<?xml version='1.0' encoding='UTF-8'?><?xml-stylesheet href='static/style.xsl' type='text/xsl'?><OAI-PMH xmlns="http://www.openarchives.org/OAI/2.0/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/ http://www.openarchives.org/OAI/2.0/OAI-PMH.xsd"><responseDate>2026-11-22T10:56:16Z</responseDate><request verb="GetRecord" identifier="oai:ecommons.cornell.edu:1813/117157" metadataPrefix="dim">https://ecommons.cornell.edu/server/oai/request</request><GetRecord><record><header><identifier>oai:ecommons.cornell.edu:1813/117157</identifier><datestamp>2026-05-15T19:44:40Z</datestamp><setSpec>com_1813_35</setSpec><setSpec>col_1813_47</setSpec></header><metadata><dim:dim xmlns:dim="http://www.dspace.org/xmlns/dspace/dim" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:doc="http://www.lyncode.com/xoai" xsi:schemaLocation="http://www.dspace.org/xmlns/dspace/dim http://www.dspace.org/schema/dim.xsd">
   <dim:field mdschema="dc" element="contributor" qualifier="author">Yan, Jing Nathan</dim:field>
   <dim:field mdschema="dc" element="contributor" qualifier="chair" lang="en_US">Rzeszotarski, Jeffrey</dim:field>
   <dim:field mdschema="dc" element="contributor" qualifier="committeeMember" lang="en_US">Rush, Alexander</dim:field>
   <dim:field mdschema="dc" element="contributor" qualifier="committeeMember" lang="en_US">Fussell, Susan</dim:field>
   <dim:field mdschema="dc" element="contributor" qualifier="committeeMember" lang="en_US">Joachims, Thorsten</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="accessioned">2025-06-30T22:03:00Z</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="available">2025-06-30T22:03:00Z</dim:field>
   <dim:field mdschema="dc" element="date" qualifier="issued">2024-12</dim:field>
   <dim:field mdschema="dc" element="identifier" qualifier="other">ProQuest Submission ID: 14786</dim:field>
   <dim:field mdschema="dc" element="identifier" qualifier="other">ProQuest Publication ID: 31641537</dim:field>
   <dim:field mdschema="dc" element="identifier" qualifier="uri">https://hdl.handle.net/1813/117157</dim:field>
   <dim:field mdschema="dc" element="identifier" qualifier="doi">http://doi.org/10.7298/5ceh-ch09</dim:field>
   <dim:field mdschema="dc" element="identifier" qualifier="bibid">16922036</dim:field>
   <dim:field mdschema="dc" element="description" lang="en_US">309 pages</dim:field>
   <dim:field mdschema="dc" element="description" qualifier="abstract" lang="en_US">The rapid evolution of generative models has significantly advanced artificial intelligence, enabling the creation of human-like text, realistic images, and even scientific discoveries. Powered by transformer architectures and vast datasets, these models demonstrate remarkable proficiency across various domains. However, as these models become increasingly large and sophisticated, they face substantial computational challenges. The quadratic complexity inherent in attention mechanisms during inference poses a significant bottleneck, restricting practical deployment—especially in resource-constrained environments. Consequently, despite their impressive capabilities, the scalability and accessibility of current generative models remain constrained. This thesis addresses these challenges by designing new models and inference paradigms for future generative models that retain or surpass high-quality outputs while significantly improving computational efficiency. We investigate how increasing context complexity impacts computational efficiency and model performance, identifying bottlenecks in current architectures (RQ1). Leveraging architectural innovations and advanced model compression techniques, we develop novel models that optimize inference for detailed and high-resolution outputs without incurring prohibitive computational costs (RQ2). We adapt these models for deployment in resource-constrained environments without sacrificing performance or capabilities, broadening their accessibility and practicality (RQ3). Additionally, we enhance interpretability and usability by studying how users make sense of interactions with generative models, ensuring fairness and facilitating meaningful human engagement in user-facing applications (RQ4). Our contributions advance the design of inference paradigms for future generative models, paving the way for scalable, efficient, and user-friendly systems that bridge the gap between advanced capabilities and real-world deployment.</dim:field>
   <dim:field mdschema="dc" element="language" qualifier="iso">en</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en_US">Computational Efficiency</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en_US">Generative Models</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en_US">Human Computer Interaction</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en_US">Model Architecture</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en_US">Natural Language Processing</dim:field>
   <dim:field mdschema="dc" element="subject" lang="en_US">Usability</dim:field>
   <dim:field mdschema="dc" element="title" lang="en_US">Designing for Inference in Future Generative Models</dim:field>
   <dim:field mdschema="dc" element="type" lang="en_US">dissertation or thesis</dim:field>
   <dim:field mdschema="dc" element="relation" qualifier="localuri">https://newcatalog.library.cornell.edu/catalog/16922036</dim:field>
   <dim:field mdschema="dc" element="format" qualifier="mimetype">application/pdf</dim:field>
   <dim:field mdschema="thesis" element="degree" qualifier="discipline">Information Science</dim:field>
   <dim:field mdschema="thesis" element="degree" qualifier="grantor">Cornell University</dim:field>
   <dim:field mdschema="thesis" element="degree" qualifier="level">Doctor of Philosophy</dim:field>
   <dim:field mdschema="thesis" element="degree" qualifier="name">Ph. D., Information Science</dim:field>
   <dim:field mdschema="dcterms" element="license">https://hdl.handle.net/1813/59810.2</dim:field>
   <dim:field mdschema="dspace" element="entity" qualifier="type">Publication</dim:field>
   <dim:field mdschema="cris" element="virtual" qualifier="collection" authority="https://cornell-ecommons.eks.prod.4science.cloud/handle/1813/47" confidence="600">Cornell Theses and Dissertations</dim:field>
   <dim:field mdschema="cris" element="virtual" qualifier="author">Yan, Jing Nathan</dim:field>
   <dim:field mdschema="cris" element="virtualsource" qualifier="collection">5893a6ea-7af3-41d7-abc6-04bcd26ab5df</dim:field>
   <dim:field mdschema="others" element="access-status">open.access</dim:field>
   <dim:field mdschema="others" element="access-status">open.access</dim:field>
   <dim:field mdschema="cerif" element="openaire" authority="" confidence="-1">&lt;Publication xmlns="https://www.openaire.eu/cerif-profile/1.1/" id="229be33f-2429-465b-951e-513c24ea21d5">
	&lt;Type xmlns="https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types">http://purl.org/coar/resource_type/c_1843&lt;/Type>
	&lt;Language>en&lt;/Language>
   	&lt;Title>Designing for Inference in Future Generative Models&lt;/Title>
   	&lt;PublishedIn>
    	&lt;Publication>
      	&lt;/Publication>
   	&lt;/PublishedIn>
   	&lt;PublicationDate>2024-12&lt;/PublicationDate>
   	&lt;DOI>http://doi.org/10.7298/5ceh-ch09&lt;/DOI>
   	&lt;Authors>
      	&lt;Author>
        	&lt;DisplayName>Yan, Jing Nathan&lt;/DisplayName>
         	&lt;Affiliation>
         		&lt;OrgUnit>
         		&lt;/OrgUnit>
         	&lt;/Affiliation>
      	&lt;/Author>
	&lt;/Authors>
   	&lt;Editors>
	&lt;/Editors>
    &lt;Publishers>
        &lt;Publisher>
            &lt;OrgUnit />
        &lt;/Publisher>
    &lt;/Publishers>
    &lt;Keyword>Computational Efficiency&lt;/Keyword>
    &lt;Keyword>Generative Models&lt;/Keyword>
    &lt;Keyword>Human Computer Interaction&lt;/Keyword>
    &lt;Keyword>Model Architecture&lt;/Keyword>
    &lt;Keyword>Natural Language Processing&lt;/Keyword>
    &lt;Keyword>Usability&lt;/Keyword>
   	&lt;Abstract>The rapid evolution of generative models has significantly advanced artificial intelligence, enabling the creation of human-like text, realistic images, and even scientific discoveries. Powered by transformer architectures and vast datasets, these models demonstrate remarkable proficiency across various domains. However, as these models become increasingly large and sophisticated, they face substantial computational challenges. The quadratic complexity inherent in attention mechanisms during inference poses a significant bottleneck, restricting practical deployment—especially in resource-constrained environments. Consequently, despite their impressive capabilities, the scalability and accessibility of current generative models remain constrained. This thesis addresses these challenges by designing new models and inference paradigms for future generative models that retain or surpass high-quality outputs while significantly improving computational efficiency. We investigate how increasing context complexity impacts computational efficiency and model performance, identifying bottlenecks in current architectures (RQ1). Leveraging architectural innovations and advanced model compression techniques, we develop novel models that optimize inference for detailed and high-resolution outputs without incurring prohibitive computational costs (RQ2). We adapt these models for deployment in resource-constrained environments without sacrificing performance or capabilities, broadening their accessibility and practicality (RQ3). Additionally, we enhance interpretability and usability by studying how users make sense of interactions with generative models, ensuring fairness and facilitating meaningful human engagement in user-facing applications (RQ4). Our contributions advance the design of inference paradigms for future generative models, paving the way for scalable, efficient, and user-friendly systems that bridge the gap between advanced capabilities and real-world deployment.&lt;/Abstract>
	&lt;Access xmlns="http://purl.org/coar/access_right" 
    >
    &lt;/Access>
&lt;/Publication>
</dim:field>
</dim:dim>
</metadata></record></GetRecord></OAI-PMH>