{"id":"https://openalex.org/W7165944925","doi":"https://doi.org/10.48550/arxiv.2606.24970","title":"Don't Go Breaking My LLM: The Impact of Pruning Attention Layers on Explanation Faithfulness and Confidence Calibration","display_name":"Don't Go Breaking My LLM: The Impact of Pruning Attention Layers on Explanation Faithfulness and Confidence Calibration","publication_year":2026,"publication_date":"2026-06-23","ids":{"openalex":"https://openalex.org/W7165944925","doi":"https://doi.org/10.48550/arxiv.2606.24970"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.24970","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.24970","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,"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":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.2606.24970","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5117865140","display_name":"Pietro Tropeano","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tropeano, Pietro","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139371972","display_name":"Maria Maistro","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Maistro, Maria","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5064247235","display_name":"Tuukka Ruotsalo","orcid":"https://orcid.org/0000-0002-2203-4928"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ruotsalo, Tuukka","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5045425016","display_name":"Christina Lioma","orcid":"https://orcid.org/0000-0003-2600-2701"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lioma, Christina","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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.8309999704360962,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.8309999704360962,"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/T11636","display_name":"Artificial Intelligence in Healthcare and Education","score":0.03020000085234642,"subfield":{"id":"https://openalex.org/subfields/2718","display_name":"Health Informatics"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}},{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.02019999921321869,"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/interpretability","display_name":"Interpretability","score":0.8465999960899353},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.8295000195503235},{"id":"https://openalex.org/keywords/calibration","display_name":"Calibration","score":0.7167999744415283},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5863000154495239},{"id":"https://openalex.org/keywords/reliability","display_name":"Reliability (semiconductor)","score":0.3481000065803528},{"id":"https://openalex.org/keywords/confidence-interval","display_name":"Confidence interval","score":0.3375999927520752}],"concepts":[{"id":"https://openalex.org/C2781067378","wikidata":"https://www.wikidata.org/wiki/Q17027399","display_name":"Interpretability","level":2,"score":0.8465999960899353},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.8295000195503235},{"id":"https://openalex.org/C165838908","wikidata":"https://www.wikidata.org/wiki/Q736777","display_name":"Calibration","level":2,"score":0.7167999744415283},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6035000085830688},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5863000154495239},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4968999922275543},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.49410000443458557},{"id":"https://openalex.org/C43214815","wikidata":"https://www.wikidata.org/wiki/Q7310987","display_name":"Reliability (semiconductor)","level":3,"score":0.3481000065803528},{"id":"https://openalex.org/C44249647","wikidata":"https://www.wikidata.org/wiki/Q208498","display_name":"Confidence interval","level":2,"score":0.3375999927520752},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.33149999380111694},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.32120001316070557},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.2872999906539917},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.26759999990463257},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.25690001249313354},{"id":"https://openalex.org/C2776035688","wikidata":"https://www.wikidata.org/wiki/Q1606558","display_name":"Affect (linguistics)","level":2,"score":0.25529998540878296},{"id":"https://openalex.org/C150921843","wikidata":"https://www.wikidata.org/wiki/Q1170431","display_name":"Resampling","level":2,"score":0.25279998779296875}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.24970","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.24970","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,"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":"doi:10.48550/arxiv.2606.24970","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.24970","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,"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":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Pruning":[0],"Large":[1],"Language":[2],"Models":[3],"(LLMs)":[4],"reduces":[5],"memory":[6],"and":[7,70,88,94,110,116,133,145,153,161],"inference":[8],"costs":[9],"by":[10,151],"removing":[11],"parts":[12,32],"of":[13,22,33,49,62],"the":[14,29,60,98],"network,":[15],"producing":[16],"smaller":[17],"models":[18,100],"that":[19,45,107,138],"retain":[20],"most":[21,30],"their":[23,108],"accuracy.":[24,134],"As":[25],"attention":[26,50,63,83],"layers":[27,51,84],"are":[28],"resource-intensive":[31],"LLMs,":[34],"pruning":[35,64,82,140],"them":[36],"is":[37],"a":[38,127],"promising":[39],"compression":[40],"strategy.":[41],"Prior":[42],"work":[43],"shows":[44],"up":[46],"to":[47],"33%":[48],"can":[52,118,141],"be":[53],"pruned":[54,99,166],"with":[55],"minimal":[56],"accuracy":[57,123,152],"loss.":[58],"Nevertheless,":[59],"impact":[61],"on":[65],"model":[66,130],"interpretability,":[67,132],"specifically":[68],"faithfulness":[69,87,109,115],"confidence":[71,89],"calibration,":[72],"remains":[73,124],"unstudied.":[74],"To":[75],"address":[76],"this":[77],"gap,":[78],"we":[79,105],"study":[80],"how":[81],"affects":[85],"explanation":[86],"calibration":[90,111,117,162],"across":[91],"five":[92],"LLMs":[93],"eight":[95],"datasets.":[96],"While":[97],"often":[101,112],"maintain":[102],"high":[103],"accuracy,":[104],"find":[106],"degrade.":[113],"Notably,":[114],"fluctuate":[119],"significantly,":[120],"even":[121],"when":[122,164],"stable,":[125],"highlighting":[126],"misalignment":[128],"between":[129],"confidence,":[131],"Our":[135],"findings":[136],"suggest":[137],"layer":[139],"affect":[142],"LLMs'":[143],"interpretability":[144],"reliability":[146],"in":[147],"ways":[148],"not":[149],"captured":[150],"efficiency":[154],"measures":[155],"alone.":[156],"We":[157],"recommend":[158],"including":[159],"explainability":[160],"metrics":[163],"evaluating":[165],"models.":[167]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-06-26T00:00:00"}