{"id":"https://openalex.org/W7225184162","doi":"https://doi.org/10.3390/app162110895","title":"QLoRA Fine-Tuning for Theory-of-Mind Reasoning in a 3B-Parameter Small Language Model: A ToMBench-Based Cross-Benchmark Generalization Study","display_name":"QLoRA Fine-Tuning for Theory-of-Mind Reasoning in a 3B-Parameter Small Language Model: A ToMBench-Based Cross-Benchmark Generalization Study","publication_year":2026,"publication_date":"2026-11-03","ids":{"openalex":"https://openalex.org/W7225184162","doi":"https://doi.org/10.3390/app162110895"},"language":"en","primary_location":{"id":"doi:10.3390/app162110895","doi":"https://doi.org/10.3390/app162110895","is_oa":true,"landing_page_url":"https://doi.org/10.3390/app162110895","pdf_url":null,"source":{"id":"https://openalex.org/S4210205812","display_name":"Applied Sciences","issn_l":"2076-3417","issn":["2076-3417"],"is_oa":true,"is_in_doaj":true,"is_core":true,"listed_in":["anvur-class-a","anvur-scientific","cwts-core","doaj","norway-1","oarl","poland-100","russia-white-list-2","vabb-shw"],"host_organization":"https://openalex.org/P4310310987","host_organization_name":"Multidisciplinary Digital Publishing Institute","host_organization_lineage":["https://openalex.org/P4310310987"],"host_organization_lineage_names":["Multidisciplinary Digital Publishing Institute"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Applied Sciences","raw_type":"journal-article"},"volume":null,"type":"article","is_traditional":true,"indexed_in":["crossref","doaj"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.3390/app162110895","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5160358047","display_name":"Ji-Hyeong Hong","orcid":"https://orcid.org/0009-0007-1244-7669","observed_orcids":["https://orcid.org/0009-0007-1244-7669"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ji-Hyeong Hong","raw_affiliation_strings":["Division of ICT Convergence Security, Seoul Hoseo Occupational Training College, 420 Gangseo-ro, Gangseo-gu, Seoul 07583, Republic of Korea"],"raw_orcid":"https://orcid.org/0009-0007-1244-7669","affiliations":[{"raw_affiliation_string":"Division of ICT Convergence Security, Seoul Hoseo Occupational Training College, 420 Gangseo-ro, Gangseo-gu, Seoul 07583, Republic of Korea","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5032371858","display_name":"Sanghyun Yoo","orcid":"https://orcid.org/0009-0005-4644-7793","observed_orcids":["https://orcid.org/0009-0005-4644-7793","https://orcid.org/0000-0002-8097-7975","https://orcid.org/0009-0000-3447-4624","https://orcid.org/0009-0008-9199-8238"]},"institutions":[{"id":"https://openalex.org/I141371507","display_name":"Soongsil University","ror":"https://ror.org/017xnm587","country_code":"KR","type":"education","lineage":["https://openalex.org/I141371507"]}],"countries":["KR"],"is_corresponding":true,"raw_author_name":"Sang-Hyun Yoo","raw_affiliation_strings":["School of Computer Science and Engineering, Soongsil University, 50 Sadang-ro, Dongjak-gu, Seoul 07027, Republic of Korea"],"raw_orcid":"https://orcid.org/0009-0008-9199-8238","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Soongsil University, 50 Sadang-ro, Dongjak-gu, Seoul 07027, Republic of Korea","institution_ids":["https://openalex.org/I141371507"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":["https://openalex.org/A5032371858"],"corresponding_institution_ids":["https://openalex.org/I141371507"],"apc_list":{"value":2400,"currency":"CHF","value_usd":2896},"apc_paid":{"value":2400,"currency":"CHF","value_usd":2896},"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.90271596,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":null,"biblio":{"volume":"16","issue":"21","first_page":"10895","last_page":"10895"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10656","display_name":"Child and Animal Learning Development","score":0.3659999966621399,"subfield":{"id":"https://openalex.org/subfields/3204","display_name":"Developmental and Educational Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},"topics":[{"id":"https://openalex.org/T10656","display_name":"Child and Animal Learning Development","score":0.3659999966621399,"subfield":{"id":"https://openalex.org/subfields/3204","display_name":"Developmental and Educational Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.21960000693798065,"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":0.1582999974489212,"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/qlora","display_name":"QLoRA","score":0.843999981880188},{"id":"https://openalex.org/keywords/low-rank-adaptation","display_name":"low-rank adaptation","score":0.8220000267028809},{"id":"https://openalex.org/keywords/fine-tuning","display_name":"fine-tuning","score":0.6570000052452087},{"id":"https://openalex.org/keywords/small-language-models","display_name":"small language models","score":0.8970000147819519},{"id":"https://openalex.org/keywords/tomis","display_name":"Tomis","score":0.6100000143051147}],"concepts":[{"id":"https://openalex.org/C21308566","wikidata":"https://www.wikidata.org/wiki/Q7169365","display_name":"Permutation (music)","level":2,"score":0.6105999946594238},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.5519999861717224},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5461999773979187},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5081999897956848},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.439300000667572},{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.42570000886917114},{"id":"https://openalex.org/C2776175482","wikidata":"https://www.wikidata.org/wiki/Q1195816","display_name":"Transfer (computing)","level":2,"score":0.37630000710487366},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3659000098705292},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3483999967575073},{"id":"https://openalex.org/C2778067643","wikidata":"https://www.wikidata.org/wiki/Q166507","display_name":"Interval (graph theory)","level":2,"score":0.3418999910354614},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.33730000257492065},{"id":"https://openalex.org/C198082294","wikidata":"https://www.wikidata.org/wiki/Q3399648","display_name":"Position (finance)","level":2,"score":0.31790000200271606},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.31790000200271606},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2971999943256378},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.2808000147342682},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.25279998779296875},{"id":"https://openalex.org/C2776434776","wikidata":"https://www.wikidata.org/wiki/Q19246213","display_name":"Domain adaptation","level":3,"score":0.2513999938964844}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.3390/app162110895","doi":"https://doi.org/10.3390/app162110895","is_oa":true,"landing_page_url":"https://doi.org/10.3390/app162110895","pdf_url":null,"source":{"id":"https://openalex.org/S4210205812","display_name":"Applied Sciences","issn_l":"2076-3417","issn":["2076-3417"],"is_oa":true,"is_in_doaj":true,"is_core":true,"listed_in":["anvur-class-a","anvur-scientific","cwts-core","doaj","norway-1","oarl","poland-100","russia-white-list-2","vabb-shw"],"host_organization":"https://openalex.org/P4310310987","host_organization_name":"Multidisciplinary Digital Publishing Institute","host_organization_lineage":["https://openalex.org/P4310310987"],"host_organization_lineage_names":["Multidisciplinary Digital Publishing Institute"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Applied Sciences","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.3390/app162110895","doi":"https://doi.org/10.3390/app162110895","is_oa":true,"landing_page_url":"https://doi.org/10.3390/app162110895","pdf_url":null,"source":{"id":"https://openalex.org/S4210205812","display_name":"Applied Sciences","issn_l":"2076-3417","issn":["2076-3417"],"is_oa":true,"is_in_doaj":true,"is_core":true,"listed_in":["anvur-class-a","anvur-scientific","cwts-core","doaj","norway-1","oarl","poland-100","russia-white-list-2","vabb-shw"],"host_organization":"https://openalex.org/P4310310987","host_organization_name":"Multidisciplinary Digital Publishing Institute","host_organization_lineage":["https://openalex.org/P4310310987"],"host_organization_lineage_names":["Multidisciplinary Digital Publishing Institute"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Applied Sciences","raw_type":"journal-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4416644609","https://openalex.org/W7197036620","https://openalex.org/W7163997117","https://openalex.org/W7202211372","https://openalex.org/W4402671125","https://openalex.org/W7128698738","https://openalex.org/W7131673088","https://openalex.org/W7128034750","https://openalex.org/W7131793712","https://openalex.org/W7162848887"],"abstract_inverted_index":{"We":[0,27,168],"investigate":[1],"whether":[2,18],"quantized":[3],"low-rank":[4,32],"adaptation":[5],"(QLoRA)":[6],"fine-tuning":[7,45,189],"improves":[8],"theory-of-mind":[9],"(ToM)":[10],"reasoning":[11],"in":[12,68,115],"a":[13,38,126,191,195],"small":[14],"language":[15],"model,":[16],"and":[17,121,156,181,194],"apparent":[19],"cross-benchmark":[20,185],"transfer":[21,177],"survives":[22],"control":[23],"for":[24,187],"response-position":[25],"bias.":[26],"fine-tuned":[28],"4-bit":[29],"Qwen2.5-3B-Instruct":[30],"with":[31,41,91,152,158],"(LoRA)":[33],"adapters":[34],"on":[35,51,107],"ToMBench":[36,73],"under":[37,58,86,129,135,142,198],"story-disjoint":[39,137],"split":[40,132],"five":[42],"seeds.":[43],"Because":[44],"shifted":[46],"the":[47,95,111,116,130,136,143,165,173,199],"model\u2019s":[48],"answer-letter":[49],"preference":[50],"every":[52,69],"benchmark,":[53],"we":[54,182],"evaluated":[55],"all":[56],"models":[57],"cyclic":[59],"permutation":[60,145],"of":[61,178],"answer":[62,67,113],"options,":[63],"placing":[64],"each":[65],"gold":[66,112],"position":[70],"once.":[71],"Held-out":[72],"accuracy":[74,148],"rose":[75],"by":[76,149],"+8.15":[77],"points":[78,151],"(between-seed":[79],"95%":[80],"interval":[81],"[+7.46,":[82],"+8.85]),":[83],"against":[84],"+14.01":[85],"an":[87],"item-level":[88,96,131],"split,":[89],"consistent":[90],"story":[92],"overlap":[93],"inflating":[94],"estimate.":[97],"On":[98],"external":[99],"benchmarks,":[100],"ToMi":[101,147,174],"improved":[102],"(+14.19":[103],"[+11.87,":[104],"+16.50]),":[105],"mostly":[106],"items":[108],"where":[109],"only":[110],"appears":[114],"story,":[117],"whereas":[118],"SocialIQa,":[119],"OpenToM":[120],"Hi-ToM":[122],"showed":[123],"no":[124],"direction;":[125],"SocialIQa":[127],"decline":[128],"(\u22124.14)":[133],"disappeared":[134],"split.":[138],"Training-free":[139],"three-shot":[140],"prompting":[141,196],"same":[144,200],"raised":[146],"+7.23":[150],"content-free":[153],"format":[154],"exemplars":[155,160],"+12.96":[157],"rationale":[159],"but":[161],"did":[162],"not":[163,171],"reproduce":[164],"in-domain":[166],"gain.":[167],"therefore":[169],"do":[170],"read":[172],"gain":[175],"as":[176],"ToM":[179,188],"reasoning,":[180],"conclude":[183],"that":[184],"claims":[186],"require":[190],"position-invariant":[192],"evaluation":[193],"baseline":[197],"control.":[201]},"counts_by_year":[],"updated_date":"2026-11-05T12:25:35.769452","created_date":"2026-11-04T00:00:00"}