{"id":"https://openalex.org/W7204144899","doi":"https://doi.org/10.48550/arxiv.2608.20961","title":"TreeWY: Speculative Verification for Gated DeltaNet Hybrids","display_name":"TreeWY: Speculative Verification for Gated DeltaNet Hybrids","publication_year":2026,"publication_date":"2026-08-21","ids":{"openalex":"https://openalex.org/W7204144899","doi":"https://doi.org/10.48550/arxiv.2608.20961"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.20961","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.20961","pdf_url":"https://arxiv.org/pdf/2608.20961","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":"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/2608.20961","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5147989727","display_name":"Sneha Murthy Ghantasala","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Ghantasala, Sneha Murthy","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5147989727"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":4,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.6851000189781189,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.6851000189781189,"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/T10028","display_name":"Topic Modeling","score":0.14270000159740448,"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.03099999949336052,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/speculative-decoding","display_name":"speculative decoding","score":0.9459999799728394},{"id":"https://openalex.org/keywords/hybrid-language-models","display_name":"hybrid language models","score":0.4959999918937683},{"id":"https://openalex.org/keywords/key-value-cache","display_name":"key-value cache","score":0.5899999737739563},{"id":"https://openalex.org/keywords/memory-efficiency","display_name":"memory efficiency","score":0.5289999842643738},{"id":"https://openalex.org/keywords/time-to-first-token","display_name":"time-to-first-token","score":0.7860000133514404},{"id":"https://openalex.org/keywords/qwen3-5","display_name":"Qwen3.5","score":0.7269999980926514}],"concepts":[{"id":"https://openalex.org/C55282118","wikidata":"https://www.wikidata.org/wiki/Q252683","display_name":"Snapshot (computer storage)","level":2,"score":0.5770999789237976},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5666000247001648},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.5238000154495239},{"id":"https://openalex.org/C113174947","wikidata":"https://www.wikidata.org/wiki/Q2859736","display_name":"Tree (set theory)","level":2,"score":0.4964999854564667},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.45329999923706055},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.4521999955177307},{"id":"https://openalex.org/C198082294","wikidata":"https://www.wikidata.org/wiki/Q3399648","display_name":"Position (finance)","level":2,"score":0.435699999332428},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.37700000405311584},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.37630000710487366},{"id":"https://openalex.org/C94375191","wikidata":"https://www.wikidata.org/wiki/Q11205","display_name":"Arithmetic","level":1,"score":0.3319000005722046},{"id":"https://openalex.org/C37423430","wikidata":"https://www.wikidata.org/wiki/Q6750281","display_name":"Hull","level":2,"score":0.31139999628067017},{"id":"https://openalex.org/C106487976","wikidata":"https://www.wikidata.org/wiki/Q685816","display_name":"Matrix (chemical analysis)","level":2,"score":0.2831000089645386},{"id":"https://openalex.org/C164624739","wikidata":"https://www.wikidata.org/wiki/Q754331","display_name":"Activity-based costing","level":2,"score":0.2770000100135803},{"id":"https://openalex.org/C104528550","wikidata":"https://www.wikidata.org/wiki/Q506265","display_name":"Triangular matrix","level":3,"score":0.25859999656677246},{"id":"https://openalex.org/C107814960","wikidata":"https://www.wikidata.org/wiki/Q611957","display_name":"Solid-state","level":2,"score":0.25699999928474426},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.25360000133514404}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.20961","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.20961","pdf_url":"https://arxiv.org/pdf/2608.20961","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.20961","doi":"https://doi.org/10.48550/arxiv.2608.20961","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.20961","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:2608.20961","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.20961","pdf_url":"https://arxiv.org/pdf/2608.20961","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":"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/W7204144899.pdf","grobid_xml":"https://content.openalex.org/works/W7204144899.grobid-xml"},"referenced_works_count":6,"referenced_works":["https://openalex.org/W4319166707","https://openalex.org/W4391124766","https://openalex.org/W4412875469","https://openalex.org/W4415327314","https://openalex.org/W4415795816","https://openalex.org/W7172426877"],"related_works":["https://openalex.org/W7214037119","https://openalex.org/W7161849267","https://openalex.org/W7214759301","https://openalex.org/W7172426877","https://openalex.org/W7161697858","https://openalex.org/W7204098307","https://openalex.org/W7167793395","https://openalex.org/W7163216738","https://openalex.org/W7222100008","https://openalex.org/W7204895662"],"abstract_inverted_index":{"Modern":[0],"open":[1],"models":[2],"are":[3,7],"hybrids:":[4],"most":[5],"layers":[6,12],"linear-attention":[8],"(Gated":[9],"DeltaNet,":[10],"GDN)":[11],"carrying":[13],"a":[14,21,37,72,76,87,103,117,185,200,209],"small":[15,118],"fixed-size":[16],"recurrent":[17,54],"state":[18,55,113],"instead":[19,121],"of":[20,39,71,91,122,146],"growing":[22],"key-value":[23],"(KV)":[24],"cache.":[25],"This":[26],"makes":[27],"ordinary":[28],"decoding":[29],"memory-efficient,":[30],"but":[31],"hurts":[32],"speculative":[33,157],"decoding.":[34],"To":[35],"verify":[36],"batch":[38],"draft":[40,58,73,99,203],"tokens":[41],"and":[42,63,107,153,160,175,183],"then":[43],"roll":[44],"back":[45],"the":[46,52,84,92,110,125,130,169,195],"rejected":[47],"ones,":[48],"today's":[49],"systems":[50],"snapshot":[51],"full":[53],"at":[56,163],"every":[57,98],"position":[59],"for":[60],"GDN":[61],"layers,":[62],"those":[64],"snapshots":[65],"cannot":[66],"be":[67],"shared":[68],"across":[69],"branches":[70],"tree,":[74],"so":[75],"wide,":[77],"high-acceptance":[78],"tree":[79,193],"becomes":[80,204],"memory-infeasible.":[81],"We":[82],"remove":[83],"snapshots.":[85],"Using":[86],"tree-structured":[88],"WY":[89],"transform":[90],"gated":[93,131],"delta":[94,132],"rule,":[95,133],"we":[96],"compute":[97],"node's":[100],"output":[101],"with":[102],"single":[104],"triangular":[105],"solve":[106],"reconstruct":[108],"only":[109,128],"one":[111,147],"accepted":[112],"on":[114,129,135,143],"commit,":[115],"storing":[116],"pseudo-value":[119],"matrix":[120],"per-node":[123],"states;":[124],"derivation":[126],"depends":[127],"not":[134,207],"any":[136],"other":[137],"architectural":[138],"detail.":[139],"In":[140],"serving":[141],"benchmarks":[142],"two":[144],"scales":[145],"hybrid":[148],"model":[149],"family":[150],"(Qwen3.5":[151],"35B":[152],"397B)":[154],"this":[155],"cuts":[156],"recurrent-state":[158],"memory":[159,181,197],"KV-cache":[161],"pressure":[162],"essentially":[164],"identical":[165],"acceptance":[166],"length,":[167],"turning":[168],"freed":[170],"HBM":[171],"into":[172],"higher":[173],"throughput":[174,210],"much":[176],"lower":[177],"time-to-first-token":[178],"(TTFT)":[179],"wherever":[180],"binds,":[182],"costing":[184],"few":[186],"percent":[187],"where":[188],"it":[189],"does":[190],"not.":[191],"For":[192],"width":[194],"same":[196],"buys":[198],"affordability:":[199],"wider,":[201],"higher-acceptance":[202],"possible,":[205],"though":[206],"yet":[208],"win.":[211]},"counts_by_year":[{"year":2026,"cited_by_count":4}],"updated_date":"2026-11-09T13:22:03.289127","created_date":"2026-08-25T00:00:00"}