{"id":"https://openalex.org/W4414973451","doi":"https://doi.org/10.48550/arxiv.2504.07086","title":"A Sober Look at Progress in Language Model Reasoning: Pitfalls and Paths to Reproducibility","display_name":"A Sober Look at Progress in Language Model Reasoning: Pitfalls and Paths to Reproducibility","publication_year":2025,"publication_date":"2025-04-09","ids":{"openalex":"https://openalex.org/W4414973451","doi":"https://doi.org/10.48550/arxiv.2504.07086"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2504.07086","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2504.07086","pdf_url":"https://arxiv.org/pdf/2504.07086","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/2504.07086","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5107077701","display_name":"Andreas Hochlehnert","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4401726832","display_name":"T\u00fcbingen AI Center","ror":"https://ror.org/0107nyd78","country_code":"DE","type":"education","lineage":["https://openalex.org/I149899117","https://openalex.org/I4210135521","https://openalex.org/I4401726832","https://openalex.org/I8087733"]},{"id":"https://openalex.org/I8087733","display_name":"University of T\u00fcbingen","ror":"https://ror.org/03a1kwz48","country_code":"DE","type":"education","lineage":["https://openalex.org/I8087733"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Hochlehnert, Andreas","raw_affiliation_strings":["T\u00fcbingen AI Center , University of T\u00fcbingen"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"T\u00fcbingen AI Center , University of T\u00fcbingen","institution_ids":["https://openalex.org/I4401726832","https://openalex.org/I8087733"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5114438807","display_name":"Hardik Bhatnagar","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4401726832","display_name":"T\u00fcbingen AI Center","ror":"https://ror.org/0107nyd78","country_code":"DE","type":"education","lineage":["https://openalex.org/I149899117","https://openalex.org/I4210135521","https://openalex.org/I4401726832","https://openalex.org/I8087733"]},{"id":"https://openalex.org/I8087733","display_name":"University of T\u00fcbingen","ror":"https://ror.org/03a1kwz48","country_code":"DE","type":"education","lineage":["https://openalex.org/I8087733"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Bhatnagar, Hardik","raw_affiliation_strings":["T\u00fcbingen AI Center , University of T\u00fcbingen"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"T\u00fcbingen AI Center , University of T\u00fcbingen","institution_ids":["https://openalex.org/I4401726832","https://openalex.org/I8087733"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5108719787","display_name":"Vishaal Udandarao","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]},{"id":"https://openalex.org/I4401726832","display_name":"T\u00fcbingen AI Center","ror":"https://ror.org/0107nyd78","country_code":"DE","type":"education","lineage":["https://openalex.org/I149899117","https://openalex.org/I4210135521","https://openalex.org/I4401726832","https://openalex.org/I8087733"]},{"id":"https://openalex.org/I8087733","display_name":"University of T\u00fcbingen","ror":"https://ror.org/03a1kwz48","country_code":"DE","type":"education","lineage":["https://openalex.org/I8087733"]}],"countries":["DE","GB"],"is_corresponding":false,"raw_author_name":"Udandarao, Vishaal","raw_affiliation_strings":["T\u00fcbingen AI Center , University of T\u00fcbingen","University of Cambridge Leaderboard Code Eval"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"T\u00fcbingen AI Center , University of T\u00fcbingen","institution_ids":["https://openalex.org/I4401726832","https://openalex.org/I8087733"]},{"raw_affiliation_string":"University of Cambridge Leaderboard Code Eval","institution_ids":["https://openalex.org/I241749"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102018867","display_name":"Samuel Albanie","orcid":"https://orcid.org/0000-0003-1732-9198","observed_orcids":["https://orcid.org/0000-0003-1732-9198"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Albanie, Samuel","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113274335","display_name":"Ameya Prabhu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4401726832","display_name":"T\u00fcbingen AI Center","ror":"https://ror.org/0107nyd78","country_code":"DE","type":"education","lineage":["https://openalex.org/I149899117","https://openalex.org/I4210135521","https://openalex.org/I4401726832","https://openalex.org/I8087733"]},{"id":"https://openalex.org/I8087733","display_name":"University of T\u00fcbingen","ror":"https://ror.org/03a1kwz48","country_code":"DE","type":"education","lineage":["https://openalex.org/I8087733"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Prabhu, Ameya","raw_affiliation_strings":["T\u00fcbingen AI Center , University of T\u00fcbingen"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"T\u00fcbingen AI Center , University of T\u00fcbingen","institution_ids":["https://openalex.org/I4401726832","https://openalex.org/I8087733"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5061457780","display_name":"Matthias Bethge","orcid":"https://orcid.org/0000-0002-6417-7812","observed_orcids":["https://orcid.org/0000-0002-6417-7812","https://orcid.org/0009-0008-5601-6120"]},"institutions":[{"id":"https://openalex.org/I4401726832","display_name":"T\u00fcbingen AI Center","ror":"https://ror.org/0107nyd78","country_code":"DE","type":"education","lineage":["https://openalex.org/I149899117","https://openalex.org/I4210135521","https://openalex.org/I4401726832","https://openalex.org/I8087733"]},{"id":"https://openalex.org/I8087733","display_name":"University of T\u00fcbingen","ror":"https://ror.org/03a1kwz48","country_code":"DE","type":"education","lineage":["https://openalex.org/I8087733"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Bethge, Matthias","raw_affiliation_strings":["T\u00fcbingen AI Center , University of T\u00fcbingen"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"T\u00fcbingen AI Center , University of T\u00fcbingen","institution_ids":["https://openalex.org/I4401726832","https://openalex.org/I8087733"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"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":0.7562999725341797,"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":0.7562999725341797,"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":0.1054999977350235,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.013700000010430813,"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/language-model-reasoning","display_name":"language model reasoning","score":0.8500000238418579},{"id":"https://openalex.org/keywords/reproducibility","display_name":"reproducibility","score":0.9319999814033508},{"id":"https://openalex.org/keywords/reinforcement-learning-for-reasoning","display_name":"reinforcement learning for reasoning","score":0.9150000214576721},{"id":"https://openalex.org/keywords/supervised-fine-tuning","display_name":"supervised fine-tuning","score":0.8740000128746033},{"id":"https://openalex.org/keywords/overfitting","display_name":"overfitting","score":0.9390000104904175},{"id":"https://openalex.org/keywords/generalization","display_name":"generalization","score":0.4519999921321869}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7063999772071838},{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.6437000036239624},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.5960999727249146},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5152000188827515},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4772999882698059},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.4180999994277954},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.41200000047683716},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.4090000092983246},{"id":"https://openalex.org/C184356942","wikidata":"https://www.wikidata.org/wiki/Q830382","display_name":"Best practice","level":2,"score":0.38850000500679016},{"id":"https://openalex.org/C539667460","wikidata":"https://www.wikidata.org/wiki/Q2414942","display_name":"Management science","level":1,"score":0.38609999418258667},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.3734000027179718},{"id":"https://openalex.org/C49937458","wikidata":"https://www.wikidata.org/wiki/Q2599292","display_name":"Probabilistic logic","level":2,"score":0.35989999771118164},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.3422999978065491},{"id":"https://openalex.org/C2778571376","wikidata":"https://www.wikidata.org/wiki/Q1355821","display_name":"Frontier","level":2,"score":0.3093999922275543},{"id":"https://openalex.org/C152752567","wikidata":"https://www.wikidata.org/wiki/Q116877","display_name":"Code refactoring","level":3,"score":0.2922999858856201},{"id":"https://openalex.org/C158154518","wikidata":"https://www.wikidata.org/wiki/Q7310970","display_name":"Relevance (law)","level":2,"score":0.28189998865127563},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.28040000796318054},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.27810001373291016},{"id":"https://openalex.org/C188147891","wikidata":"https://www.wikidata.org/wiki/Q147638","display_name":"Cognitive science","level":1,"score":0.2777000069618225},{"id":"https://openalex.org/C2777904410","wikidata":"https://www.wikidata.org/wiki/Q7397","display_name":"Software","level":2,"score":0.2554999887943268},{"id":"https://openalex.org/C2777673361","wikidata":"https://www.wikidata.org/wiki/Q5281228","display_name":"Disadvantage","level":2,"score":0.2506999969482422},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.25040000677108765}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2504.07086","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2504.07086","pdf_url":"https://arxiv.org/pdf/2504.07086","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.2504.07086","doi":"https://doi.org/10.48550/arxiv.2504.07086","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2504.07086","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:2504.07086","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2504.07086","pdf_url":"https://arxiv.org/pdf/2504.07086","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":[{"id":"https://openalex.org/F4320315389","display_name":"Open Philanthropy Project","ror":"https://ror.org/004d1k391"},{"id":"https://openalex.org/F4320317967","display_name":"International Max Planck Research School for Environmental, Cellular and Molecular Microbiology","ror":null},{"id":"https://openalex.org/F4320319740","display_name":"Good Ventures Foundation","ror":"https://ror.org/01t5wp533"},{"id":"https://openalex.org/F4320321114","display_name":"Bundesministerium f\u00fcr Bildung und Forschung","ror":"https://ror.org/04pz7b180"},{"id":"https://openalex.org/F4320327553","display_name":"International Max Planck Research School for Advanced Methods in Process and Systems Engineering","ror":"https://ror.org/04hazks31"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4414973451.pdf","grobid_xml":"https://content.openalex.org/works/W4414973451.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7161132819","https://openalex.org/W4416435617","https://openalex.org/W7154037293","https://openalex.org/W4416430148","https://openalex.org/W7153982152","https://openalex.org/W7138944128","https://openalex.org/W7162434131","https://openalex.org/W4415336078","https://openalex.org/W4417431391","https://openalex.org/W4416036113"],"abstract_inverted_index":{"Reasoning":[0],"has":[1],"emerged":[2],"as":[3],"the":[4,154],"next":[5],"major":[6],"frontier":[7],"for":[8,169,176],"language":[9],"models":[10],"(LMs),":[11],"with":[12,28,103],"rapid":[13],"advances":[14],"from":[15],"both":[16],"academic":[17],"and":[18,51,71,74,108,118,166],"industrial":[19],"labs.":[20],"However,":[21],"this":[22,43,112],"progress":[23],"often":[24],"outpaces":[25],"methodological":[26],"rigor,":[27],"many":[29],"evaluations":[30],"relying":[31],"on":[32,85,138],"benchmarking":[33],"practices":[34,107],"that":[35,53,120],"lack":[36],"transparency,":[37],"robustness,":[38],"or":[39,88],"statistical":[40],"grounding.":[41],"In":[42,143],"work,":[44],"we":[45,97,114,156,161],"conduct":[46],"a":[47,99],"comprehensive":[48],"empirical":[49],"study":[50],"find":[52,119],"current":[54],"mathematical":[55],"reasoning":[56,170],"benchmarks":[57,140],"are":[58,133],"highly":[59],"sensitive":[60],"to":[61,135],"subtle":[62],"implementation":[63],"choices--including":[64],"decoding":[65],"parameters,":[66],"random":[67],"seeds,":[68],"prompt":[69],"formatting,":[70],"even":[72],"hardware":[73],"software":[75],"configurations.":[76],"Performance":[77],"gains":[78],"reported":[79],"in":[80,153],"recent":[81,116],"studies":[82],"frequently":[83],"hinge":[84],"unclear":[86],"comparisons":[87],"unreported":[89],"sources":[90],"of":[91],"variance.":[92],"To":[93,158],"address":[94],"these":[95],"issues,":[96],"propose":[98],"standardized":[100],"evaluation":[101],"framework":[102],"clearly":[104],"defined":[105],"best":[106],"reporting":[109],"standards.":[110],"Using":[111],"framework,":[113],"reassess":[115],"methods":[117,148],"most":[121],"reinforcement":[122],"learning":[123],"(RL)":[124],"approaches":[125],"yield":[126],"only":[127],"modest":[128],"improvements--far":[129],"below":[130],"prior":[131],"claims--and":[132],"prone":[134],"overfitting,":[136],"especially":[137],"small-scale":[139],"like":[141],"AIME'24.":[142],"contrast,":[144],"supervised":[145],"finetuning":[146],"(SFT)":[147],"show":[149],"consistently":[150],"stronger":[151],"generalization":[152],"settings":[155],"study.":[157],"foster":[159],"reproducibility,":[160],"release":[162],"all":[163],"code,":[164],"prompts,":[165],"model":[167],"outputs,":[168],"benchmarks,":[171],"establishing":[172],"more":[173],"rigorous":[174],"foundations":[175],"future":[177],"work.":[178]},"counts_by_year":[],"updated_date":"2026-10-27T15:51:10.671819","created_date":"2025-10-09T00:00:00"}