{"id":"https://openalex.org/W7128518386","doi":"https://doi.org/10.48550/arxiv.2602.08321","title":"Improving Data and Reward Design for Scientific Reasoning in Large Language Models","display_name":"Improving Data and Reward Design for Scientific Reasoning in Large Language Models","publication_year":2026,"publication_date":"2026-02-09","ids":{"openalex":"https://openalex.org/W7128518386","doi":"https://doi.org/10.48550/arxiv.2602.08321"},"language":null,"primary_location":{"id":"pmh:doi:10.48550/arxiv.2602.08321","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"type":"article","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":null,"any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5125552892","display_name":"Zijie Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Zijie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125537544","display_name":"Zhenghao Lin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lin, Zhenghao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125496954","display_name":"Xiao Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Xiao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125569646","display_name":"Zhenzhong Lan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lan, Zhenzhong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125532930","display_name":"Yeyun Gong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gong, Yeyun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5125505168","display_name":"Peng cheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cheng, Peng","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":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.1713629,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"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":0.5741999745368958,"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":0.5741999745368958,"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/T11948","display_name":"Machine Learning in Materials Science","score":0.07270000129938126,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.05649999901652336,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/pipeline","display_name":"Pipeline (software)","score":0.6741999983787537},{"id":"https://openalex.org/keywords/workflow","display_name":"Workflow","score":0.6485000252723694},{"id":"https://openalex.org/keywords/operationalization","display_name":"Operationalization","score":0.5590999722480774},{"id":"https://openalex.org/keywords/bottleneck","display_name":"Bottleneck","score":0.534600019454956},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.4948999881744385},{"id":"https://openalex.org/keywords/heuristics","display_name":"Heuristics","score":0.4909000098705292},{"id":"https://openalex.org/keywords/scientific-reasoning","display_name":"Scientific reasoning","score":0.35350000858306885},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.3416999876499176}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7303000092506409},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.6741999983787537},{"id":"https://openalex.org/C177212765","wikidata":"https://www.wikidata.org/wiki/Q627335","display_name":"Workflow","level":2,"score":0.6485000252723694},{"id":"https://openalex.org/C9354725","wikidata":"https://www.wikidata.org/wiki/Q286017","display_name":"Operationalization","level":2,"score":0.5590999722480774},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.534600019454956},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4999000132083893},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.4948999881744385},{"id":"https://openalex.org/C127705205","wikidata":"https://www.wikidata.org/wiki/Q5748245","display_name":"Heuristics","level":2,"score":0.4909000098705292},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.39739999175071716},{"id":"https://openalex.org/C2992562121","wikidata":"https://www.wikidata.org/wiki/Q3817808","display_name":"Scientific reasoning","level":2,"score":0.35350000858306885},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.3416999876499176},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3411000072956085},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.33410000801086426},{"id":"https://openalex.org/C138379479","wikidata":"https://www.wikidata.org/wiki/Q1116876","display_name":"Scientific modelling","level":2,"score":0.33410000801086426},{"id":"https://openalex.org/C124304363","wikidata":"https://www.wikidata.org/wiki/Q673661","display_name":"Abstraction","level":2,"score":0.29010000824928284},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.289900004863739},{"id":"https://openalex.org/C111640148","wikidata":"https://www.wikidata.org/wiki/Q847349","display_name":"Rubric","level":2,"score":0.2761000096797943},{"id":"https://openalex.org/C75684735","wikidata":"https://www.wikidata.org/wiki/Q858810","display_name":"Big data","level":2,"score":0.2703000009059906},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.2669999897480011},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.25679999589920044}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.48550/arxiv.2602.08321","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},{"id":"doi:10.48550/arxiv.2602.08321","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.08321","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":"pmh:doi:10.48550/arxiv.2602.08321","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"sustainable_development_goals":[{"score":0.5263628363609314,"display_name":"Quality Education","id":"https://metadata.un.org/sdg/4"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Solving":[0],"open-ended":[1,72,133,176],"science":[2,43],"questions":[3,53,135],"remains":[4],"challenging":[5],"for":[6,28,71],"large":[7],"language":[8],"models,":[9],"particularly":[10],"due":[11],"to":[12,107,117],"inherently":[13],"unreliable":[14],"supervision":[15],"and":[16,25,65,123,153,166],"evaluation.":[17],"The":[18],"bottleneck":[19],"lies":[20],"in":[21,171,175],"the":[22,80,87,101,118],"data":[23,36,44,116],"construction":[24],"reward":[26],"design":[27],"scientific":[29,121,134,172],"post-training.":[30],"We":[31],"develop":[32],"a":[33],"large-scale,":[34],"systematic":[35],"processing":[37],"pipeline":[38,148],"that":[39,68],"transforms":[40],"heterogeneous":[41],"open-source":[42],"into":[45],"Dr.":[46,81,146],"SCI":[47,82,147],"dataset,":[48,77],"which":[49,85,99,113,127],"comprises":[50],"of":[51],"1M":[52],"across":[54],"eight":[55],"STEM":[56],"subjects,":[57],"with":[58,139],"explicit":[59,140],"verifiable/open-ended":[60],"splits,":[61],"scalable":[62],"difficulty":[63],"annotation,":[64],"fine-grained":[66],"rubrics":[67],"operationalize":[69],"evaluation":[70,138],"answers.":[73],"Building":[74],"on":[75,132,151,155],"this":[76],"we":[78],"propose":[79],"post-training":[83],"pipeline,":[84],"redesigns":[86],"standard":[88],"SFT":[89],"->":[90],"RL":[91],"workflow":[92],"through":[93],"three":[94],"components:":[95],"(i)":[96],"Exploration-Expanding":[97],"SFT,":[98],"broadens":[100],"model's":[102,119],"reasoning":[103],"pattern":[104],"coverage":[105],"prior":[106],"RL;":[108],"(ii)":[109],"Dynamic":[110],"Difficulty":[111],"Curriculum,":[112],"adapts":[114],"training":[115],"evolving":[120],"capability;":[122],"(iii)":[124],"SciRubric-Guided":[125],"RL,":[126],"enables":[128],"stable":[129],"reinforcement":[130],"learning":[131],"via":[136],"rubric-based":[137],"answer":[141],"correctness.":[142],"Qwen3-4B-Base":[143],"trained":[144],"using":[145],"achieves":[149],"63.2":[150],"GPQA-diamond":[152],"32.4":[154],"GPQA-general,":[156],"consistently":[157],"improves":[158],"over":[159],"strong":[160],"post-trained":[161],"baselines":[162],"such":[163],"as":[164],"o1-mini":[165],"GPT-4o,":[167],"demonstrating":[168],"substantial":[169],"gains":[170],"reasoning,":[173],"especially":[174],"settings.":[177]},"counts_by_year":[],"updated_date":"2026-07-15T18:14:33.161393","created_date":"2026-02-11T00:00:00"}