{"id":"https://openalex.org/W4412888161","doi":"https://doi.org/10.18653/v1/2025.findings-acl.830","title":"BOSE: A Systematic Evaluation Method Optimized for Base Models","display_name":"BOSE: A Systematic Evaluation Method Optimized for Base Models","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412888161","doi":"https://doi.org/10.18653/v1/2025.findings-acl.830"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.830","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.830","pdf_url":"https://aclanthology.org/2025.findings-acl.830.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-acl.830.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5069329865","display_name":"Hongzhi Luan","orcid":"https://orcid.org/0009-0002-4168-4640"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hongzhi Luan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5088471665","display_name":"Changxin Tian","orcid":"https://orcid.org/0000-0002-3013-9439"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Changxin Tian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5001638913","display_name":"Zhaoxin Huan","orcid":"https://orcid.org/0000-0002-3611-0901"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhaoxin Huan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100746437","display_name":"Xiaolu Zhang","orcid":"https://orcid.org/0000-0002-4044-8550"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiaolu Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5055381638","display_name":"Kunlong Chen","orcid":"https://orcid.org/0000-0002-8828-9595"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kunlong Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5063672467","display_name":"Zhiqiang Zhang","orcid":"https://orcid.org/0000-0002-0083-7455"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhiqiang Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5054711269","display_name":"Jun Zhou","orcid":"https://orcid.org/0000-0003-0346-1390"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jun Zhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":7,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.17486556,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"16147","last_page":"16158"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11450","display_name":"Model-Driven Software Engineering Techniques","score":0.07360000163316727,"subfield":{"id":"https://openalex.org/subfields/1712","display_name":"Software"},"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/T11450","display_name":"Model-Driven Software Engineering Techniques","score":0.07360000163316727,"subfield":{"id":"https://openalex.org/subfields/1712","display_name":"Software"},"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/computer-science","display_name":"Computer science","score":0.6222289204597473},{"id":"https://openalex.org/keywords/base","display_name":"Base (topology)","score":0.5347703099250793},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.13795846700668335}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6222289204597473},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.5347703099250793},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.13795846700668335},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.830","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.830","pdf_url":"https://aclanthology.org/2025.findings-acl.830.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-acl.830","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.830","pdf_url":"https://aclanthology.org/2025.findings-acl.830.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412888161.pdf","grobid_xml":"https://content.openalex.org/works/W4412888161.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"This":[0],"paper":[1],"poses":[2,80],"two":[3,124],"critical":[4],"issues":[5],"in":[6,17],"evaluating":[7],"base":[8,63,67,87,120,192],"models":[9,24,68],"(without":[10],"post-training):":[11],"(1)":[12],"Unstable":[13],"evaluation":[14,36,72,91,118,156,171],"during":[15,186],"training:":[16],"the":[18,23,26,48,117,144,160,170,182,189,202],"early":[19],"stages":[20],"of":[21,119,184],"pre-training,":[22],"lack":[25],"capability":[27],"to":[28,34,42,46,75,95,115,153,162,167],"answer":[29],"questions":[30],"as":[31,55],"required,":[32],"leading":[33],"unstable":[35],"results.This":[37],"instability":[38],"makes":[39],"it":[40],"difficult":[41],"provide":[43],"solid":[44],"conclusions":[45],"guide":[47],"training,":[49],"especially":[50],"for":[51,83,131,136,201],"key":[52,125],"experiments":[53],"such":[54],"data":[56],"ablation":[57],"and":[58,64,134,173,188,193],"scaling":[59],"law.(2)":[60],"Inconsistency":[61],"between":[62,191],"instruct":[65,77,98,194],"models:":[66],"generally":[69],"exhibit":[70],"poorer":[71],"performance":[73],"compared":[74],"corresponding":[76],"models.This":[78],"gap":[79],"a":[81,86,96,111,150],"challenge":[82],"assessing":[84],"whether":[85],"model":[88,106],"with":[89,139],"better":[90,97],"can":[92],"truly":[93],"lead":[94],"model.To":[99],"address":[100],"these":[101],"issues,":[102],"we":[103,158],"propose":[104,163],"Base":[105],"Oriented":[107],"Systematic":[108],"Evaluation":[109],"(BOSE),":[110],"method":[112],"specifically":[113],"designed":[114],"optimize":[116],"models.Specifically,":[121],"BOSE":[122,178],"introduces":[123],"innovations:":[126],"In-Context":[127],"Light-instruction":[128],"Prompt":[129],"(ICLiP)":[130],"open-ended":[132],"tasks":[133,138],"Blank-ppl":[135],"multichoice":[137],"candidate":[140],"options,":[141],"which":[142],"transforms":[143],"standard":[145],"perplexity":[146],"(ppl)":[147],"metric":[148],"into":[149],"fill-in-the-blank":[151],"format":[152],"mitigate":[154],"early-stage":[155],"fluctuations.Furthermore,":[157],"are":[159],"first":[161],"Kendall's":[164],"rank":[165],"correlation":[166],"quantitatively":[168],"measure":[169],"stability":[172,183],"consistency.Experimental":[174],"results":[175],"demonstrate":[176],"that":[177],"significantly":[179],"enhances":[180],"both":[181],"evaluations":[185],"pre-training":[187],"consistency":[190],"models,":[195],"thereby":[196],"providing":[197],"more":[198],"reliable":[199],"guidance":[200],"LLMs'":[203],"training.":[204]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}