{"id":"https://openalex.org/W7128552430","doi":"https://doi.org/10.48550/arxiv.2602.07397","title":"Scout Before You Attend: Sketch-and-Walk Sparse Attention for Efficient LLM Inference","display_name":"Scout Before You Attend: Sketch-and-Walk Sparse Attention for Efficient LLM Inference","publication_year":2026,"publication_date":"2026-02-07","ids":{"openalex":"https://openalex.org/W7128552430","doi":"https://doi.org/10.48550/arxiv.2602.07397"},"language":null,"primary_location":{"id":"pmh:doi:10.48550/arxiv.2602.07397","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":null,"any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5125378550","display_name":"Hoang Anh Duy Le","orcid":null},"institutions":[{"id":"https://openalex.org/I74775410","display_name":"Rice University","ror":"https://ror.org/008zs3103","country_code":"US","type":"education","lineage":["https://openalex.org/I74775410"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Le, Hoang Anh Duy","raw_affiliation_strings":["Rice University"],"affiliations":[{"raw_affiliation_string":"Rice University","institution_ids":["https://openalex.org/I74775410"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125405010","display_name":"Sahil Joshi","orcid":null},"institutions":[{"id":"https://openalex.org/I74775410","display_name":"Rice University","ror":"https://ror.org/008zs3103","country_code":"US","type":"education","lineage":["https://openalex.org/I74775410"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Joshi, Sahil","raw_affiliation_strings":["Rice University"],"affiliations":[{"raw_affiliation_string":"Rice University","institution_ids":["https://openalex.org/I74775410"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125581292","display_name":"Zeyu Yang","orcid":null},"institutions":[{"id":"https://openalex.org/I74775410","display_name":"Rice University","ror":"https://ror.org/008zs3103","country_code":"US","type":"education","lineage":["https://openalex.org/I74775410"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yang, Zeyu","raw_affiliation_strings":["Rice University"],"affiliations":[{"raw_affiliation_string":"Rice University","institution_ids":["https://openalex.org/I74775410"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5065317043","display_name":"Zhuoqun Xu","orcid":"https://orcid.org/0000-0002-3535-4402"},"institutions":[{"id":"https://openalex.org/I108468826","display_name":"Stevens Institute of Technology","ror":"https://ror.org/02z43xh36","country_code":"US","type":"education","lineage":["https://openalex.org/I108468826"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Xu, Zhaozhuo","raw_affiliation_strings":["Stevens Institute of Technology"],"affiliations":[{"raw_affiliation_string":"Stevens Institute of Technology","institution_ids":["https://openalex.org/I108468826"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5125587771","display_name":"Anshumali Shrivastava","orcid":null},"institutions":[{"id":"https://openalex.org/I74775410","display_name":"Rice University","ror":"https://ror.org/008zs3103","country_code":"US","type":"education","lineage":["https://openalex.org/I74775410"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Shrivastava, Anshumali","raw_affiliation_strings":["Rice University"],"affiliations":[{"raw_affiliation_string":"Rice University","institution_ids":["https://openalex.org/I74775410"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5125378550"],"corresponding_institution_ids":["https://openalex.org/I74775410"],"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/T10036","display_name":"Advanced Neural Network Applications","score":0.21279999613761902,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.21279999613761902,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.20990000665187836,"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/T10028","display_name":"Topic Modeling","score":0.16369999945163727,"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/inference","display_name":"Inference","score":0.8104000091552734},{"id":"https://openalex.org/keywords/range","display_name":"Range (aeronautics)","score":0.5530999898910522},{"id":"https://openalex.org/keywords/approximate-inference","display_name":"Approximate inference","score":0.45910000801086426},{"id":"https://openalex.org/keywords/hadamard-transform","display_name":"Hadamard transform","score":0.4293000102043152},{"id":"https://openalex.org/keywords/encoding","display_name":"Encoding (memory)","score":0.32580000162124634},{"id":"https://openalex.org/keywords/mechanism","display_name":"Mechanism (biology)","score":0.3240000009536743}],"concepts":[{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.8104000091552734},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7059000134468079},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.5530999898910522},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.46459999680519104},{"id":"https://openalex.org/C2777472644","wikidata":"https://www.wikidata.org/wiki/Q16968992","display_name":"Approximate inference","level":3,"score":0.45910000801086426},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4438000023365021},{"id":"https://openalex.org/C60292330","wikidata":"https://www.wikidata.org/wiki/Q1014065","display_name":"Hadamard transform","level":2,"score":0.4293000102043152},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.36800000071525574},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3425000011920929},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.32580000162124634},{"id":"https://openalex.org/C89611455","wikidata":"https://www.wikidata.org/wiki/Q6804646","display_name":"Mechanism (biology)","level":2,"score":0.3240000009536743},{"id":"https://openalex.org/C134261354","wikidata":"https://www.wikidata.org/wiki/Q938438","display_name":"Statistical inference","level":2,"score":0.31470000743865967},{"id":"https://openalex.org/C56372850","wikidata":"https://www.wikidata.org/wiki/Q1050404","display_name":"Sparse matrix","level":3,"score":0.31310001015663147},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.29190000891685486},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.2827000021934509},{"id":"https://openalex.org/C77637269","wikidata":"https://www.wikidata.org/wiki/Q7002051","display_name":"Neural coding","level":2,"score":0.2581000030040741},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.25270000100135803}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.48550/arxiv.2602.07397","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},{"id":"doi:10.48550/arxiv.2602.07397","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.07397","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":"article"}],"best_oa_location":{"id":"pmh:doi:10.48550/arxiv.2602.07397","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Self-attention":[0],"dominates":[1],"the":[2,93],"computational":[3],"and":[4,14,36,95,110,120],"memory":[5],"cost":[6],"of":[7,47,108],"long-context":[8],"LLM":[9],"inference":[10,134],"across":[11,54],"both":[12,92],"prefill":[13,94],"decode":[15,96],"phases.":[16],"To":[17],"address":[18],"this":[19],"challenge,":[20],"we":[21],"introduce":[22],"Sketch&Walk":[23,39,112],"Attention,":[24],"a":[25,57,84,105],"training-free":[26,86],"sparse":[27,101],"attention":[28,48,62,78,102,118,125],"method":[29],"that":[30,60,88],"determines":[31],"sparsity":[32,82],"with":[33,83,99],"lightweight":[34],"sketches":[35],"deterministic":[37],"walk.":[38],"applies":[40,89],"Hadamard":[41],"sketching":[42],"to":[43,75,91,132],"get":[44],"inexpensive":[45],"approximations":[46],"scores,":[49],"then":[50],"aggregates":[51],"these":[52],"estimates":[53],"layers":[55],"via":[56],"walk":[58,71],"mechanism":[59],"captures":[61],"influence":[63],"beyond":[64],"direct":[65],"interactions":[66],"between":[67],"tokens.":[68],"The":[69],"accumulated":[70],"scores":[72],"are":[73],"used":[74],"select":[76],"top-k":[77],"blocks,":[79],"enabling":[80],"dynamic":[81],"single":[85],"algorithm":[87],"uniformly":[90],"phases,":[97],"together":[98],"custom":[100],"kernels.":[103],"Across":[104],"wide":[106],"range":[107],"models":[109],"tasks,":[111],"maintains":[113],"near-lossless":[114],"accuracy":[115],"at":[116],"20%":[117],"density":[119],"can":[121],"slightly":[122],"outperform":[123],"dense":[124],"in":[126],"some":[127],"settings,":[128],"while":[129],"achieving":[130],"up":[131],"6x":[133],"speedup.":[135]},"counts_by_year":[],"updated_date":"2026-04-04T16:13:02.066488","created_date":"2026-02-11T00:00:00"}