{"id":"https://openalex.org/W7126253111","doi":"https://doi.org/10.48550/arxiv.2601.21590","title":"Scalable Power Sampling: Unlocking Efficient, Training-Free Reasoning for LLMs via Distribution Sharpening","display_name":"Scalable Power Sampling: Unlocking Efficient, Training-Free Reasoning for LLMs via Distribution Sharpening","publication_year":2026,"publication_date":"2026-01-29","ids":{"openalex":"https://openalex.org/W7126253111","doi":"https://doi.org/10.48550/arxiv.2601.21590"},"language":null,"primary_location":{"id":"pmh:doi:10.48550/arxiv.2601.21590","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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":"Article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":null,"any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5124441334","display_name":"Xiaotong Ji","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ji, Xiaotong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5000651751","display_name":"Rasul Tutunov","orcid":"https://orcid.org/0000-0002-6495-6437"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tutunov, Rasul","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5028712584","display_name":"Matthieu Zimmer","orcid":"https://orcid.org/0000-0002-8029-308X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zimmer, Matthieu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5102431252","display_name":"Haitham Bou Ammar","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ammar, Haitham Bou","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":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":0.2766000032424927,"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":0.2766000032424927,"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":0.24469999969005585,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.08079999685287476,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/sharpening","display_name":"Sharpening","score":0.7875000238418579},{"id":"https://openalex.org/keywords/markov-chain-monte-carlo","display_name":"Markov chain Monte Carlo","score":0.6682000160217285},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6141999959945679},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.5705999732017517},{"id":"https://openalex.org/keywords/markov-chain","display_name":"Markov chain","score":0.4235000014305115},{"id":"https://openalex.org/keywords/trajectory","display_name":"Trajectory","score":0.4088999927043915},{"id":"https://openalex.org/keywords/distribution","display_name":"Distribution (mathematics)","score":0.3878999948501587},{"id":"https://openalex.org/keywords/importance-sampling","display_name":"Importance sampling","score":0.37619999051094055},{"id":"https://openalex.org/keywords/prior-probability","display_name":"Prior probability","score":0.3605000078678131}],"concepts":[{"id":"https://openalex.org/C2781137444","wikidata":"https://www.wikidata.org/wiki/Q237105","display_name":"Sharpening","level":2,"score":0.7875000238418579},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.708299994468689},{"id":"https://openalex.org/C111350023","wikidata":"https://www.wikidata.org/wiki/Q1191869","display_name":"Markov chain Monte Carlo","level":3,"score":0.6682000160217285},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6141999959945679},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5705999732017517},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.436599999666214},{"id":"https://openalex.org/C98763669","wikidata":"https://www.wikidata.org/wiki/Q176645","display_name":"Markov chain","level":2,"score":0.4235000014305115},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.4088999927043915},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.4074999988079071},{"id":"https://openalex.org/C110121322","wikidata":"https://www.wikidata.org/wiki/Q865811","display_name":"Distribution (mathematics)","level":2,"score":0.3878999948501587},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.37619999051094055},{"id":"https://openalex.org/C177769412","wikidata":"https://www.wikidata.org/wiki/Q278090","display_name":"Prior probability","level":3,"score":0.3605000078678131},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.357699990272522},{"id":"https://openalex.org/C163258240","wikidata":"https://www.wikidata.org/wiki/Q25342","display_name":"Power (physics)","level":2,"score":0.3521000146865845},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3393000066280365},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.3314000070095062},{"id":"https://openalex.org/C19499675","wikidata":"https://www.wikidata.org/wiki/Q232207","display_name":"Monte Carlo method","level":2,"score":0.32749998569488525},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.3163999915122986},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3100000023841858},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.3005000054836273},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.296999990940094},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.28380000591278076},{"id":"https://openalex.org/C159886148","wikidata":"https://www.wikidata.org/wiki/Q176645","display_name":"Markov process","level":2,"score":0.272599995136261},{"id":"https://openalex.org/C2777472644","wikidata":"https://www.wikidata.org/wiki/Q16968992","display_name":"Approximate inference","level":3,"score":0.272599995136261},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.27250000834465027},{"id":"https://openalex.org/C149441793","wikidata":"https://www.wikidata.org/wiki/Q200726","display_name":"Probability distribution","level":2,"score":0.2567000091075897},{"id":"https://openalex.org/C43091099","wikidata":"https://www.wikidata.org/wiki/Q1067788","display_name":"Through-the-lens metering","level":3,"score":0.25040000677108765}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.48550/arxiv.2601.21590","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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":"Article"},{"id":"doi:10.48550/arxiv.2601.21590","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2601.21590","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":"article"}],"best_oa_location":{"id":"pmh:doi:10.48550/arxiv.2601.21590","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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":"Article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Reinforcement":[0],"learning":[1],"(RL)":[2],"post-training":[3,61],"is":[4],"a":[5,86,99,112,130],"dominant":[6],"approach":[7],"for":[8,78,94],"improving":[9],"the":[10,32,44,68,92,104,118,137],"reasoning":[11],"performance":[12,57],"of":[13,34,47,72],"large":[14],"language":[15],"models":[16],"(LLMs),":[17],"yet":[18],"growing":[19],"evidence":[20],"suggests":[21],"that":[22,41,90,103,135,159],"its":[23],"gains":[24],"arise":[25],"primarily":[26],"from":[27,43],"distribution":[28,46,107,141],"sharpening":[29],"rather":[30],"than":[31],"acquisition":[33],"new":[35],"capabilities.":[36],"Recent":[37],"work":[38],"has":[39],"shown":[40],"sampling":[42],"power":[45,106],"LLMs":[48],"using":[49],"Markov":[50],"chain":[51],"Monte":[52],"Carlo":[53],"(MCMC)":[54],"can":[55,108],"recover":[56],"comparable":[58],"to":[59,181],"RL":[60],"without":[62,167],"relying":[63,168],"on":[64,148,169],"external":[65,171],"rewards;":[66],"however,":[67],"high":[69],"computational":[70],"cost":[71],"MCMC":[73],"makes":[74],"such":[75],"approaches":[76],"impractical":[77],"widespread":[79],"adoption.":[80],"In":[81],"this":[82,126],"work,":[83],"we":[84,128,144],"propose":[85],"theoretically":[87],"grounded":[88],"alternative":[89],"eliminates":[91],"need":[93],"iterative":[95],"MCMC.":[96],"We":[97],"derive":[98],"novel":[100],"formulation":[101],"showing":[102],"global":[105],"be":[109],"approximated":[110],"by":[111,177],"token-level":[113],"scaled":[114],"low-temperature":[115],"one,":[116],"where":[117],"scaling":[119],"factor":[120],"captures":[121],"future":[122],"trajectory":[123],"quality.":[124],"Leveraging":[125],"insight,":[127],"introduce":[129],"training-free":[131],"and":[132,151,157],"verifier-free":[133],"algorithm":[134],"sharpens":[136],"base":[138],"model's":[139],"generative":[140],"autoregressively.":[142],"Empirically,":[143],"evaluate":[145],"our":[146,160],"method":[147,161],"math,":[149],"QA,":[150],"code":[152],"tasks":[153],"across":[154],"four":[155],"LLMs,":[156],"show":[158],"matches":[162],"or":[163],"surpasses":[164],"one-shot":[165],"GRPO":[166],"any":[170],"rewards,":[172],"while":[173],"reducing":[174],"inference":[175],"latency":[176],"over":[178],"10x":[179],"compared":[180],"MCMC-based":[182],"sampling.":[183]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-02-01T00:00:00"}