The Optimal BERT Surgeon: Scalable and Accurate Second-Order Pruning for Large Language Models

• URL: http://arxiv.org/abs/2203.07259v1
• Date: Mon, 14 Mar 2022 16:40:31 GMT
• Title: The Optimal BERT Surgeon: Scalable and Accurate Second-Order Pruning for Large Language Models
• Authors: Eldar Kurtic, Daniel Campos, Tuan Nguyen, Elias Frantar, Mark Kurtz, Benjamin Fineran, Michael Goin, Dan Alistarh
• Abstract summary: This paper studies the accuracy-compression trade-off for unstructured weight pruning in the context of BERT models. We introduce Optimal BERT Surgeon (O-BERT-S), an efficient and accurate weight pruning method based on approximate second-order information. We investigate the impact of this pruning method when compounding compression approaches for Transformer-based models.
• Score: 23.12519490211362
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Pre-trained Transformer-based language models have become a key building block for natural language processing (NLP) tasks. While these models are extremely accurate, they can be too large and computationally intensive to run on standard deployments. A variety of compression methods, including distillation, quantization, structured and unstructured pruning are known to be applicable to decrease model size and increase inference speed. In this context, this paper's contributions are two-fold. We begin with an in-depth study of the accuracy-compression trade-off for unstructured weight pruning in the context of BERT models, and introduce Optimal BERT Surgeon (O-BERT-S), an efficient and accurate weight pruning method based on approximate second-order information, which we show to yield state-of-the-art results in terms of the compression/accuracy trade-off. Specifically, Optimal BERT Surgeon extends existing work on second-order pruning by allowing for pruning blocks of weights, and by being applicable at BERT scale. Second, we investigate the impact of this pruning method when compounding compression approaches for Transformer-based models, which allows us to combine state-of-the-art structured and unstructured pruning together with quantization, in order to obtain highly compressed, but accurate models. The resulting compression framework is powerful, yet general and efficient: we apply it to both the fine-tuning and pre-training stages of language tasks, to obtain state-of-the-art results on the accuracy-compression trade-off with relatively simple compression recipes. For example, we obtain 10x model size compression with < 1% relative drop in accuracy to the dense BERT-base, 10x end-to-end CPU-inference speedup with < 2% relative drop in accuracy, and 29x inference speedups with < 7.5% relative accuracy drop.

Related papers

• oBERTa: Improving Sparse Transfer Learning via improved initialization, distillation, and pruning regimes [82.99830498937729]
  oBERTa is an easy-to-use set of language models for Natural Language Processing. It allows NLP practitioners to obtain between 3.8 and 24.3 times faster models without expertise in model compression. We explore the use of oBERTa on seven representative NLP tasks.
  arXiv  Detail & Related papers  (2023-03-30T01:37:19Z)
• Optimal Brain Compression: A Framework for Accurate Post-Training Quantization and Pruning [29.284147465251685]
  We introduce a new compression framework which covers both weight pruning and quantization in a unified setting. We show that it can improve significantly upon the compression-accuracy trade-offs of existing post-training methods.
  arXiv  Detail & Related papers  (2022-08-24T14:33:35Z)
• CrAM: A Compression-Aware Minimizer [103.29159003723815]
  We propose a new compression-aware minimizer dubbed CrAM that modifies the optimization step in a principled way. CrAM produces dense models that can be more accurate than the standard SGD/Adam-based baselines, but which are stable under weight pruning. CrAM can produce sparse models which perform well for transfer learning, and it also works for semi-structured 2:4 pruning patterns supported by GPU hardware.
  arXiv  Detail & Related papers  (2022-07-28T16:13:28Z)
• Compression of Generative Pre-trained Language Models via Quantization [62.80110048377957]
  We find that previous quantization methods fail on generative tasks due to the textithomogeneous word embeddings We propose a token-level contrastive distillation to learn distinguishable word embeddings, and a module-wise dynamic scaling to make quantizers adaptive to different modules.
  arXiv  Detail & Related papers  (2022-03-21T02:11:35Z)
• Automatic Mixed-Precision Quantization Search of BERT [62.65905462141319]
  Pre-trained language models such as BERT have shown remarkable effectiveness in various natural language processing tasks. These models usually contain millions of parameters, which prevents them from practical deployment on resource-constrained devices. We propose an automatic mixed-precision quantization framework designed for BERT that can simultaneously conduct quantization and pruning in a subgroup-wise level.
  arXiv  Detail & Related papers  (2021-12-30T06:32:47Z)
• Prune Once for All: Sparse Pre-Trained Language Models [0.6063525456640462]
  We present a new method for training sparse pre-trained Transformer language models by integrating weight pruning and model distillation. These sparse pre-trained models can be used to transfer learning for a wide range of tasks while maintaining their sparsity pattern. We show how the compressed sparse pre-trained models we trained transfer their knowledge to five different downstream natural language tasks with minimal accuracy loss.
  arXiv  Detail & Related papers  (2021-11-10T15:52:40Z)
• ROSITA: Refined BERT cOmpreSsion with InTegrAted techniques [10.983311133796745]
  Pre-trained language models of the BERT family have defined the state-of-the-arts in a wide range of NLP tasks. Performance of BERT-based models is mainly driven by the enormous amount of parameters, which hinders their application to resource-limited scenarios. We introduce three kinds of compression methods (weight pruning, low-rank factorization and knowledge distillation) and explore a range of designs concerning model architecture. Our best compressed model, dubbed Refined BERT cOmpreSsion with InTegrAted techniques (ROSITA), is $7.5 times$ smaller than
  arXiv  Detail & Related papers  (2021-03-21T11:33:33Z)
• BinaryBERT: Pushing the Limit of BERT Quantization [74.65543496761553]
  We propose BinaryBERT, which pushes BERT quantization to the limit with weight binarization. We find that a binary BERT is hard to be trained directly than a ternary counterpart due to its complex and irregular loss landscapes. Empirical results show that BinaryBERT has negligible performance drop compared to the full-precision BERT-base.
  arXiv  Detail & Related papers  (2020-12-31T16:34:54Z)
• Efficient Transformer-based Large Scale Language Representations using Hardware-friendly Block Structured Pruning [12.761055946548437]
  We propose an efficient transformer-based large-scale language representation using hardware-friendly block structure pruning. Besides the significantly reduced weight storage and computation, the proposed approach achieves high compression rates. It is suitable to deploy the final compressed model on resource-constrained edge devices.
  arXiv  Detail & Related papers  (2020-09-17T04:45:47Z)
• Training with Quantization Noise for Extreme Model Compression [57.51832088938618]
  We tackle the problem of producing compact models, maximizing their accuracy for a given model size. A standard solution is to train networks with Quantization Aware Training, where the weights are quantized during training and the gradients approximated with the Straight-Through Estimator. In this paper, we extend this approach to work beyond int8 fixed-point quantization with extreme compression methods.
  arXiv  Detail & Related papers  (2020-04-15T20:10:53Z)

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.