CoChat
Sign Up
Cited by 305Advanced Neural Network ApplicationsParallel Computing and Optimization TechniquesTopic Modeling

LLM.int8(): 8-Bit Matrix Multiplication for Transformers at Scale

Open access
Dettmers, Tim, Lewis, Mike, Belkada, Younes, Zettlemoyer, Luke·2022·Advances in Neural Information Processing Systems 35
DOI 10.52202/068431-2198
Open this paper in CoChat — free accountOpen PDF

Abstract

Large language models have been widely adopted but require significant GPU memory for inference. We develop a procedure for Int8 matrix multiplication for feed-forward and attention projection layers in transformers, which cut the memory needed for inference by half while retaining full precision performance. With our method, a 175B parameter 16/32-bit checkpoint can be loaded, converted to Int8, and used immediately without performance degradation. This is made possible by understanding and working around properties of highly systematic emergent features in transformer language models that dominate attention and transformer predictive performance. To cope with these features, we develop a two-part quantization procedure, LLM.int8(). We first use vector-wise quantization with separate normalization constants for each inner product in the matrix multiplication, to quantize most of the features. However, for the emergent outliers, we also include a new mixed-precision decomposition scheme, which isolates the outlier feature dimensions into a 16-bit matrix multiplication while still more than 99.9% of values are multiplied in 8-bit. Using LLM.int8(), we show empirically it is possible to perform inference in LLMs with up to 175B parameters without any performance degradation. This result makes such models much more accessible, for example making it possible to use OPT-175B/BLOOM on a single server with consumer GPUs. We open-source our software.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.52202/068431-2198

Cites

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • KV management optimizes memory efficiency and inference performance through techniques such as KV cache organization, compression, and quantization.

    “KV management optimizes memory efficiency and inference performance through techniques such as KV cache organization [111], compression [110], and quantization [116].”

    From From Human Memory to AI Memory: A Survey on Memory Mechanisms in the Era of LLMs (2025) · cites this paper as [116]

    The Abstract (p. 1) reports that 8-bit quantization of transformer projection layers halves inference memory use, and Main Results (p. 6) report faster large matrix multiplications. But the paper does not describe quantizing or managing a KV cache; its Discussion (p. 10) says it does not use Int8 multiplication for the attention function. Thus it supports the quantization benefit in general, not the KV-management context attributed to it. (LLM, conf=0.86)

    Experimental — accuracy not yet measured

  • Transformer architectures have 'emergent outliers'.

    “As transformer architectures have”

    From The Platonic Representation Hypothesis (2024) · cites this paper as Dettmers et al., 2022

  • OpenVLA codebase has full support for training on the Open X dataset, integrates with HuggingFace's AutoModel class, and supports LoRA fine-tuning and quantized model inference.

    “Out of the box, the OpenVLA codebase has full support for training on the Open X dataset, integrates with HuggingFace's [21] AutoModel class, and supports LoRA fine-tuning [26] and quantized model inference [27, 88].”

    From OpenVLA: An Open-Source Vision-Language-Action Model (2024) · cites this paper as [88]

  • Quantization techniques load the weights of the network at lower precision, trading off reduced memory requirements for potentially reduced inference speed and accuracy.

    “These approaches load the weights of the network at lower precision, thereby trading off reduced memory requirements for potentially reduced inference speed and accuracy.”

    From OpenVLA: An Open-Source Vision-Language-Action Model (2024) · cites this paper as [88]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

No references could be resolved yet.

Research works printed but not linked: 60

  1. Artetxe, M., Bhosale, S., Goyal, N., Mihaylov, T., Ott, M., Shleifer, S., Lin, X. V., Du, J., Iyer, S., Pasunuru, R., et al. (2021). Efficient large scale language modeling with mixtures of experts. arXiv preprint arXiv:2112.10684 . · arXiv
  2. Bai, H., Zhang, W., Hou, L., Shang, L., Jin, J., Jiang, X., Liu, Q., Lyu, M. R., and King, I. (2021). Binarybert: Pushing the limit of bert quantization. ArXiv , abs/2012.15701. · arXiv
  3. Bondarenko, Y., Nagel, M., and Blankevoort, T. (2021). Understanding and overcoming the challenges of efficient transformer quantization. arXiv preprint arXiv:2109.12948 . · arXiv
  4. Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020). Language models are few-shot learners. arXiv preprint arXiv:2005.14165 . · arXiv
  5. Cambier, L., Bhiwandiwalla, A., Gong, T., Elibol, O. H., Nekuii, M., and Tang, H. (2020). Shifted and squeezed 8-bit floating point format for low-precision training of deep neural networks. In 8th International Conference on Learning Representations, ICLR 2020, Addis Ababa, Ethiopia, April 26-30, 2020 . OpenReview.net.
  6. Chen, J., Gai, Y., Yao, Z., Mahoney, M. W., and Gonzalez, J. E. (2020). A statistical framework for low-bitwidth training of deep neural networks. Advances in Neural Information Processing Systems , 33:883-894.
  7. Choi, J., Venkataramani, S., Srinivasan, V., Gopalakrishnan, K., Wang, Z., and Chuang, P. (2019). Accurate and efficient 2-bit quantized neural networks. In Talwalkar, A., Smith, V., and Zaharia, M., editors, Proceedings of Machine Learning and Systems 2019, MLSys 2019, Stanford, CA, USA, March 31 - April 2, 2019 . mlsys.org.
  8. Courbariaux, M. and Bengio, Y. (2016). Binarynet: Training deep neural networks with weights and activations constrained to +1 or -1. CoRR , abs/1602.02830. · arXiv
  9. Courbariaux, M., Bengio, Y., and David, J. (2015). Binaryconnect: Training deep neural networks with binary weights during propagations. In Cortes, C., Lawrence, N. D., Lee, D. D., Sugiyama, M., and Garnett, R., editors, Advances in Neural Information Processing Systems 28: Annual Conference on Neural Information Processing Systems 2015, December 7-12, 2015, Montreal, Quebec, Canada , pages 3123-3131.
  10. Courbariaux, M., Bengio, Y., and David, J.-P. (2014). Training deep neural networks with low precision multiplications. arXiv preprint arXiv:1412.7024 . · arXiv
  11. Dettmers, T., Lewis, M., Shleifer, S., and Zettlemoyer, L. (2022). 8-bit optimizers via block-wise quantization. 9th International Conference on Learning Representations, ICLR .
  12. Devlin, J., Chang, M.-W., Lee, K., and Toutanova, K. (2018). Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805 . · arXiv
  13. Dong, Z., Yao, Z., Gholami, A., Mahoney, M. W., and Keutzer, K. (2019). Hawq: Hessian aware quantization of neural networks with mixed-precision. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 293-302.
  14. Esser, S. K., McKinstry, J. L., Bablani, D., Appuswamy, R., and Modha, D. S. (2019). Learned step size quantization. arXiv preprint arXiv:1902.08153 . · arXiv
  15. Fan, A., Stock, P., Graham, B., Grave, E., Gribonval, R., Jegou, H., and Joulin, A. (2020). Training with quantization noise for extreme model compression. arXiv preprint arXiv:2004.07320 . · arXiv
  16. Gao, J., He, D., Tan, X., Qin, T., Wang, L., and Liu, T.-Y. (2019). Representation degeneration problem in training natural language generation models. arXiv preprint arXiv:1907.12009 . · arXiv
  17. Gholami, A., Kim, S., Dong, Z., Yao, Z., Mahoney, M. W., and Keutzer, K. (2021). A survey of quantization methods for efficient neural network inference. arXiv preprint arXiv:2103.13630 . · arXiv
  18. Gong, R., Liu, X., Jiang, S., Li, T., Hu, P., Lin, J., Yu, F., and Yan, J. (2019). Differentiable soft quantization: Bridging full-precision and low-bit neural networks. In 2019 IEEE/CVF International Conference on Computer Vision, ICCV 2019, Seoul, Korea (South), October 27 - November 2, 2019 , pages 4851-4860. IEEE.
  19. Henighan, T., Kaplan, J., Katz, M., Chen, M., Hesse, C., Jackson, J., Jun, H., Brown, T. B., Dhariwal, P., Gray, S., et al. (2020). Scaling laws for autoregressive generative modeling. arXiv preprint arXiv:2010.14701 . · arXiv
  20. Hoffmann, J., Borgeaud, S., Mensch, A., Buchatskaya, E., Cai, T., Rutherford, E., Casas, D. d. L., Hendricks, L. A., Welbl, J., Clark, A., et al. (2022). Training compute-optimal large language models. arXiv preprint arXiv:2203.15556 . · arXiv
  21. Ilharco, G., Ilharco, C., Turc, I., Dettmers, T., Ferreira, F., and Lee, K. (2020). High performance natural language processing. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: Tutorial Abstracts , pages 24-27, Online. Association for Computational Linguistics.
  22. Jin, Q., Ren, J., Zhuang, R., Hanumante, S., Li, Z., Chen, Z., Wang, Y., Yang, K., and Tulyakov, S. (2022). F8net: Fixed-point 8-bit only multiplication for network quantization. arXiv preprint arXiv:2202.05239 . · arXiv
  23. Khudia, D., Huang, J., Basu, P., Deng, S., Liu, H., Park, J., and Smelyanskiy, M. (2021). Fbgemm: Enabling high-performance low-precision deep learning inference. arXiv preprint arXiv:2101.05615 . · arXiv
  24. Kovaleva, O., Kulshreshtha, S., Rogers, A., and Rumshisky, A. (2021). Bert busters: Outlier dimensions that disrupt transformers. arXiv preprint arXiv:2105.06990 . · arXiv
  25. Li, R., Wang, Y., Liang, F., Qin, H., Yan, J., and Fan, R. (2019). Fully quantized network for object detection. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019 , pages 2810-2819. Computer Vision Foundation / IEEE.
  26. Lin, Y., Li, Y ., Liu, T., Xiao, T., Liu, T., and Zhu, J. (2020). Towards fully 8-bit integer inference for the transformer model. arXiv preprint arXiv:2009.08034 . · arXiv
  27. Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., Chen, D., Levy, O., Lewis, M., Zettlemoyer, L., and Stoyanov, V. (2019). Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692 . · arXiv
  28. Luo, Z., Kulmizev, A., and Mao, X. (2021). Positional artefacts propagate through masked language model embeddings. In 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) , pages 5312-5327, Online. Association for Computational Linguistics.
  29. Machᡠcek, M. and Bojar, O. (2014). Results of the wmt14 metrics shared task. In Proceedings of the Ninth Workshop on Statistical Machine Translation , pages 293-301.
  30. Mellempudi, N., Srinivasan, S., Das, D., and Kaul, B. (2019). Mixed precision training with 8-bit floating point. CoRR , abs/1905.12334. · arXiv
  31. Ott, M., Edunov, S., Baevski, A., Fan, A., Gross, S., Ng, N., Grangier, D., and Auli, M. (2019). fairseq: A fast, extensible toolkit for sequence modeling. arXiv preprint arXiv:1904.01038 . · arXiv
  32. Ott, M., Edunov, S., Grangier, D., and Auli, M. (2018). Scaling neural machine translation. arXiv preprint arXiv:1806.00187 . · arXiv
  33. Park, G., Park, B., Kwon, S. J., Kim, B., Lee, Y ., and Lee, D. (2022). nuqmm: Quantized matmul for efficient inference of large-scale generative language models. arXiv preprint arXiv:2206.09557 . · arXiv
  34. Puccetti, G., Rogers, A., Drozd, A., and Dell'Orletta, F. (2022). Outliers dimensions that disrupt transformers are driven by frequency. arXiv preprint arXiv:2205.11380 . · arXiv
  35. Qin, H., Gong, R., Liu, X., Bai, X., Song, J., and Sebe, N. (2020). Binary neural networks: A survey. CoRR , abs/2004.03333. · arXiv
  36. Radford, A., Wu, J., Child, R., Luan, D., Amodei, D., and Sutskever, I. (2019). Language models are unsupervised multitask learners. OpenAI blog , 1(8):9.
  37. Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., and Liu, P. J. (2019). Exploring the limits of transfer learning with a unified text-to-text transformer. arXiv preprint arXiv:1910.10683 . · arXiv
  38. Rastegari, M., Ordonez, V., Redmon, J., and Farhadi, A. (2016). Xnor-net: Imagenet classification using binary convolutional neural networks. In Leibe, B., Matas, J., Sebe, N., and Welling, M., editors, Computer Vision - ECCV 2016 - 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part IV , volume 9908 of Lecture Notes in Computer Science , pages 525-542. Springer.
  39. Sennrich, R., Haddow, B., and Birch, A. (2016). Edinburgh neural machine translation systems for wmt 16. arXiv preprint arXiv:1606.02891 . · arXiv
  40. Shazeer, N., Cheng, Y., Parmar, N., Tran, D., Vaswani, A., Koanantakool, P., Hawkins, P., Lee, H., Hong, M., Young, C., et al. (2018). Mesh-tensorflow: Deep learning for supercomputers. Advances in neural information processing systems , 31.
  41. Shen, S., Dong, Z., Ye, J., Ma, L., Yao, Z., Gholami, A., Mahoney, M. W., and Keutzer, K. (2020). Q-bert: Hessian based ultra low precision quantization of bert. In Proceedings of the AAAI Conference on Artificial Intelligence , volume 34, pages 8815-8821.
  42. Sun, X., Choi, J., Chen, C., Wang, N., Venkataramani, S., Srinivasan, V., Cui, X., Zhang, W., and Gopalakrishnan, K. (2019). Hybrid 8-bit floating point (HFP8) training and inference for deep neural networks. In Wallach, H. M., Larochelle, H., Beygelzimer, A., d'Alché-Buc, F., Fox, E. B., and Garnett, R., editors, Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems 2019, NeurIPS 2019, December 8-14, 2019, Vancouver, BC, Canada , pages 4901-4910.
  43. Timkey, W. and van Schijndel, M. (2021). All bark and no bite: Rogue dimensions in transformer language models obscure representational quality. arXiv preprint arXiv:2109.04404 . · arXiv
  44. Trinh, T. H. and Le, Q. V. (2018). A simple method for commonsense reasoning. arXiv preprint arXiv:1806.02847 . · arXiv
  45. Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., and Polosukhin, I. (2017). Attention is all you need. arXiv preprint arXiv:1706.03762 . · arXiv
  46. Wang, N., Choi, J., Brand, D., Chen, C., and Gopalakrishnan, K. (2018). Training deep neural networks with 8-bit floating point numbers. In Bengio, S., Wallach, H. M., Larochelle, H., Grauman, K., Cesa-Bianchi, N., and Garnett, R., editors, Advances in Neural Information Processing Systems 31: Annual Conference on Neural Information Processing Systems 2018, NeurIPS 2018, December 3-8, 2018, Montréal, Canada , pages 7686-7695.
  47. Wei, X., Zhang, Y., Zhang, X., Gong, R., Zhang, S., Zhang, Q., Yu, F., and Liu, X. (2022). Outlier suppression: Pushing the limit of low-bit transformer language models. arXiv preprint arXiv:2209.13325 . · arXiv
  48. Wenzek, G., Lachaux, M.-A., Conneau, A., Chaudhary, V., Guzmán, F., Joulin, A., and Grave, E. (2020). CCNet: Extracting high quality monolingual datasets from web crawl data. In Proceedings of the 12th Language Resources and Evaluation Conference , pages 4003-4012, Marseille, France. European Language Resources Association.
  49. Wolf, T., Debut, L., Sanh, V., Chaumond, J., Delangue, C., Moi, A., Cistac, P., Rault, T., Louf, R., Funtowicz, M., et al. (2019). Huggingface's transformers: State-of-the-art natural language processing. arXiv preprint arXiv:1910.03771 . · arXiv
  50. Wu, H., Judd, P., Zhang, X., Isaev, M., and Micikevicius, P. (2020). Integer quantization for deep learning inference: Principles and empirical evaluation. arXiv preprint arXiv:2004.09602 . · arXiv
  51. Yao, Z., Aminabadi, R. Y., Zhang, M., Wu, X., Li, C., and He, Y. (2022). Zeroquant: Efficient and affordable post-training quantization for large-scale transformers. arXiv preprint arXiv:2206.01861 . · arXiv
  52. Yao, Z., Dong, Z., Zheng, Z., Gholami, A., Yu, J., Tan, E., Wang, L., Huang, Q., Wang, Y., Mahoney, M., et al. (2021). Hawq-v3: Dyadic neural network quantization. In International Conference on Machine Learning , pages 11875-11886. PMLR.
  53. Zafrir, O., Boudoukh, G., Izsak, P., and Wasserblat, M. (2019). Q8bert: Quantized 8bit bert. In 2019 Fifth Workshop on Energy Efficient Machine Learning and Cognitive Computing-NeurIPS Edition (EMC2-NIPS) , pages 36-39. IEEE.
  54. Zeng, A., Liu, X., Du, Z., Wang, Z., Lai, H., Ding, M., Yang, Z., Xu, Y., Zheng, W., Xia, X., et al. (2022). Glm-130b: An open bilingual pre-trained model. arXiv preprint arXiv:2210.02414 . · arXiv
  55. Zhang, D., Yang, J., Ye, D., and Hua, G. (2018). Lq-nets: Learned quantization for highly accurate and compact deep neural networks. In Proceedings of the European conference on computer vision (ECCV) , pages 365-382.
  56. Zhang, S., Roller, S., Goyal, N., Artetxe, M., Chen, M., Chen, S., Dewan, C., Diab, M., Li, X., Lin, X. V., et al. (2022). Opt: Open pre-trained transformer language models. arXiv preprint arXiv:2205.01068 . · arXiv
  57. Zhang, W., Hou, L., Yin, Y., Shang, L., Chen, X., Jiang, X., and Liu, Q. (2020). Ternarybert: Distillation-aware ultra-low bit bert. In EMNLP .
  58. Zhao, C., Hua, T., Shen, Y., Lou, Q., and Jin, H. (2021). Automatic mixed-precision quantization search of bert. arXiv preprint arXiv:2112.14938 . · arXiv
  59. Zhu, C., Han, S., Mao, H., and Dally, W. J. (2017). Trained ternary quantization. In 5th International Conference on Learning Representations, ICLR 2017, Toulon, France, April 24-26, 2017, Conference Track Proceedings . OpenReview.net.
  60. Zhu, Y., Kiros, R., Zemel, R., Salakhutdinov, R., Urtasun, R., Torralba, A., and Fidler, S. (2015). Aligning books and movies: Towards story-like visual explanations by watching movies and reading books. In Proceedings of the IEEE international conference on computer vision , pages 19-27.

Cited by

Sorted by citations

Authors
Dettmers, T., Lewis, M., Belkada, Y., & Zettlemoyer, L.
ORCID
Zettlemoyer, Luke
Venue
Advances in Neural Information Processing Systems 35
Publisher
Neural Information Processing Systems Foundation, Inc. (NeurIPS)
Published
2022
Pages
30318-30332
ISSN
1049-5258 (ISSN-L)
DOI
10.52202/068431-2198
OpenAlex
W7133198810
Cited by
305
References
0
Citations by year
2026: 183 · 2025: 37 · 2024: 43 · 2023: 41 · 2022: 1
FWCI
11.06
Type
conference paper
Language
en
Open access
Locations
  • neural information processing systems (publishedVersion)
  • arXiv (Cornell University) (publishedVersion)
  • arXiv (Cornell University) (cc-by)
Topics
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
  • Physical Sciences › Computer Science › Hardware and Architecture › Parallel Computing and Optimization Techniques
  • Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
Keywords
low-bit quantization, transformer models, large language models, mixed-precision training

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