{"id":"https://openalex.org/W7164573556","doi":"https://doi.org/10.48550/arxiv.2606.12941","title":"Multi-Turn Reasoning When Context Arrives in Pieces: Scalable Sharding and Memory-Augmented RL","display_name":"Multi-Turn Reasoning When Context Arrives in Pieces: Scalable Sharding and Memory-Augmented RL","publication_year":2026,"publication_date":"2026-06-11","ids":{"openalex":"https://openalex.org/W7164573556","doi":"https://doi.org/10.48550/arxiv.2606.12941"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.12941","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.12941","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.2606.12941","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5138542907","display_name":"Shu Tong Luo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Luo, Shu Tong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138529412","display_name":"Wenqin Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Wenqin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138517323","display_name":"Rui Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Rui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138530625","display_name":"Mingming Gong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gong, Mingming","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5084631859","display_name":"Jiaxian Guo","orcid":"https://orcid.org/0000-0002-0212-8329"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Guo, Jiaxian","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":0.6740999817848206,"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":0.6740999817848206,"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/T12031","display_name":"Speech and dialogue systems","score":0.07419999688863754,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.05510000139474869,"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/context","display_name":"Context (archaeology)","score":0.723800003528595},{"id":"https://openalex.org/keywords/conversation","display_name":"Conversation","score":0.7199000120162964},{"id":"https://openalex.org/keywords/pipeline","display_name":"Pipeline (software)","score":0.7178999781608582},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.6432999968528748},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.4609000086784363},{"id":"https://openalex.org/keywords/test","display_name":"Test (biology)","score":0.3483999967575073}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7728000283241272},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.723800003528595},{"id":"https://openalex.org/C2777200299","wikidata":"https://www.wikidata.org/wiki/Q52943","display_name":"Conversation","level":2,"score":0.7199000120162964},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.7178999781608582},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.6432999968528748},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5372999906539917},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.4609000086784363},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.41690000891685486},{"id":"https://openalex.org/C2777267654","wikidata":"https://www.wikidata.org/wiki/Q3519023","display_name":"Test (biology)","level":2,"score":0.3483999967575073},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.3463999927043915},{"id":"https://openalex.org/C2779679103","wikidata":"https://www.wikidata.org/wiki/Q5251805","display_name":"Degradation (telecommunications)","level":2,"score":0.31459999084472656},{"id":"https://openalex.org/C183322885","wikidata":"https://www.wikidata.org/wiki/Q17007702","display_name":"Context model","level":3,"score":0.30390000343322754},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.30219998955726624},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.26969999074935913},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.25360000133514404}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.12941","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.12941","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.2606.12941","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.12941","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":[{"id":"https://metadata.un.org/sdg/4","score":0.4396517276763916,"display_name":"Quality Education"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"When":[0],"a":[1,38,46,56],"user":[2],"reveals":[3],"task-critical":[4],"information":[5],"across":[6],"several":[7],"conversation":[8],"turns,":[9],"LLM":[10],"accuracy":[11,88],"drops":[12],"by":[13,33],"up":[14],"to":[15,36,45,92,117],"65%":[16],"despite":[17],"full":[18,109],"context":[19],"availability.":[20],"We":[21],"show":[22],"that":[23,60,115],"this":[24],"Lost":[25],"in":[26],"Conversation":[27],"degradation":[28],"can":[29],"be":[30],"substantially":[31],"mitigated":[32],"training":[34,52],"models":[35,101],"maintain":[37],"compact":[39],"rolling":[40],"memory":[41],"instead":[42],"of":[43,74],"attending":[44],"growing":[47],"history.":[48],"To":[49],"make":[50],"such":[51],"scalable,":[53],"we":[54],"introduce":[55],"low-cost":[57],"sharding":[58],"pipeline":[59],"converts":[61],"single-turn":[62],"QA":[63],"datasets":[64],"into":[65],"multi-turn":[66,87],"fragmented-information":[67],"episodes,":[68],"eliminating":[69],"the":[70,108],"need":[71],"for":[72],"hours":[73],"manual":[75],"annotation.":[76],"Training":[77],"only":[78],"on":[79],"sharded":[80],"GSM8K,":[81],"our":[82],"memory-augmented":[83],"policy":[84],"significantly":[85],"improves":[86],"and":[89,95],"generalises":[90],"zero-shot":[91],"harder":[93],"math":[94],"out-of-domain":[96],"long-context":[97],"QA.":[98],"Moreover,":[99],"memory-trained":[100],"outperform":[102],"full-history":[103],"baselines":[104],"even":[105],"when":[106],"given":[107],"history":[110],"at":[111],"test":[112],"time,":[113],"suggesting":[114],"learning":[116],"compress":[118],"induces":[119],"more":[120],"robust":[121],"incremental":[122],"reasoning":[123],"than":[124],"full-context":[125],"exposure":[126],"alone.":[127]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-13T00:00:00"}