[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-327713-105":59,"doc-detail-327713-en":131},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":124,"head_meta":126,"extra_data":128,"updated_unix":130},105,"en","spectral-signatures-of-large-language-models-paper","Spectral Signatures of Large Language Models - Paper","","Rapid growth of public large language models (LLMs) creates major obstacles for systematic management and quantification, including lineage tracing, licensing, and consistent evaluation. Task-specific benchmarks are not sufficient because LLMs vary greatly in architectures, scales, and training procedures. The work introduces spectral shape-based metrics grounded in Heavy-Tailed Self-Regularization theory, using weight empirical spectral density as compact, data-free signatures robust to post-training changes. The signature enables large-scale benchmarking, lineage tracking, unsupervised clustering, and performance quantification across diverse open-source model families, supported by available code.",{"@graph":69,"@context":123},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/spectral-signatures-of-large-language-models-paper/327713/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/spectral-signatures-of-large-language-models-paper/327713.png","ImageObject",300,407,{"name":92,"@type":93},"Oliver","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-24","2026-09-21",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",12,{"@type":107,"mainEntity":108},"FAQPage",[109,115,119],{"name":110,"@type":111,"acceptedAnswer":112},"What problem does the paper address for large language models (LLMs)?","Question",{"text":113,"@type":114},"It addresses the difficulty of systematically managing and quantifying LLMs at scale, including tasks like model lineage tracing, licensing, and evaluation consistency. It argues that task-specific benchmarks are insufficient due to wide diversity in LLM architectures and training.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"What is the core idea of the proposed method?",{"text":118,"@type":114},"The method adopts spectral shape-based metrics derived from Heavy-Tailed Self-Regularization theory. It uses the shape of the weight empirical spectral density as a compact spectral signature for each model.",{"name":120,"@type":111,"acceptedAnswer":121},"What can the spectral signature be used for?",{"text":122,"@type":114},"The spectral signature supports tracking model lineage, unsupervised clustering of similar models, and quantifying model performance across models and downstream tasks. The paper also emphasizes that the metric is data-free, computationally efficient, and scale-invariant.","https://schema.org",{"og:url":83,"og:type":125,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":127,"canonical":83},"index,follow",{"doc_id":129,"site_id":62},327713,1790269526,{"code":4,"msg":5,"data":132},{"doc_id":129,"user_id":133,"nickname":92,"user_avatar":134,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":105,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":144,"read_time":41,"preview_page_count":4},8796095461610,"https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c","Spectral Signatures of Large Language Models  \narXiv :2607 .03377v 1 [ cs .CL] 3 Jul 2026  \nZhuoying Zhang∗ [zz183@rice.edu](zz183@rice.edu)[ ](zz183@rice.edu)Rice University Houston, TX, USA  \nZihang Liu  \n[zihang.liu@berkeley.edu](zihang.liu@berkeley.edu)[ ](zihang.liu@berkeley.edu)University of California, Berkeley Berkeley, CA, USA  \nIshan V. Prasad∗  \n[ishan.v.prasad.26@dartmouth.edu](ishan.v.prasad.26@dartmouth.edu)[ ](ishan.v.prasad.26@dartmouth.edu)Dartmouth College Hanover, NH, USA  \nHengrui Luo  \n[hl180@rice.edu](hl180@rice.edu)[ ](hl180@rice.edu)Rice University Houston, TX, USA  \nYuanzhe Hu  \n[yzhu.ml@outlook.com](yzhu.ml@outlook.com)[ ](yzhu.ml@outlook.com)Georgia Institute of Technology Atlanta, GA, USA  \nPu Ren† [pren@lbl.gov](pren@lbl.gov)[ ](pren@lbl.gov)Lawrence Berkeley National Laboratory Berkeley, CA, USA  \nYaoqing Yang  \n[yaoqing.yang@dartmouth.edu](yaoqing.yang@dartmouth.edu)[ ](yaoqing.yang@dartmouth.edu)Dartmouth College Hanover, NH, USA  \nAbstract  \nThe rapidly growing repository of publicly available large language models (LLMs) presents significant challenges for systematic management and quantification at scale, such as model lineage tracing, licensing, and evaluation. However, task-specific benchmarks are insufficient for this setting, as LLMs differ widely in architectures, scales, and training procedures. To address this challenge, we adopt spectral shape-based metrics for managing and quantifying LLMs based on Heavy-Tailed Self-Regularization theory. Our approach uses the shape information of the weight empirical spectral density as a compact spectral signature of each model. This signature captures intrinsic properties of pretrained models and remains robust during post-training, making it suitable for model-level analysis. In addition, this metric is data-free, computationally-efficient, and scale-invariant, enabling large-scale analysis in practice. Moreover, we curate a large and diverse model corpus consisting of major open-source LLM families, and use it to systematically benchmark spectral and non-spectral metrics across models and downstream tasks. We show that our spectral signature supports the tracking of the model lineage, the unsupervised clustering of similar models, and the quantification of the model performance. Overall, the proposed spectral signature provides a meaningful proxy for broad performance trends across LLMs, enabling efficient organization, comparison, and analysis of large model collections. The code is available at [https://github.com/Ingrid-505/Spectral_Signature](https://github.com/Ingrid-505/Spectral_Signature).  \nKeywords  \nLarge language models; Model lineage; Model ranking; Weight space learning  \n∗ Both authors contribute equally to this research.†Corresponding author.  \n[https://creativecommons.org/licenses/by/4.0This work is](https://creativecommons.org/licenses/by/4.0This work is) licensed under a Creative Commons Attribution 4 .0 International License.  \nKDD’26, Jeju Island, Republic of Korea  \n© 2026 Copyright held by the owner/author(s) .  \nACM ISBN 979-8-4007-2259-2/2026/08  \n[https://doi.org/10.1145/3770855.3818090](https://doi.org/10.1145/3770855.3818090)  \nACM Reference Format:  \nZhuoying Zhang, Ishan V. Prasad, Yuanzhe Hu, Zihang Liu, Hengrui Luo, Pu Ren, and Yaoqing Yang. 2026. Spectral Signatures of Large Language Models. In Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2 (KDD’26), August 09–13, 2026, Jeju Island, Republic of Korea. ACM, New York, NY, USA, 12 pages. [https://doi.org/10](https://doi.org/10) . 1145/3770855.3818090  \n1 Introduction  \nPublicly available large language models (LLMs) have been growing rapidly in recent years, with hundreds of thousands of models now available through open-source platforms such as Hugging Face (HF) [59] . These models exhibit significant diversity in their training trajectories, including variations in model architecture, parameter scale, training data, and emp","cbCaiubEaJM6oBGi","https://ap.wps.com/l/cbCaiubEaJM6oBGi","pdf",3261565,"English","# Abstract\n# Introduction\n## Motivation\n## Spectral shape-based perspective\n## Spectral signature metric","[{\"question\":\"What problem does the paper address for large language models (LLMs)?\",\"answer\":\"It addresses the difficulty of systematically managing and quantifying LLMs at scale, including tasks like model lineage tracing, licensing, and evaluation consistency. It argues that task-specific benchmarks are insufficient due to wide diversity in LLM architectures and training.\"},{\"question\":\"What is the core idea of the proposed method?\",\"answer\":\"The method adopts spectral shape-based metrics derived from Heavy-Tailed Self-Regularization theory. It uses the shape of the weight empirical spectral density as a compact spectral signature for each model.\"},{\"question\":\"What can the spectral signature be used for?\",\"answer\":\"The spectral signature supports tracking model lineage, unsupervised clustering of similar models, and quantifying model performance across models and downstream tasks. The paper also emphasizes that the metric is data-free, computationally efficient, and scale-invariant.\"}]","Spectral Signatures of Large Language Models - Paper | PDF",1789977891]