{"id":"https://openalex.org/W7204883267","doi":"https://doi.org/10.48550/arxiv.2608.29357","title":"LiteSearch-VL: Small Multimodal Search Agents via Trajectory Distillation and Synthetic Step-DPO","display_name":"LiteSearch-VL: Small Multimodal Search Agents via Trajectory Distillation and Synthetic Step-DPO","publication_year":2026,"publication_date":"2026-08-29","ids":{"openalex":"https://openalex.org/W7204883267","doi":"https://doi.org/10.48550/arxiv.2608.29357"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.29357","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.29357","pdf_url":"https://arxiv.org/pdf/2608.29357","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","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2608.29357","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5040137532","display_name":"Saeed Khaki","orcid":"https://orcid.org/0000-0003-0095-8588","observed_orcids":["https://orcid.org/0000-0003-0095-8588"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Khaki, Saeed","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123605288","display_name":"Nima Safaei","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I52357470","display_name":"The Ohio State University","ror":"https://ror.org/00rs6vg23","country_code":"US","type":"education","lineage":["https://openalex.org/I52357470"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Safaei, Nima","raw_affiliation_strings":["Ohio State University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Ohio State University","institution_ids":["https://openalex.org/I52357470"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5123571972","display_name":"Kamal Ginotra","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1290206253","display_name":"Microsoft (United States)","ror":"https://ror.org/00d0nc645","country_code":"US","type":"company","lineage":["https://openalex.org/I1290206253"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ginotra, Kamal","raw_affiliation_strings":["Microsoft AI"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microsoft AI","institution_ids":["https://openalex.org/I1290206253"]}]}],"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":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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9955999851226807,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9955999851226807,"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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.0017000000225380063,"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"}},{"id":"https://openalex.org/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.00039999998989515007,"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/lora-adapters","display_name":"LoRA adapters","score":0.7400000095367432},{"id":"https://openalex.org/keywords/visual-question-answering","display_name":"visual question answering","score":0.7829999923706055},{"id":"https://openalex.org/keywords/answer-verification","display_name":"answer verification","score":0.5849999785423279},{"id":"https://openalex.org/keywords/search-depth","display_name":"search depth","score":0.4339999854564667},{"id":"https://openalex.org/keywords/qwen3-vl","display_name":"Qwen3-VL","score":0.75}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6444000005722046},{"id":"https://openalex.org/C28034677","wikidata":"https://www.wikidata.org/wiki/Q17092530","display_name":"Interleaving","level":2,"score":0.5586000084877014},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5246999859809875},{"id":"https://openalex.org/C166955791","wikidata":"https://www.wikidata.org/wiki/Q629579","display_name":"Macro","level":2,"score":0.5044999718666077},{"id":"https://openalex.org/C2778671685","wikidata":"https://www.wikidata.org/wiki/Q219239","display_name":"Recipe","level":2,"score":0.4749999940395355},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.47049999237060547},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.46160000562667847},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.4106000065803528},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.3707999885082245},{"id":"https://openalex.org/C160920958","wikidata":"https://www.wikidata.org/wiki/Q7662746","display_name":"Synthetic data","level":2,"score":0.3666999936103821},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.3443000018596649},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.3425999879837036},{"id":"https://openalex.org/C2779955035","wikidata":"https://www.wikidata.org/wiki/Q4686785","display_name":"Advice (programming)","level":2,"score":0.3269999921321869},{"id":"https://openalex.org/C62611344","wikidata":"https://www.wikidata.org/wiki/Q1062658","display_name":"Node (physics)","level":2,"score":0.2962999939918518},{"id":"https://openalex.org/C2780615836","wikidata":"https://www.wikidata.org/wiki/Q2471869","display_name":"USable","level":2,"score":0.2904999852180481},{"id":"https://openalex.org/C90329073","wikidata":"https://www.wikidata.org/wiki/Q914232","display_name":"Ask price","level":2,"score":0.2867000102996826},{"id":"https://openalex.org/C173801870","wikidata":"https://www.wikidata.org/wiki/Q201413","display_name":"Heuristic","level":2,"score":0.2800999879837036},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.27950000762939453},{"id":"https://openalex.org/C19889080","wikidata":"https://www.wikidata.org/wiki/Q2835852","display_name":"Beam search","level":3,"score":0.27730000019073486},{"id":"https://openalex.org/C204030448","wikidata":"https://www.wikidata.org/wiki/Q101017","display_name":"Distillation","level":2,"score":0.27469998598098755},{"id":"https://openalex.org/C77967617","wikidata":"https://www.wikidata.org/wiki/Q4677561","display_name":"Active learning (machine learning)","level":2,"score":0.259799987077713},{"id":"https://openalex.org/C75553542","wikidata":"https://www.wikidata.org/wiki/Q178161","display_name":"A priori and a posteriori","level":2,"score":0.2590000033378601},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.2572000026702881},{"id":"https://openalex.org/C41550386","wikidata":"https://www.wikidata.org/wiki/Q529909","display_name":"Multi-agent system","level":2,"score":0.2538999915122986},{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.2517000138759613}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.29357","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.29357","pdf_url":"https://arxiv.org/pdf/2608.29357","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"},{"id":"doi:10.48550/arxiv.2608.29357","doi":"https://doi.org/10.48550/arxiv.2608.29357","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.29357","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,"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":"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":"pmh:oai:arXiv.org:2608.29357","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.29357","pdf_url":"https://arxiv.org/pdf/2608.29357","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/W7204883267.pdf","grobid_xml":"https://content.openalex.org/works/W7204883267.grobid-xml"},"referenced_works_count":20,"referenced_works":["https://openalex.org/W4226227340","https://openalex.org/W4391631327","https://openalex.org/W4392019816","https://openalex.org/W4400141325","https://openalex.org/W4403365357","https://openalex.org/W4403748154","https://openalex.org/W4403853618","https://openalex.org/W4407759977","https://openalex.org/W4414990013","https://openalex.org/W4415023020","https://openalex.org/W4415272886","https://openalex.org/W4416271960","https://openalex.org/W4416526576","https://openalex.org/W4416679470","https://openalex.org/W4416780398","https://openalex.org/W4417422257","https://openalex.org/W7117964748","https://openalex.org/W7126150514","https://openalex.org/W7127415714","https://openalex.org/W7160533687"],"related_works":["https://openalex.org/W7166840362","https://openalex.org/W7152558338","https://openalex.org/W7160533687","https://openalex.org/W7154613953","https://openalex.org/W7133517971","https://openalex.org/W7151627296","https://openalex.org/W7133354148","https://openalex.org/W7155358945","https://openalex.org/W7222652627","https://openalex.org/W7133318532"],"abstract_inverted_index":{"Multimodal":[0],"search":[1,198,213],"agents":[2],"answer":[3,152,210],"visual":[4],"questions":[5],"by":[6],"interleaving":[7],"image":[8],"understanding,":[9],"web":[10],"retrieval,":[11],"tool":[12,174],"use,":[13],"and":[14,31,43,79,90,122,172],"evidence":[15],"synthesis.":[16],"Strong":[17],"systems":[18],"exist,":[19],"but":[20],"in":[21],"two":[22],"expensive":[23],"regimes:":[24],"proprietary":[25],"frontier":[26],"models":[27],"such":[28],"as":[29,177,215],"GPT-5":[30],"Gemini,":[32],"or":[33,161],"large":[34],"open":[35],"vision-language":[36],"backbones":[37,60],"trained":[38],"with":[39,72],"substantial":[40],"agentic":[41],"data":[42],"reinforcement":[44],"learning.":[45],"We":[46,69],"ask":[47],"a":[48,62,74,131,150,190],"different":[49],"question:":[50],"when":[51],"released":[52,84],"agent":[53,140],"trajectories":[54],"are":[55],"distilled":[56],"into":[57,202],"much":[58],"smaller":[59],"under":[61],"single-node":[63],"budget,":[64],"what":[65],"is":[66,127],"actually":[67],"transferred?":[68],"study":[70],"this":[71],"LiteSearch-VL,":[73],"low-compute":[75],"recipe":[76],"for":[77,219],"Qwen3-VL-2B":[78],"Qwen3-VL-4B":[80],"that":[81,196],"uses":[82],"only":[83],"OpenSearch-VL":[85],"trajectories,":[86],"parameter-efficient":[87],"LoRA":[88],"adapters,":[89],"synthetic":[91],"step-level":[92],"preferences:":[93],"DPO":[94],"on":[95,118],"GPT-5-generated":[96],"hard":[97],"negatives":[98],"targeting":[99],"five":[100],"local":[101],"failure":[102],"modes":[103],"(premature":[104],"answer,":[105],"wrong":[106],"tool,":[107],"weak":[108],"query,":[109,111],"repeated":[110],"ignored":[112],"image).":[113],"Across":[114],"12,400":[115],"GPT-5-judged":[116],"rollouts":[117],"SimpleVQA,":[119],"FVQA,":[120],"LiveVQA,":[121],"VDR-Bench-testmini,":[123],"the":[124,139,143,164,216],"dominant":[125],"effect":[126],"behavioral":[128],"rather":[129,179,205],"than":[130,180,206],"uniform":[132],"accuracy":[133],"lift:":[134],"full-trajectory":[135],"supervised":[136],"fine-tuning":[137],"transfers":[138],"contract,":[141],"taking":[142],"2B":[144],"model":[145],"from":[146],"almost":[147],"never":[148],"emitting":[149],"usable":[151],"(1,237/1,240":[153],"no_answer":[154],"rollouts)":[155],"to":[156],"28.4%":[157],"macro":[158,187],"Pass@1,":[159],"matching":[160],"slightly":[162],"exceeding":[163],"off-the-shelf":[165],"4B":[166,184],"base":[167],"(25.6%).":[168],"Synthetic":[169],"preference":[170],"learning":[171],"compact":[173],"distillation":[175],"act":[176],"refinements":[178],"phase":[181],"transitions":[182],"(best":[183],"configuration:":[185],"30.8%":[186],"Pass@1).":[188],"Finally,":[189],"controlled":[191],"VDR":[192],"step-budget":[193],"ablation":[194],"shows":[195],"extra":[197],"turns":[199],"convert":[200],"abstentions":[201],"wrong_entity":[203],"errors":[204],"correct":[207],"answers,":[208],"identifying":[209],"verification,":[211],"not":[212],"depth,":[214],"next":[217],"bottleneck":[218],"small":[220],"multimodal":[221],"agents.":[222]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-09-02T00:00:00"}