{"id":"https://openalex.org/W4403586357","doi":"https://doi.org/10.48550/arxiv.2409.04168","title":"From Calculation to Adjudication: Examining LLM judges on Mathematical Reasoning Tasks","display_name":"From Calculation to Adjudication: Examining LLM judges on Mathematical Reasoning Tasks","publication_year":2024,"publication_date":"2024-09-06","ids":{"openalex":"https://openalex.org/W4403586357","doi":"https://doi.org/10.48550/arxiv.2409.04168"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2409.04168","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2409.04168","pdf_url":"https://arxiv.org/pdf/2409.04168","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2409.04168","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5058746097","display_name":"Andreas Stephan","orcid":"https://orcid.org/0000-0001-5776-9396","observed_orcids":["https://orcid.org/0000-0001-5776-9396"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Stephan, Andreas","raw_affiliation_strings":["Faculty of Computer Science ,","UniVie Doctoral School Computer Science ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Faculty of Computer Science ,","institution_ids":[]},{"raw_affiliation_string":"UniVie Doctoral School Computer Science ,","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100567405","display_name":"Dawei Zhu","orcid":"https://orcid.org/0000-0003-1534-8854","observed_orcids":["https://orcid.org/0000-0003-1534-8854"]},"institutions":[{"id":"https://openalex.org/I91712215","display_name":"Saarland University","ror":"https://ror.org/01jdpyv68","country_code":"DE","type":"education","lineage":["https://openalex.org/I91712215"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Zhu, Dawei","raw_affiliation_strings":["Saarland University , Saarland Informatics Campus ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Saarland University , Saarland Informatics Campus ,","institution_ids":["https://openalex.org/I91712215"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5069469652","display_name":"Matthias A\u00dfenmacher","orcid":"https://orcid.org/0000-0003-2154-5774","observed_orcids":["https://orcid.org/0000-0003-2154-5774"]},"institutions":[{"id":"https://openalex.org/I4403386549","display_name":"Munich Center for Machine Learning","ror":"https://ror.org/02nfy3535","country_code":"DE","type":"education","lineage":["https://openalex.org/I4403386549","https://openalex.org/I62916508","https://openalex.org/I8204097"]},{"id":"https://openalex.org/I8204097","display_name":"Ludwig-Maximilians-Universit\u00e4t M\u00fcnchen","ror":"https://ror.org/05591te55","country_code":"DE","type":"education","lineage":["https://openalex.org/I8204097"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"A\u00dfenmacher, Matthias","raw_affiliation_strings":["Department of Statistics , LMU Munich ,","Munich Center for Machine Learning (MCML)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Statistics , LMU Munich ,","institution_ids":["https://openalex.org/I8204097"]},{"raw_affiliation_string":"Munich Center for Machine Learning (MCML)","institution_ids":["https://openalex.org/I4403386549"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103000631","display_name":"Xiaoyu Shen","orcid":"https://orcid.org/0000-0002-0217-2469","observed_orcids":["https://orcid.org/0000-0002-0217-2469"]},"institutions":[{"id":"https://openalex.org/I4394709157","display_name":"Eastern Institute of Technology, Ningbo","ror":"https://ror.org/036mbz113","country_code":"CN","type":"education","lineage":["https://openalex.org/I4394709157"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shen, Xiaoyu","raw_affiliation_strings":["Eastern Institute of Technology , Ningbo"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Eastern Institute of Technology , Ningbo","institution_ids":["https://openalex.org/I4394709157"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5046895021","display_name":"Benjamin Roth","orcid":"https://orcid.org/0000-0003-0362-0267","observed_orcids":["https://orcid.org/0000-0003-0362-0267"]},"institutions":[{"id":"https://openalex.org/I129774422","display_name":"University of Vienna","ror":"https://ror.org/03prydq77","country_code":"AT","type":"education","lineage":["https://openalex.org/I129774422"]}],"countries":["AT"],"is_corresponding":false,"raw_author_name":"Roth, Benjamin","raw_affiliation_strings":["Faculty of Computer Science ,","Faculty of Philological and Cultural Studies , University of Vienna , Vienna , Austria"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Faculty of Computer Science ,","institution_ids":[]},{"raw_affiliation_string":"Faculty of Philological and Cultural Studies , University of Vienna , Vienna , Austria","institution_ids":["https://openalex.org/I129774422"]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":5,"corresponding_author_ids":[],"corresponding_institution_ids":[],"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/T10028","display_name":"Topic Modeling","score":0.7470999956130981,"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":0.7470999956130981,"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":0.1282999962568283,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.012000000104308128,"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/llm-as-a-judge","display_name":"LLM-as-a-judge","score":0.9279999732971191},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.6480000019073486},{"id":"https://openalex.org/keywords/mathematical-reasoning","display_name":"mathematical reasoning","score":0.8690000176429749},{"id":"https://openalex.org/keywords/multi-step-reasoning","display_name":"multi-step reasoning","score":0.7009999752044678},{"id":"https://openalex.org/keywords/model-quality-assessment","display_name":"model quality assessment","score":0.5590000152587891},{"id":"https://openalex.org/keywords/adjudication","display_name":"adjudication","score":0.4650000035762787},{"id":"https://openalex.org/keywords/part-of-speech-features","display_name":"part-of-speech features","score":0.6940000057220459}],"concepts":[{"id":"https://openalex.org/C204434341","wikidata":"https://www.wikidata.org/wiki/Q357789","display_name":"Adjudication","level":2,"score":0.9416676759719849},{"id":"https://openalex.org/C97364631","wikidata":"https://www.wikidata.org/wiki/Q484284","display_name":"Deductive reasoning","level":2,"score":0.44780221581459045},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4191990792751312},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.39858362078666687},{"id":"https://openalex.org/C17744445","wikidata":"https://www.wikidata.org/wiki/Q36442","display_name":"Political science","level":0,"score":0.28191637992858887},{"id":"https://openalex.org/C199539241","wikidata":"https://www.wikidata.org/wiki/Q7748","display_name":"Law","level":1,"score":0.2186082899570465},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.21119114756584167}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2409.04168","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2409.04168","pdf_url":"https://arxiv.org/pdf/2409.04168","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2409.04168","doi":"https://doi.org/10.48550/arxiv.2409.04168","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2409.04168","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":"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2409.04168","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2409.04168","pdf_url":"https://arxiv.org/pdf/2409.04168","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","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/W4403586357.pdf","grobid_xml":"https://content.openalex.org/works/W4403586357.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4409362112","https://openalex.org/W7127610788","https://openalex.org/W4400612109","https://openalex.org/W7152705476","https://openalex.org/W7164175590","https://openalex.org/W4417299031","https://openalex.org/W4404782624","https://openalex.org/W7138076159","https://openalex.org/W7166879487","https://openalex.org/W4403583114"],"abstract_inverted_index":{"To":[0],"reduce":[1],"the":[2,17,33,63,107,135,172],"need":[3],"for":[4],"human":[5,36],"annotations,":[6],"large":[7],"language":[8],"models":[9,119],"(LLMs)":[10],"have":[11],"been":[12],"proposed":[13],"as":[14,42,144],"judges":[15,27,52,114,139,169],"of":[16,19,25,65,137,155],"quality":[18],"other":[20],"candidate":[21,108],"models.":[22],"The":[23],"performance":[24,79,105],"LLM":[26,51,138,168],"is":[28,68,124],"typically":[29],"evaluated":[30],"by":[31,161],"measuring":[32],"correlation":[34,102],"with":[35],"judgments":[37],"on":[38,53],"generative":[39],"tasks":[40,58],"such":[41,143],"summarization":[43],"or":[44],"machine":[45],"translation.":[46],"In":[47],"contrast,":[48],"we":[49,129,132,150,180],"study":[50,160],"mathematical":[54],"reasoning":[55],"tasks.":[56],"These":[57],"require":[59],"multi-step":[60],"reasoning,":[61],"and":[62,81,90,106,147],"correctness":[64],"their":[66,122],"solutions":[67],"verifiable,":[69],"enabling":[70],"a":[71,77,100,127],"more":[72],"objective":[73],"evaluation.":[74],"We":[75,157],"perform":[76],"detailed":[78],"analysis":[80,98],"find":[82,148],"that":[83,113,149,167],"easy":[84,87],"samples":[85,92],"are":[86,93],"to":[88,95,116,183],"judge,":[89],"difficult":[91,94],"judge.":[96],"Our":[97],"uncovers":[99],"strong":[101],"between":[103],"judgment":[104],"model":[109,175],"task":[110,185],"performance,":[111],"indicating":[112],"tend":[115],"favor":[117],"higher-quality":[118],"even":[120],"if":[121,179],"answer":[123],"incorrect.":[125],"As":[126],"consequence,":[128],"test":[130],"whether":[131],"can":[133,151],"predict":[134,153],"behavior":[136],"using":[140],"simple":[141],"features":[142],"part-of-speech":[145],"tags":[146],"correctly":[152],"70%-75%":[154],"judgments.":[156],"conclude":[158],"this":[159],"analyzing":[162],"practical":[163],"use":[164,181],"cases,":[165],"showing":[166],"consistently":[170],"detect":[171],"on-average":[173],"better":[174],"but":[176],"largely":[177],"fail":[178],"them":[182],"improve":[184],"performance.":[186]},"counts_by_year":[],"updated_date":"2026-10-27T15:51:10.671819","created_date":"2024-10-21T00:00:00"}