{"id":"https://openalex.org/W7133329382","doi":"https://doi.org/10.48550/arxiv.2603.01571","title":"Beyond Length Scaling: Synergizing Breadth and Depth for Generative Reward Models","display_name":"Beyond Length Scaling: Synergizing Breadth and Depth for Generative Reward Models","publication_year":2026,"publication_date":"2026-03-02","ids":{"openalex":"https://openalex.org/W7133329382","doi":"https://doi.org/10.48550/arxiv.2603.01571"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.01571","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.01571","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":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2603.01571","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5127896638","display_name":"Qiyuan Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Zhang, Qiyuan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5127991606","display_name":"Yufei Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Yufei","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128024849","display_name":"Tianhe Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Tianhe","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5127885085","display_name":"Can Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Can","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5127900029","display_name":"Qingfeng Sun","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sun, Qingfeng","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5127941618","display_name":"Kai Zheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Kai","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128012243","display_name":"Xue Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Xue","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5128012729","display_name":"Chen Ma","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ma, Chen","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":8,"corresponding_author_ids":["https://openalex.org/A5127896638"],"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/T10203","display_name":"Recommender Systems and Techniques","score":0.1940000057220459,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T10203","display_name":"Recommender Systems and Techniques","score":0.1940000057220459,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.1809999942779541,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.05530000105500221,"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/correctness","display_name":"Correctness","score":0.642300009727478},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.5489000082015991},{"id":"https://openalex.org/keywords/generative-grammar","display_name":"Generative grammar","score":0.5230000019073486},{"id":"https://openalex.org/keywords/generative-model","display_name":"Generative model","score":0.49239999055862427},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.4255000054836273},{"id":"https://openalex.org/keywords/modular-design","display_name":"Modular design","score":0.4106999933719635},{"id":"https://openalex.org/keywords/reliability","display_name":"Reliability (semiconductor)","score":0.3905999958515167},{"id":"https://openalex.org/keywords/verifiable-secret-sharing","display_name":"Verifiable secret sharing","score":0.37950000166893005}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6761999726295471},{"id":"https://openalex.org/C55439883","wikidata":"https://www.wikidata.org/wiki/Q360812","display_name":"Correctness","level":2,"score":0.642300009727478},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5659000277519226},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.5489000082015991},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.5230000019073486},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.49239999055862427},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.4255000054836273},{"id":"https://openalex.org/C101468663","wikidata":"https://www.wikidata.org/wiki/Q1620158","display_name":"Modular design","level":2,"score":0.4106999933719635},{"id":"https://openalex.org/C43214815","wikidata":"https://www.wikidata.org/wiki/Q7310987","display_name":"Reliability (semiconductor)","level":3,"score":0.3905999958515167},{"id":"https://openalex.org/C85847156","wikidata":"https://www.wikidata.org/wiki/Q59015987","display_name":"Verifiable secret sharing","level":3,"score":0.37950000166893005},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.36899998784065247},{"id":"https://openalex.org/C2781067378","wikidata":"https://www.wikidata.org/wiki/Q17027399","display_name":"Interpretability","level":2,"score":0.364300012588501},{"id":"https://openalex.org/C89611455","wikidata":"https://www.wikidata.org/wiki/Q6804646","display_name":"Mechanism (biology)","level":2,"score":0.34380000829696655},{"id":"https://openalex.org/C207390915","wikidata":"https://www.wikidata.org/wiki/Q1230525","display_name":"Divergence (linguistics)","level":2,"score":0.3353999853134155},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.3280999958515167},{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.3206999897956848},{"id":"https://openalex.org/C159032336","wikidata":"https://www.wikidata.org/wiki/Q2488768","display_name":"Non-monotonic logic","level":2,"score":0.30070000886917114},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.29919999837875366},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.26100000739097595}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.01571","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.01571","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":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2603.01571","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.01571","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":"article"},"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":{"Recent":[0],"advancements":[1],"in":[2,120,130],"Generative":[3],"Reward":[4],"Models":[5],"(GRMs)":[6],"have":[7],"demonstrated":[8],"that":[9,61,96,148],"scaling":[10],"the":[11,19,33,136,140,160,182],"length":[12,30],"of":[13,21,36,112],"Chain-of-Thought":[14],"(CoT)":[15],"reasoning":[16,38,137,165],"considerably":[17],"enhances":[18],"reliability":[20],"evaluation.":[22],"However,":[23],"current":[24],"works":[25],"predominantly":[26],"rely":[27],"on":[28],"unstructured":[29],"scaling,":[31],"ignoring":[32],"divergent":[34],"efficacy":[35],"different":[37],"mechanisms:":[39],"Breadth-CoT":[40],"(B-CoT,":[41],"i.e.,":[42,49],"multi-dimensional":[43],"principle":[44],"coverage)":[45],"and":[46,68,80,89,174,181],"Depth-CoT":[47],"(D-CoT,":[48],"substantive":[50],"judgment":[51],"soundness).":[52],"To":[53],"address":[54],"this,":[55],"we":[56,146],"introduce":[57],"Mix-GRM,":[58],"a":[59,71,99,117,152],"framework":[60],"reconfigures":[62],"raw":[63],"rationales":[64],"into":[65],"structured":[66],"B-CoT":[67,122],"D-CoT":[69,128],"through":[70],"modular":[72],"synthesis":[73],"pipeline,":[74],"subsequently":[75],"employing":[76],"Supervised":[77],"Fine-Tuning":[78],"(SFT)":[79],"Reinforcement":[81],"Learning":[82],"with":[83,139],"Verifiable":[84],"Rewards":[85],"(RLVR)":[86],"to":[87,167],"internalize":[88],"optimize":[90],"these":[91],"mechanisms.":[92],"Comprehensive":[93],"experiments":[94],"demonstrate":[95,147],"Mix-GRM":[97],"establishes":[98],"new":[100],"state-of-the-art":[101],"across":[102],"five":[103],"benchmarks,":[104],"surpassing":[105],"leading":[106],"open-source":[107],"RMs":[108],"by":[109],"an":[110,156],"average":[111],"8.2\\%.":[113],"Our":[114],"results":[115],"reveal":[116],"clear":[118],"divergence":[119],"reasoning:":[121],"benefits":[123],"subjective":[124],"preference":[125],"tasks,":[126],"whereas":[127],"excels":[129],"objective":[131],"correctness":[132],"tasks.":[133],"Consequently,":[134],"misaligning":[135],"mechanism":[138],"task":[141,169],"directly":[142],"degrades":[143],"performance.":[144],"Furthermore,":[145],"RLVR":[149],"acts":[150],"as":[151],"switching":[153],"amplifier,":[154],"inducing":[155],"emergent":[157],"polarization":[158],"where":[159],"model":[161],"spontaneously":[162],"allocates":[163],"its":[164],"style":[166],"match":[168],"demands.":[170],"The":[171],"synthesized":[172],"data":[173],"models":[175],"are":[176],"released":[177,185],"at":[178,186],"\\href{https://huggingface.co/collections/DonJoey/mix-grm}{Hugging":[179],"Face},":[180],"code":[183],"is":[184],"\\href{https://github.com/Don-Joey/Mix-GRM}{Github}.":[187]},"counts_by_year":[],"updated_date":"2026-03-04T07:09:34.246503","created_date":"2026-03-04T00:00:00"}