{"id":"https://openalex.org/W7103879537","doi":"https://doi.org/10.48550/arxiv.2510.27598","title":"InnovatorBench: Evaluating Agents' Ability to Conduct Innovative LLM Research","display_name":"InnovatorBench: Evaluating Agents' Ability to Conduct Innovative LLM Research","publication_year":2025,"publication_date":"2025-10-31","ids":{"openalex":"https://openalex.org/W7103879537","doi":"https://doi.org/10.48550/arxiv.2510.27598"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2510.27598","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.27598","pdf_url":"https://arxiv.org/pdf/2510.27598","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/2510.27598","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Wu, Yunze","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Yunze","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Fu, Dayuan","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fu, Dayuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Si, Weiye","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Si, Weiye","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Huang, Zhen","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Zhen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Jiang, Mohan","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiang, Mohan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Li, Keyu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Keyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Xia, Shijie","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xia, Shijie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Sun, Jie","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sun, Jie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Xu, Tianze","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Tianze","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Hu, Xiangkun","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hu, Xiangkun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Lu, Pengrui","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lu, Pengrui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Cai, Xiaojie","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cai, Xiaojie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Ye, Lyumanshan","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ye, Lyumanshan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Zhu, Wenhong","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhu, Wenhong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Xiao, Yang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiao, Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Liu, Pengfei","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Pengfei","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/T13904","display_name":"Artificial Intelligence Applications","score":0.6664999723434448,"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/T13904","display_name":"Artificial Intelligence Applications","score":0.6664999723434448,"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.13420000672340393,"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/T10028","display_name":"Topic Modeling","score":0.05640000104904175,"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/ai-agents","display_name":"AI agents","score":0.6309999823570251},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.8029999732971191},{"id":"https://openalex.org/keywords/benchmarking","display_name":"benchmarking","score":0.5609999895095825},{"id":"https://openalex.org/keywords/react-agents","display_name":"ReAct agents","score":0.6779999732971191},{"id":"https://openalex.org/keywords/hypothesis-generation","display_name":"hypothesis generation","score":0.5989999771118164},{"id":"https://openalex.org/keywords/experimental-design","display_name":"experimental design","score":0.8429999947547913},{"id":"https://openalex.org/keywords/data-augmentation","display_name":"data augmentation","score":0.6850000023841858},{"id":"https://openalex.org/keywords/loss-function-design","display_name":"loss function design","score":0.8109999895095825},{"id":"https://openalex.org/keywords/reward-function-design","display_name":"reward function design","score":0.9919999837875366},{"id":"https://openalex.org/keywords/scaffold-fabrication","display_name":"scaffold fabrication","score":0.5709999799728394},{"id":"https://openalex.org/keywords/long-horizon-decision-making","display_name":"long-horizon decision-making","score":0.7170000076293945}],"concepts":[{"id":"https://openalex.org/C160145156","wikidata":"https://www.wikidata.org/wiki/Q778586","display_name":"Executable","level":2,"score":0.7221999764442444},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7185999751091003},{"id":"https://openalex.org/C55282118","wikidata":"https://www.wikidata.org/wiki/Q252683","display_name":"Snapshot (computer storage)","level":2,"score":0.49079999327659607},{"id":"https://openalex.org/C151319957","wikidata":"https://www.wikidata.org/wiki/Q752739","display_name":"Asynchronous communication","level":2,"score":0.42809998989105225},{"id":"https://openalex.org/C2778571376","wikidata":"https://www.wikidata.org/wiki/Q1355821","display_name":"Frontier","level":2,"score":0.4142000079154968},{"id":"https://openalex.org/C206345919","wikidata":"https://www.wikidata.org/wiki/Q20380951","display_name":"Resource (disambiguation)","level":2,"score":0.4129999876022339},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.4072999954223633},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.34060001373291016},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3346000015735626},{"id":"https://openalex.org/C56739046","wikidata":"https://www.wikidata.org/wiki/Q192060","display_name":"Knowledge management","level":1,"score":0.3296000063419342},{"id":"https://openalex.org/C195094911","wikidata":"https://www.wikidata.org/wiki/Q14167904","display_name":"Process management","level":1,"score":0.30959999561309814},{"id":"https://openalex.org/C184356942","wikidata":"https://www.wikidata.org/wiki/Q830382","display_name":"Best practice","level":2,"score":0.2888999879360199},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.2822999954223633},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.2777999937534332},{"id":"https://openalex.org/C133462117","wikidata":"https://www.wikidata.org/wiki/Q4929239","display_name":"Data collection","level":2,"score":0.2696000039577484},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.2696000039577484},{"id":"https://openalex.org/C114073186","wikidata":"https://www.wikidata.org/wiki/Q2631895","display_name":"Automated planning and scheduling","level":2,"score":0.2639999985694885},{"id":"https://openalex.org/C53173841","wikidata":"https://www.wikidata.org/wiki/Q1897444","display_name":"Action research","level":2,"score":0.25529998540878296},{"id":"https://openalex.org/C177212765","wikidata":"https://www.wikidata.org/wiki/Q627335","display_name":"Workflow","level":2,"score":0.25450000166893005}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2510.27598","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.27598","pdf_url":"https://arxiv.org/pdf/2510.27598","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.2510.27598","doi":"https://doi.org/10.48550/arxiv.2510.27598","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2510.27598","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":"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2510.27598","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.27598","pdf_url":"https://arxiv.org/pdf/2510.27598","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/W4414587939","https://openalex.org/W7117547917","https://openalex.org/W6891935834","https://openalex.org/W7163890116","https://openalex.org/W7090704326","https://openalex.org/W7118539294","https://openalex.org/W4416967341","https://openalex.org/W7128395166","https://openalex.org/W7130324807","https://openalex.org/W7127605006"],"abstract_inverted_index":{"AI":[0],"agents":[1,40,158],"could":[2],"accelerate":[3],"scientific":[4],"discovery":[5],"by":[6],"automating":[7],"hypothesis":[8],"formation,":[9],"experiment":[10],"design,":[11],"coding,":[12],"execution,":[13,93],"and":[14,60,67,74,91,96,121,142,152,174],"analysis,":[15],"yet":[16],"existing":[17],"benchmarks":[18],"probe":[19],"narrow":[20],"skills":[21],"in":[22,132],"simplified":[23],"settings.":[24],"To":[25,76],"address":[26],"this":[27],"gap,":[28],"we":[29,80],"introduce":[30],"InnovatorBench,":[31,169],"a":[32,83,102],"benchmark-platform":[33],"pair":[34],"for":[35],"realistic,":[36],"end-to-end":[37],"assessment":[38,68],"of":[39,69,178,185],"performing":[41],"Large":[42],"Language":[43],"Model":[44],"(LLM)":[45],"research.":[46],"It":[47],"comprises":[48],"20":[49],"tasks":[50,141],"spanning":[51],"Data":[52],"Construction,":[53,62],"Filtering,":[54],"Augmentation,":[55],"Loss":[56],"Design,":[57,59],"Reward":[58],"Scaffold":[61],"which":[63],"require":[64,159],"runnable":[65],"artifacts":[66],"correctness,":[70],"performance,":[71],"output":[72],"quality,":[73],"uncertainty.":[75],"support":[77],"agent":[78,105],"operation,":[79],"develop":[81],"ResearchGym,":[82],"research":[84,134,187],"environment":[85],"offering":[86],"rich":[87],"action":[88],"spaces,":[89],"distributed":[90],"long-horizon":[92,143],"asynchronous":[94],"monitoring,":[95],"snapshot":[97],"saving.":[98],"We":[99],"also":[100],"implement":[101],"lightweight":[103],"ReAct":[104],"that":[106,126],"couples":[107],"explicit":[108],"reasoning":[109],"with":[110,138],"executable":[111],"planning":[112],"using":[113],"frontier":[114,128],"models":[115,129],"such":[116,146],"as":[117,147],"Claude-4,":[118],"GPT-5,":[119],"GLM-4.5,":[120],"Kimi-K2.":[122],"Our":[123],"experiments":[124],"demonstrate":[125],"while":[127],"show":[130],"promise":[131],"code-driven":[133],"tasks,":[135],"they":[136],"struggle":[137],"fragile":[139],"algorithm-related":[140],"decision":[144],"making,":[145],"impatience,":[148],"poor":[149],"resource":[150],"management,":[151],"overreliance":[153],"on":[154,168],"template-based":[155],"reasoning.":[156],"Furthermore,":[157],"over":[160],"11":[161],"hours":[162],"to":[163,180],"achieve":[164],"their":[165],"best":[166],"performance":[167],"underscoring":[170],"the":[171,176,182],"benchmark's":[172],"difficulty":[173],"showing":[175],"potential":[177],"InnovatorBench":[179],"be":[181],"next":[183],"generation":[184],"code-based":[186],"benchmark.":[188]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2025-11-05T00:00:00"}