OpenAI：2023年GPT-4技术报告（英文版）

GPT-4TechnicalReportOpenAI∗AbstractWereportthedevelopmentofGPT-4,alarge-scale,multimodalmodelwhichcanacceptimageandtextinputsandproducetextoutputs.Whilelesscapablethanhumansinmanyreal-worldscenarios,GPT-4exhibitshuman-levelperformanceonvariousprofessionalandacademicbenchmarks,includingpassingasimulatedbarexamwithascorearoundthetop10%oftesttakers.GPT-4isaTransformer-basedmodelpre-trainedtopredictthenexttokeninadocument.Thepost-trainingalignmentprocessresultsinimprovedperformanceonmeasuresoffactualityandadherencetodesiredbehavior.Acorecomponentofthisprojectwasdevelopinginfrastructureandoptimizationmethodsthatbehavepredictablyacrossawiderangeofscales.ThisallowedustoaccuratelypredictsomeaspectsofGPT-4’sperformancebasedonmodelstrainedwithnomorethan1/1,000ththecomputeofGPT-4.1IntroductionThistechnicalreportpresentsGPT-4,alargemultimodalmodelcapableofprocessingimageandtextinputsandproducingtextoutputs.Suchmodelsareanimportantareaofstudyastheyhavethepotentialtobeusedinawiderangeofapplications,