{"id":"https://openalex.org/W7155161495","doi":"https://doi.org/10.48550/arxiv.2604.18966","title":"Self-Improving Tabular Language Models via Iterative Reward-Guided Post-Training","display_name":"Self-Improving Tabular Language Models via Iterative Reward-Guided Post-Training","publication_year":2026,"publication_date":"2026-04-21","ids":{"openalex":"https://openalex.org/W7155161495","doi":"https://doi.org/10.48550/arxiv.2604.18966"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.18966","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.18966","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":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.2604.18966","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5134358152","display_name":"Yunbo Long","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Long, Yunbo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5077949475","display_name":"Tejumade Afonja","orcid":"https://orcid.org/0000-0003-0639-9668"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Afonja, Tejumade","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134268542","display_name":"Alexandra Brintrup","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hao, Guangya","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5003887059","display_name":"Mario Fritz","orcid":"https://orcid.org/0000-0001-8949-9896"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Brintrup, Alexandra","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Fritz, Mario","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fritz, Mario","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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.17059999704360962,"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"}},"topics":[{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.17059999704360962,"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"}},{"id":"https://openalex.org/T12535","display_name":"Machine Learning and Data Classification","score":0.12110000103712082,"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":0.11050000041723251,"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/language-model","display_name":"Language model","score":0.5435000061988831},{"id":"https://openalex.org/keywords/modular-design","display_name":"Modular design","score":0.5192999839782715},{"id":"https://openalex.org/keywords/classifier","display_name":"Classifier (UML)","score":0.4973999857902527},{"id":"https://openalex.org/keywords/statistical-model","display_name":"Statistical model","score":0.450300008058548},{"id":"https://openalex.org/keywords/autoregressive-model","display_name":"Autoregressive model","score":0.43220001459121704},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.41909998655319214},{"id":"https://openalex.org/keywords/probabilistic-logic","display_name":"Probabilistic logic","score":0.4047999978065491},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.3952000141143799}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8014000058174133},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5462999939918518},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.5435000061988831},{"id":"https://openalex.org/C101468663","wikidata":"https://www.wikidata.org/wiki/Q1620158","display_name":"Modular design","level":2,"score":0.5192999839782715},{"id":"https://openalex.org/C95623464","wikidata":"https://www.wikidata.org/wiki/Q1096149","display_name":"Classifier (UML)","level":2,"score":0.4973999857902527},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.450300008058548},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.43220001459121704},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.41909998655319214},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4097999930381775},{"id":"https://openalex.org/C49937458","wikidata":"https://www.wikidata.org/wiki/Q2599292","display_name":"Probabilistic logic","level":2,"score":0.4047999978065491},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.3952000141143799},{"id":"https://openalex.org/C2781181686","wikidata":"https://www.wikidata.org/wiki/Q4226068","display_name":"Coherence (philosophical gambling strategy)","level":2,"score":0.39250001311302185},{"id":"https://openalex.org/C42812","wikidata":"https://www.wikidata.org/wiki/Q1082910","display_name":"Partition (number theory)","level":2,"score":0.33160001039505005},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.3149000108242035},{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.3050000071525574},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2992999851703644},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.296999990940094},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.28780001401901245},{"id":"https://openalex.org/C179518139","wikidata":"https://www.wikidata.org/wiki/Q5140297","display_name":"Coding (social sciences)","level":2,"score":0.2847000062465668},{"id":"https://openalex.org/C55439883","wikidata":"https://www.wikidata.org/wiki/Q360812","display_name":"Correctness","level":2,"score":0.27959999442100525},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.2752000093460083},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.25429999828338623}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.18966","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.18966","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":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.18966","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.18966","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":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":{"Tabular":[0],"language":[1,56],"models":[2,57],"can":[3,183],"generate":[4],"synthetic":[5,47,66],"tables":[6],"by":[7],"modeling":[8],"rows":[9],"as":[10,25,206],"token":[11],"sequences,":[12],"but":[13],"they":[14],"are":[15],"typically":[16],"trained":[17],"once":[18],"with":[19],"supervised":[20,82,125,152],"fine-tuning":[21,126],"and":[22,41,73,101,127,137,142,164,174,187,190],"then":[23],"used":[24,44],"static":[26,218],"synthesizers.":[27,220],"This":[28],"is":[29,76,195],"limiting":[30],"because":[31],"next-token":[32],"likelihood":[33],"does":[34],"not":[35],"directly":[36],"optimize":[37],"the":[38,74,129,151,157,180,199],"distributional,":[39],"utility,":[40,144],"indistinguishability":[42],"properties":[43],"to":[45,79,216],"evaluate":[46],"data.":[48],"We":[49],"study":[50],"iterative":[51],"reward-guided":[52],"post-training":[53,181,209],"for":[54,197,211],"tabular":[55,212,219],"through":[58],"a":[59,63,68,80,94,120,207],"generate--score--align":[60],"protocol,":[61,86],"where":[62],"generator":[64],"samples":[65],"rows,":[67],"task-specified":[69],"reward":[70,162],"ranks":[71],"them,":[72],"model":[75],"updated":[77],"relative":[78],"fixed":[81],"reference.":[83],"Within":[84],"this":[85],"we":[87],"propose":[88],"\\textbf{TabGRAA}":[89],"(\\textbf{Tab}ular":[90],"\\textbf{G}roup-\\textbf{R}elative":[91],"\\textbf{A}dvantage":[92],"\\textbf{A}lignment),":[93],"group-relative":[95],"alignment":[96],"method":[97,210],"that":[98,156,179,191],"compares":[99],"high-":[100],"low-reward":[102],"generated":[103],"groups":[104],"using":[105],"group-averaged":[106],"policy/reference":[107],"log-ratios":[108],"rather":[109,168],"than":[110,169],"one-to-one":[111],"preference":[112],"pairs.":[113],"Across":[114],"five":[115],"mixed-type":[116],"benchmarks,":[117],"TabGRAA":[118,205],"improves":[119],"GReaT":[121],"backbone":[122],"beyond":[123],"additional":[124],"achieves":[128],"strongest":[130],"average":[131],"trade-off":[132],"among":[133],"adapted":[134],"DPO,":[135],"KTO,":[136],"NPO":[138],"baselines":[139],"on":[140,160],"fidelity":[141],"downstream":[143],"while":[145],"maintaining":[146],"empirical":[147],"privacy":[148],"diagnostics":[149],"near":[150],"baseline.":[153],"Ablations":[154],"show":[155,178],"gains":[158],"depend":[159],"meaningful":[161],"ranking":[163],"stable":[165],"group-level":[166],"updates":[167],"extra":[170],"training":[171],"alone.":[172],"Reward-substitution":[173],"scorer-separation":[175],"studies":[176],"further":[177],"loop":[182],"use":[184],"both":[185],"classifier-based":[186],"classifier-free":[188],"rewards,":[189],"proper":[192],"scorer":[193],"separation":[194],"important":[196],"preserving":[198],"fidelity--utility--privacy":[200],"trade-off.":[201],"These":[202],"results":[203],"position":[204],"self-improving":[208],"language-model":[213],"generators,":[214],"complementary":[215],"strong":[217]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-04-23T00:00:00"}