본문

서브메뉴

Interpretation Errors: Extracting Functionality From Generative Models of Language by Understanding Them Better- [electronic resource]
Interpretation Errors: Extracting Functionality From Generative Models of Language by Unde...
Interpretation Errors: Extracting Functionality From Generative Models of Language by Understanding Them Better- [electronic resource]

상세정보

자료유형  
 학위논문파일 국외
최종처리일시  
20240214101651
ISBN  
9798380328883
DDC  
004
저자명  
Holtzman, Ari.
서명/저자  
Interpretation Errors: Extracting Functionality From Generative Models of Language by Understanding Them Better - [electronic resource]
발행사항  
[S.l.]: : University of Washington., 2023
발행사항  
Ann Arbor : : ProQuest Dissertations & Theses,, 2023
형태사항  
1 online resource(129 p.)
주기사항  
Source: Dissertations Abstracts International, Volume: 85-03, Section: A.
주기사항  
Advisor: Zettlemoyer, Luke.
학위논문주기  
Thesis (Ph.D.)--University of Washington, 2023.
사용제한주기  
This item must not be sold to any third party vendors.
초록/해제  
요약The rise of large language models as the workhorse of NLP, and the continuous release of better models (OpenAI, 2023; Pichai, 2023; Schulman et al., 2022, inter alia) has created a strange situation: we have models that are more powerful language generators than ever before, but since we did not design them for a specific purpose we struggle to understand how they should be used or what their idiosyncracies are.This dissertation describes three empirical projects that sought to characterize the underlying behavior of language models and, importantly, to make them more reliable tools for generating and selecting text where this behavior does not match up with the tasks we would like models to complete. Each project attempts to understand what language models and accompanying inference methods currently optimize for, to characterize the gap between that and the true objective of a potential user, and to close it with some new inference method. An emergent theme through these works is that models are already doing what we trained them to do quite well-and it is often the experimenters and practitioners who misunderstand precisely what we trained models to do in the first place. We conclude with a conceptual analysis of how we should study generative models going forward-as models keep improving and new, unanticipated uses and misuses become ever more available.The first half of this dissertation concerns two works, Neural Text Degeneration and Surface Form Competition-two failure modes of generative models that occur when probability is viewed as equivalent to "correctness" in text generation and multiple choice scenarios, respectively. For these works we describe the resultant issues, and propose inference methods that largely alleviate them.The second half of this dissertation goes deeper into the question of how generative models of language capture the communicative goals that humans are optimizing: first with Learning to Write, operationalizing communicative goals into auxiliary search objectives for text decoding, and then with Generative Models as a Complex Systems Science, which presents a framework to think about the study of generative models as NLP shifts to analyzing systems that are often infeasible to replicate.How does a model that is predicting the distribution of next tokens understand-and fail to understand-the structure of an essay? This is precisely the kind of question we must face head-on in the new science of generative models. 
일반주제명  
Computer science.
일반주제명  
Computer engineering.
일반주제명  
Linguistics.
키워드  
Language models
키워드  
Communicative goals
키워드  
Text decoding
키워드  
Analyzing systems
키워드  
Interpretation errors
기타저자  
University of Washington Computer Science and Engineering
기본자료저록  
Dissertations Abstracts International. 85-03A.
기본자료저록  
Dissertation Abstract International
전자적 위치 및 접속  
로그인 후 원문을 볼 수 있습니다.

MARC

 008240612s2023      us  |||||||||||||||c||eng  d
■001000016934766
■00520240214101651
■006m          o    d                
■007cr#unu||||||||
■020    ▼a9798380328883
■035    ▼a(MiAaPQ)AAI30634344
■040    ▼aMiAaPQ▼cMiAaPQ
■0820  ▼a004
■1001  ▼aHoltzman,  Ari.
■24510▼aInterpretation  Errors:  Extracting  Functionality  From  Generative  Models  of  Language  by  Understanding  Them  Better▼h[electronic  resource]
■260    ▼a[S.l.]:▼bUniversity  of  Washington.  ▼c2023
■260  1▼aAnn  Arbor  :▼bProQuest  Dissertations  &  Theses,  ▼c2023
■300    ▼a1  online  resource(129  p.)
■500    ▼aSource:  Dissertations  Abstracts  International,  Volume:  85-03,  Section:  A.
■500    ▼aAdvisor:  Zettlemoyer,  Luke.
■5021  ▼aThesis  (Ph.D.)--University  of  Washington,  2023.
■506    ▼aThis  item  must  not  be  sold  to  any  third  party  vendors.
■520    ▼aThe  rise  of  large  language  models  as  the  workhorse  of  NLP,  and  the  continuous  release  of  better  models  (OpenAI,  2023;  Pichai,  2023;  Schulman  et  al.,  2022,  inter  alia)  has  created  a  strange  situation:  we  have  models  that  are  more  powerful  language  generators  than  ever  before,  but  since  we  did  not  design  them  for  a  specific  purpose  we  struggle  to  understand  how  they  should  be  used  or  what  their  idiosyncracies  are.This  dissertation  describes  three  empirical  projects  that  sought  to  characterize  the  underlying  behavior  of  language  models  and,  importantly,  to  make  them  more  reliable  tools  for  generating  and  selecting  text  where  this  behavior  does  not  match  up  with  the  tasks  we  would  like  models  to  complete.  Each  project  attempts  to  understand  what  language  models  and  accompanying  inference  methods  currently  optimize  for,  to  characterize  the  gap  between  that  and  the  true  objective  of  a  potential  user,  and  to  close  it  with  some  new  inference  method.  An  emergent  theme  through  these  works  is  that  models  are  already  doing  what  we  trained  them  to  do  quite  well-and  it  is  often  the  experimenters  and  practitioners  who  misunderstand  precisely  what  we  trained  models  to  do  in  the  first  place.  We  conclude  with  a  conceptual  analysis  of  how  we  should  study  generative  models  going  forward-as  models  keep  improving  and  new,  unanticipated  uses  and  misuses  become  ever  more  available.The  first  half  of  this  dissertation  concerns  two  works,  Neural  Text  Degeneration  and  Surface  Form  Competition-two  failure  modes  of  generative  models  that  occur  when  probability  is  viewed  as  equivalent  to  "correctness"  in  text  generation  and  multiple  choice  scenarios,  respectively.  For  these  works  we  describe  the  resultant  issues,  and  propose  inference  methods  that  largely  alleviate  them.The  second  half  of  this  dissertation  goes  deeper  into  the  question  of  how  generative  models  of  language  capture  the  communicative  goals  that  humans  are  optimizing:  first  with  Learning  to  Write,  operationalizing  communicative  goals  into  auxiliary  search  objectives  for  text  decoding,  and  then  with  Generative  Models  as  a  Complex  Systems  Science,  which  presents  a  framework  to  think  about  the  study  of  generative  models  as  NLP  shifts  to  analyzing  systems  that  are  often  infeasible  to  replicate.How  does  a  model  that  is  predicting  the  distribution  of  next  tokens  understand-and  fail  to  understand-the  structure  of  an  essay?  This  is  precisely  the  kind  of  question  we  must  face  head-on  in  the  new  science  of  generative  models. 
■590    ▼aSchool  code:  0250.
■650  4▼aComputer  science.
■650  4▼aComputer  engineering.
■650  4▼aLinguistics.
■653    ▼aLanguage  models
■653    ▼aCommunicative  goals
■653    ▼aText  decoding
■653    ▼aAnalyzing  systems
■653    ▼aInterpretation  errors
■690    ▼a0984
■690    ▼a0464
■690    ▼a0290
■71020▼aUniversity  of  Washington▼bComputer  Science  and  Engineering.
■7730  ▼tDissertations  Abstracts  International▼g85-03A.
■773    ▼tDissertation  Abstract  International
■790    ▼a0250
■791    ▼aPh.D.
■792    ▼a2023
■793    ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T16934766▼nKERIS▼z이  자료의  원문은  한국교육학술정보원에서  제공합니다.
■980    ▼a202402▼f2024

미리보기

내보내기

chatGPT토론

Ai 추천 관련 도서


    신착도서 더보기
    최근 3년간 통계입니다.

    소장정보

    • 예약
    • 소재불명신고
    • 나의폴더
    • 우선정리요청
    • 비도서대출신청
    • 야간 도서대출신청
    소장자료
    등록번호 청구기호 소장처 대출가능여부 대출정보
    TF09205 전자도서 마이폴더 부재도서신고 비도서대출신청

    * 대출중인 자료에 한하여 예약이 가능합니다. 예약을 원하시면 예약버튼을 클릭하십시오.

    해당 도서를 다른 이용자가 함께 대출한 도서

    관련 인기도서

    로그인 후 이용 가능합니다.