{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,23]],"date-time":"2026-01-23T15:49:26Z","timestamp":1769183366560,"version":"3.49.0"},"reference-count":69,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"2","license":[{"start":{"date-parts":[[2026,2,1]],"date-time":"2026-02-01T00:00:00Z","timestamp":1769904000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2026,2,1]],"date-time":"2026-02-01T00:00:00Z","timestamp":1769904000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,2,1]],"date-time":"2026-02-01T00:00:00Z","timestamp":1769904000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"name":"Research Grants Council of Hong Kong, SAR","award":["CUHK14208021"],"award-info":[{"award-number":["CUHK14208021"]}]},{"name":"MIND Project","award":["MINDXZ202404"],"award-info":[{"award-number":["MINDXZ202404"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Comput.-Aided Des. Integr. Circuits Syst."],"published-print":{"date-parts":[[2026,2]]},"DOI":"10.1109\/tcad.2025.3585023","type":"journal-article","created":{"date-parts":[[2025,7,1]],"date-time":"2025-07-01T13:39:10Z","timestamp":1751377150000},"page":"929-942","source":"Crossref","is-referenced-by-count":0,"title":["Oiso: Outlier-Isolated Data Format for Low-Bit Large Language Model Quantization"],"prefix":"10.1109","volume":"45","author":[{"ORCID":"https:\/\/orcid.org\/0009-0004-6820-7064","authenticated-orcid":false,"given":"Lancheng","family":"Zou","sequence":"first","affiliation":[{"name":"Department of Computer Science and Engineering, The Chinese University of Hong Kong, Shatin, SAR, Hong Kong"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4927-0194","authenticated-orcid":false,"given":"Shuo","family":"Yin","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Engineering, The Chinese University of Hong Kong, Shatin, SAR, Hong Kong"}]},{"ORCID":"https:\/\/orcid.org\/0009-0007-6740-8413","authenticated-orcid":false,"given":"Mingjun","family":"Li","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Engineering, The Chinese University of Hong Kong, Shatin, SAR, Hong Kong"}]},{"given":"Mingzi","family":"Wang","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Engineering, The Chinese University of Hong Kong, Shatin, SAR, Hong Kong"}]},{"given":"Chen","family":"Bai","sequence":"additional","affiliation":[{"name":"Department of Electronic and Computer Engineering, The Hong Kong University of Science and Technology, Sai Kung, Hong Kong"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9501-9254","authenticated-orcid":false,"given":"Wenqian","family":"Zhao","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Engineering, The Chinese University of Hong Kong, Shatin, SAR, Hong Kong"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6406-4810","authenticated-orcid":false,"given":"Bei","family":"Yu","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Engineering, The Chinese University of Hong Kong, Shatin, SAR, Hong Kong"}]}],"member":"263","reference":[{"key":"ref1","first-page":"1877","article-title":"Language models are few-shot learners","volume-title":"Proc. 34th NeurIPS","author":"Brown"},{"key":"ref2","article-title":"OPT: Open pre-trained transformer language models","author":"Zhang","year":"2022","journal-title":"arXiv:2205.01068"},{"key":"ref3","article-title":"Llama 2: Open foundation and fine-tuned chat models","author":"Touvron","year":"2023","journal-title":"arXiv:2307.09288"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1016\/j.joule.2023.09.004"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00286"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR42600.2020.01318"},{"key":"ref7","first-page":"1","article-title":"BRECQ: Pushing the limit of post-training quantization by block reconstruction","volume-title":"Proc. ICLR","author":"Li"},{"key":"ref8","first-page":"1","article-title":"QDrop: Randomly dropping quantization for extremely low-bit post-training quantization","volume-title":"Proc. ICLR","author":"Wei"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/MICRO.2014.58"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2016.2616357"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1109\/ISCA.2018.00069"},{"key":"ref12","first-page":"382","article-title":"Bit-pragmatic deep neural network computing","volume-title":"Proc. 50th IEEE\/ACM MICRO","author":"Albericio"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1145\/3079856.3080246"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/MICRO50266.2020.00071"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/ISCA.2018.00063"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1109\/MICRO56248.2022.00095"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1145\/3579371.3589038"},{"key":"ref18","first-page":"1","article-title":"BiScaled-DNN: Quantizing long-tailed datastructures with two scale factors for deep neural networks","volume-title":"Proc. 56th ACM\/IEEE DAC","author":"Jain"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/ISCA45697.2020.00086"},{"key":"ref20","first-page":"10271","article-title":"Pushing the limits of narrow precision inferencing at cloud scale with Microsoft floating point","volume-title":"Proc. 34th NeurIPS","author":"Rouhani"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.emnlp-main.617"},{"key":"ref22","first-page":"62978","article-title":"BiE: Bi-exponent block floating-point for large language models quantization","volume-title":"Proc. 41st ICML","author":"Zou"},{"key":"ref23","article-title":"Accurate block quantization in LLMs with outliers","author":"Trukhanov","year":"2024","journal-title":"arXiv:2403.20137"},{"key":"ref24","first-page":"1","article-title":"OPAL: Outlier-preserved microscaling quantization accelerator for generative large language models","volume-title":"Proc. 61st ACM\/IEEE DAC","author":"Koo"},{"key":"ref25","article-title":"LLM.int8(): 8-bit matrix multiplication for transformers at scale","author":"Dettmers","year":"2022","journal-title":"arXiv:2208.07339"},{"key":"ref26","first-page":"38087","article-title":"SmoothQuant: Accurate and efficient post-training quantization for large language models","volume-title":"Proc. 40th ICML","author":"Xiao"},{"key":"ref27","article-title":"Outlier suppression+: Accurate quantization of large language models by equivalent and optimal shifting and scaling","author":"Wei","year":"2023","journal-title":"arXiv:2304.09145"},{"key":"ref28","article-title":"Microscaling data formats for deep learning","author":"Rouhani","year":"2023","journal-title":"arXiv:2310.10537"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v32i1.11334"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA53966.2022.00067"},{"key":"ref31","first-page":"1002","article-title":"Bucket getter: A bucket-based processing engine for low-bit block floating point (BFP) DNNs","volume-title":"Proc. 56th Annu. IEEE\/ACM MICRO","author":"Lo"},{"key":"ref32","volume-title":"d-Matrix Jayhawk II Architecture","year":"2024"},{"key":"ref33","volume-title":"Tenstorrent Grayskull Architecture","year":"2024"},{"key":"ref34","volume-title":"Tenstorrent Wormhole Architecture","year":"2024"},{"key":"ref35","volume-title":"AMD Ryzen AI 300 Series Architecture","year":"2024"},{"key":"ref36","article-title":"Deep compression: Compressing deep neural networks with pruning, trained quantization and Huffman coding","author":"Han","year":"2016","journal-title":"arXiv:1510.00149"},{"key":"ref37","first-page":"1","article-title":"Learning both weights and connections for efficient neural network","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Han"},{"key":"ref38","first-page":"10323","article-title":"SparseGPT: Massive language models can be accurately pruned in one-shot","volume-title":"Proc. 40th ICML","author":"Frantar"},{"key":"ref39","first-page":"21702","article-title":"LLM-pruner: On the structural pruning of large language models","volume-title":"Proc. 37th NeurIPS","author":"Ma"},{"key":"ref40","first-page":"1","article-title":"A simple and effective pruning approach for large language models","volume-title":"Proc. ICLR","author":"Sun"},{"key":"ref41","first-page":"7950","article-title":"Post training 4-bit quantization of convolutional networks for rapid-deployment","volume-title":"Proc. 33rd Int. Conf. Neural Inf. Process. Syst.","author":"Banner"},{"key":"ref42","first-page":"1","article-title":"PyTorch: An imperative style, high-performance deep learning library","volume-title":"Proc. 33rd NeurIPS","author":"Paszke"},{"key":"ref43","article-title":"The Llama 3 herd of models","author":"Grattafiori","year":"2024","journal-title":"arXiv:2407.21783"},{"key":"ref44","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.emnlp-demos.6"},{"key":"ref45","article-title":"Pointer sentinel mixture models","author":"Merity","year":"2016","journal-title":"arXiv:1609.07843"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P16-1144"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v34i05.6239"},{"key":"ref48","article-title":"Think you have solved question answering? Try arc, the AI2 reasoning challenge","author":"Clark","year":"2018","journal-title":"arXiv:1803.05457"},{"key":"ref49","first-page":"90","article-title":"Choice of plausible alternatives: An evaluation of commonsense causal reasoning","volume-title":"Proc. AAAI Spring Symp. Ser.","author":"Roemmele"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/w18-5446"},{"key":"ref51","volume-title":"A framework for few-shot language model evaluation","author":"Gao","year":"2023"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.1145\/2228360.2228584"},{"key":"ref53","volume-title":"Synopsys Design Compiler: Concurrent Timing, Area, Power, and Test Optimization","year":"2024"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.1016\/j.mejo.2016.04.006"},{"key":"ref55","first-page":"28","article-title":"Cacti 6.0: A tool to model large caches","volume":"27","author":"Muralimanohar","year":"2009","journal-title":"HP Lab."},{"key":"ref56","doi-asserted-by":"publisher","DOI":"10.1109\/DAC18072.2020.9218516"},{"key":"ref57","article-title":"The pile: An 800GB dataset of diverse text for language modeling","author":"Gao","year":"2020","journal-title":"arXiv:2101.00027"},{"key":"ref58","first-page":"27168","article-title":"ZeroQuant: Efficient and affordable post-training quantization for large-scale transformers","volume-title":"Proc. 36th NeurIPS","author":"Yao"},{"key":"ref59","article-title":"OmniQuant: Omnidirectionally calibrated quantization for large language models","author":"Shao","year":"2024","journal-title":"arXiv:2308.13137"},{"key":"ref60","article-title":"GPTQ: Accurate post-training quantization for generative pre-trained transformers","author":"Frantar","year":"2023","journal-title":"arXiv:2210.17323"},{"key":"ref61","article-title":"AWQ: Activation-aware weight quantization for LLM compression and acceleration","author":"Lin","year":"2024","journal-title":"arXiv:2306.00978"},{"key":"ref62","first-page":"20023","article-title":"BiLLM: Pushing the limit of post-training quantization for LLMs","volume-title":"Proc. 41st ICML","author":"Huang"},{"key":"ref63","first-page":"1","article-title":"BiBERT: Accurate fully binarized BERT","volume-title":"Proc. ICLR","author":"Qin"},{"key":"ref64","first-page":"1","article-title":"ARB-LLM: Alternating refined binarizations for large language models","volume-title":"Proc. ICLR","author":"Li"},{"key":"ref65","first-page":"41498","article-title":"Accurate LoRA-finetuning quantization of LLMs via information retention","volume-title":"Proc. 41st ICML","author":"Qin"},{"key":"ref66","first-page":"10088","article-title":"QLoRA: Efficient finetuning of quantized LLMs","volume-title":"Proc. 37th Adv. Neural Inf. Process. Syst.","author":"Dettmers"},{"key":"ref67","doi-asserted-by":"publisher","DOI":"10.1109\/PROC.1972.8820"},{"key":"ref68","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2019.2913958"},{"key":"ref69","first-page":"25490","article-title":"Be like water: Adaptive floating point for machine learning","volume-title":"Proc. 39th ICML","author":"Yeh"}],"container-title":["IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/43\/11360509\/11062568.pdf?arnumber=11062568","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,1,22]],"date-time":"2026-01-22T21:02:20Z","timestamp":1769115740000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11062568\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,2]]},"references-count":69,"journal-issue":{"issue":"2"},"URL":"https:\/\/doi.org\/10.1109\/tcad.2025.3585023","relation":{},"ISSN":["0278-0070","1937-4151"],"issn-type":[{"value":"0278-0070","type":"print"},{"value":"1937-4151","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,2]]}}}