{"id":"https://openalex.org/W7161763122","doi":"https://doi.org/10.48550/arxiv.2605.18797","title":"Simply Stabilizing the Loop via Fully Looped Transformer","display_name":"Simply Stabilizing the Loop via Fully Looped Transformer","publication_year":2026,"publication_date":"2026-05-11","ids":{"openalex":"https://openalex.org/W7161763122","doi":"https://doi.org/10.48550/arxiv.2605.18797"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.18797","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.18797","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":"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.18797","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5136576062","display_name":"Rao Fu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fu, Rao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136563754","display_name":"Zixuan Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Zixuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136577867","display_name":"Jiankun Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Jiankun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136587710","display_name":"Jing Ma","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ma, Jing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136605788","display_name":"Hechang Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Hechang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136562690","display_name":"Yu Li","orcid":"https://orcid.org/0000-0003-3227-4420"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5136507065","display_name":"Yi Chang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chang, Yi","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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.24650000035762787,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.24650000035762787,"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/T11206","display_name":"Model Reduction and Neural Networks","score":0.13740000128746033,"subfield":{"id":"https://openalex.org/subfields/3109","display_name":"Statistical and Nonlinear Physics"},"field":{"id":"https://openalex.org/fields/31","display_name":"Physics and Astronomy"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10502","display_name":"Advanced Memory and Neural Computing","score":0.06400000303983688,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.6700000166893005},{"id":"https://openalex.org/keywords/control-theory","display_name":"Control theory (sociology)","score":0.5796999931335449},{"id":"https://openalex.org/keywords/residual","display_name":"Residual","score":0.567799985408783},{"id":"https://openalex.org/keywords/adaptability","display_name":"Adaptability","score":0.5335999727249146},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.5223000049591064},{"id":"https://openalex.org/keywords/scaling","display_name":"Scaling","score":0.47209998965263367},{"id":"https://openalex.org/keywords/flyback-transformer","display_name":"Flyback transformer","score":0.3601999878883362},{"id":"https://openalex.org/keywords/distribution-transformer","display_name":"Distribution transformer","score":0.32820001244544983}],"concepts":[{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6700000166893005},{"id":"https://openalex.org/C47446073","wikidata":"https://www.wikidata.org/wiki/Q5165890","display_name":"Control theory (sociology)","level":3,"score":0.5796999931335449},{"id":"https://openalex.org/C155512373","wikidata":"https://www.wikidata.org/wiki/Q287450","display_name":"Residual","level":2,"score":0.567799985408783},{"id":"https://openalex.org/C177606310","wikidata":"https://www.wikidata.org/wiki/Q5674297","display_name":"Adaptability","level":2,"score":0.5335999727249146},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.5223000049591064},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5023999810218811},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.47209998965263367},{"id":"https://openalex.org/C17321440","wikidata":"https://www.wikidata.org/wiki/Q184909","display_name":"Flyback transformer","level":4,"score":0.3601999878883362},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.3370000123977661},{"id":"https://openalex.org/C117323899","wikidata":"https://www.wikidata.org/wiki/Q1907989","display_name":"Distribution transformer","level":4,"score":0.32820001244544983},{"id":"https://openalex.org/C58716799","wikidata":"https://www.wikidata.org/wiki/Q6035648","display_name":"Inner loop","level":3,"score":0.3246000111103058},{"id":"https://openalex.org/C2779624466","wikidata":"https://www.wikidata.org/wiki/Q838119","display_name":"For loop","level":3,"score":0.31709998846054077},{"id":"https://openalex.org/C180850520","wikidata":"https://www.wikidata.org/wiki/Q2166404","display_name":"Isolation transformer","level":4,"score":0.3012999892234802},{"id":"https://openalex.org/C207821765","wikidata":"https://www.wikidata.org/wiki/Q405372","display_name":"Instability","level":2,"score":0.27480000257492065},{"id":"https://openalex.org/C2983447341","wikidata":"https://www.wikidata.org/wiki/Q1413083","display_name":"Model parameter","level":2,"score":0.27219998836517334},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.2651999890804291},{"id":"https://openalex.org/C30403606","wikidata":"https://www.wikidata.org/wiki/Q2981904","display_name":"Electromagnetic coil","level":2,"score":0.26170000433921814},{"id":"https://openalex.org/C206588197","wikidata":"https://www.wikidata.org/wiki/Q846574","display_name":"Reuse","level":2,"score":0.26170000433921814},{"id":"https://openalex.org/C125884559","wikidata":"https://www.wikidata.org/wiki/Q369791","display_name":"Linear variable differential transformer","level":5,"score":0.25850000977516174}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.18797","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.18797","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.18797","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.18797","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":"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"},"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":{"Scaling":[0],"model":[1,6],"performance":[2,26,53,175],"typically":[3],"requires":[4],"increasing":[5,28],"size.":[7],"Looped":[8,58,96,105,139,163,169,186],"Transformer":[9,19,59,140,164,170,187],"offers":[10],"a":[11,48],"compelling":[12],"alternative":[13],"by":[14,176,203],"iteratively":[15],"reusing":[16],"the":[17,35,66,94,123,137],"same":[18],"blocks,":[20],"trading":[21],"additional":[22],"computation":[23],"for":[24,51],"improved":[25],"without":[27],"parameter":[29],"count":[30],"or":[31],"context":[32],"length.":[33],"Because":[34],"number":[36,67],"of":[37,68],"loop":[38,69,148,205],"iterations":[39,70,206],"can":[40],"be":[41,142],"adjusted":[42],"at":[43,207],"inference,":[44],"it":[45],"also":[46],"provides":[47,195],"natural":[49],"mechanism":[50],"balancing":[52],"and":[54,84,194],"test-time":[55,200],"compute.":[56],"However,":[57],"still":[60,171],"suffers":[61],"from":[62,79],"training":[63,134,189],"instability":[64,77],"when":[65],"increases.":[71],"Our":[72],"analysis":[73],"reveals":[74],"that":[75,184],"this":[76,157],"stems":[78],"two":[80,90,100],"sources:":[81],"gradient":[82,129],"oscillation":[83],"residual":[85,116],"explosion.":[86],"To":[87],"address":[88],"these":[89],"problems,":[91],"we":[92],"propose":[93],"Fully":[95,104,138,168,185],"Transformer,":[97],"which":[98,107,121],"introduces":[99],"parameter-free":[101],"modifications:":[102],"(1)":[103],"Architecture,":[106],"distributes":[108],"inter-loop":[109],"signals":[110],"across":[111],"all":[112],"layers":[113],"to":[114,127,141,146,178],"mitigate":[115],"explosion;":[117],"(2)":[118],"Attention":[119],"Injection,":[120],"reuses":[122],"existing":[124],"attention":[125],"block":[126],"suppress":[128],"oscillation.":[130],"These":[131],"modifications":[132],"stabilize":[133],"dynamics,":[135],"enabling":[136],"trained":[143],"stably":[144],"up":[145,177],"12":[147],"iterations,":[149],"whereas":[150],"other":[151],"baseline":[152],"looped":[153],"models":[154],"collapse":[155],"in":[156],"regime.":[158],"In":[159],"milder":[160],"settings":[161],"where":[162],"does":[165],"not":[166],"collapse,":[167],"improves":[172,188],"average":[173],"downstream-task":[174],"13.2\\%.":[179],"Overall,":[180],"our":[181],"experiments":[182],"demonstrate":[183],"stability,":[190],"enhances":[191],"downstream":[192],"performance,":[193],"preliminary":[196],"adaptability":[197],"under":[198],"different":[199],"compute":[201],"budgets":[202],"varying":[204],"inference.":[208]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-21T00:00:00"}