Nearly Lossless Adaptive Bit Switching

• URL: http://arxiv.org/abs/2502.01199v1
• Date: Mon, 03 Feb 2025 09:46:26 GMT
• Title: Nearly Lossless Adaptive Bit Switching
• Authors: Haiduo Huang, Zhenhua Liu, Tian Xia, Wenzhe zhao, Pengju Ren,
• Abstract summary: Experimental results on the ImageNet-1K classification demonstrate that our methods have enough advantages to state-of-the-art one-shot joint QAT in both multi-precision and mixed-precision.
• Score: 8.485009775430411
• License:
• Abstract: Model quantization is widely applied for compressing and accelerating deep neural networks (DNNs). However, conventional Quantization-Aware Training (QAT) focuses on training DNNs with uniform bit-width. The bit-width settings vary across different hardware and transmission demands, which induces considerable training and storage costs. Hence, the scheme of one-shot joint training multiple precisions is proposed to address this issue. Previous works either store a larger FP32 model to switch between different precision models for higher accuracy or store a smaller INT8 model but compromise accuracy due to using shared quantization parameters. In this paper, we introduce the Double Rounding quantization method, which fully utilizes the quantized representation range to accomplish nearly lossless bit-switching while reducing storage by using the highest integer precision instead of full precision. Furthermore, we observe a competitive interference among different precisions during one-shot joint training, primarily due to inconsistent gradients of quantization scales during backward propagation. To tackle this problem, we propose an Adaptive Learning Rate Scaling (ALRS) technique that dynamically adapts learning rates for various precisions to optimize the training process. Additionally, we extend our Double Rounding to one-shot mixed precision training and develop a Hessian-Aware Stochastic Bit-switching (HASB) strategy. Experimental results on the ImageNet-1K classification demonstrate that our methods have enough advantages to state-of-the-art one-shot joint QAT in both multi-precision and mixed-precision. We also validate the feasibility of our method on detection and segmentation tasks, as well as on LLMs task. Our codes are available at https://github.com/haiduo/Double-Rounding.

Related papers

• SKIM: Any-bit Quantization Pushing The Limits of Post-Training Quantization [7.198819240352308]
  Large Language Models (LLMs) exhibit impressive performance across various tasks, but deploying them for inference poses challenges. We propose a new method called SKIM: Scaled K-means clustering wIth Mixed precision. In terms of model perplexity, our method narrows the gap between 3-bit quantized LLaMA models and their full precision counterparts by 16.3% on average.
  arXiv  Detail & Related papers  (2024-12-05T14:19:59Z)
• AdaBM: On-the-Fly Adaptive Bit Mapping for Image Super-Resolution [53.23803932357899]
  We introduce the first on-the-fly adaptive quantization framework that accelerates the processing time from hours to seconds. We achieve competitive performance with the previous adaptive quantization methods, while the processing time is accelerated by x2000.
  arXiv  Detail & Related papers  (2024-04-04T08:37:27Z)
• FLIQS: One-Shot Mixed-Precision Floating-Point and Integer Quantization Search [50.07268323597872]
  We propose the first one-shot mixed-precision quantization search that eliminates the need for retraining in both integer and low-precision floating point models. With integer models, we increase the accuracy of ResNet-18 on ImageNet by 1.31% and ResNet-50 by 0.90% with equivalent model cost over previous methods. For the first time, we explore a novel mixed-precision floating-point search and improve MobileNetV2 by up to 0.98% compared to prior state-of-the-art FP8 models.
  arXiv  Detail & Related papers  (2023-08-07T04:17:19Z)
• Augmenting Hessians with Inter-Layer Dependencies for Mixed-Precision Post-Training Quantization [7.392278887917975]
  We propose a mixed-precision post training quantization approach that assigns different numerical precisions to tensors in a network based on their specific needs. Our experiments demonstrate latency reductions compared to a 16-bit baseline of $25.48%$, $21.69%$, and $33.28%$ respectively.
  arXiv  Detail & Related papers  (2023-06-08T02:18:58Z)
• Mixed Precision of Quantization of Transformer Language Models for Speech Recognition [67.95996816744251]
  State-of-the-art neural language models represented by Transformers are becoming increasingly complex and expensive for practical applications. Current low-bit quantization methods are based on uniform precision and fail to account for the varying performance sensitivity at different parts of the system to quantization errors. The optimal local precision settings are automatically learned using two techniques. Experiments conducted on Penn Treebank (PTB) and a Switchboard corpus trained LF-MMI TDNN system.
  arXiv  Detail & Related papers  (2021-11-29T09:57:00Z)
• Improving Calibration for Long-Tailed Recognition [68.32848696795519]
  We propose two methods to improve calibration and performance in such scenarios. For dataset bias due to different samplers, we propose shifted batch normalization. Our proposed methods set new records on multiple popular long-tailed recognition benchmark datasets.
  arXiv  Detail & Related papers  (2021-04-01T13:55:21Z)
• Differentiable Joint Pruning and Quantization for Hardware Efficiency [16.11027058505213]
  DJPQ incorporates variational information bottleneck based structured pruning and mixed-bit precision quantization into a single differentiable loss function. We show that DJPQ significantly reduces the number of Bit-Operations (BOPs) for several networks while maintaining the top-1 accuracy of original floating-point models.
  arXiv  Detail & Related papers  (2020-07-20T20:45:47Z)
• Learning to Learn Parameterized Classification Networks for Scalable Input Images [76.44375136492827]
  Convolutional Neural Networks (CNNs) do not have a predictable recognition behavior with respect to the input resolution change. We employ meta learners to generate convolutional weights of main networks for various input scales. We further utilize knowledge distillation on the fly over model predictions based on different input resolutions.
  arXiv  Detail & Related papers  (2020-07-13T04:27:25Z)
• FracBits: Mixed Precision Quantization via Fractional Bit-Widths [29.72454879490227]
  Mixed precision quantization is favorable with customized hardwares supporting arithmetic operations at multiple bit-widths. We propose a novel learning-based algorithm to derive mixed precision models end-to-end under target computation constraints.
  arXiv  Detail & Related papers  (2020-07-04T06:09:09Z)
• APQ: Joint Search for Network Architecture, Pruning and Quantization Policy [49.3037538647714]
  We present APQ for efficient deep learning inference on resource-constrained hardware. Unlike previous methods that separately search the neural architecture, pruning policy, and quantization policy, we optimize them in a joint manner. With the same accuracy, APQ reduces the latency/energy by 2x/1.3x over MobileNetV2+HAQ.
  arXiv  Detail & Related papers  (2020-06-15T16:09:17Z)
• Efficient Bitwidth Search for Practical Mixed Precision Neural Network [33.80117489791902]
  Network quantization has rapidly become one of the most widely used methods to compress and accelerate deep neural networks. Recent efforts propose to quantize weights and activations from different layers with different precision to improve the overall performance. It is challenging to find the optimal bitwidth (i.e., precision) for weights and activations of each layer efficiently. It is yet unclear how to perform convolution for weights and activations of different precision efficiently on generic hardware platforms.
  arXiv  Detail & Related papers  (2020-03-17T08:27:48Z)

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.