{"id":"https://openalex.org/W7163329770","doi":"https://doi.org/10.48550/arxiv.2606.03197","title":"MemTrain: Self-Supervised Context Memory Training","display_name":"MemTrain: Self-Supervised Context Memory Training","publication_year":2026,"publication_date":"2026-06-02","ids":{"openalex":"https://openalex.org/W7163329770","doi":"https://doi.org/10.48550/arxiv.2606.03197"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.03197","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.03197","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":null,"license_id":null,"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.2606.03197","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137749442","display_name":"Ziheng Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Ziheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137803448","display_name":"Xingrun Xing","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xing, Xingrun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137787218","display_name":"Haoqing Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Haoqing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137789180","display_name":"Zhi-Hong Deng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Deng, Zhi-Hong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137771548","display_name":"Yehui Tang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tang, Yehui","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/T10028","display_name":"Topic Modeling","score":0.896399974822998,"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/T10028","display_name":"Topic Modeling","score":0.896399974822998,"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":0.03709999844431877,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.012500000186264515,"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/recall","display_name":"Recall","score":0.5748999714851379},{"id":"https://openalex.org/keywords/memory-model","display_name":"Memory model","score":0.5667999982833862},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.4341000020503998},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.4221999943256378},{"id":"https://openalex.org/keywords/context-dependent-memory","display_name":"Context-dependent memory","score":0.41659998893737793},{"id":"https://openalex.org/keywords/downstream","display_name":"Downstream (manufacturing)","score":0.3817000091075897},{"id":"https://openalex.org/keywords/overlay","display_name":"Overlay","score":0.34599998593330383},{"id":"https://openalex.org/keywords/extended-memory","display_name":"Extended memory","score":0.33730000257492065}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.765500009059906},{"id":"https://openalex.org/C100660578","wikidata":"https://www.wikidata.org/wiki/Q18733","display_name":"Recall","level":2,"score":0.5748999714851379},{"id":"https://openalex.org/C12186640","wikidata":"https://www.wikidata.org/wiki/Q6815743","display_name":"Memory model","level":3,"score":0.5667999982833862},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.4341000020503998},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.4221999943256378},{"id":"https://openalex.org/C76679254","wikidata":"https://www.wikidata.org/wiki/Q5165163","display_name":"Context-dependent memory","level":4,"score":0.41659998893737793},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4034000039100647},{"id":"https://openalex.org/C2776207758","wikidata":"https://www.wikidata.org/wiki/Q5303302","display_name":"Downstream (manufacturing)","level":2,"score":0.3817000091075897},{"id":"https://openalex.org/C136085584","wikidata":"https://www.wikidata.org/wiki/Q910289","display_name":"Overlay","level":2,"score":0.34599998593330383},{"id":"https://openalex.org/C171675096","wikidata":"https://www.wikidata.org/wiki/Q1143380","display_name":"Extended memory","level":4,"score":0.33730000257492065},{"id":"https://openalex.org/C2985957978","wikidata":"https://www.wikidata.org/wiki/Q492","display_name":"Human memory","level":3,"score":0.32899999618530273},{"id":"https://openalex.org/C88576662","wikidata":"https://www.wikidata.org/wiki/Q18646","display_name":"Episodic memory","level":3,"score":0.31040000915527344},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.3070000112056732},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3057999908924103},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.3043999969959259},{"id":"https://openalex.org/C197914299","wikidata":"https://www.wikidata.org/wiki/Q18650","display_name":"Semantic memory","level":3,"score":0.2825999855995178},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.27900001406669617},{"id":"https://openalex.org/C74426580","wikidata":"https://www.wikidata.org/wiki/Q719484","display_name":"Memory map","level":3,"score":0.26930001378059387},{"id":"https://openalex.org/C176649486","wikidata":"https://www.wikidata.org/wiki/Q2308807","display_name":"Memory management","level":3,"score":0.259799987077713},{"id":"https://openalex.org/C81669768","wikidata":"https://www.wikidata.org/wiki/Q2359161","display_name":"Precision and recall","level":2,"score":0.2581999897956848},{"id":"https://openalex.org/C148220186","wikidata":"https://www.wikidata.org/wiki/Q7111912","display_name":"Outcome (game theory)","level":2,"score":0.2517000138759613}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.03197","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.03197","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.03197","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.03197","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"score":0.4516802728176117,"display_name":"Quality Education","id":"https://metadata.un.org/sdg/4"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Memory":[0],"is":[1,41],"an":[2,92,122],"indispensable":[3],"capability":[4,72],"for":[5,38,67,76],"long-horizon":[6],"LLM":[7,74],"agents,":[8],"enabling":[9],"them":[10],"to":[11,52,101,131,183],"preserve":[12],"and":[13,43,120,143,163],"utilize":[14],"information":[15,135],"accumulated":[16],"across":[17,176],"extended":[18],"interactions.":[19],"Existing":[20],"memory-agent":[21],"approaches":[22],"are":[23,153],"typically":[24],"trained":[25],"end-to-end":[26,93],"with":[27],"reinforcement":[28],"learning":[29],"on":[30,160],"downstream":[31,79,172],"tasks.":[32],"However,":[33],"collecting":[34],"high-quality":[35],"annotated":[36],"problems":[37],"memory-intensive":[39,173],"scenarios":[40],"costly,":[42],"the":[44,70,99,116,129,147],"resulting":[45],"training":[46,65],"data":[47],"often":[48],"lack":[49],"sufficient":[50],"diversity":[51],"cover":[53],"general":[54],"memory":[55,109,113,124,138,144],"behaviors.":[56],"In":[57],"this":[58],"work,":[59],"we":[60],"propose":[61],"MemTrain,":[62],"a":[63],"self-supervised":[64],"framework":[66],"generally":[68],"enhancing":[69],"context-memory":[71],"of":[73,108,181],"agents":[75],"more":[77],"effective":[78],"post-training.":[80,189],"MemTrain":[81,169],"introduces":[82],"two":[83,151],"coupled":[84],"proxy":[85],"tasks":[86],"over":[87,186],"unlabeled":[88],"Wikipedia":[89],"corpora:":[90],"(1)":[91],"masked":[94,103,133],"reconstruction":[95],"objective,":[96,126],"which":[97,127],"requires":[98,128],"model":[100,130],"recover":[102],"entities":[104],"after":[105],"multiple":[106],"rounds":[107],"updates,":[110],"thereby":[111],"encouraging":[112,140],"maintenance":[114],"from":[115],"final":[117],"outcome":[118],"perspective;":[119],"(2)":[121],"intermediate":[123,137],"recall":[125],"reconstruct":[132],"historical":[134],"using":[136,156],"states,":[139],"faithful":[141],"compression":[142],"completeness":[145],"throughout":[146],"interaction":[148],"process.":[149],"The":[150],"objectives":[152],"jointly":[154],"optimized":[155],"GRPO.":[157],"Extensive":[158],"experiments":[159],"long-text":[161],"QA":[162,165],"search-based":[164],"benchmarks":[166],"demonstrate":[167],"that":[168],"consistently":[170],"improves":[171],"reasoning":[174],"performance":[175],"different":[177],"models,":[178],"achieving":[179],"gains":[180],"up":[182],"17.67":[184],"points":[185],"direct":[187],"task-specific":[188]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-04T00:00:00"}