{"id":"https://openalex.org/W7220748520","doi":null,"title":"How To Train Your World Model: Fine-tuning vs RAG for LM-based World Modeling","display_name":"How To Train Your World Model: Fine-tuning vs RAG for LM-based World Modeling","publication_year":2026,"publication_date":"2026-10-01","ids":{"openalex":"https://openalex.org/W7220748520"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2610.02542","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.02542","pdf_url":"https://arxiv.org/pdf/2610.02542","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,"listed_in":[],"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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2610.02542","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5031475293","display_name":"Dhananjay Ashok","orcid":"https://orcid.org/0000-0001-7500-3730","observed_orcids":["https://orcid.org/0000-0001-7500-3730"]},"institutions":[{"id":"https://openalex.org/I1174212","display_name":"University of Southern California","ror":"https://ror.org/03taz7m60","country_code":"US","type":"education","lineage":["https://openalex.org/I1174212"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ashok, Dhananjay","raw_affiliation_strings":["Information Sciences Institute University of Southern California"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Information Sciences Institute University of Southern California","institution_ids":["https://openalex.org/I1174212"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102186934","display_name":"Shantanu Agarwal","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1305444813","display_name":"Capital One (United States)","ror":"https://ror.org/00svp7168","country_code":"US","type":"company","lineage":["https://openalex.org/I1305444813"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Agarwal, Shantanu","raw_affiliation_strings":["Capital One"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Capital One","institution_ids":["https://openalex.org/I1305444813"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120034789","display_name":"Vivek Datla","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1305444813","display_name":"Capital One (United States)","ror":"https://ror.org/00svp7168","country_code":"US","type":"company","lineage":["https://openalex.org/I1305444813"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Datla, Vivek","raw_affiliation_strings":["Capital One"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Capital One","institution_ids":["https://openalex.org/I1305444813"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5000874697","display_name":"Jonathan May","orcid":"https://orcid.org/0000-0002-5284-477X","observed_orcids":["https://orcid.org/0000-0002-5284-477X"]},"institutions":[{"id":"https://openalex.org/I1174212","display_name":"University of Southern California","ror":"https://ror.org/03taz7m60","country_code":"US","type":"education","lineage":["https://openalex.org/I1174212"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"May, Jonathan","raw_affiliation_strings":["Information Sciences Institute University of Southern California"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Information Sciences Institute University of Southern California","institution_ids":["https://openalex.org/I1174212"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5066674792","display_name":"Alfy Samuel","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1305444813","display_name":"Capital One (United States)","ror":"https://ror.org/00svp7168","country_code":"US","type":"company","lineage":["https://openalex.org/I1305444813"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Samuel, Alfy","raw_affiliation_strings":["Capital One"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Capital One","institution_ids":["https://openalex.org/I1305444813"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":1,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.9018999934196472,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.9018999934196472,"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.07100000232458115,"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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.006899999920278788,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/world-models","display_name":"world models","score":0.921999990940094},{"id":"https://openalex.org/keywords/language-model-fine-tuning","display_name":"language model fine-tuning","score":0.7139999866485596},{"id":"https://openalex.org/keywords/retrieval-augmented-generation","display_name":"retrieval-augmented generation","score":0.9739999771118164},{"id":"https://openalex.org/keywords/transitional-dynamics","display_name":"transition dynamics","score":0.5270000100135803},{"id":"https://openalex.org/keywords/embodied-agents","display_name":"embodied agents","score":0.7929999828338623},{"id":"https://openalex.org/keywords/website-navigation","display_name":"web navigation","score":0.8270000219345093},{"id":"https://openalex.org/keywords/social-actors","display_name":"social actors","score":0.5789999961853027},{"id":"https://openalex.org/keywords/counterfactual-intervention","display_name":"counterfactual intervention","score":0.8960000276565552},{"id":"https://openalex.org/keywords/query-reformulation","display_name":"query rewriting","score":0.871999979019165},{"id":"https://openalex.org/keywords/data-efficiency","display_name":"data efficiency","score":0.5289999842643738},{"id":"https://openalex.org/keywords/exploration","display_name":"exploration","score":0.3499999940395355}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7235999703407288},{"id":"https://openalex.org/C108650721","wikidata":"https://www.wikidata.org/wiki/Q1783253","display_name":"Counterfactual thinking","level":2,"score":0.6852999925613403},{"id":"https://openalex.org/C136197465","wikidata":"https://www.wikidata.org/wiki/Q1729295","display_name":"Variety (cybernetics)","level":2,"score":0.5938000082969666},{"id":"https://openalex.org/C557471498","wikidata":"https://www.wikidata.org/wiki/Q554950","display_name":"Recommender system","level":2,"score":0.5547000169754028},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.5453000068664551},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5144000053405762},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4399999976158142},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.39149999618530273},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3849000036716461},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.3806999921798706},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.32120001316070557},{"id":"https://openalex.org/C518677369","wikidata":"https://www.wikidata.org/wiki/Q202833","display_name":"Social media","level":2,"score":0.26809999346733093},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.265500009059906},{"id":"https://openalex.org/C145912823","wikidata":"https://www.wikidata.org/wiki/Q113558","display_name":"Dynamics (music)","level":2,"score":0.2551000118255615},{"id":"https://openalex.org/C2164484","wikidata":"https://www.wikidata.org/wiki/Q5170150","display_name":"Core (optical fiber)","level":2,"score":0.2524000108242035}],"mesh":[],"locations_count":1,"locations":[{"id":"pmh:oai:arXiv.org:2610.02542","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.02542","pdf_url":"https://arxiv.org/pdf/2610.02542","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,"listed_in":[],"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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2610.02542","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.02542","pdf_url":"https://arxiv.org/pdf/2610.02542","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,"listed_in":[],"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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7220748520.pdf","grobid_xml":"https://content.openalex.org/works/W7220748520.grobid-xml"},"referenced_works_count":14,"referenced_works":["https://openalex.org/W2795843265","https://openalex.org/W2992977009","https://openalex.org/W3030163527","https://openalex.org/W3091507139","https://openalex.org/W3126325318","https://openalex.org/W4390963073","https://openalex.org/W4401306886","https://openalex.org/W4404401752","https://openalex.org/W4407091933","https://openalex.org/W4415255786","https://openalex.org/W4416372179","https://openalex.org/W6947991961","https://openalex.org/W7139967446","https://openalex.org/W7166714766"],"related_works":["https://openalex.org/W4415255786","https://openalex.org/W7117153217","https://openalex.org/W4416034645","https://openalex.org/W7165749951","https://openalex.org/W7164049912","https://openalex.org/W7216147366","https://openalex.org/W7127888457","https://openalex.org/W7170111426","https://openalex.org/W7220686551","https://openalex.org/W7221134068"],"abstract_inverted_index":{"World":[0],"models":[1],"(WMs)":[2],"simulate":[3],"the":[4,14,40,127,148,152,164,186,232,235,238],"transition":[5],"dynamics":[6],"of":[7,16,43,129,151,176,234,240],"environments,":[8,21],"enabling":[9,95],"agents":[10,96],"to":[11,27,97,146,168,209],"plan":[12],"over":[13],"consequences":[15],"their":[17],"actions.":[18],"In":[19],"text-based":[20],"fine-tuning":[22,76,88,121],"a":[23,30,35,61,133,140,174,182,196],"Language":[24],"Model":[25],"(LM)":[26],"serve":[28],"as":[29,34,47],"WM":[31],"has":[32],"emerged":[33],"dominant":[36],"paradigm.":[37],"However,":[38],"despite":[39],"widespread":[41],"success":[42],"non-parametric":[44],"approaches":[45,79,116,122],"such":[46],"Retrieval":[48],"Augmented":[49],"Generation":[50],"(RAG),":[51],"retrieval":[52,153,188,212],"for":[53,80],"LM-based":[54,81],"world":[55,82,198],"modelling":[56,199],"remains":[57],"underexplored.":[58],"We":[59],"conduct":[60],"systematic":[62],"evaluation":[63],"across":[64,226],"five":[65],"diverse":[66,113],"environments":[67,228],"spanning":[68],"embodied,":[69],"web":[70],"navigation":[71],"and":[72,77,111,120,155,229,237],"social":[73],"settings,":[74],"comparing":[75],"RAG-based":[78,115,136],"modelling.":[83],"Our":[84,219],"study":[85,173],"reveals":[86],"that":[87,142,157,181,201],"often":[89],"outperforms":[90,185,223],"RAG,":[91],"with":[92],"fine-tuned":[93],"WMs":[94],"obtain":[98],"higher":[99],"rewards":[100],"on":[101,135,211],"15/20":[102],"settings.":[103],"While":[104],"both":[105],"construction":[106],"paradigms":[107],"benefit":[108,124],"from":[109,125,163,213],"additional":[110],"more":[112,118],"exploration,":[114],"prove":[117],"data-efficient,":[119],"disproportionately":[123],"scaling":[126],"amount":[128],"experience":[130,165],"collected.":[131],"With":[132],"focus":[134],"WMs,":[137],"we":[138,172,191],"devise":[139],"procedure":[141],"uses":[143],"counterfactual":[144],"intervention":[145],"estimate":[147],"error":[149],"rate":[150],"stage,":[154],"show":[156],"retrievers":[158],"consistently":[159,222],"surface":[160],"suboptimal":[161],"transitions":[162],"buffer.":[166],"Hoping":[167],"address":[169],"this":[170],"failing,":[171],"variety":[175],"query":[177],"reformulation":[178],"strategies,":[179],"demonstrating":[180],"hierarchical":[183],"approach":[184,236],"traditional":[187],"pipeline.":[189],"Finally,":[190],"compose":[192],"our":[193,241],"findings":[194],"into":[195],"hybrid":[197,220],"system":[200,221],"parametrically":[202],"captures":[203],"core":[204],"environment":[205],"dynamics,":[206],"while":[207],"learning":[208],"rely":[210],"an":[214],"actively":[215],"maintained":[216],"memory":[217],"store.":[218],"other":[224],"methods":[225],"multiple":[227],"models,":[230],"showcasing":[231],"robustness":[233],"applicability":[239],"findings.":[242]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-10-07T00:00:00"}