{"id":"https://openalex.org/W7214733154","doi":"https://doi.org/10.48550/arxiv.2609.35303","title":"PIVOT: Pivot-Aware On Policy Self Distillation for Multi-Turn VLM Agents","display_name":"PIVOT: Pivot-Aware On Policy Self Distillation for Multi-Turn VLM Agents","publication_year":2026,"publication_date":"2026-09-28","ids":{"openalex":"https://openalex.org/W7214733154","doi":"https://doi.org/10.48550/arxiv.2609.35303"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.35303","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.35303","pdf_url":"https://arxiv.org/pdf/2609.35303","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/2609.35303","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5074096594","display_name":"Jiazhou Zhou","orcid":"https://orcid.org/0009-0008-5258-1675","observed_orcids":["https://orcid.org/0009-0008-5258-1675"]},"institutions":[{"id":"https://openalex.org/I4405261251","display_name":"The Hong Kong University of Science and Technology (Guangzhou)","ror":"https://ror.org/050h0vm43","country_code":"CN","type":"education","lineage":["https://openalex.org/I4405261251"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhou, Jiazhou","raw_affiliation_strings":["AI Thrust , The Hong Kong University of Science and Technology (Guangzhou )","International Digital Economy Academy (IDEA)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AI Thrust , The Hong Kong University of Science and Technology (Guangzhou )","institution_ids":["https://openalex.org/I4405261251"]},{"raw_affiliation_string":"International Digital Economy Academy (IDEA)","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133079970","display_name":"Hu Zhou","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I14243506","display_name":"Hong Kong Polytechnic University","ror":"https://ror.org/0030zas98","country_code":"HK","type":"education","lineage":["https://openalex.org/I14243506"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Zhou, Hu","raw_affiliation_strings":["The Hong Kong Polytechnic University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Hong Kong Polytechnic University","institution_ids":["https://openalex.org/I14243506"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5153463839","display_name":"Yucheng Chen","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]},{"id":"https://openalex.org/I4405261251","display_name":"The Hong Kong University of Science and Technology (Guangzhou)","ror":"https://ror.org/050h0vm43","country_code":"CN","type":"education","lineage":["https://openalex.org/I4405261251"]},{"id":"https://openalex.org/I4405278195","display_name":"Centre of AI in Medicine","ror":"https://ror.org/058w28210","country_code":"SG","type":"facility","lineage":["https://openalex.org/I4405278195"]}],"countries":["CN","SG"],"is_corresponding":false,"raw_author_name":"Chen, Yucheng","raw_affiliation_strings":["AI Thrust , The Hong Kong University of Science and Technology (Guangzhou )","MedVisAI Lab , Lee Kong Chian School of Medicine , Nanyang Technological University , and Centre of AI in Medicine"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AI Thrust , The Hong Kong University of Science and Technology (Guangzhou )","institution_ids":["https://openalex.org/I4405261251"]},{"raw_affiliation_string":"MedVisAI Lab , Lee Kong Chian School of Medicine , Nanyang Technological University , and Centre of AI in Medicine","institution_ids":["https://openalex.org/I172675005","https://openalex.org/I4405278195"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5055164625","display_name":"Jinyuan Qu","orcid":"https://orcid.org/0009-0000-9963-3032","observed_orcids":["https://orcid.org/0009-0000-9963-3032"]},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Qu, Jinyuan","raw_affiliation_strings":["International Digital Economy Academy (IDEA)","Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"International Digital Economy Academy (IDEA)","institution_ids":[]},{"raw_affiliation_string":"Tsinghua University","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101938761","display_name":"Ying-Cong Chen","orcid":"https://orcid.org/0000-0002-9565-8205","observed_orcids":["https://orcid.org/0000-0002-9565-8205"]},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]},{"id":"https://openalex.org/I4405261251","display_name":"The Hong Kong University of Science and Technology (Guangzhou)","ror":"https://ror.org/050h0vm43","country_code":"CN","type":"education","lineage":["https://openalex.org/I4405261251"]},{"id":"https://openalex.org/I4405278195","display_name":"Centre of AI in Medicine","ror":"https://ror.org/058w28210","country_code":"SG","type":"facility","lineage":["https://openalex.org/I4405278195"]}],"countries":["CN","SG"],"is_corresponding":false,"raw_author_name":"Chen, Ying-Cong","raw_affiliation_strings":["AI Thrust , The Hong Kong University of Science and Technology (Guangzhou )","MedVisAI Lab , Lee Kong Chian School of Medicine , Nanyang Technological University , and Centre of AI in Medicine"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AI Thrust , The Hong Kong University of Science and Technology (Guangzhou )","institution_ids":["https://openalex.org/I4405261251"]},{"raw_affiliation_string":"MedVisAI Lab , Lee Kong Chian School of Medicine , Nanyang Technological University , and Centre of AI in Medicine","institution_ids":["https://openalex.org/I172675005","https://openalex.org/I4405278195"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5153503362","display_name":"Lei Zhang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Lei","raw_affiliation_strings":["International Digital Economy Academy (IDEA)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"International Digital Economy Academy (IDEA)","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":5,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.9922000169754028,"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":0.9922000169754028,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.0035000001080334187,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.000699999975040555,"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/on-policy-self-distillation","display_name":"on-policy self-distillation","score":0.7789999842643738},{"id":"https://openalex.org/keywords/group-relative-policy-optimization","display_name":"Group Relative Policy Optimization","score":0.7730000019073486},{"id":"https://openalex.org/keywords/reinforcement-learning-with-verifiable-rewards","display_name":"reinforcement learning with verifiable rewards","score":0.8410000205039978},{"id":"https://openalex.org/keywords/sparse-rewards","display_name":"sparse rewards","score":0.36500000953674316},{"id":"https://openalex.org/keywords/failure-analysis","display_name":"failure analysis","score":0.5490000247955322},{"id":"https://openalex.org/keywords/state-restoration","display_name":"state restoration","score":0.5740000009536743},{"id":"https://openalex.org/keywords/embodied-navigation","display_name":"embodied navigation","score":0.3449999988079071},{"id":"https://openalex.org/keywords/3d-navigation","display_name":"3D navigation","score":0.6499999761581421},{"id":"https://openalex.org/keywords/generative-reasoning","display_name":"generative reasoning","score":0.8299999833106995},{"id":"https://openalex.org/keywords/qwen2-5-vl","display_name":"Qwen2.5-VL","score":0.652999997138977},{"id":"https://openalex.org/keywords/qwen3-vl","display_name":"Qwen3-VL","score":0.9210000038146973}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7660999894142151},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.5468000173568726},{"id":"https://openalex.org/C108650721","wikidata":"https://www.wikidata.org/wiki/Q1783253","display_name":"Counterfactual thinking","level":2,"score":0.5425000190734863},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.5185999870300293},{"id":"https://openalex.org/C10347200","wikidata":"https://www.wikidata.org/wiki/Q1960297","display_name":"Hindsight bias","level":2,"score":0.5065000057220459},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.46320000290870667},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.4544999897480011},{"id":"https://openalex.org/C43214815","wikidata":"https://www.wikidata.org/wiki/Q7310987","display_name":"Reliability (semiconductor)","level":3,"score":0.4507000148296356},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.43560001254081726},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3668000102043152},{"id":"https://openalex.org/C174220543","wikidata":"https://www.wikidata.org/wiki/Q395307","display_name":"Rollback","level":3,"score":0.35510000586509705},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.35409998893737793},{"id":"https://openalex.org/C49937458","wikidata":"https://www.wikidata.org/wiki/Q2599292","display_name":"Probabilistic logic","level":2,"score":0.34950000047683716},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.3490999937057495},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.3416999876499176},{"id":"https://openalex.org/C166109690","wikidata":"https://www.wikidata.org/wiki/Q4677422","display_name":"Action selection","level":3,"score":0.33660000562667847},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.32600000500679016},{"id":"https://openalex.org/C80519477","wikidata":"https://www.wikidata.org/wiki/Q3532236","display_name":"Scenario testing","level":3,"score":0.31119999289512634},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.3052999973297119},{"id":"https://openalex.org/C2779662365","wikidata":"https://www.wikidata.org/wiki/Q5416694","display_name":"Event (particle physics)","level":2,"score":0.30219998955726624},{"id":"https://openalex.org/C100609095","wikidata":"https://www.wikidata.org/wiki/Q1335050","display_name":"Embodied cognition","level":2,"score":0.3012999892234802},{"id":"https://openalex.org/C65909025","wikidata":"https://www.wikidata.org/wiki/Q1945033","display_name":"Monocular","level":2,"score":0.28760001063346863},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.28600001335144043},{"id":"https://openalex.org/C2777267654","wikidata":"https://www.wikidata.org/wiki/Q3519023","display_name":"Test (biology)","level":2,"score":0.2856000065803528},{"id":"https://openalex.org/C2779199153","wikidata":"https://www.wikidata.org/wiki/Q43965","display_name":"Quadcopter","level":2,"score":0.274399995803833},{"id":"https://openalex.org/C58315980","wikidata":"https://www.wikidata.org/wiki/Q179436","display_name":"Asymptote","level":2,"score":0.2694000005722046},{"id":"https://openalex.org/C44154836","wikidata":"https://www.wikidata.org/wiki/Q45045","display_name":"Simulation","level":1,"score":0.26809999346733093},{"id":"https://openalex.org/C13687954","wikidata":"https://www.wikidata.org/wiki/Q4826847","display_name":"Autonomous agent","level":2,"score":0.26649999618530273},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.266400009393692},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.26579999923706055},{"id":"https://openalex.org/C47446073","wikidata":"https://www.wikidata.org/wiki/Q5165890","display_name":"Control theory (sociology)","level":3,"score":0.263700008392334},{"id":"https://openalex.org/C2778012447","wikidata":"https://www.wikidata.org/wiki/Q1034415","display_name":"Scope (computer science)","level":2,"score":0.2635999917984009},{"id":"https://openalex.org/C101104100","wikidata":"https://www.wikidata.org/wiki/Q1063540","display_name":"Heteroscedasticity","level":2,"score":0.2535000145435333}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.35303","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.35303","pdf_url":"https://arxiv.org/pdf/2609.35303","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.2609.35303","doi":"https://doi.org/10.48550/arxiv.2609.35303","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.35303","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:2609.35303","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.35303","pdf_url":"https://arxiv.org/pdf/2609.35303","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/W7214733154.pdf","grobid_xml":"https://content.openalex.org/works/W7214733154.grobid-xml"},"referenced_works_count":17,"referenced_works":["https://openalex.org/W2736601468","https://openalex.org/W2776202271","https://openalex.org/W4310290453","https://openalex.org/W4378505261","https://openalex.org/W4391631327","https://openalex.org/W4407590542","https://openalex.org/W4407794264","https://openalex.org/W4414281281","https://openalex.org/W4416048648","https://openalex.org/W4416780398","https://openalex.org/W6948439803","https://openalex.org/W7125861034","https://openalex.org/W7126278298","https://openalex.org/W7154299763","https://openalex.org/W7161258139","https://openalex.org/W7169575253","https://openalex.org/W7204908525"],"related_works":["https://openalex.org/W7215033270","https://openalex.org/W7161950017","https://openalex.org/W7213501694","https://openalex.org/W7155416398","https://openalex.org/W7140105616","https://openalex.org/W7214894101","https://openalex.org/W7203440145","https://openalex.org/W7171936987","https://openalex.org/W7167619710","https://openalex.org/W7160876190"],"abstract_inverted_index":{"Reinforcement":[0],"learning":[1],"with":[2],"verifiable":[3],"rewards":[4,43],"(RLVR)":[5],"via":[6],"Group-Relative":[7],"Policy":[8],"Optimization":[9],"(GRPO)":[10],"is":[11],"widely":[12],"used":[13],"for":[14],"multi-turn":[15,60,206],"VLM":[16,61,207],"agent":[17,62,208],"training,":[18],"yet":[19],"it":[20],"suffers":[21],"from":[22,165],"zero-gradient":[23],"silence":[24],"on":[25,204,227,242],"uniform":[26],"failures":[27],"and":[28,37,99,123,137,161,169,183,189,198,217,219,233,238],"coarse":[29],"episode-level":[30],"credit":[31],"assignment.":[32],"While":[33],"On-Policy":[34,38,113],"Distillation":[35],"(OPD)":[36],"Self-Distillation":[39],"(OPSD)":[40],"mitigate":[41],"sparse":[42],"using":[44],"hindsight":[45],"information,":[46],"their":[47],"underlying":[48],"mechanisms":[49],"remain":[50],"poorly":[51],"understood.":[52],"Through":[53],"controlled":[54],"counterfactual":[55],"rollback":[56,77],"probes":[57],"across":[58,210],"five":[59,205],"benchmarks,":[63],"we":[64,108],"reveal":[65],"that":[66,119],"performance":[67],"gains":[68],"in":[69,101],"OPSD/OPD":[70],"are":[71,96,201],"largely":[72],"driven":[73],"by":[74],"physical":[75,93],"state":[76,94,124],"at":[78,141],"the":[79,84,158],"pivot":[80,121,159],"step,":[81],"defined":[82],"as":[83],"first":[85],"unrecoverable":[86],"action":[87,170],"without":[88],"remaining":[89],"step":[90,160],"budget.":[91],"However,":[92],"rollbacks":[95,133],"computationally":[97],"prohibitive":[98],"infeasible":[100],"real-world":[102],"environments.":[103],"To":[104],"bridge":[105],"this":[106,179],"gap,":[107],"present":[109],"Pivot-Aware":[110],"Internalized":[111],"Visual":[112],"Training":[114],"(PIVOT),":[115],"an":[116],"RL":[117,135],"framework":[118],"internalizes":[120],"localization":[122],"restoration":[125],"directly":[126],"into":[127],"token-level":[128],"parameter":[129],"updates,":[130],"eliminating":[131],"environment":[132],"during":[134],"training":[136],"additional":[138],"skill":[139],"hints":[140],"test":[142,194],"time.":[143],"PIVOT":[144,222],"unifies":[145],"three":[146],"functional":[147],"roles":[148],"within":[149],"a":[150,153,172,184],"single":[151],"architecture:":[152],"failure":[154,163],"Analyzer":[155,199],"non-invasively":[156],"localizes":[157],"diagnoses":[162],"modes":[164],"visual":[166],"trajectory":[167],"collages":[168],"logs;":[171],"detached":[173],"Teacher":[174,197],"re-scores":[175],"failed":[176],"tokens":[177],"under":[178],"privileged":[180],"diagnostic":[181],"context;":[182],"Student":[185],"optimizes":[186],"joint":[187],"GRPO":[188],"confidence-gated":[190],"OPD":[191],"objectives.":[192],"At":[193],"time,":[195],"both":[196],"branches":[200],"stripped.":[202],"Evaluated":[203],"tasks":[209],"cognitive":[211],"grid":[212],"puzzles,":[213],"3D":[214],"embodied":[215],"control":[216],"navigation,":[218],"generative":[220],"reasoning,":[221],"achieves":[223],"0.90":[224],"overall":[225],"accuracy":[226],"Qwen2.5-VL-3B":[228],"(+8%":[229],"over":[230,235,245],"SFT+GRPO":[231,246],"baseline":[232],"+5%":[234],"previous":[236],"SOTA)":[237],"scales":[239],"to":[240],"0.92":[241],"Qwen3-VL-2B":[243],"(+12%":[244],"baseline).":[247]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-09-30T00:00:00"}