{"id":"https://openalex.org/W7127350043","doi":"https://doi.org/10.48550/arxiv.2602.01997","title":"On the Limits of Layer Pruning for Generative Reasoning in Large Language Models","display_name":"On the Limits of Layer Pruning for Generative Reasoning in Large Language Models","publication_year":2026,"publication_date":"2026-02-02","ids":{"openalex":"https://openalex.org/W7127350043","doi":"https://doi.org/10.48550/arxiv.2602.01997"},"language":null,"primary_location":{"id":"pmh:doi:10.48550/arxiv.2602.01997","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"type":"article","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":null,"any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5116269531","display_name":"Safal Shrestha","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shrestha, Safal","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5052181179","display_name":"Anima Shrestha","orcid":"https://orcid.org/0000-0002-9919-476X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shrestha, Anubhav","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5115863336","display_name":"Aadim Nepal","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Nepal, Aadim","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5109709876","display_name":"Minwu Kim","orcid":"https://orcid.org/0009-0005-8104-2142"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kim, Minwu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Ross, Keith","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ross, Keith","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":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.13459383,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"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":0.2689000070095062,"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":0.2689000070095062,"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/T11948","display_name":"Machine Learning in Materials Science","score":0.10520000010728836,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.10509999841451645,"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/pruning","display_name":"Pruning","score":0.8219000101089478},{"id":"https://openalex.org/keywords/generative-grammar","display_name":"Generative grammar","score":0.8116999864578247},{"id":"https://openalex.org/keywords/generative-model","display_name":"Generative model","score":0.6459000110626221},{"id":"https://openalex.org/keywords/layer","display_name":"Layer (electronics)","score":0.565500020980835},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.47589999437332153},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.4406000077724457},{"id":"https://openalex.org/keywords/simple","display_name":"Simple (philosophy)","score":0.4077000021934509}],"concepts":[{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.8219000101089478},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.8116999864578247},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6638000011444092},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.6459000110626221},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6060000061988831},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5968999862670898},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.565500020980835},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.47589999437332153},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.4406000077724457},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.4077000021934509},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.39800000190734863},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.36739999055862427},{"id":"https://openalex.org/C2776542307","wikidata":"https://www.wikidata.org/wiki/Q16912820","display_name":"Parenthesis","level":2,"score":0.3237999975681305},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.2935999929904938},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.28369998931884766},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.27950000762939453},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.27149999141693115}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.48550/arxiv.2602.01997","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},{"id":"doi:10.48550/arxiv.2602.01997","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.01997","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":"pmh:doi:10.48550/arxiv.2602.01997","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","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":{"Recent":[0],"work":[1],"has":[2],"shown":[3],"that":[4,43,125],"layer":[5,156],"pruning":[6,48,116,157],"can":[7],"effectively":[8],"compress":[9],"large":[10],"language":[11],"models":[12,110],"(LLMs)":[13],"while":[14],"retaining":[15],"strong":[16],"performance":[17,96],"on":[18,82,97,112,137,164],"classification":[19,98],"benchmarks,":[20],"often":[21],"with":[22,85],"little":[23],"or":[24,73],"no":[25],"finetuning.":[26],"In":[27],"contrast,":[28],"generative":[29,103,159],"reasoning":[30,104,122,160],"tasks,":[31,99],"such":[32,126,140],"as":[33,130,141],"GSM8K":[34],"and":[35,60,161],"HumanEval\\textsuperscript{+},":[36],"exhibit":[37],"substantially":[38],"weaker":[39],"recovery.":[40],"We":[41],"show":[42],"beyond":[44],"surface-level":[45],"text":[46],"degradation,":[47],"leads":[49],"to":[50,70,92,118],"a":[51,77],"loss":[52],"of":[53,94,155],"key":[54],"algorithmic":[55],"capabilities,":[56],"including":[57],"arithmetic":[58],"computation":[59],"balanced":[61],"parenthesis":[62],"generation.":[63,148],"Under":[64],"realistic":[65],"post-training":[66,172],"constraints,":[67],"without":[68],"access":[69],"pretraining-scale":[71],"data":[72],"compute,":[74],"we":[75,150],"evaluate":[76],"minimal":[78],"recovery":[79,101],"strategy":[80],"based":[81],"supervised":[83],"finetuning":[84],"self-generated":[86],"responses.":[87],"This":[88,133],"approach":[89],"recovers":[90],"up":[91],"90\\%":[93],"baseline":[95],"but":[100],"for":[102,158],"remains":[105],"fundamentally":[106],"limited.":[107],"Notably,":[108],"even":[109,136],"finetuned":[111],"$\\sim$400B":[113],"tokens":[114],"after":[115],"fail":[117],"recover":[119],"their":[120],"original":[121],"performance,":[123],"suggesting":[124],"capabilities":[127],"are":[128],"not":[129,145],"easily":[131],"restored.":[132],"limitation":[134],"persists":[135],"simple":[138],"tasks":[139],"arithmetic,":[142],"which":[143],"do":[144],"require":[146],"multi-step":[147],"Overall,":[149],"characterize":[151],"the":[152],"practical":[153],"limits":[154],"provide":[162],"guidance":[163],"when":[165],"depth":[166],"reduction":[167],"is":[168],"effective":[169],"under":[170],"constrained":[171],"regimes.":[173]},"counts_by_year":[],"updated_date":"2026-07-15T18:14:33.161393","created_date":"2026-02-04T00:00:00"}