methods in most scenarios but also surpasses full fine-tuning models in effectiveness.
  arXiv  Detail & Related papers  (2023-12-16T17:13:08Z)
• Efficient Neural PDE-Solvers using Quantization Aware Training [71.0934372968972]
  We show that quantization can successfully lower the computational cost of inference while maintaining performance. Our results on four standard PDE datasets and three network architectures show that quantization-aware training works across settings and three orders of FLOPs magnitudes.
  arXiv  Detail & Related papers  (2023-08-14T09:21:19Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.