{"id":"https://openalex.org/W4415249779","doi":"https://doi.org/10.48550/arxiv.2504.09696","title":"GRPO-LEAD: A Difficulty-Aware Reinforcement Learning Approach for Concise Mathematical Reasoning in Language Models","display_name":"GRPO-LEAD: A Difficulty-Aware Reinforcement Learning Approach for Concise Mathematical Reasoning in Language Models","publication_year":2025,"publication_date":"2025-04-13","ids":{"openalex":"https://openalex.org/W4415249779","doi":"https://doi.org/10.48550/arxiv.2504.09696"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2504.09696","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2504.09696","pdf_url":"https://arxiv.org/pdf/2504.09696","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":null,"license_id":null,"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/2504.09696","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5086913695","display_name":"Jixiao Zhang","orcid":"https://orcid.org/0000-0002-8302-4731","observed_orcids":["https://orcid.org/0000-0002-8302-4731"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Jixiao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5093917254","display_name":"Chunsheng Zuo","orcid":"https://orcid.org/0009-0000-4988-2401","observed_orcids":["https://orcid.org/0009-0000-4988-2401"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zuo, Chunsheng","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/T10028","display_name":"Topic Modeling","score":0.26440000534057617,"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":0.26440000534057617,"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/T12522","display_name":"Mathematics Education and Pedagogy","score":0.10970000177621841,"subfield":{"id":"https://openalex.org/subfields/2604","display_name":"Applied Mathematics"},"field":{"id":"https://openalex.org/fields/26","display_name":"Mathematics"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T13523","display_name":"Mathematics, Computing, and Information Processing","score":0.08299999684095383,"subfield":{"id":"https://openalex.org/subfields/1703","display_name":"Computational Theory and Mathematics"},"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/group-relative-policy-optimization","display_name":"Group Relative Policy Optimization","score":0.8500000238418579},{"id":"https://openalex.org/keywords/mathematical-reasoning","display_name":"mathematical reasoning","score":0.8820000290870667},{"id":"https://openalex.org/keywords/language-models","display_name":"language models","score":0.6079999804496765},{"id":"https://openalex.org/keywords/reward-sparsity","display_name":"reward sparsity","score":0.8640000224113464},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.503000020980835}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7498999834060669},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.7103000283241272},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6488000154495239},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.6207000017166138},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.4745999872684479},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4632999897003174},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4465999901294708},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.3939000070095062},{"id":"https://openalex.org/C76969082","wikidata":"https://www.wikidata.org/wiki/Q486902","display_name":"Mathematical model","level":2,"score":0.3540000021457672},{"id":"https://openalex.org/C2983448237","wikidata":"https://www.wikidata.org/wiki/Q1078276","display_name":"Language understanding","level":2,"score":0.3021000027656555},{"id":"https://openalex.org/C195344581","wikidata":"https://www.wikidata.org/wiki/Q2555318","display_name":"Automated reasoning","level":2,"score":0.2685999870300293}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2504.09696","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2504.09696","pdf_url":"https://arxiv.org/pdf/2504.09696","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2504.09696","doi":"https://doi.org/10.48550/arxiv.2504.09696","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2504.09696","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2504.09696","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2504.09696","pdf_url":"https://arxiv.org/pdf/2504.09696","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":null,"license_id":null,"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":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7128535360","https://openalex.org/W4416962804","https://openalex.org/W7127640081","https://openalex.org/W7147282947","https://openalex.org/W4417092538","https://openalex.org/W7124358607","https://openalex.org/W7126115229","https://openalex.org/W7138884082","https://openalex.org/W7140217152","https://openalex.org/W7137986045"],"abstract_inverted_index":{"Group":[0],"Relative":[1],"Policy":[2],"Optimization":[3],"(GRPO),":[4],"which":[5],"is":[6],"widely":[7],"adopted":[8],"by":[9],"R1-like":[10],"reasoning":[11,74],"models,":[12,86],"has":[13],"advanced":[14],"mathematical":[15],"reasoning.":[16],"Nevertheless,":[17],"GRPO":[18,35],"faces":[19],"challenges":[20],"in":[21],"reward":[22],"sparsity,":[23],"verbosity,":[24],"and":[25,56,77,97,105],"inadequate":[26],"focus":[27],"on":[28,64],"problem":[29],"difficulty.":[30],"We":[31],"propose":[32],"GRPO-LEAD,":[33],"enhancing":[34],"with:":[36],"(1)":[37],"length-regularized":[38],"rewards":[39],"to":[40,52],"encourage":[41],"conciseness":[42],"while":[43],"maintaining":[44],"accuracy;":[45],"(2)":[46],"explicit":[47],"penalties":[48],"for":[49,61,84],"incorrect":[50],"solutions":[51],"improve":[53],"model":[54,95],"precision;":[55],"(3)":[57],"difficulty-aware":[58],"advantage":[59],"reweighting":[60],"robust":[62],"generalization":[63],"challenging":[65],"problems.":[66],"Comprehensive":[67],"evaluations":[68],"demonstrate":[69],"that":[70],"GRPO-LEAD":[71],"significantly":[72],"improves":[73],"accuracy,":[75],"conciseness,":[76],"efficiency.":[78],"Our":[79,100],"approach":[80],"achieves":[81],"state-of-the-art":[82],"performance":[83],"14B-scale":[85],"underscoring":[87],"the":[88],"synergy":[89],"of":[90],"our":[91],"methods":[92],"with":[93],"appropriate":[94],"scale":[96],"high-quality":[98],"data.":[99],"source":[101],"code,":[102],"generated":[103],"dataset,":[104],"models":[106],"are":[107],"available":[108],"at":[109],"https://github.com/aeroplanepaper/GRPO-LEAD.":[110]},"counts_by_year":[],"updated_date":"2026-09-21T07:43:02.447757","created_date":"2025-10-16T00:00:00"}