{"id":"https://openalex.org/W7196942663","doi":"https://doi.org/10.1145/3770855.3818131","title":"NumCache: KV Cache Compression and Retrieval for Financial Document QA","display_name":"NumCache: KV Cache Compression and Retrieval for Financial Document QA","publication_year":2026,"publication_date":"2026-08-06","ids":{"openalex":"https://openalex.org/W7196942663","doi":"https://doi.org/10.1145/3770855.3818131"},"language":"en","primary_location":{"id":"doi:10.1145/3770855.3818131","doi":"https://doi.org/10.1145/3770855.3818131","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3770855.3818131","pdf_url":null,"source":{"id":"https://openalex.org/S4363608767","display_name":"ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","is_traditional":true,"indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://doi.org/10.1145/3770855.3818131","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5088625984","display_name":"Eftychia Makri","orcid":"https://orcid.org/0009-0003-0469-4812","observed_orcids":["https://orcid.org/0009-0003-0469-4812"]},"institutions":[{"id":"https://openalex.org/I32971472","display_name":"Yale University","ror":"https://ror.org/03v76x132","country_code":"US","type":"education","lineage":["https://openalex.org/I32971472"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Eftychia Makri","raw_affiliation_strings":["Yale University, New Haven, CT, USA"],"raw_orcid":"https://orcid.org/0009-0003-0469-4812","affiliations":[{"raw_affiliation_string":"Yale University, New Haven, CT, USA","institution_ids":["https://openalex.org/I32971472"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101534635","display_name":"Peiwen Li","orcid":"https://orcid.org/0009-0009-8318-2420","observed_orcids":["https://orcid.org/0009-0009-8318-2420"]},"institutions":[{"id":"https://openalex.org/I32971472","display_name":"Yale University","ror":"https://ror.org/03v76x132","country_code":"US","type":"education","lineage":["https://openalex.org/I32971472"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Peiwen Li","raw_affiliation_strings":["Yale University, New Haven, CT, USA"],"raw_orcid":"https://orcid.org/0009-0009-8318-2420","affiliations":[{"raw_affiliation_string":"Yale University, New Haven, CT, USA","institution_ids":["https://openalex.org/I32971472"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126120620","display_name":"Yidong Jiang","orcid":"https://orcid.org/0009-0000-4033-8709","observed_orcids":["https://orcid.org/0009-0000-4033-8709"]},"institutions":[{"id":"https://openalex.org/I116953780","display_name":"Tongji University","ror":"https://ror.org/03rc6as71","country_code":"CN","type":"education","lineage":["https://openalex.org/I116953780"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yidong Jiang","raw_affiliation_strings":["Tongji University, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0000-4033-8709","affiliations":[{"raw_affiliation_string":"Tongji University, Shanghai, China","institution_ids":["https://openalex.org/I116953780"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125496080","display_name":"Junrong Chen","orcid":"https://orcid.org/0009-0009-2529-3114","observed_orcids":["https://orcid.org/0009-0009-2529-3114"]},"institutions":[{"id":"https://openalex.org/I32971472","display_name":"Yale University","ror":"https://ror.org/03v76x132","country_code":"US","type":"education","lineage":["https://openalex.org/I32971472"]},{"id":"https://openalex.org/I36258959","display_name":"University of California San Diego","ror":"https://ror.org/0168r3w48","country_code":"US","type":"education","lineage":["https://openalex.org/I36258959"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Junrong Chen","raw_affiliation_strings":["University of California, San Diego, San Diego, CA, USA","Yale University, New Haven, CT, USA"],"raw_orcid":"https://orcid.org/0009-0009-2529-3114","affiliations":[{"raw_affiliation_string":"University of California, San Diego, San Diego, CA, USA","institution_ids":["https://openalex.org/I36258959"]},{"raw_affiliation_string":"Yale University, New Haven, CT, USA","institution_ids":["https://openalex.org/I32971472"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100339552","display_name":"Jialin Chen","orcid":"https://orcid.org/0009-0007-0909-4620","observed_orcids":["https://orcid.org/0009-0007-0909-4620"]},"institutions":[{"id":"https://openalex.org/I32971472","display_name":"Yale University","ror":"https://ror.org/03v76x132","country_code":"US","type":"education","lineage":["https://openalex.org/I32971472"]},{"id":"https://openalex.org/I36258959","display_name":"University of California San Diego","ror":"https://ror.org/0168r3w48","country_code":"US","type":"education","lineage":["https://openalex.org/I36258959"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jialin Chen","raw_affiliation_strings":["University of California, San Diego, San Diego, CA, USA","Yale University, New Haven, CT, USA"],"raw_orcid":"https://orcid.org/0009-0007-0909-4620","affiliations":[{"raw_affiliation_string":"University of California, San Diego, San Diego, CA, USA","institution_ids":["https://openalex.org/I36258959"]},{"raw_affiliation_string":"Yale University, New Haven, CT, USA","institution_ids":["https://openalex.org/I32971472"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5072222232","display_name":"Ali Maatouk","orcid":"https://orcid.org/0000-0002-3436-7068","observed_orcids":["https://orcid.org/0000-0002-3436-7068"]},"institutions":[{"id":"https://openalex.org/I32971472","display_name":"Yale University","ror":"https://ror.org/03v76x132","country_code":"US","type":"education","lineage":["https://openalex.org/I32971472"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ali Maatouk","raw_affiliation_strings":["Yale University, New Haven, CT, USA"],"raw_orcid":"https://orcid.org/0000-0002-3436-7068","affiliations":[{"raw_affiliation_string":"Yale University, New Haven, CT, USA","institution_ids":["https://openalex.org/I32971472"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5014892027","display_name":"Leandros Tassiulas","orcid":"https://orcid.org/0000-0003-0932-774X","observed_orcids":["https://orcid.org/0000-0003-0932-774X"]},"institutions":[{"id":"https://openalex.org/I32971472","display_name":"Yale University","ror":"https://ror.org/03v76x132","country_code":"US","type":"education","lineage":["https://openalex.org/I32971472"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Leandros Tassiulas","raw_affiliation_strings":["Yale University, New Haven, CT, USA"],"raw_orcid":"https://orcid.org/0000-0003-0932-774X","affiliations":[{"raw_affiliation_string":"Yale University, New Haven, CT, USA","institution_ids":["https://openalex.org/I32971472"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125514656","display_name":"Eliot Brenner","orcid":"https://orcid.org/0000-0002-9176-0595","observed_orcids":["https://orcid.org/0000-0002-9176-0595"]},"institutions":[{"id":"https://openalex.org/I40713646","display_name":"Goldman Sachs (United States)","ror":"https://ror.org/031rnv444","country_code":"US","type":"company","lineage":["https://openalex.org/I40713646"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Eliot Brenner","raw_affiliation_strings":["Goldman Sachs, New York, New York, USA"],"raw_orcid":"https://orcid.org/0000-0002-9176-0595","affiliations":[{"raw_affiliation_string":"Goldman Sachs, New York, New York, USA","institution_ids":["https://openalex.org/I40713646"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5107249743","display_name":"Bing Xiang","orcid":"https://orcid.org/0009-0006-4028-4935","observed_orcids":["https://orcid.org/0009-0006-4028-4935"]},"institutions":[{"id":"https://openalex.org/I40713646","display_name":"Goldman Sachs (United States)","ror":"https://ror.org/031rnv444","country_code":"US","type":"company","lineage":["https://openalex.org/I40713646"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Bing Xiang","raw_affiliation_strings":["Goldman Sachs, New York, USA"],"raw_orcid":"https://orcid.org/0009-0006-4028-4935","affiliations":[{"raw_affiliation_string":"Goldman Sachs, New York, USA","institution_ids":["https://openalex.org/I40713646"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5078337825","display_name":"Rex Ying","orcid":"https://orcid.org/0000-0002-5856-5229","observed_orcids":["https://orcid.org/0000-0002-5856-5229"]},"institutions":[{"id":"https://openalex.org/I32971472","display_name":"Yale University","ror":"https://ror.org/03v76x132","country_code":"US","type":"education","lineage":["https://openalex.org/I32971472"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Rex Ying","raw_affiliation_strings":["Yale University, New Haven, CT, USA"],"raw_orcid":"https://orcid.org/0000-0002-5856-5229","affiliations":[{"raw_affiliation_string":"Yale University, New Haven, CT, USA","institution_ids":["https://openalex.org/I32971472"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.0,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"3655","last_page":"3666"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9643999934196472,"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":0.9643999934196472,"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":0.025299999862909317,"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/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"}}],"keywords":[{"id":"https://openalex.org/keywords/kv-cache-compression","display_name":"KV cache compression","score":0.9110000133514404},{"id":"https://openalex.org/keywords/sec-filings","display_name":"SEC filings","score":0.9919999837875366},{"id":"https://openalex.org/keywords/long-context-retrieval","display_name":"long-context retrieval","score":0.6340000033378601},{"id":"https://openalex.org/keywords/retrieval-augmented-generation","display_name":"retrieval-augmented generation","score":0.8240000009536743},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.7319999933242798}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7736999988555908},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.6420000195503235},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.6378999948501587},{"id":"https://openalex.org/C774472","wikidata":"https://www.wikidata.org/wiki/Q6760393","display_name":"Margin (machine learning)","level":2,"score":0.5135999917984009},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.49900001287460327},{"id":"https://openalex.org/C2776459999","wikidata":"https://www.wikidata.org/wiki/Q2119376","display_name":"Fidelity","level":2,"score":0.4607999920845032},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.44699999690055847},{"id":"https://openalex.org/C147494362","wikidata":"https://www.wikidata.org/wiki/Q2078905","display_name":"Troubleshooting","level":2,"score":0.4399000108242035},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.3912999927997589},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.3612000048160553},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.31839999556541443},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.3154999911785126},{"id":"https://openalex.org/C2776034101","wikidata":"https://www.wikidata.org/wiki/Q1509347","display_name":"Commission","level":2,"score":0.30959999561309814},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.305400013923645},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.2939000129699707},{"id":"https://openalex.org/C10138342","wikidata":"https://www.wikidata.org/wiki/Q43015","display_name":"Finance","level":1,"score":0.29190000891685486},{"id":"https://openalex.org/C77088390","wikidata":"https://www.wikidata.org/wiki/Q8513","display_name":"Database","level":1,"score":0.29109999537467957},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.28679999709129333},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.2863999903202057},{"id":"https://openalex.org/C180016635","wikidata":"https://www.wikidata.org/wiki/Q2712821","display_name":"Compression (physics)","level":2,"score":0.2858000099658966},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.2802000045776367},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2743000090122223},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.26809999346733093},{"id":"https://openalex.org/C189783530","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"CPU cache","level":3,"score":0.26190000772476196},{"id":"https://openalex.org/C199776023","wikidata":"https://www.wikidata.org/wiki/Q202875","display_name":"Negotiation","level":2,"score":0.2565999925136566},{"id":"https://openalex.org/C113364801","wikidata":"https://www.wikidata.org/wiki/Q26674","display_name":"High fidelity","level":2,"score":0.2540999948978424},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.2508000135421753}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3770855.3818131","doi":"https://doi.org/10.1145/3770855.3818131","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3770855.3818131","pdf_url":null,"source":{"id":"https://openalex.org/S4363608767","display_name":"ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3770855.3818131","doi":"https://doi.org/10.1145/3770855.3818131","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3770855.3818131","pdf_url":null,"source":{"id":"https://openalex.org/S4363608767","display_name":"ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":27,"referenced_works":["https://openalex.org/W4252076394","https://openalex.org/W4297808394","https://openalex.org/W4310923309","https://openalex.org/W4320813768","https://openalex.org/W4384392961","https://openalex.org/W4387321091","https://openalex.org/W4388926382","https://openalex.org/W4391876619","https://openalex.org/W4399794499","https://openalex.org/W4399836596","https://openalex.org/W4400611829","https://openalex.org/W4402670856","https://openalex.org/W4404781754","https://openalex.org/W4404782265","https://openalex.org/W4404792869","https://openalex.org/W4406272066","https://openalex.org/W4410636796","https://openalex.org/W4412945187","https://openalex.org/W4414836780","https://openalex.org/W4416035984","https://openalex.org/W4416037017","https://openalex.org/W4416620348","https://openalex.org/W4417114725","https://openalex.org/W6929482003","https://openalex.org/W7075994024","https://openalex.org/W7106696182","https://openalex.org/W7164653744"],"related_works":["https://openalex.org/W7125025533","https://openalex.org/W7160615865","https://openalex.org/W7163017587","https://openalex.org/W7160244788","https://openalex.org/W7106657789","https://openalex.org/W7123454532","https://openalex.org/W7104392137","https://openalex.org/W7151464617","https://openalex.org/W4416035421","https://openalex.org/W7223353990"],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"are":[4,30,111],"increasingly":[5],"deployed":[6],"in":[7],"financial":[8,21,148,246],"applications,":[9],"particularly":[10],"for":[11,86,205,244],"interpreting":[12],"U.S.":[13],"Securities":[14],"and":[15,35,52,76,99,114,121,144,188,211],"Exchange":[16],"Commission":[17],"(SEC)":[18],"filings.":[19],"However,":[20],"QA":[22,181],"over":[23],"these":[24,153],"filings":[25,135],"is":[26],"challenging,":[27],"as":[28,103,240],"they":[29],"extremely":[31],"long,":[32],"numerically":[33,141],"dense,":[34],"often":[36],"require":[37],"cross-document":[38],"reasoning.":[39,88],"Existing":[40],"approaches":[41],"struggle":[42],"to":[43,45,49,119,195],"scale":[44],"such":[46],"settings":[47],"due":[48],"long-context":[50,245],"degradation":[51],"loss":[53],"of":[54,152],"numerical":[55],"fidelity":[56],"under":[57],"context":[58,197],"compression.":[59],"Long-document":[60],"Retrieval-Augmented":[61],"Generation":[62,90],"(RAG)":[63],"improves":[64],"evidence":[65],"coverage":[66],"through":[67],"coarse-to-fine":[68],"retrieval,":[69],"yet":[70],"semantic":[71],"retrievers":[72],"collapse":[73],"fine-grained":[74],"magnitudes":[75],"units,":[77],"returning":[78],"passages":[79],"that":[80,161],"lack":[81],"the":[82,175,227],"precise":[83],"values":[84],"needed":[85],"correct":[87],"Cache-Augmented":[89],"(CAG)":[91],"projects":[92],"attention":[93],"states":[94],"into":[95,136],"compact":[96],"KV":[97,137],"representations":[98,239],"treats":[100],"precomputed":[101],"caches":[102,110,138],"reusable":[104],"internal":[105],"memory,":[106],"but":[107],"typically":[108],"assumes":[109],"already":[112],"well-formed":[113],"query-relevant,":[115],"leaving":[116],"open":[117],"how":[118],"build":[120],"select":[122],"number-faithful":[123],"caches.":[124],"To":[125],"address":[126],"this":[127],"gap,":[128],"we":[129,155],"propose":[130],"NumCache,":[131],"which":[132,178],"compresses":[133],"SEC":[134],"initialized":[139],"from":[140],"dense":[142],"regions":[143],"trained":[145],"directly":[146],"on":[147,174,208,223],"QAs.":[149],"On":[150],"top":[151],"caches,":[154],"then":[156],"train":[157],"a":[158,212],"contrastive":[159,218],"retriever":[160,219],"aligns":[162],"questions":[163],"with":[164,199,237],"cache":[165],"representations,":[166],"thus":[167],"improving":[168],"retrieval":[169,236],"performance.":[170],"We":[171],"evaluate":[172],"NumCache":[173,192],"Fin-RATE":[176],"benchmark,":[177],"comprises":[179],"SEC-filing":[180],"tasks":[182],"covering":[183],"single-filing":[184,209,224],"reasoning,":[185],"cross-firm":[186],"comparison,":[187],"longitudinal":[189],"trend":[190],"analysis.":[191],"achieves":[193],"up":[194],"4\u00d7":[196],"compression":[198],"competitive":[200],"accuracy":[201],"(41.4%":[202],"vs.\\":[203],"44.8%":[204],"uncompressed":[206],"Qwen3-4B":[207],"reasoning)":[210],"6.4\u00d7":[213],"inference":[214],"speedup,":[215],"while":[216],"its":[217],"attains":[220],"68.3%":[221],"Recall@1":[222],"retrieval;":[225],"2.3\u00d7":[226],"strongest":[228],"text-based":[229],"baseline":[230],"(29.7%).":[231],"These":[232],"results":[233],"highlight":[234],"cache-based":[235],"number-preserving":[238],"an":[241],"effective":[242],"approach":[243],"QA.":[247]},"counts_by_year":[],"updated_date":"2026-11-04T10:54:05.112534","created_date":"2026-08-07T00:00:00"}