{"id":"https://openalex.org/W4389520756","doi":"https://doi.org/10.18653/v1/2023.emnlp-main.494","title":"Automatic Prompt Optimization with \u201cGradient Descent\u201d and Beam Search","display_name":"Automatic Prompt Optimization with \u201cGradient Descent\u201d and Beam Search","publication_year":2023,"publication_date":"2023-01-01","ids":{"openalex":"https://openalex.org/W4389520756","doi":"https://doi.org/10.18653/v1/2023.emnlp-main.494"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2023.emnlp-main.494","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.emnlp-main.494","pdf_url":"https://aclanthology.org/2023.emnlp-main.494.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2023.emnlp-main.494.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5065744065","display_name":"Reid Pryzant","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Reid Pryzant","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5037853792","display_name":"Dan Iter","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dan Iter","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101439736","display_name":"Jerry Li","orcid":"https://orcid.org/0000-0002-8774-826X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jerry Li","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5028759029","display_name":"Yin Lee","orcid":"https://orcid.org/0000-0002-8661-8607"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yin Lee","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100355246","display_name":"Chenguang Zhu","orcid":"https://orcid.org/0000-0002-3549-0826"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chenguang Zhu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5089195158","display_name":"Michael Zeng","orcid":"https://orcid.org/0000-0001-5302-5883"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Michael Zeng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5065744065"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":23.8084,"has_fulltext":true,"cited_by_count":145,"citation_normalized_percentile":{"value":0.99685056,"is_in_top_1_percent":true,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":99,"max":100},"biblio":{"volume":null,"issue":null,"first_page":"7957","last_page":"7968"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9990000128746033,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9990000128746033,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10260","display_name":"Software Engineering Research","score":0.9929999709129333,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9925000071525574,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.8079355359077454},{"id":"https://openalex.org/keywords/gradient-descent","display_name":"Gradient descent","score":0.7998919486999512},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.7474480271339417},{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.49903154373168945},{"id":"https://openalex.org/keywords/point","display_name":"Point (geometry)","score":0.4962611794471741},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4946396052837372},{"id":"https://openalex.org/keywords/stochastic-gradient-descent","display_name":"Stochastic gradient descent","score":0.4922737181186676},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.4768444299697876},{"id":"https://openalex.org/keywords/selection","display_name":"Selection (genetic algorithm)","score":0.4711524248123169},{"id":"https://openalex.org/keywords/optimization-problem","display_name":"Optimization problem","score":0.41705724596977234},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.3663637042045593},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.2761734127998352},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.0865631103515625}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8079355359077454},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.7998919486999512},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.7474480271339417},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.49903154373168945},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.4962611794471741},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4946396052837372},{"id":"https://openalex.org/C206688291","wikidata":"https://www.wikidata.org/wiki/Q7617819","display_name":"Stochastic gradient descent","level":3,"score":0.4922737181186676},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.4768444299697876},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.4711524248123169},{"id":"https://openalex.org/C137836250","wikidata":"https://www.wikidata.org/wiki/Q984063","display_name":"Optimization problem","level":2,"score":0.41705724596977234},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3663637042045593},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2761734127998352},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.0865631103515625},{"id":"https://openalex.org/C187736073","wikidata":"https://www.wikidata.org/wiki/Q2920921","display_name":"Management","level":1,"score":0.0},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.0},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.0},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.0},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.0},{"id":"https://openalex.org/C13280743","wikidata":"https://www.wikidata.org/wiki/Q131089","display_name":"Geodesy","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2023.emnlp-main.494","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.emnlp-main.494","pdf_url":"https://aclanthology.org/2023.emnlp-main.494.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2023.emnlp-main.494","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.emnlp-main.494","pdf_url":"https://aclanthology.org/2023.emnlp-main.494.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.5099999904632568}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4389520756.pdf"},"referenced_works_count":29,"referenced_works":["https://openalex.org/W1544426622","https://openalex.org/W1553290137","https://openalex.org/W2126559945","https://openalex.org/W2950929549","https://openalex.org/W2953320089","https://openalex.org/W3035309367","https://openalex.org/W3088728183","https://openalex.org/W3098267758","https://openalex.org/W3160638507","https://openalex.org/W3166846774","https://openalex.org/W3173777717","https://openalex.org/W3174784402","https://openalex.org/W4205991051","https://openalex.org/W4225165463","https://openalex.org/W4226352076","https://openalex.org/W4287207937","https://openalex.org/W4300952844","https://openalex.org/W4308244910","https://openalex.org/W4311887664","https://openalex.org/W4312053697","https://openalex.org/W4312091890","https://openalex.org/W4360836968","https://openalex.org/W4365205411","https://openalex.org/W4366548330","https://openalex.org/W4366736191","https://openalex.org/W4385571789","https://openalex.org/W4385573003","https://openalex.org/W4385574139","https://openalex.org/W4386566526"],"related_works":["https://openalex.org/W2895097035","https://openalex.org/W4206903459","https://openalex.org/W2754816816","https://openalex.org/W4366280654","https://openalex.org/W3160167280","https://openalex.org/W3020853991","https://openalex.org/W4362706668","https://openalex.org/W4231621013","https://openalex.org/W3171021120","https://openalex.org/W3008318776"],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"have":[4],"shown":[5],"impressive":[6],"performance":[7,164],"as":[8],"general":[9],"purpose":[10],"agents,":[11],"but":[12],"their":[13],"abilities":[14],"remain":[15],"highly":[16],"dependent":[17],"on":[18],"prompts":[19],"which":[20,44,128],"are":[21,96,118],"hand":[22],"written":[23],"with":[24,40],"onerous":[25],"trial-and-error":[26],"effort.":[27],"We":[28],"propose":[29],"a":[30,121],"simple":[31],"and":[32,60,124,140,159],"nonparametric":[33],"solution":[34],"to":[35,51,57,70,167,172],"this":[36],"problem,":[37],"Prompt":[38,151],"Optimization":[39,152],"Textual":[41],"Gradients":[42],"(ProTeGi),":[43],"is":[45],"inspired":[46],"by":[47,102,120,165,169],"numerical":[48,83],"gradient":[49,115],"descent":[50,116],"automatically":[52],"improve":[53,160],"prompts,":[54],"assuming":[55],"access":[56],"training":[58],"data":[59,69,171],"an":[61,161],"LLM":[62,145],"API.":[63],"The":[64,92],"algorithm":[65],"uses":[66],"minibatches":[67],"of":[68,89,111,144],"form":[71],"natural":[72,93],"language":[73,94],"\u201cgradients\u201d":[74],"that":[75,149],"criticize":[76],"the":[77,87,100,104,107,112,141],"current":[78],"prompt,":[79],"much":[80],"like":[81],"how":[82],"gradients":[84,95],"point":[85],"in":[86,106],"direction":[88,110],"error":[90],"ascent.":[91],"then":[97],"\u201cpropagated\u201d":[98],"into":[99,177],"prompt":[101,105,156],"editing":[103,157],"opposite":[108],"semantic":[109],"gradient.":[113],"These":[114],"steps":[117],"guided":[119],"beam":[122],"search":[123],"bandit":[125],"selection":[126],"procedure":[127],"significantly":[129],"improves":[130],"algorithmic":[131],"efficiency.":[132],"Preliminary":[133],"results":[134],"across":[135],"three":[136],"benchmark":[137],"NLP":[138],"tasks":[139],"novel":[142],"problem":[143],"jailbreak":[146],"detection":[147],"suggest":[148],"Automatic":[150],"can":[153],"outperform":[154],"prior":[155],"techniques":[158],"initial":[162],"prompt\u2019s":[163],"up":[166],"31%,":[168],"using":[170],"rewrite":[173],"vague":[174],"task":[175],"descriptions":[176],"more":[178],"precise":[179],"annotation":[180],"instructions.":[181]},"counts_by_year":[{"year":2026,"cited_by_count":29},{"year":2025,"cited_by_count":75},{"year":2024,"cited_by_count":29},{"year":2023,"cited_by_count":12}],"updated_date":"2026-06-04T09:04:59.091469","created_date":"2025-10-10T00:00:00"}