{"id":"https://openalex.org/W7162328741","doi":"https://doi.org/10.48550/arxiv.2605.22869","title":"FuRA: Full-Rank Parameter-Efficient Fine-Tuning with Spectral Preconditioning","display_name":"FuRA: Full-Rank Parameter-Efficient Fine-Tuning with Spectral Preconditioning","publication_year":2026,"publication_date":"2026-05-19","ids":{"openalex":"https://openalex.org/W7162328741","doi":"https://doi.org/10.48550/arxiv.2605.22869"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.22869","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.22869","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.2605.22869","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5112593845","display_name":"Yequan Zhao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Yequan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136910999","display_name":"Ruijie Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Ruijie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136991286","display_name":"Liyan Tan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tan, Liyan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136980299","display_name":"Niall Moran","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Moran, Niall","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5082261032","display_name":"Tong Qin","orcid":"https://orcid.org/0000-0002-0994-9816"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qin, Tong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5136982374","display_name":"Zheng Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Zheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":6,"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/T12303","display_name":"Tensor decomposition and applications","score":0.4864000082015991,"subfield":{"id":"https://openalex.org/subfields/2605","display_name":"Computational Mathematics"},"field":{"id":"https://openalex.org/fields/26","display_name":"Mathematics"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T12303","display_name":"Tensor decomposition and applications","score":0.4864000082015991,"subfield":{"id":"https://openalex.org/subfields/2605","display_name":"Computational Mathematics"},"field":{"id":"https://openalex.org/fields/26","display_name":"Mathematics"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.08479999750852585,"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":0.0617000013589859,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/singular-value-decomposition","display_name":"Singular value decomposition","score":0.7444000244140625},{"id":"https://openalex.org/keywords/matrix-decomposition","display_name":"Matrix decomposition","score":0.5856999754905701},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.5511000156402588},{"id":"https://openalex.org/keywords/singular-value","display_name":"Singular value","score":0.5364000201225281},{"id":"https://openalex.org/keywords/block","display_name":"Block (permutation group theory)","score":0.5343999862670898},{"id":"https://openalex.org/keywords/matrix","display_name":"Matrix (chemical analysis)","score":0.4677000045776367},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.4659999907016754},{"id":"https://openalex.org/keywords/core","display_name":"Core (optical fiber)","score":0.4147000014781952},{"id":"https://openalex.org/keywords/basis","display_name":"Basis (linear algebra)","score":0.4081000089645386}],"concepts":[{"id":"https://openalex.org/C22789450","wikidata":"https://www.wikidata.org/wiki/Q420904","display_name":"Singular value decomposition","level":2,"score":0.7444000244140625},{"id":"https://openalex.org/C42355184","wikidata":"https://www.wikidata.org/wiki/Q1361088","display_name":"Matrix decomposition","level":3,"score":0.5856999754905701},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.5511000156402588},{"id":"https://openalex.org/C109282560","wikidata":"https://www.wikidata.org/wiki/Q4166054","display_name":"Singular value","level":3,"score":0.5364000201225281},{"id":"https://openalex.org/C2777210771","wikidata":"https://www.wikidata.org/wiki/Q4927124","display_name":"Block (permutation group theory)","level":2,"score":0.5343999862670898},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5184000134468079},{"id":"https://openalex.org/C106487976","wikidata":"https://www.wikidata.org/wiki/Q685816","display_name":"Matrix (chemical analysis)","level":2,"score":0.4677000045776367},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.4659999907016754},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.45980000495910645},{"id":"https://openalex.org/C2164484","wikidata":"https://www.wikidata.org/wiki/Q5170150","display_name":"Core (optical fiber)","level":2,"score":0.4147000014781952},{"id":"https://openalex.org/C12426560","wikidata":"https://www.wikidata.org/wiki/Q189569","display_name":"Basis (linear algebra)","level":2,"score":0.4081000089645386},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3885999917984009},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.35910001397132874},{"id":"https://openalex.org/C2781170535","wikidata":"https://www.wikidata.org/wiki/Q30587856","display_name":"Noisy data","level":2,"score":0.3427000045776367},{"id":"https://openalex.org/C124681953","wikidata":"https://www.wikidata.org/wiki/Q339062","display_name":"Decomposition","level":2,"score":0.3345000147819519},{"id":"https://openalex.org/C2780551164","wikidata":"https://www.wikidata.org/wiki/Q2306599","display_name":"Column (typography)","level":3,"score":0.3237999975681305},{"id":"https://openalex.org/C152822103","wikidata":"https://www.wikidata.org/wiki/Q7575207","display_name":"Spectral shape analysis","level":3,"score":0.3188000023365021},{"id":"https://openalex.org/C23463724","wikidata":"https://www.wikidata.org/wiki/Q2308831","display_name":"Spectral method","level":2,"score":0.3142000138759613},{"id":"https://openalex.org/C126042441","wikidata":"https://www.wikidata.org/wiki/Q1324888","display_name":"Frame (networking)","level":2,"score":0.3075000047683716},{"id":"https://openalex.org/C187834632","wikidata":"https://www.wikidata.org/wiki/Q188804","display_name":"Factorization","level":2,"score":0.30660000443458557},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.28630000352859497},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.28200000524520874},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.2777000069618225},{"id":"https://openalex.org/C2776291640","wikidata":"https://www.wikidata.org/wiki/Q2912517","display_name":"Value (mathematics)","level":2,"score":0.2696000039577484},{"id":"https://openalex.org/C156778621","wikidata":"https://www.wikidata.org/wiki/Q1365748","display_name":"Spectrum (functional analysis)","level":2,"score":0.26100000739097595},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.25519999861717224}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.22869","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.22869","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.2605.22869","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.22869","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":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Both":[0],"full":[1],"fine-tuning":[2,7,31,166],"(Full":[3],"FT)":[4],"and":[5,57,127,146,150,178],"parameter-efficient":[6],"methods":[8],"such":[9],"as":[10,42],"LoRA":[11],"introduce":[12],"weight":[13,48],"updates":[14,63],"without":[15],"accounting":[16],"for":[17,175,182],"the":[18,43,65,80,109,116,123,128,185],"spectral":[19,40,140],"structure":[20],"established":[21],"during":[22],"pretraining.":[23],"As":[24],"a":[25,70,101],"result,":[26],"noisy":[27],"gradients":[28],"from":[29],"limited":[30],"data":[32],"can":[33],"perturb":[34],"robust":[35],"pretrained":[36,66,117],"features.":[37],"We":[38],"identify":[39],"preconditioning":[41],"missing":[44],"ingredient:":[45],"reparameterizing":[46],"each":[47],"matrix":[49],"through":[50],"its":[51],"full-rank":[52,96,139,143],"singular":[53,60,130],"value":[54],"decomposition":[55],"(SVD)":[56],"freezing":[58],"one":[59],"basis":[61],"constrains":[62],"to":[64,115,154],"column":[67],"space,":[68],"yielding":[69],"preconditioned":[71],"optimization":[72],"scheme":[73],"that":[74],"outperforms":[75,158],"unconstrained":[76],"Full":[77,159],"FT":[78,160],"at":[79,196],"same":[81],"trainable":[82],"parameter":[83],"count.":[84],"Building":[85],"on":[86,100,168],"this":[87],"insight,":[88],"we":[89],"propose":[90],"FuRA":[91,156],"(Full-Rank":[92],"Adaptation),":[93],"an":[94],"efficient":[95],"adaptation":[97],"framework":[98],"based":[99],"block":[102],"tensor-train":[103],"factorization":[104],"W":[105],"=":[106],"LSR,":[107],"where":[108],"large":[110],"core":[111,125],"L":[112],"is":[113,194],"fixed":[114],"block-wise":[118,129],"SVD":[119],"basis,":[120],"while":[121],"only":[122],"compact":[124],"R":[126],"values":[131],"S":[132],"are":[133],"optimized.":[134],"This":[135],"design":[136],"simultaneously":[137],"provides":[138],"preconditioning,":[141],"preserves":[142],"update":[144],"expressivity,":[145],"achieves":[147],"parameter,":[148],"memory,":[149],"step-time":[151],"efficiency":[152],"comparable":[153],"LoRA.":[155],"consistently":[157],"across":[161],"multiple":[162],"settings,":[163],"including":[164],"LLM":[165,172],"(+1.37":[167],"LLaMA-3-8B":[169],"commonsense":[170],"reasoning),":[171],"reinforcement":[173],"learning":[174],"mathematical":[176],"reasoning,":[177],"visual":[179],"instruction":[180],"tuning":[181],"VLMs.":[183],"Furthermore,":[184],"4-bit":[186],"quantized":[187],"variant,":[188],"QFuRA,":[189],"also":[190],"surpasses":[191],"QLoRA.":[192],"Code":[193],"available":[195],"https://github.com/olokevin/FuRA-NIPS":[197]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-05-26T00:00:00"}