{"id":"https://openalex.org/W7224133648","doi":"https://doi.org/10.48550/arxiv.2610.36113","title":"Scaling and Renormalization in Neural Networks: An Effective Renormalization-Group Description of Grokking","display_name":"Scaling and Renormalization in Neural Networks: An Effective Renormalization-Group Description of Grokking","publication_year":2026,"publication_date":"2026-10-26","ids":{"openalex":"https://openalex.org/W7224133648","doi":"https://doi.org/10.48550/arxiv.2610.36113"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2610.36113","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.36113","pdf_url":"https://arxiv.org/pdf/2610.36113","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2610.36113","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5160050291","display_name":"Min Chen","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I170201317","display_name":"University of Pittsburgh","ror":"https://ror.org/01an3r305","country_code":"US","type":"education","lineage":["https://openalex.org/I170201317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Chen, Min","raw_affiliation_strings":["Department of Computer Science , University of Pittsburgh , Pittsburgh , PA , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science , University of Pittsburgh , Pittsburgh , PA , USA","institution_ids":["https://openalex.org/I170201317"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5160025489","display_name":"Junqi Wang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I170201317","display_name":"University of Pittsburgh","ror":"https://ror.org/01an3r305","country_code":"US","type":"education","lineage":["https://openalex.org/I170201317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Wang, Junqi","raw_affiliation_strings":["Department of Computer Science , University of Pittsburgh , Pittsburgh , PA , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science , University of Pittsburgh , Pittsburgh , PA , USA","institution_ids":["https://openalex.org/I170201317"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5002993177","display_name":"Zeguan Wu","orcid":"https://orcid.org/0000-0002-5695-7579","observed_orcids":["https://orcid.org/0000-0002-5695-7579"]},"institutions":[{"id":"https://openalex.org/I170201317","display_name":"University of Pittsburgh","ror":"https://ror.org/01an3r305","country_code":"US","type":"education","lineage":["https://openalex.org/I170201317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Wu, Zeguan","raw_affiliation_strings":["Department of Computer Science , University of Pittsburgh , Pittsburgh , PA , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science , University of Pittsburgh , Pittsburgh , PA , USA","institution_ids":["https://openalex.org/I170201317"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5160060754","display_name":"Wenhao Huang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I170201317","display_name":"University of Pittsburgh","ror":"https://ror.org/01an3r305","country_code":"US","type":"education","lineage":["https://openalex.org/I170201317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Huang, Wenhao","raw_affiliation_strings":["Department of Computer Science , University of Pittsburgh , Pittsburgh , PA , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science , University of Pittsburgh , Pittsburgh , PA , USA","institution_ids":["https://openalex.org/I170201317"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5160017473","display_name":"Junyu Liu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I170201317","display_name":"University of Pittsburgh","ror":"https://ror.org/01an3r305","country_code":"US","type":"education","lineage":["https://openalex.org/I170201317"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Liu, Junyu","raw_affiliation_strings":["Department of Computer Science , University of Pittsburgh , Pittsburgh , PA , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Science , University of Pittsburgh , Pittsburgh , PA , USA","institution_ids":["https://openalex.org/I170201317"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I170201317"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"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/T10320","display_name":"Neural Networks and Applications","score":0.7426999807357788,"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/T10320","display_name":"Neural Networks and Applications","score":0.7426999807357788,"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/T10591","display_name":"Theoretical and Computational Physics","score":0.2013999968767166,"subfield":{"id":"https://openalex.org/subfields/3104","display_name":"Condensed Matter 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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.008299999870359898,"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/grokking","display_name":"grokking","score":0.996999979019165},{"id":"https://openalex.org/keywords/renormalization-group","display_name":"renormalization group","score":0.8450000286102295},{"id":"https://openalex.org/keywords/real-space-renormalization","display_name":"real-space renormalization","score":0.8109999895095825},{"id":"https://openalex.org/keywords/scaling-exponents","display_name":"scaling exponents","score":0.7480000257492065},{"id":"https://openalex.org/keywords/critical-exponents","display_name":"critical exponents","score":0.9810000061988831},{"id":"https://openalex.org/keywords/fiedler-vector","display_name":"Fiedler vector","score":0.8130000233650208},{"id":"https://openalex.org/keywords/coupling-matrix","display_name":"coupling matrix","score":0.6439999938011169},{"id":"https://openalex.org/keywords/weight-decay","display_name":"weight decay","score":0.75},{"id":"https://openalex.org/keywords/modular-addition","display_name":"modular addition","score":0.5600000023841858},{"id":"https://openalex.org/keywords/two-layer-neural-networks","display_name":"two-layer neural networks","score":0.7990000247955322},{"id":"https://openalex.org/keywords/neural-network-dynamics","display_name":"neural network dynamics","score":0.43799999356269836}],"concepts":[{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.7542999982833862},{"id":"https://openalex.org/C166124518","wikidata":"https://www.wikidata.org/wiki/Q1047702","display_name":"Renormalization","level":2,"score":0.578499972820282},{"id":"https://openalex.org/C121864883","wikidata":"https://www.wikidata.org/wiki/Q677916","display_name":"Statistical physics","level":1,"score":0.5683000087738037},{"id":"https://openalex.org/C68532491","wikidata":"https://www.wikidata.org/wiki/Q1203669","display_name":"Renormalization group","level":2,"score":0.5616000294685364},{"id":"https://openalex.org/C72169020","wikidata":"https://www.wikidata.org/wiki/Q194404","display_name":"Monotonic function","level":2,"score":0.5142999887466431},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.5042999982833862},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.4645000100135803},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.423799991607666},{"id":"https://openalex.org/C131584629","wikidata":"https://www.wikidata.org/wiki/Q4308705","display_name":"Coupling (piping)","level":2,"score":0.3741999864578247},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.34450000524520874},{"id":"https://openalex.org/C106487976","wikidata":"https://www.wikidata.org/wiki/Q685816","display_name":"Matrix (chemical analysis)","level":2,"score":0.3287999927997589},{"id":"https://openalex.org/C207740291","wikidata":"https://www.wikidata.org/wiki/Q5508844","display_name":"Functional renormalization group","level":3,"score":0.3001999855041504},{"id":"https://openalex.org/C64812099","wikidata":"https://www.wikidata.org/wiki/Q176604","display_name":"Random matrix","level":3,"score":0.2851000130176544},{"id":"https://openalex.org/C186633575","wikidata":"https://www.wikidata.org/wiki/Q845060","display_name":"Maxima and minima","level":2,"score":0.2770000100135803},{"id":"https://openalex.org/C202213908","wikidata":"https://www.wikidata.org/wiki/Q626011","display_name":"Mean field theory","level":2,"score":0.2768000066280365},{"id":"https://openalex.org/C135593079","wikidata":"https://www.wikidata.org/wiki/Q1750766","display_name":"Scale invariance","level":2,"score":0.26910001039505005},{"id":"https://openalex.org/C164154869","wikidata":"https://www.wikidata.org/wiki/Q901927","display_name":"Critical exponent","level":3,"score":0.26660001277923584},{"id":"https://openalex.org/C2986219828","wikidata":"https://www.wikidata.org/wiki/Q1193095","display_name":"Coupling strength","level":2,"score":0.2653000056743622},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.2535000145435333}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2610.36113","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.36113","pdf_url":"https://arxiv.org/pdf/2610.36113","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2610.36113","doi":"https://doi.org/10.48550/arxiv.2610.36113","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2610.36113","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2610.36113","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.36113","pdf_url":"https://arxiv.org/pdf/2610.36113","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7224133648.pdf","grobid_xml":"https://content.openalex.org/works/W7224133648.grobid-xml"},"referenced_works_count":11,"referenced_works":["https://openalex.org/W1890000358","https://openalex.org/W4226434736","https://openalex.org/W4283026156","https://openalex.org/W4315588185","https://openalex.org/W4386559757","https://openalex.org/W4401202627","https://openalex.org/W4403963952","https://openalex.org/W4416039616","https://openalex.org/W7133219754","https://openalex.org/W7141734836","https://openalex.org/W7151550933"],"related_works":["https://openalex.org/W7151550933","https://openalex.org/W7212269809","https://openalex.org/W4417094362","https://openalex.org/W7141173969","https://openalex.org/W7166698946","https://openalex.org/W4310627449","https://openalex.org/W4402502926","https://openalex.org/W7102704356","https://openalex.org/W7131811383","https://openalex.org/W7134256001"],"abstract_inverted_index":{"Grokking":[0],"is":[1,126],"a":[2,28,40,93,118,177,191],"delayed":[3],"transition":[4],"from":[5],"memorization":[6],"to":[7,83,101,109,193],"generalization":[8,34],"in":[9],"deep":[10],"neural":[11,33,81],"networks":[12],"(NNs).":[13],"We":[14,131],"examine":[15],"whether":[16],"scaling":[17,157,185],"and":[18,56,63,69,75,104,141,159,186],"renormalization":[19,187],"group":[20,188],"(RG)":[21],"theories":[22],"can":[23],"describe":[24],"this":[25,151],"dynamics,":[26],"establishing":[27],"quantitative":[29],"physical":[30],"framework":[31,175],"for":[32,59],"transitions.":[35],"For":[36],"modular":[37],"addition":[38],"on":[39],"two-layer":[41],"NN,":[42],"an":[43,49,88,111],"ensemble":[44],"of":[45,96,122,181],"trained":[46],"models":[47],"defines":[48],"order":[50],"parameter":[51],"(the":[52],"squared":[53],"weight":[54,70],"norm)":[55],"fluctuation":[57],"proxies":[58],"the":[60,64,97,143,184],"heat":[61],"capacity":[62],"susceptibility,":[65],"with":[66,99,164],"training":[67],"epoch":[68],"decay":[71,121],"serving":[72],"as":[73,190],"thermal":[74],"field":[76],"control":[77],"coordinates.":[78],"To":[79],"bridge":[80],"architectures":[82],"real-space":[84,134],"RG,":[85],"we":[86],"construct":[87],"exact":[89],"coupling":[90,123,144],"matrix":[91,145],"via":[92],"second-order":[94],"expansion":[95],"loss":[98],"respect":[100],"per-neuron":[102],"norms":[103],"use":[105],"its":[106],"Fiedler":[107],"vector":[108],"define":[110],"effective":[112,179],"geometry.":[113],"This":[114],"spectral":[115],"ordering":[116],"yields":[117,155],"monotonic":[119],"spatial":[120],"strength":[124],"that":[125],"absent":[127],"under":[128],"random":[129],"permutations.":[130],"then":[132],"execute":[133],"RG":[135,153,169],"by":[136],"block-spin":[137],"aggregating":[138],"adjacent":[139],"neurons":[140],"parameterizing":[142],"into":[146],"off-diagonal":[147],"band":[148],"strengths.":[149],"Linearizing":[150],"local":[152],"map":[154],"relevant":[156],"exponents":[158],"five":[160],"derived":[161],"critical":[162],"exponents,":[163],"direct":[165],"empirical":[166],"estimates":[167],"matching":[168],"predictions":[170],"within":[171],"1-10%.":[172],"Overall,":[173],"our":[174],"provides":[176],"consistent":[178],"description":[180],"grokking,":[182],"suggesting":[183],"formalism":[189],"tool":[192],"investigate":[194],"NN":[195],"dynamics.":[196]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-10-29T00:00:00"}