{"id":"https://openalex.org/W7131381237","doi":"https://doi.org/10.48550/arxiv.2602.21103","title":"Prompt-Level Distillation: A Non-Parametric Alternative to Model Fine-Tuning for Efficient Reasoning","display_name":"Prompt-Level Distillation: A Non-Parametric Alternative to Model Fine-Tuning for Efficient Reasoning","publication_year":2026,"publication_date":"2026-02-24","ids":{"openalex":"https://openalex.org/W7131381237","doi":"https://doi.org/10.48550/arxiv.2602.21103"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2602.21103","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2602.21103","pdf_url":"https://arxiv.org/pdf/2602.21103","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2602.21103","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5124166415","display_name":"Sanket Badhe","orcid":"https://orcid.org/0000-0002-1808-6602","observed_orcids":["https://orcid.org/0000-0002-1808-6602"]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Badhe, Sanket","raw_affiliation_strings":["Google Mountain View , California , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Google Mountain View , California , USA","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5135914644","display_name":"Deep Shah","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Shah, Deep","raw_affiliation_strings":["Google Mountain View , California , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Google Mountain View , California , USA","institution_ids":["https://openalex.org/I1291425158"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I1291425158"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"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/T10028","display_name":"Topic Modeling","score":0.7670999765396118,"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":0.7670999765396118,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.10170000046491623,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.03579999879002571,"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/chain-of-thought-prompting","display_name":"chain-of-thought prompting","score":0.9210000038146973},{"id":"https://openalex.org/keywords/efficient-reasoning","display_name":"efficient reasoning","score":0.5799999833106995},{"id":"https://openalex.org/keywords/reasoning-patterns","display_name":"reasoning patterns","score":0.44600000977516174},{"id":"https://openalex.org/keywords/inference-latency-reduction","display_name":"inference latency reduction","score":0.7039999961853027},{"id":"https://openalex.org/keywords/interpretability","display_name":"interpretability","score":0.7179999947547913},{"id":"https://openalex.org/keywords/cross-architecture-generalization","display_name":"cross-architecture generalization","score":0.7889999747276306},{"id":"https://openalex.org/keywords/edge-deployment","display_name":"edge deployment","score":0.3880000114440918}],"concepts":[{"id":"https://openalex.org/C2781067378","wikidata":"https://www.wikidata.org/wiki/Q17027399","display_name":"Interpretability","level":2,"score":0.8840000033378601},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8001000285148621},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6712999939918518},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.5400000214576721},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5285999774932861},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.46630001068115234},{"id":"https://openalex.org/C166955791","wikidata":"https://www.wikidata.org/wiki/Q629579","display_name":"Macro","level":2,"score":0.4507000148296356},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.41839998960494995},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4041999876499176},{"id":"https://openalex.org/C89288958","wikidata":"https://www.wikidata.org/wiki/Q7301504","display_name":"Reasoning system","level":2,"score":0.399399995803833},{"id":"https://openalex.org/C112313634","wikidata":"https://www.wikidata.org/wiki/Q7886648","display_name":"Complement (music)","level":5,"score":0.37560001015663147},{"id":"https://openalex.org/C86827895","wikidata":"https://www.wikidata.org/wiki/Q7098582","display_name":"Opportunistic reasoning","level":4,"score":0.373199999332428},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.3391999900341034},{"id":"https://openalex.org/C37335422","wikidata":"https://www.wikidata.org/wiki/Q6888134","display_name":"Model-based reasoning","level":3,"score":0.3352000117301941},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.328000009059906},{"id":"https://openalex.org/C133112747","wikidata":"https://www.wikidata.org/wiki/Q7251931","display_name":"Protocol analysis","level":2,"score":0.3199000060558319},{"id":"https://openalex.org/C2776639384","wikidata":"https://www.wikidata.org/wiki/Q840396","display_name":"Ideal (ethics)","level":2,"score":0.31540000438690186},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3118000030517578},{"id":"https://openalex.org/C206345919","wikidata":"https://www.wikidata.org/wiki/Q20380951","display_name":"Resource (disambiguation)","level":2,"score":0.2903999984264374},{"id":"https://openalex.org/C20162079","wikidata":"https://www.wikidata.org/wiki/Q1151406","display_name":"Case-based reasoning","level":2,"score":0.2831999957561493},{"id":"https://openalex.org/C195344581","wikidata":"https://www.wikidata.org/wiki/Q2555318","display_name":"Automated reasoning","level":2,"score":0.2759000062942505},{"id":"https://openalex.org/C159032336","wikidata":"https://www.wikidata.org/wiki/Q2488768","display_name":"Non-monotonic logic","level":2,"score":0.27459999918937683},{"id":"https://openalex.org/C29202148","wikidata":"https://www.wikidata.org/wiki/Q287260","display_name":"Resource allocation","level":2,"score":0.2615000009536743},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.25920000672340393},{"id":"https://openalex.org/C83725634","wikidata":"https://www.wikidata.org/wiki/Q7268699","display_name":"Qualitative reasoning","level":2,"score":0.2590000033378601},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.2515999972820282}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2602.21103","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2602.21103","pdf_url":"https://arxiv.org/pdf/2602.21103","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2602.21103","doi":"https://doi.org/10.48550/arxiv.2602.21103","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.21103","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2602.21103","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2602.21103","pdf_url":"https://arxiv.org/pdf/2602.21103","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7131381237.pdf","grobid_xml":"https://content.openalex.org/works/W7131381237.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4416947700","https://openalex.org/W4414329205","https://openalex.org/W7165132526","https://openalex.org/W4385571219","https://openalex.org/W7163179378","https://openalex.org/W4414586969","https://openalex.org/W7137968040","https://openalex.org/W4378506904","https://openalex.org/W7166802716","https://openalex.org/W4312045629"],"abstract_inverted_index":{"Advanced":[0],"reasoning":[1,46],"typically":[2],"requires":[3],"Chain-of-Thought":[4],"prompting,":[5],"which":[6],"is":[7],"accurate":[8],"but":[9],"incurs":[10],"prohibitive":[11],"latency":[12,112],"and":[13,31,52,82,140,149],"substantial":[14],"test-time":[15],"inference":[16],"costs.":[17],"The":[18],"standard":[19],"alternative,":[20],"fine-tuning":[21],"smaller":[22],"models,":[23],"often":[24],"sacrifices":[25],"interpretability":[26],"while":[27,87],"introducing":[28],"significant":[29],"resource":[30],"operational":[32],"overhead.":[33,113],"To":[34],"address":[35],"these":[36,103],"limitations,":[37],"we":[38],"introduce":[39],"Prompt-Level":[40],"Distillation":[41],"(PLD).":[42],"We":[43],"extract":[44],"explicit":[45],"patterns":[47],"from":[48],"a":[49,56],"Teacher":[50],"model":[51],"organize":[53],"them":[54],"into":[55],"structured":[57],"list":[58],"of":[59,127],"expressive":[60,115],"instructions":[61,116],"for":[62,123,133],"the":[63,118],"Student":[64],"model's":[65],"System":[66],"Prompt.":[67],"Evaluated":[68],"using":[69],"Gemma-3":[70],"4B,":[71],"PLD":[72],"improved":[73],"Macro":[74],"F1":[75],"scores":[76],"on":[77,95],"StereoSet":[78],"(57\\%":[79],"to":[80,85,91,106],"90.0\\%)":[81],"Contract-NLI":[83],"(67\\%":[84],"83\\%),":[86],"increasing":[88],"LogiQA":[89],"accuracy":[90],"70\\%.":[92],"Similar":[93],"results":[94],"Mistral":[96],"Small":[97],"3.1":[98],"demonstrate":[99],"cross-architecture":[100],"generalizability,":[101],"enabling":[102],"compact":[104],"models":[105],"match":[107],"frontier":[108],"performance":[109],"with":[110],"negligible":[111],"These":[114],"render":[117],"decision-making":[119],"process":[120],"transparent,":[121],"allowing":[122],"full":[124],"human":[125],"verification":[126],"logic,":[128],"making":[129],"this":[130],"approach":[131],"ideal":[132],"regulated":[134],"industries":[135],"such":[136],"as":[137,143,145],"law,":[138],"finance,":[139],"content":[141],"moderation,":[142],"well":[144],"high-volume":[146],"use":[147],"cases":[148],"edge":[150],"devices.":[151]},"counts_by_year":[],"updated_date":"2026-10-31T12:09:11.096373","created_date":"2026-02-26T00:00:00"}