A Survey on Model Compression and Acceleration for Pretrained Language Models
Open accessThe DOI matches its Crossref record; 42 of the 42 references it cites were found in a registry or an index.
Abstract
Despite achieving state-of-the-art performance on many NLP tasks, the high energy cost and long inference delay prevent Transformer-based pretrained language models (PLMs) from seeing broader adoption including for edge and mobile computing. Efficient NLP research aims to comprehensively consider computation, time and carbon emission for the entire life-cycle of NLP, including data preparation, model training and inference. In this survey, we focus on the inference stage and review the current state of model compression and acceleration for pretrained language models, including benchmarks, metrics and methodology.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018) · arXiv · doi:10.48550/arxiv.1810.04805 · Cited by 73,173
Deep Contextualized Word Representations
Peters, M., Neumann, M., Iyyer, M., Gardner, M., Clark, C., Lee, K., & Zettlemoyer, L. (2018) · Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers) · doi:10.18653/v1/n18-1202 · Cited by 11,983
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., & Liu, P. J. (2019) · arXiv · doi:10.48550/arxiv.1910.10683 · Cited by 10,929
Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
Finn, C., Abbeel, P., & Levine, S. (2017) · arXiv · doi:10.48550/arxiv.1703.03400 · Cited by 8,512 · cited 1×
“One promising direction could be applying Meta Learning (Finn, Abbeel, and Levine 2017) or Neural Architecture Search (Liu, Simonyan, and Yang 2019) to model compression and acceleration, to minimize the need for hyperparameters and human design.”
MethodModel Selection and Estimation in Regression with Grouped Variables
Yuan, M., & Lin, Y. (2006) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.1467-9868.2005.00532.x · Cited by 7,261
SQuAD: 100,000+ Questions for Machine Comprehension of Text
Rajpurkar, P., Zhang, J., Lopyrev, K., & Liang, P. (2016) · Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d16-1264 · Cited by 6,208 · cited 1×
“SQuAD (Rajpurkar et al. 2016) is used for machine reading comprehension (MRC).”
MethodDistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
Sanh, V., Debut, L., Chaumond, J., & Wolf, T. (2019) · arXiv · doi:10.48550/arxiv.1910.01108 · Cited by 4,653
ALBERT: A Lite BERT for Self-supervised Learning of Language Representations
Lan, Z., Chen, M., Goodman, S., Gimpel, K., Sharma, P., & Soricut, R. (2019) · arXiv · doi:10.48550/arxiv.1909.11942 · Cited by 4,308 · cited 4×
“However, since all Transformer layers in ALBERT share the same weights, the model size of n -layer ALBERT is only 1 /n of n -layer BERT.”
BackgroundMethodInfluentialGLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding
Wang, A., Singh, A., Michael, J., Hill, F., Levy, O., & Bowman, S. (2018) · Proceedings of the 2018 EMNLP Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP · doi:10.18653/v1/w18-5446 · Cited by 4,222
Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference
Jacob, B., Kligys, S., Chen, B., Zhu, M., Tang, M., Howard, A., Adam, H., & Kalenichenko, D. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00286 · Cited by 4,114
DARTS: Differentiable architecture search
Liu, H., Simonyan, K., & Yang, Y. (2018) · arXiv · doi:10.48550/arxiv.1806.09055 · Cited by 3,248
Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computation
Bengio, Y., Léonard, N., & Courville, A. (2013) · arXiv · doi:10.48550/arxiv.1308.3432 · Cited by 1,910
TinyBERT: Distilling BERT for natural language understanding
Jiao, X., Yin, Y., Shang, L., Jiang, X., Chen, X., Li, L., Wang, F., & Liu, Q. (2020) · Findings of the Association for Computational Linguistics: EMNLP 2020 · doi:10.18653/v1/2020.findings-emnlp.372 · Cited by 1,786
Pre-trained models for natural language processing: A survey
Qiu, X., Sun, T., Xu, Y., Shao, Y., Dai, N., & Huang, X. (2020) · Science China Technological Sciences · doi:10.1007/s11431-020-1647-3 · Cited by 1,591 · cited 3×
“There have been surveys (Qiu et al. 2020; Han et al. 2021; Xu et al. 2021b) that cover some aspects of this topic.”
BackgroundThe Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks
Frankle, J., & Carbin, M. (2018) · arXiv · doi:10.48550/arxiv.1803.03635 · Cited by 1,529
BranchyNet: Fast inference via early exiting from deep neural networks
Teerapittayanon, S., McDanel, B., & Kung, H. T. (2016) · 2016 23rd International Conference on Pattern Recognition (ICPR) · doi:10.1109/icpr.2016.7900006 · Cited by 1,326
Analyzing Multi-Head Self-Attention: Specialized Heads Do the Heavy Lifting, the Rest Can Be Pruned
Voita, E., Talbot, D., Moiseev, F., Sennrich, R., & Titov, I. (2019) · Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/p19-1580 · Cited by 1,146 · cited 1×
“Michel, Levy, and Neubig (2019) and Voita et al. (2019) find that the multi-head attention in Transformer has redundancy.”
BackgroundSuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systems
Wang, A., Pruksachatkun, Y., Nangia, N., Singh, A., Michael, J., Hill, F., Levy, O., & Bowman, S. R. (2019) · arXiv · doi:10.48550/arxiv.1905.00537 · Cited by 1,051
Pre-trained models: Past, present and future
Han, X., Zhang, Z., Ding, N., Gu, Y., Liu, X., Huo, Y., Qiu, J., Yao, Y., Zhang, A., Zhang, L., et al. (2021) · AI Open · doi:10.1016/j.aiopen.2021.08.002 · Cited by 994
MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices
Sun, Z., Yu, H., Song, X., Liu, R., Yang, Y., & Zhou, D. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.195 · Cited by 757
MiniLM: Deep Self-Attention Distillation for Task-Agnostic Compression of Pre-Trained Transformers
Wang, W., Wei, F., Dong, L., Bao, H., Yang, N., & Zhou, M. (2020) · arXiv · doi:10.48550/arxiv.2002.10957 · Cited by 703 · cited 1×
“TinyBERT also demonstrates that performing KD in both pretraining and fine-tuning stages can improve the performance of KD. Similarly, MiniLM (Wang et al. 2020a, 2021) aligns the attention matrix and value-value scaled dot-product (i.e., value relation loss, as shown in Figure 1(f)).”
Learning Sparse Neural Networks through $L_0$ Regularization
Louizos, C., Welling, M., & Kingma, D. P. (2017) · arXiv · doi:10.48550/arxiv.1712.01312 · Cited by 628
Low-rank matrix factorization for Deep Neural Network training with high-dimensional output targets
Sainath, T. N., Kingsbury, B., Sindhwani, V., Arisoy, E., & Ramabhadran, B. (2013) · 2013 IEEE International Conference on Acoustics, Speech and Signal Processing · doi:10.1109/icassp.2013.6638949 · Cited by 593 · cited 1×
“The weight matrices in a neural network are often low-rank, indicating redundancy in model weights (Sainath et al. 2013).”
BackgroundDeep double descent: where bigger models and more data hurt*
Nakkiran, P., Kaplun, G., Bansal, Y., Yang, T., Barak, B., & Sutskever, I. (2021) · Journal of Statistical Mechanics: Theory and Experiment · doi:10.1088/1742-5468/ac3a74 · Cited by 565 · cited 1×
“This seems to be consistent with previous finding on over-parameterized models (Nakkiran et al. 2020).”
ResultHAWQ: Hessian AWare Quantization of Neural Networks With Mixed-Precision
Dong, Z., Yao, Z., Gholami, A., Mahoney, M., & Keutzer, K. (2019) · 2019 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00038 · Cited by 492
Dehghani, M., Gouws, S., Vinyals, O., Uszkoreit, J., & Kaiser, Ł. (2018) · arXiv · doi:10.48550/arxiv.1807.03819 · Cited by 467
Zafrir, O., Boudoukh, G., Izsak, P., & Wasserblat, M. (2019) · arXiv · doi:10.48550/arxiv.1910.06188 · Cited by 441
Q-BERT: Hessian Based Ultra Low Precision Quantization of BERT
Shen, S., Dong, Z., Ye, J., Ma, L., Yao, Z., Gholami, A., Mahoney, M. W., & Keutzer, K. (2020) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v34i05.6409 · Cited by 437
Quantifying the Carbon Emissions of Machine Learning
Lacoste, A., Luccioni, A., Schmidt, V., & Dandres, T. (2019) · arXiv · doi:10.48550/arxiv.1910.09700 · Cited by 408
Reducing Transformer Depth on Demand with Structured Dropout
Fan, A., Grave, E., & Joulin, A. (2019) · arXiv · doi:10.48550/arxiv.1909.11556 · Cited by 387
Compressing BERT: Studying the Effects of Weight Pruning on Transfer Learning
Gordon, M., Duh, K., & Andrews, N. (2020) · Proceedings of the 5th Workshop on Representation Learning for NLP · doi:10.18653/v1/2020.repl4nlp-1.18 · Cited by 303
Parameter-Efficient Transfer Learning with Diff Pruning
Guo, D., Rush, A., & Kim, Y. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.378 · Cited by 220 · cited 1×
“Guo, Rush, and Kim (2021) propose diff pruning , by learning a task-specific “diff” vector that extends the original pretrained parameters.”
BackgroundMovement pruning: Adaptive sparsity by fine-tuning
Sanh, V., Wolf, T., & Rush, A. M. (2020) · arXiv · doi:10.48550/arxiv.2005.07683 · Cited by 218 · cited 5×
“Combining Techniques Although there have been attempts at combining multiple model compression and acceleration techniques (Kim and Awadalla 2020; Sanh, Wolf, and Rush 2020; Xu et al. 2021a), there is a lack of comprehensive and systematic study for combining compression techniques for better…”
BackgroundMethodInfluentialI-BERT: Integer-only BERT Quantization
Kim, S., Gholami, A., Yao, Z., Mahoney, M. W., & Keutzer, K. (2021) · arXiv · doi:10.48550/arxiv.2101.01321 · Cited by 186
TernaryBERT: Distillation-aware Ultra-low Bit BERT
Zhang, W., Hou, L., Yin, Y., Shang, L., Chen, X., Jiang, X., & Liu, Q. (2020) · arXiv · doi:10.48550/arxiv.2009.12812 · Cited by 150 · cited 2×
“TernaryBERT (Zhang et al. 2020) combines approximation-aware and loss-aware ternarization (i.e., using only {− 1 , 0 , 1 } for weights) methods with different granularity for different components in BERT.”
MethodDynaBERT: Dynamic BERT with Adaptive Width and Depth
Hou, L., Huang, Z., Shang, L., Jiang, X., Chen, X., & Liu, Q. (2020) · arXiv · doi:10.48550/arxiv.2004.04037 · Cited by 143
Learned Token Pruning for Transformers
Kim, S., Shen, S., Thorsley, D., Gholami, A., Kwon, W., Hassoun, J., & Keutzer, K. (2022) · Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining · doi:10.1145/3534678.3539260 · Cited by 139
Structured Pruning Learns Compact and Accurate Models
Xia, M., Zhong, Z., & Chen, D. (2022) · Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2022.acl-long.107 · Cited by 136
Accelerating Sparse Deep Neural Networks
Mishra, A., Latorre, J. A., Pool, J., Stosic, D., Stosic, D., Venkatesh, G., Yu, C., & Micikevicius, P. (2021) · arXiv · doi:10.48550/arxiv.2104.08378 · Cited by 97
Length-Adaptive Transformer: Train Once with Length Drop, Use Anytime with Search
Kim, G., & Cho, K. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.508 · Cited by 59 · cited 1×
“Length-Adaptive Transformer (LAT, Kim and Cho 2021) introduces LengthDrop that randomly skips tokens during pretraining to mitigate the gap between pretraining and fine-tuning.”
BackgroundBinaryBERT: Pushing the Limit of BERT Quantization
Bai, H., Zhang, W., Hou, L., Shang, L., Jin, J., Jiang, X., Liu, Q., Lyu, M., & King, I. (2020) · arXiv · doi:10.48550/arxiv.2012.15701 · Cited by 42
Parameter-Efficient Transfer Learning with Diff Pruning
Guo, D., Rush, A. M., & Kim, Y. (2020) · arXiv · doi:10.48550/arxiv.2012.07463 · Cited by 20
Cited by
A Review on Edge Large Language Models: Design, Execution, and Applications
Zheng, Y., Chen, Y., Qian, B., Shi, X., Shu, Y., & Chen, J. (2025) · ACM Computing Surveys · doi:10.1145/3719664 · Cited by 169
NetLLM: adapting large language models for networking
Wu, D., Wang, X., Qiao, Y., Wang, Z., Jiang, J., Cui, S., & Wang, F. (2024) · Proceedings of the ACM SIGCOMM 2024 Conference · doi:10.1145/3651890.3672268 · Cited by 167
A Survey on the Memory Mechanism of Large Language Model based Agents
Zhang, Z., Bo, X., Ma, C., Li, R., Chen, X., Dai, Q., Zhu, J., Dong, Z., & Wen, J.-R. (2024) · arXiv · doi:10.48550/arxiv.2404.13501 · Cited by 149
Lamaakal, I., Maleh, Y., El Makkaoui, K., Ouahbi, I., Pławiak, P., Alfarraj, O., Almousa, M., & Abd El-Latif, A. A. (2025) · Sensors · doi:10.3390/s25051318 · Cited by 60
From Algorithm to Hardware: A Survey on Efficient and Safe Deployment of Deep Neural Networks
Geng, X., Wang, Z., Chen, C., Xu, Q., Xu, K., Jin, C., Gupta, M., Yang, X., Chen, Z., Sabry Aly, M. M., et al. (2025) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2024.3394494 · Cited by 26
Metadata
- Authors
- Xu, C., & McAuley, J.
- Venue
- Proceedings of the AAAI Conference on Artificial Intelligence
- Published
- 2023-06-26
- Volume
- 37
- Issue
- 9
- Pages
- 10566-10575
- ISSN
- 2159-5399 (ISSN-L)
- 2374-3468
- OpenAlex
- W4382463788
- Cited by
- 58
- References
- 42
- FWCI
- 6.48
- Type
- conference paper
- Language
- en
- Open access
- diamond
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.