{"id":"https://openalex.org/W7162432771","doi":"https://doi.org/10.48550/arxiv.2605.24998","title":"Better, Faster: Harnessing Self-Improvement in Large Reasoning Models","display_name":"Better, Faster: Harnessing Self-Improvement in Large Reasoning Models","publication_year":2026,"publication_date":"2026-05-24","ids":{"openalex":"https://openalex.org/W7162432771","doi":"https://doi.org/10.48550/arxiv.2605.24998"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.24998","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.24998","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":"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.24998","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5053183023","display_name":"Qihuang Zhong","orcid":"https://orcid.org/0009-0001-0118-5217"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhong, Qihuang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137054838","display_name":"Liang Ding","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ding, Liang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137065962","display_name":"Juhua Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Juhua","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137066382","display_name":"Bo Du","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Du, Bo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136992937","display_name":"Leszek Rutkowski","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rutkowski, Leszek","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137037131","display_name":"Dacheng Tao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tao, Dacheng","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/T11273","display_name":"Advanced Graph Neural Networks","score":0.17329999804496765,"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/T11273","display_name":"Advanced Graph Neural Networks","score":0.17329999804496765,"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":0.14710000157356262,"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":0.12120000272989273,"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/opportunistic-reasoning","display_name":"Opportunistic reasoning","score":0.6930000185966492},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6657999753952026},{"id":"https://openalex.org/keywords/reasoning-system","display_name":"Reasoning system","score":0.5317000150680542},{"id":"https://openalex.org/keywords/case-based-reasoning","display_name":"Case-based reasoning","score":0.525600016117096},{"id":"https://openalex.org/keywords/model-based-reasoning","display_name":"Model-based reasoning","score":0.4722999930381775},{"id":"https://openalex.org/keywords/non-monotonic-logic","display_name":"Non-monotonic logic","score":0.43140000104904175},{"id":"https://openalex.org/keywords/qualitative-reasoning","display_name":"Qualitative reasoning","score":0.37880000472068787},{"id":"https://openalex.org/keywords/automated-reasoning","display_name":"Automated reasoning","score":0.37439998984336853}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7041000127792358},{"id":"https://openalex.org/C86827895","wikidata":"https://www.wikidata.org/wiki/Q7098582","display_name":"Opportunistic reasoning","level":4,"score":0.6930000185966492},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6657999753952026},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6216999888420105},{"id":"https://openalex.org/C89288958","wikidata":"https://www.wikidata.org/wiki/Q7301504","display_name":"Reasoning system","level":2,"score":0.5317000150680542},{"id":"https://openalex.org/C20162079","wikidata":"https://www.wikidata.org/wiki/Q1151406","display_name":"Case-based reasoning","level":2,"score":0.525600016117096},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.49480000138282776},{"id":"https://openalex.org/C37335422","wikidata":"https://www.wikidata.org/wiki/Q6888134","display_name":"Model-based reasoning","level":3,"score":0.4722999930381775},{"id":"https://openalex.org/C159032336","wikidata":"https://www.wikidata.org/wiki/Q2488768","display_name":"Non-monotonic logic","level":2,"score":0.43140000104904175},{"id":"https://openalex.org/C83725634","wikidata":"https://www.wikidata.org/wiki/Q7268699","display_name":"Qualitative reasoning","level":2,"score":0.37880000472068787},{"id":"https://openalex.org/C195344581","wikidata":"https://www.wikidata.org/wiki/Q2555318","display_name":"Automated reasoning","level":2,"score":0.37439998984336853},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.37220001220703125},{"id":"https://openalex.org/C106131492","wikidata":"https://www.wikidata.org/wiki/Q3072260","display_name":"Filter (signal processing)","level":2,"score":0.36719998717308044},{"id":"https://openalex.org/C97364631","wikidata":"https://www.wikidata.org/wiki/Q484284","display_name":"Deductive reasoning","level":2,"score":0.31520000100135803},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.295199990272522},{"id":"https://openalex.org/C2781316041","wikidata":"https://www.wikidata.org/wiki/Q1230584","display_name":"Diversity (politics)","level":2,"score":0.2930999994277954},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.2824999988079071},{"id":"https://openalex.org/C71559656","wikidata":"https://www.wikidata.org/wiki/Q671298","display_name":"Divide and conquer algorithms","level":2,"score":0.2782999873161316},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.2775000035762787},{"id":"https://openalex.org/C2776325391","wikidata":"https://www.wikidata.org/wiki/Q6917865","display_name":"Motivated reasoning","level":3,"score":0.26750001311302185}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.24998","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.24998","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.24998","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.24998","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":"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"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Self-improvement":[0],"training":[1,16,55],"enables":[2],"the":[3,60,131,153,178,193],"large":[4,92],"reasoning":[5,13,32,75,165,179,194],"models":[6,94],"(LRMs)":[7],"to":[8,37,106,125,137,160,184,199],"improve":[9],"themselves":[10],"by":[11,110,196],"self-generating":[12],"trajectories":[14],"as":[15,156],"data":[17,51,108],"without":[18],"external":[19,158],"supervision.":[20],"However,":[21],"we":[22,46,84,143],"find":[23],"that":[24,151,172],"this":[25,82],"method":[26],"often":[27],"falls":[28],"short":[29],"in":[30,91],"complex":[31],"tasks":[33],"and":[34,119,128,163],"even":[35],"leads":[36],"model":[38],"collapse.":[39],"Through":[40],"a":[41,102],"series":[42],"of":[43],"preliminary":[44],"analyses,":[45],"reveal":[47],"two":[48,96],"problems:":[49],"(1)":[50],"imbalance,":[52],"where":[53,69],"most":[54],"samples":[56,64,72],"are":[57,65,77],"simple,":[58],"but":[59,189],"challenging":[61],"yet":[62],"crucial":[63],"scarce;":[66],"(2)":[67],"overthinking,":[68],"many":[70],"undesired":[71,132],"with":[73],"redundant":[74],"steps":[76],"used":[78],"for":[79,116],"self-training.":[80],"To":[81],"end,":[83],"propose":[85,145],"HSIR,":[86],"which":[87,142],"effectively":[88,176],"Harnesses":[89],"Self-Improvement":[90],"Reasoning":[93],"via":[95,166],"simple-yet-effective":[97],"approaches.":[98],"Specifically,":[99],"HSIR":[100,136,173],"introduces":[101],"verify-then-exit":[103],"sampling":[104],"strategy":[105],"mitigate":[107],"imbalance":[109],"efficiently":[111],"collecting":[112],"more":[113],"accurate":[114],"solutions":[115],"difficult":[117],"queries,":[118],"designs":[120],"an":[121,147,157],"Intrinsic":[122],"Diversity":[123],"score":[124],"quantify":[126],"overthinking":[127],"filter":[129],"out":[130],"solutions.":[133],"We":[134],"apply":[135],"various":[138],"post-training":[139],"paradigms,":[140],"among":[141],"further":[144],"H-GRPO,":[146],"enhanced":[148],"GRPO":[149],"algorithm":[150],"leverages":[152],"intrinsic":[154],"diversity":[155],"reward":[159],"encourage":[161],"concise":[162],"diverse":[164],"reinforcement":[167],"learning.":[168],"Extensive":[169],"results":[170],"show":[171],"not":[174],"only":[175],"enhances":[177],"performance,":[180],"i.e.,":[181],"bringing":[182],"up":[183,198],"+10.9%":[185],"average":[186],"performance":[187],"gains,":[188],"also":[190],"significantly":[191],"improves":[192],"efficiency":[195],"reducing":[197],"42.4%":[200],"relative":[201],"inference":[202],"overhead.":[203]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-27T00:00:00"}