Pushing the Limits of Low-Bit Optimizers: A Focus on EMA Dynamics

• URL: http://arxiv.org/abs/2505.00347v1
• Date: Thu, 01 May 2025 06:47:45 GMT
• Title: Pushing the Limits of Low-Bit Optimizers: A Focus on EMA Dynamics
• Authors: Cong Xu, Wenbin Liang, Mo Yu, Anan Liu, Ke-Yue Zhang, Lizhuang Ma, Jianyong Wang, Jun Wang, Wei Zhang,
• Abstract summary: We present a novel type of overload that carries with extremely lightweight state elements, achieved through ultra-low-precision quantization.<n>The proposed SOLO achieves substantial memory savings (approximately 45 GB when training a 7B model) with minimal accuracy loss.
• Score: 65.37942405146232
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: The explosion in model sizes leads to continued growth in prohibitive training/fine-tuning costs, particularly for stateful optimizers which maintain auxiliary information of even 2x the model size to achieve optimal convergence. We therefore present in this work a novel type of optimizer that carries with extremely lightweight state overloads, achieved through ultra-low-precision quantization. While previous efforts have achieved certain success with 8-bit or 4-bit quantization, our approach enables optimizers to operate at precision as low as 3 bits, or even 2 bits per state element. This is accomplished by identifying and addressing two critical challenges: the signal swamping problem in unsigned quantization that results in unchanged state dynamics, and the rapidly increased gradient variance in signed quantization that leads to incorrect descent directions. The theoretical analysis suggests a tailored logarithmic quantization for the former and a precision-specific momentum value for the latter. Consequently, the proposed SOLO achieves substantial memory savings (approximately 45 GB when training a 7B model) with minimal accuracy loss. We hope that SOLO can contribute to overcoming the bottleneck in computational resources, thereby promoting greater accessibility in fundamental research.

Related papers

• ParetoQ: Scaling Laws in Extremely Low-bit LLM Quantization [58.84018707089315]
  We present a unified framework for rigorous comparisons across 1-bit, 1.58-bit, 2-bit, 3-bit, and 4-bit quantization settings.<n>We show that ternary, 2-bit, and 3-bit quantization maintains comparable performance in the size-accuracy trade-off.<n>Considering hardware constraints, 2-bit quantization offers promising potential for memory reduction and speedup.
  arXiv  Detail & Related papers  (2025-02-04T18:59:26Z)
• Oscillations Make Neural Networks Robust to Quantization [0.16385815610837165]
  We show that oscillations in Quantization Aware Training (QAT) are undesirable artifacts caused by the Straight-Through Estimator (STE) We propose a novel regularization method that induces oscillations to improve quantization.
  arXiv  Detail & Related papers  (2025-02-01T16:39:58Z)
• 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)
• Joint Pruning and Channel-wise Mixed-Precision Quantization for Efficient Deep Neural Networks [10.229120811024162]
  deep neural networks (DNNs) pose significant challenges to their deployment on edge devices. Common approaches to address this issue are pruning and mixed-precision quantization. We propose a novel methodology to apply them jointly via a lightweight gradient-based search.
  arXiv  Detail & Related papers  (2024-07-01T08:07:02Z)
• Gradient-based Automatic Mixed Precision Quantization for Neural Networks On-Chip [0.9187138676564589]
  We present High Granularity Quantization (HGQ), an innovative quantization-aware training method. HGQ fine-tune the per-weight and per-activation precision by making them optimizable through gradient descent. This approach enables ultra-low latency and low power neural networks on hardware capable of performing arithmetic operations.
  arXiv  Detail & Related papers  (2024-05-01T17:18:46Z)
• DB-LLM: Accurate Dual-Binarization for Efficient LLMs [83.70686728471547]
  Large language models (LLMs) have significantly advanced the field of natural language processing. Existing ultra-low-bit quantization always causes severe accuracy drops. We propose a novel Dual-Binarization method for LLMs, namely DB-LLM.
  arXiv  Detail & Related papers  (2024-02-19T09:04:30Z)
• On-Chip Hardware-Aware Quantization for Mixed Precision Neural Networks [52.97107229149988]
  We propose an On-Chip Hardware-Aware Quantization framework, performing hardware-aware mixed-precision quantization on deployed edge devices. For efficiency metrics, we built an On-Chip Quantization Aware pipeline, which allows the quantization process to perceive the actual hardware efficiency of the quantization operator. For accuracy metrics, we propose Mask-Guided Quantization Estimation technology to effectively estimate the accuracy impact of operators in the on-chip scenario.
  arXiv  Detail & Related papers  (2023-09-05T04:39:34Z)
• Quantized Neural Networks for Low-Precision Accumulation with Guaranteed Overflow Avoidance [68.8204255655161]
  We introduce a quantization-aware training algorithm that guarantees avoiding numerical overflow when reducing the precision of accumulators during inference. We evaluate our algorithm across multiple quantized models that we train for different tasks, showing that our approach can reduce the precision of accumulators while maintaining model accuracy with respect to a floating-point baseline.
  arXiv  Detail & Related papers  (2023-01-31T02:46:57Z)
• VS-Quant: Per-vector Scaled Quantization for Accurate Low-Precision Neural Network Inference [7.886868529510128]
  Quantization maps floating-point weights and activations in a trained model to low-bitwidth integer values using scale factors. Excessive quantization, reducing precision too aggressively, results in accuracy degradation. Per-vector scale factors can be implemented with low-bitwidth integers when using a two-level quantization scheme.
  arXiv  Detail & Related papers  (2021-02-08T19:56:04Z)

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.