English
Related papers

Related papers: Improving the Straight-Through Estimator with Zero…

200 papers

Training activation quantized neural networks involves minimizing a piecewise constant function whose gradient vanishes almost everywhere, which is undesirable for the standard back-propagation or chain rule. An empirical way around this…

Machine Learning · Computer Science 2019-09-26 Penghang Yin , Jiancheng Lyu , Shuai Zhang , Stanley Osher , Yingyong Qi , Jack Xin

Fine-tuning large language models (LLMs) using standard first-order (FO) optimization often drives training toward sharp, poorly generalizing minima. Conversely, zeroth-order (ZO) methods offer stronger exploratory behavior without relying…

Machine Learning · Computer Science 2026-01-12 Feihu Jin , Ying Tan

Federated learning (FL), as an emerging edge artificial intelligence paradigm, enables many edge devices to collaboratively train a global model without sharing their private data. To enhance the training efficiency of FL, various…

Machine Learning · Computer Science 2022-11-23 Wenzhi Fang , Ziyi Yu , Yuning Jiang , Yuanming Shi , Colin N. Jones , Yong Zhou

Quantization has become a predominant approach for model compression, enabling deployment of large models trained on GPUs onto smaller form-factor devices for inference. Quantization-aware training (QAT) optimizes model parameters with…

Machine Learning · Computer Science 2022-12-13 Zheng Wang , Juncheng B Li , Shuhui Qu , Florian Metze , Emma Strubell

One of the major bottlenecks for deploying popular first-order differentially private (DP) machine learning algorithms (e.g., DP-SGD) lies in their high computation and memory cost, despite the existence of optimized implementations.…

Machine Learning · Computer Science 2026-01-14 Xuchen Gong , Tian Li

Zeroth-order (ZO) optimization enables memory-efficient training of neural networks by estimating gradients via forward passes only, eliminating the need for backpropagation. However, the stochastic nature of gradient estimation…

Machine Learning · Computer Science 2026-03-24 Chen Zhang , Yuxin Cheng , Chenchen Ding , Shuqi Wang , Jingreng Lei , Runsheng Yu , Yik-Chung WU , Ngai Wong

Quantized or low-bit neural networks are attractive due to their inference efficiency. However, training deep neural networks with quantized activations involves minimizing a discontinuous and piecewise constant loss function. Such a loss…

Machine Learning · Computer Science 2021-06-15 Ziang Long , Penghang Yin , Jack Xin

Optimizing large-scale nonconvex problems, common in deep learning, demands balancing rapid convergence with computational efficiency. First-order (FO) optimizers, which serve as today's baselines, provide fast convergence and good…

Machine Learning · Computer Science 2025-09-30 Jiahe Chen , Ziye Ma

Quantized neural network training optimizes a discrete, non-differentiable objective. The straight-through estimator (STE) enables backpropagation through surrogate gradients and is widely used. While previous studies have primarily focused…

Machine Learning · Statistics 2025-10-14 Yuma Ichikawa , Shuhei Kashiwamura , Ayaka Sakata

Fine-tuning Large Language Models (LLMs) has proven effective for a variety of downstream tasks. However, as LLMs grow in size, the memory demands for backpropagation become increasingly prohibitive. Zeroth-order (ZO) optimization methods…

Machine Learning · Computer Science 2025-07-25 Ziming Yu , Pan Zhou , Sike Wang , Jia Li , Mi Tian , Hua Huang

Neural network quantization aims to reduce the bit-widths of weights and activations, making it a critical technique for deploying deep neural networks on resource-constrained hardware. Most Quantization-Aware Training (QAT) methods rely on…

Machine Learning · Computer Science 2025-09-03 Kaiqi Zhao

We investigate the effectiveness of adaptive zeroth-order (ZO) optimization for memory-constrained fine-tuning of large language models (LLMs). Contrary to prior claims, we show that adaptive ZO methods such as ZO-Adam offer no convergence…

Machine Learning · Computer Science 2026-05-06 Hassan Dbouk , Nidham Gazagnadou , Matthias Reisser , Christos Louizos

In this work, we focus on the study of stochastic zeroth-order (ZO) optimization which does not require first-order gradient information and uses only function evaluations. The problem of ZO optimization has emerged in many recent machine…

Machine Learning · Statistics 2020-12-22 Pranay Sharma , Kaidi Xu , Sijia Liu , Pin-Yu Chen , Xue Lin , Pramod K. Varshney

Zeroth-order (ZO) optimization is being recognized as a simple yet powerful alternative to standard backpropagation (BP)-based training. Notably, ZO optimization allows for training with only forward passes and (almost) the same memory as…

Machine Learning · Computer Science 2025-01-09 Keisuke Sugiura , Hiroki Matsutani

Zeroth-order optimizers have recently emerged as a practical approach for fine-tuning large language models (LLMs), significantly reducing GPU memory consumption compared to traditional first-order methods. Yet, existing zeroth-order…

Machine Learning · Computer Science 2025-10-02 Kairun Zhang , Haoyu Li , Yanjun Zhao , Yifan Sun , Huan Zhang

The Straight-Through Estimator (STE) is the dominant method for training neural networks with discrete variables, enabling gradient-based optimisation by routing gradients through a differentiable surrogate. However, existing STE variants…

Machine Learning · Computer Science 2026-02-24 Rushi Shah , Mingyuan Yan , Michael Curtis Mozer , Dianbo Liu

Zeroth-order (ZO) fine-tuning is attractive for large language models because it replaces backpropagation with forward objective evaluations. Existing implementations nevertheless execute ZO algorithms inside conventional training loops,…

Machine Learning · Computer Science 2026-05-28 Zelin Li , Caiwen Ding

Zeroth-order optimization (ZO) is widely used for solving black-box optimization and control problems. In particular, single-point ZO (SZO) is well-suited to online or dynamic problem settings due to its requirement of only a single…

Optimization and Control · Mathematics 2026-02-06 Xin Chen , Zhaolin Ren

Weight quantization is an effective technique to compress deep neural networks for their deployment on edge devices with limited resources. Traditional loss-aware quantization methods commonly use the quantized gradient to replace the…

Machine Learning · Computer Science 2024-01-31 Lianbo Ma , Yuee Zhou , Jianlun Ma , Guo Yu , Qing Li

Gradient descent and backpropagation have enabled neural networks to achieve remarkable results in many real-world applications. Despite ongoing success, training a neural network with gradient descent can be a slow and strenuous affair. We…

Machine Learning · Computer Science 2020-11-19 Varun Ranganathan , Alex Lewandowski