{"id":"https://openalex.org/W7170195842","doi":"https://doi.org/10.48550/arxiv.2607.20129","title":"CUSUM-Shaped Inference-Time Monitoring and Targeted Re-Decoding for Quantized Small Language Model Reasoning","display_name":"CUSUM-Shaped Inference-Time Monitoring and Targeted Re-Decoding for Quantized Small Language Model Reasoning","publication_year":2026,"publication_date":"2026-07-22","ids":{"openalex":"https://openalex.org/W7170195842","doi":"https://doi.org/10.48550/arxiv.2607.20129"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2607.20129","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.20129","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,"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.2607.20129","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5030656896","display_name":"El Hassane Ettifouri","orcid":"https://orcid.org/0000-0001-5299-9053"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ettifouri, El Hassane","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5143488973","display_name":"Ayoub Belfatmi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Belfatmi, Ayoub","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5083133361","display_name":"Mahaman Sanoussi Yahaya Alassan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Alassan, Mahaman Sanoussi Yahaya","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5143490387","display_name":"Walid Dahhane","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dahhane, Walid","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/T10028","display_name":"Topic Modeling","score":0.400299996137619,"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":0.400299996137619,"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":0.22709999978542328,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.09459999948740005,"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/mcnemars-test","display_name":"McNemar's test","score":0.7718999981880188},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.5877000093460083},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.42260000109672546},{"id":"https://openalex.org/keywords/trajectory","display_name":"Trajectory","score":0.42089998722076416},{"id":"https://openalex.org/keywords/causality","display_name":"Causality (physics)","score":0.3950999975204468},{"id":"https://openalex.org/keywords/recall","display_name":"Recall","score":0.36820000410079956},{"id":"https://openalex.org/keywords/semantics","display_name":"Semantics (computer science)","score":0.3628999888896942},{"id":"https://openalex.org/keywords/event","display_name":"Event (particle physics)","score":0.359499990940094},{"id":"https://openalex.org/keywords/selection","display_name":"Selection (genetic algorithm)","score":0.3443000018596649}],"concepts":[{"id":"https://openalex.org/C186282968","wikidata":"https://www.wikidata.org/wiki/Q1434261","display_name":"McNemar's test","level":2,"score":0.7718999981880188},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.612500011920929},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.5877000093460083},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.44350001215934753},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.42260000109672546},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.42089998722076416},{"id":"https://openalex.org/C64357122","wikidata":"https://www.wikidata.org/wiki/Q1149766","display_name":"Causality (physics)","level":2,"score":0.3950999975204468},{"id":"https://openalex.org/C100660578","wikidata":"https://www.wikidata.org/wiki/Q18733","display_name":"Recall","level":2,"score":0.36820000410079956},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.3628999888896942},{"id":"https://openalex.org/C2779662365","wikidata":"https://www.wikidata.org/wiki/Q5416694","display_name":"Event (particle physics)","level":2,"score":0.359499990940094},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.3443000018596649},{"id":"https://openalex.org/C5274069","wikidata":"https://www.wikidata.org/wiki/Q2285707","display_name":"Categorical variable","level":2,"score":0.33739998936653137},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.334199994802475},{"id":"https://openalex.org/C112313634","wikidata":"https://www.wikidata.org/wiki/Q7886648","display_name":"Complement (music)","level":5,"score":0.3264999985694885},{"id":"https://openalex.org/C2779119184","wikidata":"https://www.wikidata.org/wiki/Q294350","display_name":"ALARM","level":2,"score":0.3246000111103058},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3228999972343445},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.31779998540878296},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.31610000133514404},{"id":"https://openalex.org/C58489278","wikidata":"https://www.wikidata.org/wiki/Q1172284","display_name":"Data set","level":2,"score":0.30309998989105225},{"id":"https://openalex.org/C2778067643","wikidata":"https://www.wikidata.org/wiki/Q166507","display_name":"Interval (graph theory)","level":2,"score":0.29989999532699585},{"id":"https://openalex.org/C81669768","wikidata":"https://www.wikidata.org/wiki/Q2359161","display_name":"Precision and recall","level":2,"score":0.2978000044822693},{"id":"https://openalex.org/C3018260909","wikidata":"https://www.wikidata.org/wiki/Q1322871","display_name":"Exploratory analysis","level":2,"score":0.28929999470710754},{"id":"https://openalex.org/C107673813","wikidata":"https://www.wikidata.org/wiki/Q812534","display_name":"Bayesian probability","level":2,"score":0.28700000047683716},{"id":"https://openalex.org/C115086926","wikidata":"https://www.wikidata.org/wiki/Q17004651","display_name":"Causal reasoning","level":3,"score":0.2842999994754791},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2840999960899353},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.28220000863075256},{"id":"https://openalex.org/C127705205","wikidata":"https://www.wikidata.org/wiki/Q5748245","display_name":"Heuristics","level":2,"score":0.2628999948501587},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.2567000091075897},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.2513999938964844}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2607.20129","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.20129","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,"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.2607.20129","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2607.20129","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,"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":{"Quantized":[0],"small":[1],"autoregressive":[2],"reasoning":[3,254],"models":[4],"can":[5],"enter":[6],"long,":[7],"repetitive,":[8],"or":[9,183,233,251],"unproductive":[10],"trajectories,":[11],"yet":[12],"inference-time":[13],"compute":[14],"is":[15,188,220],"usually":[16],"allocated":[17],"without":[18],"observing":[19],"how":[20],"a":[21,37,61,71,125,230,238,249],"trajectory":[22],"develops.":[23],"Building":[24],"on":[25,89],"an":[26,67,192],"earlier":[27],"token-level":[28],"e-CUSUM":[29],"controller,":[30],"we":[31,104],"develop":[32],"MGT-B":[33],"(Monitoring-Guided":[34],"Test-time":[35],"Backtracking),":[36],"revised":[38],"external":[39],"controller":[40],"that":[41],"maps":[42],"overlapping":[43],"windows":[44],"of":[45,98,161],"pre-sampling":[46],"uncertainty":[47],"and":[48,64,76,79,113,187,215,222],"degeneration":[49],"features":[50],"to":[51,66,136,168,205],"position-conditional":[52],"empirical":[53,224],"tail":[54],"probabilities,":[55],"accumulates":[56],"mixture":[57],"betting":[58],"factors":[59,225],"with":[60],"CUSUM-shaped":[62],"reset,":[63],"responds":[65],"alarm":[68],"by":[69],"estimating":[70],"rollback":[72],"point,":[73],"restoring":[74],"token":[75],"key-value-cache":[77],"state,":[78],"performing":[80],"constrained":[81],"re-decoding.":[82],"To":[83],"audit":[84],"whether":[85],"the":[86,95,115,200,208,223,243],"effect":[87],"persists":[88],"problem":[90],"identities":[91],"first":[92,117],"observed":[93],"after":[94],"manual":[96],"choice":[97],"log":[99],"threshold":[100,185],"h":[101],"=":[102,148,174],"10,":[103],"retrospectively":[105],"exclude":[106],"260":[107],"IDs":[108,180],"present":[109],"in":[110,199],"pre-threshold":[111],"artifacts":[112],"retain":[114],"chronologically":[116],"post-threshold":[118],"pair":[119],"for":[120,242],"each":[121],"remaining":[122],"ID,":[123],"yielding":[124],"240-pair":[126],"chronology-audit":[127],"set.":[128],"On":[129],"this":[130],"set,":[131],"accuracy":[132,165],"changes":[133,164],"from":[134,166],"82/240":[135],"88/240":[137],"(+2.50":[138],"percentage":[139],"points;":[140,171],"13":[141],"corrections,":[142],"7":[143],"regressions;":[144],"exact":[145],"McNemar":[146,172],"p":[147,173],"0.2632;":[149],"paired":[150],"bootstrap":[151],"95%":[152],"interval":[153],"[-1.25,":[154],"+6.25]).":[155],"A":[156],"broader":[157],"467-pair":[158,201],"historical-coverage":[159],"set":[160,202],"seed-matched":[162],"pairs":[163],"146/467":[167],"167/467":[169],"(+4.50":[170],"0.000753),":[175],"but":[176],"includes":[177],"200":[178],"seed-1":[179],"available":[181],"before":[182],"during":[184],"selection":[186],"reported":[189],"only":[190],"as":[191,229],"exploratory":[193],"estimate.":[194],"All":[195],"316":[196],"no-alarm":[197],"outputs":[198],"are":[203,226],"identical":[204],"vanilla,":[206],"while":[207],"151":[209],"alarmed":[210],"trajectories":[211],"contain":[212],"29":[213],"corrections":[214],"8":[216],"regressions.":[217],"Neither":[218],"analysis":[219],"confirmatory,":[221],"not":[227],"established":[228],"valid":[231],"e-process":[232],"e-detector.":[234],"The":[235],"results":[236],"support":[237],"selective":[239],"monitoring-and-repair":[240],"mechanism":[241],"studied":[244],"MATH-500":[245],"setting,":[246],"rather":[247],"than":[248],"general":[250],"theoretically":[252],"certified":[253],"improvement.":[255]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-07-24T00:00:00"}