{"id":"https://openalex.org/W7204833972","doi":"https://doi.org/10.20944/preprints202608.2150.v1","title":"EchoMuon: Cross-Timescale Gating in Muon, and the Learning-Rate Confound in Gated Optimizers","display_name":"EchoMuon: Cross-Timescale Gating in Muon, and the Learning-Rate Confound in Gated Optimizers","publication_year":2026,"publication_date":"2026-08-30","ids":{"openalex":"https://openalex.org/W7204833972","doi":"https://doi.org/10.20944/preprints202608.2150.v1"},"language":"en","primary_location":{"id":"doi:10.20944/preprints202608.2150.v1","doi":"https://doi.org/10.20944/preprints202608.2150.v1","is_oa":true,"landing_page_url":"https://doi.org/10.20944/preprints202608.2150.v1","pdf_url":null,"source":{"id":"https://openalex.org/S6309402219","display_name":"Preprints.org","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":["https://openalex.org/P4310310987"],"host_organization_lineage_names":["Multidisciplinary Digital Publishing Institute"],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":null,"raw_type":"posted-content"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.20944/preprints202608.2150.v1","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5047158395","display_name":"Stamatis Mastromichalakis","orcid":"https://orcid.org/0000-0003-1713-5078","observed_orcids":["https://orcid.org/0000-0003-1713-5078"]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Stamatis Mastromichalakis","raw_affiliation_strings":[],"raw_orcid":"https://orcid.org/0000-0003-1713-5078","affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5047158395"],"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9648000001907349,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9648000001907349,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.010700000450015068,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.004600000102072954,"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/learning-rate","display_name":"learning rate","score":0.4819999933242798},{"id":"https://openalex.org/keywords/cifar","display_name":"CIFAR","score":0.3840000033378601},{"id":"https://openalex.org/keywords/tiny-imagenet","display_name":"Tiny ImageNet","score":0.953000009059906}],"concepts":[{"id":"https://openalex.org/C187691185","wikidata":"https://www.wikidata.org/wiki/Q2020720","display_name":"Grid","level":2,"score":0.6197999715805054},{"id":"https://openalex.org/C194544171","wikidata":"https://www.wikidata.org/wiki/Q21105679","display_name":"Gating","level":2,"score":0.5935999751091003},{"id":"https://openalex.org/C60718061","wikidata":"https://www.wikidata.org/wiki/Q1414747","display_name":"Momentum (technical analysis)","level":2,"score":0.5436999797821045},{"id":"https://openalex.org/C124584101","wikidata":"https://www.wikidata.org/wiki/Q1053266","display_name":"Multiplier (economics)","level":2,"score":0.5156999826431274},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4973999857902527},{"id":"https://openalex.org/C57691317","wikidata":"https://www.wikidata.org/wiki/Q1289248","display_name":"Scalar (mathematics)","level":2,"score":0.42010000348091125},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.41929998993873596},{"id":"https://openalex.org/C205334942","wikidata":"https://www.wikidata.org/wiki/Q3151","display_name":"Muon","level":2,"score":0.38679999113082886},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.3758000135421753},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3440999984741211},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.34119999408721924},{"id":"https://openalex.org/C79581498","wikidata":"https://www.wikidata.org/wiki/Q1367530","display_name":"Suite","level":2,"score":0.3411000072956085},{"id":"https://openalex.org/C2983008078","wikidata":"https://www.wikidata.org/wiki/Q863495","display_name":"Grid cell","level":3,"score":0.3077000081539154},{"id":"https://openalex.org/C184720557","wikidata":"https://www.wikidata.org/wiki/Q7825049","display_name":"Topology (electrical circuits)","level":2,"score":0.2614000141620636},{"id":"https://openalex.org/C44154836","wikidata":"https://www.wikidata.org/wiki/Q45045","display_name":"Simulation","level":1,"score":0.25270000100135803}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.20944/preprints202608.2150.v1","doi":"https://doi.org/10.20944/preprints202608.2150.v1","is_oa":true,"landing_page_url":"https://doi.org/10.20944/preprints202608.2150.v1","pdf_url":null,"source":{"id":"https://openalex.org/S6309402219","display_name":"Preprints.org","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":["https://openalex.org/P4310310987"],"host_organization_lineage_names":["Multidisciplinary Digital Publishing Institute"],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":null,"raw_type":"posted-content"}],"best_oa_location":{"id":"doi:10.20944/preprints202608.2150.v1","doi":"https://doi.org/10.20944/preprints202608.2150.v1","is_oa":true,"landing_page_url":"https://doi.org/10.20944/preprints202608.2150.v1","pdf_url":null,"source":{"id":"https://openalex.org/S6309402219","display_name":"Preprints.org","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":["https://openalex.org/P4310310987"],"host_organization_lineage_names":["Multidisciplinary Digital Publishing Institute"],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":null,"raw_type":"posted-content"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7161139062","https://openalex.org/W7163382933","https://openalex.org/W7170391031","https://openalex.org/W7160875907","https://openalex.org/W7214710398","https://openalex.org/W7220470082","https://openalex.org/W7214763360","https://openalex.org/W7127945923","https://openalex.org/W4417156118","https://openalex.org/W4416617024"],"abstract_inverted_index":{"Muon":[0,82],"orthogonalizes":[1],"the":[2,10,62,69,110,114,118,121,127,136,144,154,166,198,218,229,273,277,304,314,321,327],"momentum":[3,28],"matrix,":[4],"granting":[5],"every":[6],"singular":[7],"direction":[8],"of":[9,169,228],"update":[11],"equal":[12],"trust.":[13],"EchoMuon":[14,290],"prices":[15],"that":[16,34,340],"trust":[17],"by":[18,131,185,329,342],"each":[19,331],"direction's":[20],"echo:":[21],"its":[22,213,335],"support":[23],"in":[24,57,99,344],"a":[25,40,47,50,84,170,179,221,238,266,338,354],"second,":[26],"slower":[27],"buffer.":[29],"This":[30],"paper":[31],"reports":[32],"what":[33,37,232],"buys,":[35],"and":[36,61,89,126,176,189,216,281,296,320],"measuring":[38],"such":[39],"gate":[41,45,145],"honestly":[42],"costs.":[43],"The":[44,246],"is":[46,76,124,129,203,234],"contraction,":[48],"not":[49,108,263],"reallocation.":[51],"Its":[52],"multiplier":[53],"never":[54],"exceeds":[55],"one":[56,93],"5,760":[58],"logged":[59],"layer-steps,":[60],"arm":[63,123],"carrying":[64],"it":[65,75,175],"sits":[66,116],"higher":[67],"on":[68,148,178,217,292,358],"learning-rate":[70,104,267],"axis:":[71],"over":[72,212],"four":[73],"cells":[74,272],"penalized":[77],"3.7x":[78],"less":[79],"than":[80,237],"ungated":[81],"for":[83,92],"step":[85],"twice":[86],"too":[87,95],"large,":[88],"1.6x":[90],"more":[91],"half":[94],"small,":[96],"without":[97],"exception":[98],"either":[100],"direction.":[101],"A":[102],"shared":[103,172],"grid":[105,115,200],"therefore":[106],"does":[107],"compare":[109],"two":[111,214],"fairly:":[112],"when":[113],"above":[117],"joint":[119],"optimum,":[120],"contracting":[122],"rescued":[125],"other":[128],"not,":[130],"an":[132,349],"amount":[133],"comparable":[134],"to":[135,187,258,264],"margins":[137,184],"being":[138],"measured.":[139,245],"Our":[140],"initial":[141],"hypotheses":[142],"about":[143],"were":[146],"wrong":[147],"this":[149],"point,":[150],"so":[151,231],"we":[152,307,324],"re-ran":[153],"full":[155],"suite":[156,316],"under":[157,299,348],"corrected":[158],"selection.":[159],"All":[160],"eight":[161],"CIFAR":[162],"arm-cells":[163],"had":[164],"picked":[165],"bottom":[167],"edge":[168],"single":[171],"grid;":[173],"widening":[174],"selecting":[177],"held-out":[180],"split":[181],"cuts":[182],"those":[183],"42":[186],"80%":[188],"leaves":[190],"none":[191],"individually":[192],"significant.":[193],"Tiny":[194],"ImageNet,":[195],"whose":[196],"optimum":[197],"original":[199],"did":[201],"contain,":[202],"unchanged":[204],"at":[205,253,260],"+1.65":[206],"pp":[207,356],"(t":[208],"=":[209],"+7.4)":[210],"pooled":[211],"cells,":[215],"clean":[219,293],"cell":[220],"norm-matched":[222],"scalar":[223],"control":[224],"reproduces":[225],"only":[226,298],"30%":[227],"gain,":[230],"remains":[233],"directional":[235],"rather":[236],"step-size":[239],"effect.":[240],"Three":[241],"further":[242],"boundaries":[243],"are":[244],"FineWeb":[247],"advantage":[248],"falls":[249],"from":[250],"-0.031":[251],"nats":[252],"0.3":[254],"tokens":[255],"per":[256],"parameter":[257],"-0.003":[259],"0.9,":[261],"verified":[262],"be":[265],"artifact.":[268],"Across":[269],"ten":[270],"vision":[271],"margin":[274,328],"scales":[275],"with":[276,353],"baseline's":[278],"error":[279],"rate":[280,333],"crosses":[282],"zero":[283],"near":[284],"90%":[285],"accuracy.":[286,360],"On":[287],"byte-level":[288],"text":[289],"wins":[291],"enwik8":[294],"(t=-10.1)":[295],"loses":[297],"input":[300],"corruption,":[301],"which":[302,330],"replaces":[303],"tokenization":[305],"boundary":[306],"first":[308],"proposed.":[309],"We":[310],"release":[311],"1,815":[312],"runs,":[313],"re-run":[315],"behind":[317],"these":[318],"numbers,":[319],"reporting":[322],"practice":[323],"recommend:":[325],"publish":[326],"learning":[332],"beat":[334],"runner-up.":[336],"Here":[337],"pick":[339],"won":[341],"0.0026":[343],"validation":[345],"loss":[346],"reversed":[347],"equally":[350],"valid":[351],"split,":[352],"1.45":[355],"consequence":[357],"test":[359]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-09-01T00:00:00"}