{"id":"https://openalex.org/W7158100516","doi":"https://doi.org/10.48550/arxiv.2604.25098","title":"Revisiting the Effectiveness of LLM Pruning for Test-Time Scaling","display_name":"Revisiting the Effectiveness of LLM Pruning for Test-Time Scaling","publication_year":2026,"publication_date":"2026-04-28","ids":{"openalex":"https://openalex.org/W7158100516","doi":"https://doi.org/10.48550/arxiv.2604.25098"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.25098","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.25098","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":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.25098","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5014340264","display_name":"Ocean Monjur","orcid":"https://orcid.org/0000-0003-1526-6808"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Monjur, Ocean","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120721021","display_name":"Shahriar Kabir Nahin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Nahin, Shahriar Kabir","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5134855431","display_name":"Anshuman Chhabra","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chhabra, Anshuman","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/T10028","display_name":"Topic Modeling","score":0.5256999731063843,"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":0.5256999731063843,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.32760000228881836,"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/T13629","display_name":"Text Readability and Simplification","score":0.01679999940097332,"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":0.8902000188827515},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6115999817848206},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.5730000138282776},{"id":"https://openalex.org/keywords/scaling","display_name":"Scaling","score":0.5246999859809875},{"id":"https://openalex.org/keywords/parametric-statistics","display_name":"Parametric statistics","score":0.5131000280380249}],"concepts":[{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.8902000188827515},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6590999960899353},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6115999817848206},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.5730000138282776},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5396999716758728},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5284000039100647},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.5246999859809875},{"id":"https://openalex.org/C117251300","wikidata":"https://www.wikidata.org/wiki/Q1849855","display_name":"Parametric statistics","level":2,"score":0.5131000280380249},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.44830000400543213},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.2992999851703644},{"id":"https://openalex.org/C24574437","wikidata":"https://www.wikidata.org/wiki/Q7135228","display_name":"Parametric model","level":3,"score":0.25290000438690186},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2513999938964844}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.25098","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.25098","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":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.25098","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.25098","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":"Preprint"},"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":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"now":[4],"exhibit":[5],"remarkable":[6],"reasoning":[7,57,77,109,113],"capabilities":[8],"through":[9],"test-time":[10],"compute":[11],"scaling":[12],"(TTS),":[13],"with":[14],"impressive":[15],"performance":[16,125,175],"across":[17,107],"math":[18],"and":[19,87,116,130,176],"coding":[20],"benchmarks.":[21],"In":[22],"parallel,":[23],"research":[24,47],"in":[25,80,177],"model":[26],"compression":[27],"has":[28,61],"developed":[29],"pruning":[30,65,91,122,171,183],"methods":[31],"that":[32,63,93,120,170,180],"seek":[33],"to":[34,56,127],"remove":[35,68,95],"redundant/detrimental":[36,98],"parameters":[37],"without":[38],"sacrificing":[39],"task":[40],"performance.":[41,78],"The":[42],"intersection":[43],"of":[44,71,147],"these":[45,161],"two":[46,112],"advancements":[48],"lays":[49],"the":[50,136,145,167],"foundation":[51],"for":[52,159],"our":[53,104],"work.":[54],"Specific":[55],"LLMs,":[58],"prior":[59],"work":[60],"shown":[62],"structured":[64,128],"(methods":[66,92],"which":[67,153],"entire":[69],"set":[70],"layer":[72],"blocks),":[73],"significantly":[74],"degrades":[75],"TTS":[76,124,174,186],"However,":[79],"this":[81,85],"work,":[82],"we":[83,141],"revisit":[84],"assumption":[86],"investigate":[88],"whether":[89],"unstructured":[90,121,162],"carefully":[94,181],"only":[96],"certain":[97],"weights)":[99],"exhibits":[100],"similar":[101],"limitations.":[102],"Surprisingly,":[103],"extensive":[105],"experiments":[106],"four":[108],"benchmarks":[110],"on":[111],"LLMs:":[114],"s1.1-7B":[115],"Qwen3-8B,":[117],"consistently":[118],"show":[119],"augments":[123],"compared":[126],"pruning,":[129],"at":[131],"times":[132],"can":[133,184],"even":[134],"outperform":[135],"unpruned":[137],"full-weight":[138],"LLMs.":[139],"Furthermore,":[140],"also":[142],"empirically":[143],"study":[144],"impact":[146],"different":[148],"layer-wise":[149],"sparsity":[150],"allocation":[151],"strategies,":[152],"are":[154],"an":[155],"important":[156],"parametric":[157],"choice":[158],"instantiating":[160],"methods.":[163],"These":[164],"findings":[165],"challenge":[166],"conventional":[168],"notion":[169],"always":[172],"reduces":[173],"fact,":[178],"suggest":[179],"undertaken":[182],"retain":[185],"effectiveness.":[187]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-04-30T00:00:00"}