{"id":"https://openalex.org/W7160204180","doi":"https://doi.org/10.48550/arxiv.2605.00433","title":"Improving LLM Code Generation via Requirement-Aware Curriculum Reinforcement Learning","display_name":"Improving LLM Code Generation via Requirement-Aware Curriculum Reinforcement Learning","publication_year":2026,"publication_date":"2026-05-01","ids":{"openalex":"https://openalex.org/W7160204180","doi":"https://doi.org/10.48550/arxiv.2605.00433"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.00433","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.00433","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":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.00433","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5124321560","display_name":"Shouyu Yin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yin, Shouyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135193475","display_name":"Zhao Tian","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tian, Zhao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135197599","display_name":"Junjie Chen","orcid":"https://orcid.org/0000-0002-3046-4821"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Junjie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135126579","display_name":"Shikai Guo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Guo, Shikai","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/T10260","display_name":"Software Engineering Research","score":0.7893000245094299,"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/T10260","display_name":"Software Engineering Research","score":0.7893000245094299,"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/T10430","display_name":"Software Engineering Techniques and Practices","score":0.08110000193119049,"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/T10743","display_name":"Software Testing and Debugging Techniques","score":0.02759999968111515,"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/reinforcement-learning","display_name":"Reinforcement learning","score":0.6715999841690063},{"id":"https://openalex.org/keywords/curriculum","display_name":"Curriculum","score":0.5554999709129333},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.5443000197410583},{"id":"https://openalex.org/keywords/construct","display_name":"Construct (python library)","score":0.49390000104904175},{"id":"https://openalex.org/keywords/code-generation","display_name":"Code generation","score":0.43549999594688416},{"id":"https://openalex.org/keywords/quality","display_name":"Quality (philosophy)","score":0.4287000000476837},{"id":"https://openalex.org/keywords/software-quality","display_name":"Software quality","score":0.4129999876022339}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7075999975204468},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.6715999841690063},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.5723999738693237},{"id":"https://openalex.org/C47177190","wikidata":"https://www.wikidata.org/wiki/Q207137","display_name":"Curriculum","level":2,"score":0.5554999709129333},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.5443000197410583},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.49390000104904175},{"id":"https://openalex.org/C133162039","wikidata":"https://www.wikidata.org/wiki/Q1061077","display_name":"Code generation","level":3,"score":0.43549999594688416},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.4287000000476837},{"id":"https://openalex.org/C117447612","wikidata":"https://www.wikidata.org/wiki/Q1412670","display_name":"Software quality","level":4,"score":0.4129999876022339},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.40220001339912415},{"id":"https://openalex.org/C529173508","wikidata":"https://www.wikidata.org/wiki/Q638608","display_name":"Software development","level":3,"score":0.3952000141143799},{"id":"https://openalex.org/C2777904410","wikidata":"https://www.wikidata.org/wiki/Q7397","display_name":"Software","level":2,"score":0.3882000148296356},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.37439998984336853},{"id":"https://openalex.org/C133237599","wikidata":"https://www.wikidata.org/wiki/Q2295111","display_name":"Code smell","level":5,"score":0.34779998660087585},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.34450000524520874},{"id":"https://openalex.org/C150292731","wikidata":"https://www.wikidata.org/wiki/Q1342704","display_name":"Code review","level":5,"score":0.26660001277923584},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.2540999948978424},{"id":"https://openalex.org/C77967617","wikidata":"https://www.wikidata.org/wiki/Q4677561","display_name":"Active learning (machine learning)","level":2,"score":0.2533000111579895},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.2531999945640564},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.25029999017715454}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.00433","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.00433","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":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.00433","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.00433","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":"Preprint"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/9","score":0.5937826037406921,"display_name":"Industry, innovation and infrastructure"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Code":[0],"generation,":[1,106],"which":[2],"aims":[3],"to":[4,16,71,114,123,157,169],"automatically":[5,149],"generate":[6],"source":[7],"code":[8,32,74,105,146,186],"from":[9,38,84,129],"given":[10],"programming":[11,45,107],"requirements,":[12],"has":[13,34],"the":[14,23,92,111,115,196],"potential":[15],"substantially":[17],"improve":[18,72,158],"software":[19,130],"development":[20],"efficiency.":[21],"With":[22],"rapid":[24],"advancement":[25],"of":[26,94,199,209],"large":[27],"language":[28],"models":[29],"(LLMs),":[30],"LLM-based":[31,145],"generation":[33,75,187],"attracted":[35],"widespread":[36],"attention":[37],"both":[39],"academia":[40],"and":[41,98,120,162],"industry.":[42],"However,":[43],"as":[44,110],"requirements":[46,108,131,156],"become":[47],"increasingly":[48],"complex,":[49],"existing":[50,80],"LLMs":[51,182],"still":[52],"exhibit":[53],"notable":[54],"performance":[55],"limitations.":[56],"To":[57],"address":[58],"this":[59],"challenge,":[60],"recent":[61],"studies":[62],"have":[63],"proposed":[64],"training-based":[65],"curriculum":[66,100,139,166],"reinforcement":[67,140],"learning":[68,141],"(CRL)":[69],"strategies":[70],"LLM":[73],"performance.":[76],"Despite":[77],"their":[78,118],"effectiveness,":[79],"CRL":[81],"approaches":[82],"suffer":[83],"several":[85],"limitations,":[86],"including":[87],"misaligned":[88],"requirement":[89,95,152],"difficulty":[90,96,121],"perception,":[91],"absence":[93],"optimization,":[97],"suboptimal":[99],"sampling":[101,167],"strategies.":[102],"In":[103],"CRL-based":[104],"serve":[109],"sole":[112],"input":[113],"model,":[116],"making":[117],"quality":[119],"critical":[122],"training":[124,159,171],"effectiveness.":[125],"Motivated":[126],"by":[127,189],"insights":[128],"engineering,":[132],"we":[133],"propose":[134],"RECRL,":[135],"a":[136],"novel":[137],"requirement-aware":[138],"framework":[142],"for":[143],"enhancing":[144],"generation.":[147],"RECRL":[148,203],"perceives":[150],"model-specific":[151],"difficulty,":[153],"optimizes":[154],"challenging":[155],"data":[160],"utilization,":[161],"employs":[163],"an":[164,205],"adaptive":[165],"strategy":[168],"construct":[170],"batches":[172],"with":[173,191],"smoothly":[174],"varying":[175],"difficulty.":[176],"Extensive":[177],"experiments":[178],"on":[179],"five":[180,184,192],"state-of-the-art":[181,193,213],"across":[183],"widely-used":[185],"benchmarks":[188],"comparing":[190],"baselines,":[194],"demonstrate":[195],"significant":[197],"effectiveness":[198],"RECRL.":[200],"For":[201],"example,":[202],"achieves":[204],"average":[206],"Pass@1":[207],"improvement":[208],"1.23%-5.62%":[210],"over":[211],"all":[212],"baselines.":[214]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-05T00:00:00"}