{"id":"https://openalex.org/W7225798571","doi":"https://doi.org/10.5281/zenodo.24136373","title":"Who Checks the Weight Editors? Controls That Change the Reading of LLM-Authored Weight Edits and of Prompt-Rule Effects in Small Language Models","display_name":"Who Checks the Weight Editors? Controls That Change the Reading of LLM-Authored Weight Edits and of Prompt-Rule Effects in Small Language Models","publication_year":2026,"publication_date":"2026-11-11","ids":{"openalex":"https://openalex.org/W7225798571","doi":"https://doi.org/10.5281/zenodo.24136373"},"language":"en","primary_location":{"id":"doi:10.5281/zenodo.24136373","doi":"https://doi.org/10.5281/zenodo.24136373","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24136373","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.5281/zenodo.24136373","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5161444832","display_name":"Vagner Campos","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I863293078","display_name":"Walsh College","ror":"https://ror.org/05s4pg032","country_code":"US","type":"education","lineage":["https://openalex.org/I863293078"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Campos, Vagner","raw_affiliation_strings":["Walsh College","Siberia Institute"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Walsh College","institution_ids":["https://openalex.org/I863293078"]},{"raw_affiliation_string":"Siberia Institute","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":["https://openalex.org/A5161444832"],"corresponding_institution_ids":["https://openalex.org/I863293078"],"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":0.5557000041007996,"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":0.5557000041007996,"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":0.15320000052452087,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.06849999725818634,"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":[],"concepts":[{"id":"https://openalex.org/C554936623","wikidata":"https://www.wikidata.org/wiki/Q199657","display_name":"Reading (process)","level":2,"score":0.7576000094413757},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6000000238418579},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.4505999982357025},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.43290001153945923},{"id":"https://openalex.org/C4438859","wikidata":"https://www.wikidata.org/wiki/Q186117","display_name":"Timeline","level":2,"score":0.39899998903274536},{"id":"https://openalex.org/C2777267654","wikidata":"https://www.wikidata.org/wiki/Q3519023","display_name":"Test (biology)","level":2,"score":0.39739999175071716},{"id":"https://openalex.org/C204434341","wikidata":"https://www.wikidata.org/wiki/Q357789","display_name":"Adjudication","level":2,"score":0.39239999651908875},{"id":"https://openalex.org/C2781162219","wikidata":"https://www.wikidata.org/wiki/Q26250693","display_name":"Replicate","level":2,"score":0.3765999972820282},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3626999855041504},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.35850000381469727},{"id":"https://openalex.org/C2779010991","wikidata":"https://www.wikidata.org/wiki/Q2720909","display_name":"Artifact (error)","level":2,"score":0.33640000224113464},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.33390000462532043},{"id":"https://openalex.org/C184670325","wikidata":"https://www.wikidata.org/wiki/Q512604","display_name":"Loop (graph theory)","level":2,"score":0.3089999854564667},{"id":"https://openalex.org/C2775922551","wikidata":"https://www.wikidata.org/wiki/Q7135033","display_name":"Parallels","level":2,"score":0.29739999771118164},{"id":"https://openalex.org/C44291984","wikidata":"https://www.wikidata.org/wiki/Q1074173","display_name":"Question answering","level":2,"score":0.2858999967575073},{"id":"https://openalex.org/C182365436","wikidata":"https://www.wikidata.org/wiki/Q50701","display_name":"Variable (mathematics)","level":2,"score":0.2754000127315521},{"id":"https://openalex.org/C2781291010","wikidata":"https://www.wikidata.org/wiki/Q178580","display_name":"Period (music)","level":2,"score":0.2689000070095062},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.2612999975681305},{"id":"https://openalex.org/C169590947","wikidata":"https://www.wikidata.org/wiki/Q47506","display_name":"Compiler","level":2,"score":0.2572999894618988},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.2563999891281128},{"id":"https://openalex.org/C2779236577","wikidata":"https://www.wikidata.org/wiki/Q5421970","display_name":"Extensive reading","level":3,"score":0.25540000200271606}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.5281/zenodo.24136373","doi":"https://doi.org/10.5281/zenodo.24136373","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24136373","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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"},{"id":"doi:10.5281/zenodo.24136374","doi":"https://doi.org/10.5281/zenodo.24136374","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24136374","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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":"doi:10.5281/zenodo.24136373","doi":"https://doi.org/10.5281/zenodo.24136373","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24136373","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"AI":[0],"agents":[1],"have":[2],"begun":[3],"to":[4,75,117,275,284],"write":[5],"edits":[6,104],"directly":[7],"into":[8,209],"the":[9,17,24,44,97,110,124,135,145,171,181,187,210,215,222,269,293,296,300,314,316,320],"weights":[10],"of":[11,26,58,127,147,170,221,299,313],"language":[12],"models.":[13],"We":[14,133],"argue":[15],"that":[16,34,140,160,179],"bottleneck":[18],"in":[19,67,151,201,254,331],"this":[20],"emerging":[21],"area":[22],"is":[23,142,304,308,339],"adjudication":[25],"claimed":[27],"gains,":[28],"and":[29,60,115,137,225,288,319,325],"propose":[30],"seven":[31,202],"inexpensive":[32],"controls":[33],"any":[35,206],"such":[36],"claim":[37],"should":[38],"pass:":[39],"hash-verified":[40],"replication,":[41],"failure-mode":[42],"decomposition,":[43],"cheapest":[45],"alternative,":[46],"dose":[47],"arms":[48],"with":[49,71,156,164,173,306],"a":[50,53,72,78,148,197,250,265],"random-direction":[51],"control,":[52],"double":[54],"placebo,":[55],"manual":[56],"reading":[57,62,263],"outcomes":[59],"pre-declared":[61],"rules.":[63],"The":[64,311],"work":[65,234],"began":[66],"late":[68],"October":[69,94],"2026":[70,239],"harness":[73,112],"built":[74],"test":[76],"whether":[77],"prompt-level":[79],"grounding":[80,252],"rule":[81],"acts":[82],"through":[83],"an":[84,194,227,289,309],"identifiable":[85],"internal":[86],"direction":[87,183,199,230,303],"(harness":[88],"archive":[89],"SHA-256":[90],"04e1e4d9\u2026,":[91],"dated":[92],"30\u201331":[93],"2026).":[95],"When":[96],"MIT":[98],"CSAIL":[99],"report":[100],"on":[101,106,236],"LLM-authored":[102,125],"weight":[103,166],"appeared":[105],"2":[107,248],"November":[108,238],"2026,":[109],"same":[111,211,216],"was":[113],"extended":[114],"used":[116],"cross-validate":[118],"it":[119,141,191],"independently.":[120],"Case":[121,247,337],"1":[122,338],"re-examines":[123],"edit":[126],"Qwen3.5-0.8B":[128],"(MATH-500":[129],"7.9%":[130],"\u2192":[131],"48.1%).":[132],"replicate":[134],"gain":[136,172,224],"(i)":[138],"quantify":[139],"almost":[143],"entirely":[144],"repair":[146],"generation":[149],"loop":[150,198],"reasoning":[152,162],"mode;":[153],"(ii)":[154],"consistent":[155],"independent":[157],"evaluations,":[158],"show":[159,178],"switching":[161],"off":[163],"no":[165,285],"change":[167],"recovers":[168,218],"most":[169],"6\u20139\u00d7":[174],"fewer":[175],"tokens;":[176],"(iii)":[177],"ablating":[180],"edited":[182],"does":[184],"not":[185],"reproduce":[186],"gain,":[188],"whereas":[189],"reflecting":[190],"does.":[192],"In":[193],"exploratory":[195],"follow-up,":[196],"found":[200],"CPU":[203],"minutes":[204],"without":[205],"agent,":[207],"written":[208],"64":[212],"columns":[213],"at":[214],"norm,":[217],"about":[219],"63%":[220],"agent's":[223],"beats":[226],"equal-norm":[228],"random":[229],"(60":[231],"problems).":[232],"Concurrent":[233],"posted":[235],"4":[237],"reached":[240],"convergent":[241],"results;":[242],"both":[243,276],"timelines":[244],"are":[245,323],"reported.":[246],"evaluates":[249],"short":[251],"instruction":[253,270],"Qwen2.5-7B-Instruct":[255],"against":[256,295],"two":[257],"length-matched":[258],"placebos.":[259,277],"Under":[260],"partially":[261],"blinded":[262],"by":[264,327],"single":[266],"LLM":[267],"reader,":[268],"reduces":[271],"invented":[272],"specifics":[273],"relative":[274,283],"Both":[278],"placebos,":[279],"however,":[280],"increase":[281],"fabrication":[282],"system":[286],"text,":[287],"automatic":[290],"extractor":[291],"biased":[292],"comparison":[294],"instruction.":[297],"70\u201390%":[298],"rule-minus-placebo":[301],"activation":[302],"shared":[305],"'this":[307],"instruction'.":[310],"text":[312],"instruction,":[315],"placebo":[317],"texts":[318],"new":[321],"items":[322],"withheld":[324],"committed":[326],"hash":[328],"(PROVENIENCIA_SHA256.txt,":[329],"anchored":[330],"Bitcoin":[332],"block":[333],"974,965":[334],"via":[335],"OpenTimestamps).":[336],"fully":[340],"reproducible":[341],"from":[342],"public":[343],"artefacts.":[344]},"counts_by_year":[],"updated_date":"2026-11-12T12:28:57.059296","created_date":"2026-11-12T00:00:00"}