基本信息
文件名称：使用 GPT-4 Turbo-128k 优化 Automuse分析.pdf
文件大小：3.75 MB
总页数：5 页
更新时间：2024-04-14
总字数：约2.08万字
文档摘要

OptmizingAutomusewithGPT-4Turbo-128k

OptmizingAutomusewithGPT-4Turbo-128k

CadeyA.RatioNicoleBrennanJessicaWilliamsAshleyKaplanStephanieWilliamsMaInsa

xn--g28hxn--g28hxn--g28hxn--g28hxn--g28hxn--g28h

cadey@xeserv.ustwi@xeserv.usjess@xeserv.usashe@xeserv.usphi@xeserv.usmai@xeserv.us

Abstract

FurtherimprovementstotheAutomusesystemdescribedinAutomuse[1]aredescribed.TheuseofGPT-4

Turbo128k[2]allowsforuniqueopportunitiesinincreasingoutputqualityandquantity.Further

adaptationstomodernizescenariosandplotsarealsodescribed.

IntroductionMotivation

ModernadvancementstolargelanguagemodelTheauthorsweremotivatedtocreatethispaper

technologyhasallowedforcreatingnewandinter-bytheexistenceoftheGPT-4Turbo128kmodel.

estingapplicationsontopofthetechnology.OneThismodelallowsfor128,000tokensofcontext,

suchapplicationistheAutomuse[1]system.Thiswhichisenoughtostoreanentirenovel’sworth

systemusedTypeScriptandGPT-3.5[3]togenerateoftextasdefinedbytheNationalNovemberWriting

scenariosandprosefor“pulpnovels”,orlow-qual-Monthrules[6].Thisallowsforthepossibilityof

ityfictionchurnedoutoncheappapertomeetmassgeneratinganentirenovelwithasinglecontext

needsforentertainment.Thispaperdescribesthewindow,whichisasignificantimprovementoverthe

improvementsmadetothesystembyusingGPT-4previoussystemhavingtomanagemultiplecontext

Turbo128k[2]andmodernizingthescenariosandwindowsandhaphazardlystitchthemtoge