{"id":"https://openalex.org/W7126455421","doi":"https://doi.org/10.18653/v1/2024.findings-eacl.143","title":"Hierarchical and Dynamic Prompt Compression for Efficient Zero-shot API Usage","display_name":"Hierarchical and Dynamic Prompt Compression for Efficient Zero-shot API Usage","publication_year":2024,"publication_date":"2024-01-01","ids":{"openalex":"https://openalex.org/W7126455421","doi":"https://doi.org/10.18653/v1/2024.findings-eacl.143"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2024.findings-eacl.143","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2024.findings-eacl.143","pdf_url":"https://aclanthology.org/2024.findings-eacl.143.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":"Findings of the Association for Computational Linguistics: EACL 2024","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2024.findings-eacl.143.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101705761","display_name":"Yichen Jiang","orcid":"https://orcid.org/0000-0002-9139-3799"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yichen Jiang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134063028","display_name":"Marco Del Vecchio","orcid":"https://orcid.org/0009-0008-3600-574X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Marco Vecchio","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5089199686","display_name":"Mohit Bansal","orcid":"https://orcid.org/0000-0001-5858-9944"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mohit Bansal","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5108243977","display_name":"Anders Johannsen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Anders Johannsen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.51552409,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"2162","last_page":"2174"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.1225999966263771,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.1225999966263771,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T10126","display_name":"Logic, programming, and type systems","score":0.09679999947547913,"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/T11269","display_name":"Algorithms and Data Compression","score":0.09619999676942825,"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/compression","display_name":"Compression (physics)","score":0.5594000220298767},{"id":"https://openalex.org/keywords/data-compression","display_name":"Data compression","score":0.44510000944137573},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.2768000066280365},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.2386000007390976},{"id":"https://openalex.org/keywords/component","display_name":"Component (thermodynamics)","score":0.23819999396800995}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6650999784469604},{"id":"https://openalex.org/C180016635","wikidata":"https://www.wikidata.org/wiki/Q2712821","display_name":"Compression (physics)","level":2,"score":0.5594000220298767},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.44510000944137573},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2768000066280365},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.24809999763965607},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.2386000007390976},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.23819999396800995},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.23520000278949738},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.23250000178813934},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.23190000653266907}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2024.findings-eacl.143","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2024.findings-eacl.143","pdf_url":"https://aclanthology.org/2024.findings-eacl.143.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":"Findings of the Association for Computational Linguistics: EACL 2024","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2024.findings-eacl.143","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2024.findings-eacl.143","pdf_url":"https://aclanthology.org/2024.findings-eacl.143.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":"Findings of the Association for Computational Linguistics: EACL 2024","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7126455421.pdf","grobid_xml":"https://content.openalex.org/works/W7126455421.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Long":[0],"prompts":[1],"present":[2],"a":[3,50,65,131,147],"significant":[4],"challenge":[5],"for":[6,23,104,116,125],"practical":[7],"LLM-based":[8],"systems":[9,26],"that":[10,27,58],"need":[11],"to":[12,29,60,86,150],"operate":[13],"with":[14],"low":[15,83],"latency":[16],"and":[17,120,169],"limited":[18],"resources.We":[19],"investigate":[20],"prompt":[21,44,63],"compression":[22,171],"zeroshot":[24],"dialogue":[25],"learn":[28],"use":[30,111],"unseen":[31],"APIs":[32],"directly":[33],"in-context":[34],"from":[35,49,155],"their":[36],"documentation,":[37,80],"which":[38],"may":[39],"take":[40],"up":[41],"hundreds":[42],"of":[43,130,176],"tokens":[45,103,139],"per":[46],"API.We":[47],"start":[48],"recently":[51],"introduced":[52],"approach":[53,180],"(Mu":[54],"et":[55],"al.,":[56],"2023)":[57],"learns":[59],"compress":[61],"the":[62,87,152,156,166,177],"into":[64],"few":[66],"\"gist":[67],"token\"":[68],"activations":[69],"during":[70],"finetuning.However,":[71],"this":[72,93],"simple":[73],"idea":[74],"is":[75],"ineffective":[76],"in":[77,82],"compressing":[78,117,126],"API":[79,153],"resulting":[81],"accuracy":[84],"compared":[85],"baseline":[88],"using":[89],"an":[90,108,118,127],"uncompressed":[91],"prompt.In":[92],"work,":[94],"we":[95,100,110,145],"introduce":[96],"two":[97],"major":[98],"improvements.First,":[99],"specialize":[101],"gist":[102,157,178],"different":[105],"hierarchies":[106],"within":[107],"API:":[109],"one":[112,121],"Gist":[113,122,137],"arg":[114],"token":[115,124,179],"argument":[119],"value":[123,129,138],"acceptable":[128],"categorical":[132],"argument.We":[133],"then":[134],"dynamically":[135],"reveal":[136],"only":[140],"when":[141],"they":[142],"are":[143],"needed.Second,":[144],"add":[146],"reconstruction":[148],"loss":[149],"predict":[151],"documentation":[154],"tokens.On":[158],"multiple":[159],"API-calling":[160],"tasks,":[161],"our":[162],"proposed":[163],"system":[164],"keeps":[165],"simplicity,":[167],"efficiency,":[168],"large":[170],"factor":[172],"(20x":[173],"on":[174],"SGD)":[175],"while":[181],"achieving":[182],"significantly":[183],"better":[184],"accuracy.":[185],"1":[186]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-02-02T00:00:00"}
