{"id":"https://openalex.org/W7128723525","doi":"https://doi.org/10.48550/arxiv.2602.10273","title":"Power-SMC: Low-Latency Sequence-Level Power Sampling for Training-Free LLM Reasoning","display_name":"Power-SMC: Low-Latency Sequence-Level Power Sampling for Training-Free LLM Reasoning","publication_year":2026,"publication_date":"2026-02-10","ids":{"openalex":"https://openalex.org/W7128723525","doi":"https://doi.org/10.48550/arxiv.2602.10273"},"language":null,"primary_location":{"id":"pmh:doi:10.48550/arxiv.2602.10273","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/A5125746437","display_name":"Seyedarmin Azizi","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Azizi, Seyedarmin","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125691204","display_name":"Erfan Baghaei Potraghloo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Potraghloo, Erfan Baghaei","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5007824421","display_name":"Minoo Ahmadi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ahmadi, Minoo","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5080397425","display_name":"Souvik Kundu","orcid":"https://orcid.org/0000-0002-1189-3073"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kundu, Souvik","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Pedram, Massoud","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pedram, Massoud","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5125746437"],"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/T10028","display_name":"Topic Modeling","score":0.3407000005245209,"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":0.3407000005245209,"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":0.2556999921798706,"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/T13629","display_name":"Text Readability and Simplification","score":0.04859999939799309,"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/monte-carlo-method","display_name":"Monte Carlo method","score":0.5759000182151794},{"id":"https://openalex.org/keywords/importance-sampling","display_name":"Importance sampling","score":0.5206000208854675},{"id":"https://openalex.org/keywords/residual","display_name":"Residual","score":0.5192999839782715},{"id":"https://openalex.org/keywords/stability","display_name":"Stability (learning theory)","score":0.5174000263214111},{"id":"https://openalex.org/keywords/sampling","display_name":"Sampling (signal processing)","score":0.5073000192642212},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.49160000681877136},{"id":"https://openalex.org/keywords/schedule","display_name":"Schedule","score":0.4244999885559082},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.4223000109195709},{"id":"https://openalex.org/keywords/sampling-distribution","display_name":"Sampling distribution","score":0.4187000095844269}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6043000221252441},{"id":"https://openalex.org/C19499675","wikidata":"https://www.wikidata.org/wiki/Q232207","display_name":"Monte Carlo method","level":2,"score":0.5759000182151794},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.5206000208854675},{"id":"https://openalex.org/C155512373","wikidata":"https://www.wikidata.org/wiki/Q287450","display_name":"Residual","level":2,"score":0.5192999839782715},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.5174000263214111},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.5073000192642212},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.49160000681877136},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4830999970436096},{"id":"https://openalex.org/C68387754","wikidata":"https://www.wikidata.org/wiki/Q7271585","display_name":"Schedule","level":2,"score":0.4244999885559082},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.4223000109195709},{"id":"https://openalex.org/C167723999","wikidata":"https://www.wikidata.org/wiki/Q3773214","display_name":"Sampling distribution","level":2,"score":0.4187000095844269},{"id":"https://openalex.org/C163258240","wikidata":"https://www.wikidata.org/wiki/Q25342","display_name":"Power (physics)","level":2,"score":0.3855000138282776},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.35760000348091125},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3449999988079071},{"id":"https://openalex.org/C149441793","wikidata":"https://www.wikidata.org/wiki/Q200726","display_name":"Probability distribution","level":2,"score":0.33640000224113464},{"id":"https://openalex.org/C25432639","wikidata":"https://www.wikidata.org/wiki/Q3711790","display_name":"Weight distribution","level":2,"score":0.33079999685287476},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.3174999952316284},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.3100999891757965},{"id":"https://openalex.org/C52421305","wikidata":"https://www.wikidata.org/wiki/Q1151499","display_name":"Particle filter","level":3,"score":0.30799999833106995},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.2921000123023987},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.29030001163482666},{"id":"https://openalex.org/C150921843","wikidata":"https://www.wikidata.org/wiki/Q1170431","display_name":"Resampling","level":2,"score":0.287200003862381},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.28690001368522644},{"id":"https://openalex.org/C110121322","wikidata":"https://www.wikidata.org/wiki/Q865811","display_name":"Distribution (mathematics)","level":2,"score":0.28540000319480896},{"id":"https://openalex.org/C75917345","wikidata":"https://www.wikidata.org/wiki/Q2725298","display_name":"Sampling bias","level":3,"score":0.2851000130176544},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.28439998626708984},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.26579999923706055},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.2556000053882599},{"id":"https://openalex.org/C134261354","wikidata":"https://www.wikidata.org/wiki/Q938438","display_name":"Statistical inference","level":2,"score":0.2506999969482422}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.48550/arxiv.2602.10273","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.2602.10273","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.10273","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.2602.10273","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":{"Many":[0],"recent":[1],"reasoning":[2,66],"gains":[3],"in":[4,100],"large":[5],"language":[6],"models":[7],"can":[8],"be":[9],"explained":[10],"as":[11],"distribution":[12,37,63],"sharpening:":[13],"biasing":[14],"generation":[15],"toward":[16],"high-likelihood":[17],"trajectories":[18],"already":[19],"supported":[20],"by":[21],"the":[22,34,84,123,149],"pretrained":[23],"model,":[24],"rather":[25],"than":[26],"modifying":[27],"its":[28],"weights.":[29],"A":[30],"natural":[31],"formalization":[32],"is":[33,122,172],"sequence-level":[35],"power":[36,158],"$\u03c0_\u03b1(y\\mid":[38],"x)\\propto":[39],"p_\u03b8(y\\mid":[40],"x)^\u03b1$":[41],"($\u03b1>1$),":[42],"which":[43],"concentrates":[44],"mass":[45],"on":[46],"whole":[47],"sequences":[48],"instead":[49],"of":[50],"adjusting":[51],"token-level":[52],"temperature.":[53],"Prior":[54],"work":[55],"shows":[56],"that":[57,82,119,143],"Metropolis--Hastings":[58],"(MH)":[59],"sampling":[60,159],"from":[61,163],"this":[62],"recovers":[64],"strong":[65],"performance,":[67],"but":[68],"at":[69,174],"order-of-magnitude":[70],"inference":[71],"slowdowns.":[72],"We":[73,117],"introduce":[74,139],"Power-SMC,":[75],"a":[76,96,112],"training-free":[77],"Sequential":[78],"Monte":[79],"Carlo":[80],"scheme":[81],"targets":[83],"same":[85],"objective":[86],"while":[87,160],"remaining":[88],"close":[89],"to":[90,165],"standard":[91],"decoding":[92],"latency.":[93],"Power-SMC":[94,153],"advances":[95],"small":[97],"particle":[98,145],"set":[99],"parallel,":[101],"corrects":[102],"importance":[103],"weights":[104],"token-by-token,":[105],"and":[106,138],"resamples":[107],"when":[108],"necessary,":[109],"all":[110],"within":[111],"single":[113],"GPU-friendly":[114],"batched":[115],"decode.":[116],"prove":[118],"temperature":[120],"$\u03c4=1/\u03b1$":[121],"unique":[124],"prefix-only":[125],"proposal":[126],"minimizing":[127],"incremental":[128],"weight":[129],"variance,":[130],"interpret":[131],"residual":[132],"instability":[133],"via":[134],"prefix-conditioned":[135],"R\u00e9nyi":[136],"entropies,":[137],"an":[140],"exponent-bridging":[141],"schedule":[142],"improves":[144],"stability":[146],"without":[147],"altering":[148],"target.":[150],"On":[151],"MATH500,":[152],"matches":[154],"or":[155],"exceeds":[156],"MH":[157],"reducing":[161],"latency":[162],"$16$--$28\\times$":[164],"$1.4$--$3.3\\times$":[166],"over":[167],"baseline":[168],"decoding.":[169],"The":[170],"code":[171],"available":[173],"https://github.com/ArminAzizi98/Power-SMC.":[175]},"counts_by_year":[],"updated_date":"2026-04-04T16:13:02.066488","created_date":"2026-02-13T00:00:00"}