{"id":"https://openalex.org/W7164004866","doi":"https://doi.org/10.48550/arxiv.2606.07690","title":"HARP: Efficient Data Selection for Finetuning Large Language Models","display_name":"HARP: Efficient Data Selection for Finetuning Large Language Models","publication_year":2026,"publication_date":"2026-06-05","ids":{"openalex":"https://openalex.org/W7164004866","doi":"https://doi.org/10.48550/arxiv.2606.07690"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.07690","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.07690","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.07690","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5138262409","display_name":"Ning Wang (108353)","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Ning","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138240349","display_name":"Zhengxin Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Zhengxin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126102419","display_name":"Maosen Tang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tang, Maosen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125202388","display_name":"Yitang Gao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gao, Yitang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138251469","display_name":"Claire Cardie","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cardie, Claire","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5138275096","display_name":"Sainyam Galhotra","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Galhotra, Sainyam","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.5960000157356262,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.5960000157356262,"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/T10028","display_name":"Topic Modeling","score":0.20499999821186066,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.03759999945759773,"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/pruning","display_name":"Pruning","score":0.6439999938011169},{"id":"https://openalex.org/keywords/selection","display_name":"Selection (genetic algorithm)","score":0.628000020980835},{"id":"https://openalex.org/keywords/embedding","display_name":"Embedding","score":0.6025999784469604},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.559499979019165},{"id":"https://openalex.org/keywords/pairwise-comparison","display_name":"Pairwise comparison","score":0.5317000150680542},{"id":"https://openalex.org/keywords/harp","display_name":"HARP","score":0.5042999982833862},{"id":"https://openalex.org/keywords/bayesian-probability","display_name":"Bayesian probability","score":0.4178999960422516},{"id":"https://openalex.org/keywords/baseline","display_name":"Baseline (sea)","score":0.39100000262260437},{"id":"https://openalex.org/keywords/similarity","display_name":"Similarity (geometry)","score":0.36959999799728394}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7820000052452087},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.6439999938011169},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.628000020980835},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.6025999784469604},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5719000101089478},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.559499979019165},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5374000072479248},{"id":"https://openalex.org/C184898388","wikidata":"https://www.wikidata.org/wiki/Q1435712","display_name":"Pairwise comparison","level":2,"score":0.5317000150680542},{"id":"https://openalex.org/C2776030095","wikidata":"https://www.wikidata.org/wiki/Q5628954","display_name":"HARP","level":2,"score":0.5042999982833862},{"id":"https://openalex.org/C107673813","wikidata":"https://www.wikidata.org/wiki/Q812534","display_name":"Bayesian probability","level":2,"score":0.4178999960422516},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.4081999957561493},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.39100000262260437},{"id":"https://openalex.org/C103278499","wikidata":"https://www.wikidata.org/wiki/Q254465","display_name":"Similarity (geometry)","level":3,"score":0.36959999799728394},{"id":"https://openalex.org/C2776207758","wikidata":"https://www.wikidata.org/wiki/Q5303302","display_name":"Downstream (manufacturing)","level":2,"score":0.35839998722076416},{"id":"https://openalex.org/C93959086","wikidata":"https://www.wikidata.org/wiki/Q6888345","display_name":"Model selection","level":2,"score":0.33469998836517334},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.3179999887943268},{"id":"https://openalex.org/C111696304","wikidata":"https://www.wikidata.org/wiki/Q2303697","display_name":"Sorting","level":2,"score":0.3034999966621399},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.287200003862381},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2799000144004822},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.27639999985694885},{"id":"https://openalex.org/C148483581","wikidata":"https://www.wikidata.org/wiki/Q446488","display_name":"Feature selection","level":2,"score":0.2759000062942505},{"id":"https://openalex.org/C102634674","wikidata":"https://www.wikidata.org/wiki/Q868473","display_name":"Smoothness","level":2,"score":0.26739999651908875},{"id":"https://openalex.org/C207201462","wikidata":"https://www.wikidata.org/wiki/Q182505","display_name":"Bayes' theorem","level":3,"score":0.26649999618530273},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.25949999690055847},{"id":"https://openalex.org/C160234255","wikidata":"https://www.wikidata.org/wiki/Q812535","display_name":"Bayesian inference","level":3,"score":0.25049999356269836}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.07690","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.07690","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.07690","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.07690","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Finetuning":[0],"data":[1,106],"selection":[2,80,136],"requires":[3],"balancing":[4],"two":[5,108],"competing":[6],"goals:":[7],"selecting":[8],"examples":[9],"that":[10,74,145],"improve":[11],"the":[12,40,84,149,154],"downstream":[13,47,76],"objective,":[14],"and":[15,95,116,130,152],"doing":[16],"so":[17],"without":[18],"repeatedly":[19],"finetuning":[20],"models.":[21],"Train-free":[22],"selectors":[23,44],"are":[24],"scalable":[25],"but":[26,57],"rely":[27],"on":[28],"proxies":[29],"such":[30],"as":[31],"embedding":[32],"similarity":[33],"or":[34,54],"clustering,":[35],"which":[36,112,118],"may":[37],"not":[38],"match":[39],"target":[41],"objective.":[42],"Train-based":[43],"better":[45],"reflect":[46],"utility":[48],"through":[49],"gradient":[50],"signals,":[51],"subset":[52],"evaluation,":[53],"Shapley":[55],"attribution,":[56],"require":[58],"many":[59],"costly":[60],"train--evaluate":[61,140],"iterations.":[62],"We":[63,123,142],"propose":[64],"Hierarchical":[65],"Active":[66],"Region":[67],"Pruning":[68],"(HARP),":[69],"an":[70],"efficient":[71],"train-based":[72],"selector":[73],"preserves":[75],"alignment":[77],"while":[78,138,162],"reducing":[79,139],"cost.":[81,141],"HARP":[82,134,146],"organizes":[83],"training":[85,167],"pool":[86],"into":[87],"a":[88],"node--leaf":[89],"hierarchy,":[90],"evaluates":[91],"only":[92],"representative":[93],"leaves,":[94],"infers":[96],"unmeasured":[97],"utilities":[98],"with":[99],"empirical":[100],"Bayes":[101],"posteriors.":[102],"It":[103],"then":[104],"selects":[105],"using":[107,163],"complementary":[109,121],"envelopes:":[110],"HARP-C,":[111],"conservatively":[113],"controls":[114,135],"redundancy,":[115],"HARP-E,":[117],"additively":[119],"rewards":[120],"regions.":[122],"theoretically":[124],"show":[125],"that,":[126],"under":[127],"local":[128],"smoothness":[129],"bounded":[131],"estimation":[132],"error,":[133],"error":[137],"further":[143],"validate":[144],"variants":[147],"achieve":[148],"best":[150],"result":[151],"outperform":[153],"strongest":[155],"baseline":[156],"by":[157],"up":[158],"to":[159],"$+8.9$":[160],"points,":[161],"roughly":[164],"$7\\times$":[165],"fewer":[166],"examples.":[168]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-10T00:00:00"}