{"id":"https://openalex.org/W7153247530","doi":"https://doi.org/10.48550/arxiv.2604.08477","title":"SUPERNOVA: Eliciting General Reasoning in LLMs with Reinforcement Learning on Natural Instructions","display_name":"SUPERNOVA: Eliciting General Reasoning in LLMs with Reinforcement Learning on Natural Instructions","publication_year":2026,"publication_date":"2026-04-09","ids":{"openalex":"https://openalex.org/W7153247530","doi":"https://doi.org/10.48550/arxiv.2604.08477"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.08477","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.08477","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.2604.08477","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5133342673","display_name":"Ashima Suvarna","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Suvarna, Ashima","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133351069","display_name":"Kendrick Phan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Phan, Kendrick","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133327412","display_name":"Mehrab Beikzadeh","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Beikzadeh, Mehrab","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112428050","display_name":"Hritik Bansal","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bansal, Hritik","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5015792268","display_name":"Saadia Gabriel","orcid":"https://orcid.org/0009-0001-9353-951X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gabriel, Saadia","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":0.3310999870300293,"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":0.3310999870300293,"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":0.1362999975681305,"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"}},{"id":"https://openalex.org/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.08590000122785568,"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/task","display_name":"Task (project management)","score":0.6567999720573425},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5662000179290771},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.5250999927520752},{"id":"https://openalex.org/keywords/reasoning-system","display_name":"Reasoning system","score":0.45840001106262207},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.4359999895095825},{"id":"https://openalex.org/keywords/causal-reasoning","display_name":"Causal reasoning","score":0.4050000011920929},{"id":"https://openalex.org/keywords/task-analysis","display_name":"Task analysis","score":0.38510000705718994},{"id":"https://openalex.org/keywords/qualitative-reasoning","display_name":"Qualitative reasoning","score":0.3822000026702881},{"id":"https://openalex.org/keywords/deductive-reasoning","display_name":"Deductive reasoning","score":0.35350000858306885}],"concepts":[{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.6567999720573425},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6266000270843506},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5662000179290771},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5250999927520752},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4902999997138977},{"id":"https://openalex.org/C89288958","wikidata":"https://www.wikidata.org/wiki/Q7301504","display_name":"Reasoning system","level":2,"score":0.45840001106262207},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.4359999895095825},{"id":"https://openalex.org/C115086926","wikidata":"https://www.wikidata.org/wiki/Q17004651","display_name":"Causal reasoning","level":3,"score":0.4050000011920929},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.38510000705718994},{"id":"https://openalex.org/C83725634","wikidata":"https://www.wikidata.org/wiki/Q7268699","display_name":"Qualitative reasoning","level":2,"score":0.3822000026702881},{"id":"https://openalex.org/C97364631","wikidata":"https://www.wikidata.org/wiki/Q484284","display_name":"Deductive reasoning","level":2,"score":0.35350000858306885},{"id":"https://openalex.org/C158600405","wikidata":"https://www.wikidata.org/wiki/Q5054566","display_name":"Causal inference","level":2,"score":0.3334999978542328},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.33000001311302185},{"id":"https://openalex.org/C2776325391","wikidata":"https://www.wikidata.org/wiki/Q6917865","display_name":"Motivated reasoning","level":3,"score":0.3269999921321869},{"id":"https://openalex.org/C183521366","wikidata":"https://www.wikidata.org/wiki/Q7256422","display_name":"Psychology of reasoning","level":4,"score":0.3255000114440918},{"id":"https://openalex.org/C195344581","wikidata":"https://www.wikidata.org/wiki/Q2555318","display_name":"Automated reasoning","level":2,"score":0.32170000672340393},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.3122999966144562},{"id":"https://openalex.org/C43971567","wikidata":"https://www.wikidata.org/wiki/Q3142865","display_name":"Logical reasoning","level":2,"score":0.2973000109195709},{"id":"https://openalex.org/C36964233","wikidata":"https://www.wikidata.org/wiki/Q7920942","display_name":"Verbal reasoning","level":3,"score":0.27570000290870667},{"id":"https://openalex.org/C20162079","wikidata":"https://www.wikidata.org/wiki/Q1151406","display_name":"Case-based reasoning","level":2,"score":0.2702000141143799},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.2678999900817871},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.26420000195503235},{"id":"https://openalex.org/C2779439875","wikidata":"https://www.wikidata.org/wiki/Q1078276","display_name":"Natural language understanding","level":3,"score":0.25780001282691956},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.2572999894618988},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.2563999891281128},{"id":"https://openalex.org/C126042441","wikidata":"https://www.wikidata.org/wiki/Q1324888","display_name":"Frame (networking)","level":2,"score":0.2547999918460846},{"id":"https://openalex.org/C193221554","wikidata":"https://www.wikidata.org/wiki/Q5153664","display_name":"Commonsense reasoning","level":2,"score":0.2533999979496002},{"id":"https://openalex.org/C85847156","wikidata":"https://www.wikidata.org/wiki/Q59015987","display_name":"Verifiable secret sharing","level":3,"score":0.25},{"id":"https://openalex.org/C180747234","wikidata":"https://www.wikidata.org/wiki/Q23373","display_name":"Cognitive psychology","level":1,"score":0.25}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.08477","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.08477","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.2604.08477","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.08477","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":[{"score":0.7915050387382507,"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Reinforcement":[0],"Learning":[1],"with":[2],"Verifiable":[3],"Rewards":[4],"(RLVR)":[5],"has":[6,121],"substantially":[7],"improved":[8],"reasoning":[9,93,127,201],"in":[10],"formal":[11],"domains":[12],"such":[13],"as":[14],"mathematics":[15],"and":[16,86,110,147,217],"code,":[17],"but":[18,65],"extending":[19],"these":[20,82,156],"gains":[21,207],"beyond":[22,28],"STEM":[23,29],"remains":[24],"challenging.":[25],"Extending":[26],"RLVR":[27,51,69,85,163,233],"is":[30],"fundamentally":[31],"constrained":[32],"by":[33,155],"the":[34,181],"lack":[35],"of":[36,62,165,188],"high-quality":[37,162],"verifiable":[38],"training":[39,176],"data.":[40],"In":[41,96],"this":[42],"work,":[43],"we":[44,76,98,158,204],"introduce":[45],"SUPERNOVA,":[46,160],"a":[47,59,122,161,185,195],"framework":[48],"for":[49,68,84,136,227],"curating":[50,228],"data":[52,64,88,101],"from":[53,169,208],"natural":[54,170],"instruction":[55,171],"datasets,":[56],"which":[57],"are":[58,66],"rich":[60],"source":[61,104,118],"expert-annotated":[63],"underexplored":[67],"training.":[70],"Through":[71],"100+":[72],"controlled":[73],"RL":[74],"experiments,":[75],"systematically":[77],"study":[78],"how":[79,87],"to":[80,211,231,234],"utilize":[81],"dataset":[83,164],"curation":[89],"decisions":[90],"affect":[91],"downstream":[92,126],"performance":[94,135,146],".":[95],"particular,":[97],"investigate":[99],"three":[100],"designs:":[102],"(a)":[103],"task":[105,108,119],"selection,":[106],"(b)":[107],"mixing,":[109],"(c)":[111],"synthetic":[112,148],"interventions.":[113],"Our":[114],"analysis":[115],"reveals":[116],"that":[117,175,206],"selection":[120],"significant":[123],"impact":[124],"on":[125,133,143,178,190],"performance.":[128],"Moreover,":[129],"selecting":[130],"tasks":[131,139],"based":[132,142],"their":[134],"individual":[137],"target":[138],"outperforms":[140,180],"strategies":[141],"overall":[144],"average":[145],"interventions":[149],"do":[150],"not":[151],"improve":[152],"reasoning.":[153,236],"Guided":[154],"insights,":[157],"construct":[159],"25K":[166],"instances":[167],"curated":[168],"datasets.":[172],"We":[173],"show":[174],"Qwen3-0.6B":[177],"SUPERNOVA":[179,209],"base":[182],"Qwen3-0.6B,":[183],"yielding":[184],"relative":[186],"gain":[187],"64.4pp":[189],"BigBench":[191],"Extra":[192],"Hard":[193],"(BBEH),":[194],"challenging":[196],"benchmark":[197],"comprising":[198],"23":[199],"complex":[200],"tasks.":[202],"Importantly,":[203],"find":[205],"generalize":[210],"unseen":[212],"benchmarks,":[213],"larger":[214],"model":[215,219],"scales,":[216],"newer":[218],"families.":[220],"Overall,":[221],"our":[222],"findings":[223],"provide":[224],"practical":[225],"insights":[226],"human-annotated":[229],"resources":[230],"extend":[232],"general":[235],"Models,":[237],"Data,":[238],"Code":[239],"at":[240],"https://github.com/asuvarna31/supernova.":[241]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-04-11T00:00:00"}