{"id":"https://openalex.org/W7134831445","doi":"https://doi.org/10.48550/arxiv.2603.08391","title":"Adaptive Loops and Memory in Transformers: Think Harder or Know More?","display_name":"Adaptive Loops and Memory in Transformers: Think Harder or Know More?","publication_year":2026,"publication_date":"2026-03-09","ids":{"openalex":"https://openalex.org/W7134831445","doi":"https://doi.org/10.48550/arxiv.2603.08391"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.08391","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.08391","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":false,"raw_source_name":null,"raw_type":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2603.08391","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5128663166","display_name":"Markus Frey","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Frey, Markus","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128651565","display_name":"Behzad Shomali","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shomali, Behzad","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5014550117","display_name":"Ali Hamza Bashir","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bashir, Ali Hamza","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128634452","display_name":"David Berghaus","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Berghaus, David","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112027775","display_name":"Joachim Koehler","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Koehler, Joachim","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5011914487","display_name":"Mehdi Ali","orcid":"https://orcid.org/0000-0003-1653-3920"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ali, Mehdi","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5128663166"],"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/T10028","display_name":"Topic Modeling","score":0.19529999792575836,"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":0.19529999792575836,"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/T11273","display_name":"Advanced Graph Neural Networks","score":0.12790000438690186,"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/T12611","display_name":"Neural Networks and Reservoir Computing","score":0.094200000166893,"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/transformer","display_name":"Transformer","score":0.6883999705314636},{"id":"https://openalex.org/keywords/block","display_name":"Block (permutation group theory)","score":0.3910999894142151},{"id":"https://openalex.org/keywords/memory-model","display_name":"Memory model","score":0.38199999928474426},{"id":"https://openalex.org/keywords/commonsense-reasoning","display_name":"Commonsense reasoning","score":0.36480000615119934},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.352400004863739},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.34389999508857727}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7329000234603882},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6883999705314636},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5026999711990356},{"id":"https://openalex.org/C2777210771","wikidata":"https://www.wikidata.org/wiki/Q4927124","display_name":"Block (permutation group theory)","level":2,"score":0.3910999894142151},{"id":"https://openalex.org/C12186640","wikidata":"https://www.wikidata.org/wiki/Q6815743","display_name":"Memory model","level":3,"score":0.38199999928474426},{"id":"https://openalex.org/C193221554","wikidata":"https://www.wikidata.org/wiki/Q5153664","display_name":"Commonsense reasoning","level":2,"score":0.36480000615119934},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.352400004863739},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.34389999508857727},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.34060001373291016},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.32170000672340393},{"id":"https://openalex.org/C30390489","wikidata":"https://www.wikidata.org/wiki/Q4680748","display_name":"Adaptive memory","level":3,"score":0.3122999966144562},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3100000023841858},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2976999878883362},{"id":"https://openalex.org/C82687282","wikidata":"https://www.wikidata.org/wiki/Q66221","display_name":"Auxiliary memory","level":2,"score":0.29100000858306885},{"id":"https://openalex.org/C2779714256","wikidata":"https://www.wikidata.org/wiki/Q25305062","display_name":"Multiple Models","level":2,"score":0.2590000033378601},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.25110000371932983}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.08391","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.08391","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":"doi:10.48550/arxiv.2603.08391","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.08391","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":false,"raw_source_name":null,"raw_type":"article"},"sustainable_development_goals":[{"score":0.5587021112442017,"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Chain-of-thought":[0],"(CoT)":[1],"prompting":[2],"enables":[3],"reasoning":[4],"in":[5],"language":[6],"models":[7,36,43,56],"but":[8],"requires":[9],"explicit":[10],"verbalization":[11],"of":[12,41,127,133],"intermediate":[13],"steps.":[14],"Looped":[15],"transformers":[16],"offer":[17],"an":[18,119],"alternative":[19],"by":[20],"iteratively":[21],"refining":[22],"representations":[23],"within":[24],"hidden":[25,71],"states.":[26],"This":[27],"parameter":[28,106],"efficiency":[29],"comes":[30],"at":[31],"a":[32,74,115],"cost,":[33],"as":[34],"looped":[35],"lack":[37],"the":[38,125],"storage":[39],"capacity":[40],"deeper":[42],"which":[44],"use":[45],"unique":[46],"weights":[47],"per":[48],"layer.":[49],"In":[50],"this":[51],"work,":[52],"we":[53],"investigate":[54],"transformer":[55,65],"that":[57,82,89,117],"feature":[58],"both":[59,112,153],"adaptive":[60],"per-layer":[61],"looping,":[62],"where":[63],"each":[64],"block":[66],"learns":[67],"to":[68,105,142],"iterate":[69],"its":[70],"state":[72],"via":[73],"learned":[75,85],"halting":[76],"mechanism,":[77],"and":[78,107,145],"gated":[79],"memory":[80,96,147],"banks,":[81],"provide":[83],"additional":[84],"storage.":[86],"We":[87],"find":[88],"looping":[90],"primarily":[91],"benefits":[92],"mathematical":[93],"reasoning,":[94],"while":[95,149],"banks":[97],"help":[98],"recover":[99],"performance":[100],"on":[101],"commonsense":[102],"tasks":[103],"compared":[104],"FLOP":[108],"matched":[109],"models.":[110],"Combining":[111],"mechanisms":[113],"yields":[114],"model":[116,134],"outperforms":[118],"iso-FLOP":[120],"baseline,":[121],"with":[122],"three":[123],"times":[124],"number":[126],"layers,":[128],"across":[129],"math":[130],"benchmarks.":[131],"Analysis":[132],"internals":[135],"reveals":[136],"layer":[137],"specialization:":[138],"early":[139],"layers":[140,151],"learn":[141],"loop":[143],"minimally":[144],"access":[146],"sparingly,":[148],"later":[150],"do":[152],"more":[154],"heavily.":[155]},"counts_by_year":[],"updated_date":"2026-03-13T14:20:09.374765","created_date":"2026-03-11T00:00:00"}