{"id":"https://openalex.org/W7164878124","doi":"https://doi.org/10.48550/arxiv.2606.16517","title":"How Post-Training Shapes Biological Reasoning Models","display_name":"How Post-Training Shapes Biological Reasoning Models","publication_year":2026,"publication_date":"2026-06-15","ids":{"openalex":"https://openalex.org/W7164878124","doi":"https://doi.org/10.48550/arxiv.2606.16517"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.16517","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.16517","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.2606.16517","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5138731384","display_name":"Lukas Fesser","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fesser, Lukas","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5029424597","display_name":"Hanlin Zhang","orcid":"https://orcid.org/0000-0001-5065-4086"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Hanlin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5033336429","display_name":"M. Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Michelle M.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138753411","display_name":"Eric Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Eric","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5005058150","display_name":"Bryan Perozzi","orcid":"https://orcid.org/0009-0002-1639-2056"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Perozzi, Bryan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5047463591","display_name":"Shekoofeh Azizi","orcid":"https://orcid.org/0000-0002-7447-6031"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Azizi, Shekoofeh","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138757348","display_name":"Sham M. Kakade","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kakade, Sham M.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5138726632","display_name":"Marinka Zitnik","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zitnik, Marinka","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/T12254","display_name":"Machine Learning in Bioinformatics","score":0.1808999925851822,"subfield":{"id":"https://openalex.org/subfields/1312","display_name":"Molecular Biology"},"field":{"id":"https://openalex.org/fields/13","display_name":"Biochemistry, Genetics and Molecular Biology"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},"topics":[{"id":"https://openalex.org/T12254","display_name":"Machine Learning in Bioinformatics","score":0.1808999925851822,"subfield":{"id":"https://openalex.org/subfields/1312","display_name":"Molecular Biology"},"field":{"id":"https://openalex.org/fields/13","display_name":"Biochemistry, Genetics and Molecular Biology"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},{"id":"https://openalex.org/T11642","display_name":"Genomics and Rare Diseases","score":0.10459999740123749,"subfield":{"id":"https://openalex.org/subfields/1311","display_name":"Genetics"},"field":{"id":"https://openalex.org/fields/13","display_name":"Biochemistry, Genetics and Molecular Biology"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},{"id":"https://openalex.org/T12090","display_name":"Language and cultural evolution","score":0.09099999815225601,"subfield":{"id":"https://openalex.org/subfields/3316","display_name":"Cultural Studies"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/generalization","display_name":"Generalization","score":0.816100001335144},{"id":"https://openalex.org/keywords/adaptation","display_name":"Adaptation (eye)","score":0.4690000116825104},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.44350001215934753},{"id":"https://openalex.org/keywords/variation","display_name":"Variation (astronomy)","score":0.3571000099182129},{"id":"https://openalex.org/keywords/mathematical-model","display_name":"Mathematical model","score":0.3463999927043915},{"id":"https://openalex.org/keywords/monotonic-function","display_name":"Monotonic function","score":0.3452000021934509},{"id":"https://openalex.org/keywords/multiple-models","display_name":"Multiple Models","score":0.29649999737739563}],"concepts":[{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.816100001335144},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6460000276565552},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.628600001335144},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.4690000116825104},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.44350001215934753},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.435699999332428},{"id":"https://openalex.org/C2778334786","wikidata":"https://www.wikidata.org/wiki/Q1586270","display_name":"Variation (astronomy)","level":2,"score":0.3571000099182129},{"id":"https://openalex.org/C76969082","wikidata":"https://www.wikidata.org/wiki/Q486902","display_name":"Mathematical model","level":2,"score":0.3463999927043915},{"id":"https://openalex.org/C72169020","wikidata":"https://www.wikidata.org/wiki/Q194404","display_name":"Monotonic function","level":2,"score":0.3452000021934509},{"id":"https://openalex.org/C2779714256","wikidata":"https://www.wikidata.org/wiki/Q25305062","display_name":"Multiple Models","level":2,"score":0.29649999737739563},{"id":"https://openalex.org/C2780966255","wikidata":"https://www.wikidata.org/wiki/Q5474306","display_name":"Foundation (evidence)","level":2,"score":0.28450000286102295},{"id":"https://openalex.org/C136197465","wikidata":"https://www.wikidata.org/wiki/Q1729295","display_name":"Variety (cybernetics)","level":2,"score":0.2831999957561493},{"id":"https://openalex.org/C66024118","wikidata":"https://www.wikidata.org/wiki/Q1122506","display_name":"Computational model","level":2,"score":0.2815999984741211},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.27239999175071716},{"id":"https://openalex.org/C2992562121","wikidata":"https://www.wikidata.org/wiki/Q3817808","display_name":"Scientific reasoning","level":2,"score":0.2597000002861023},{"id":"https://openalex.org/C195344581","wikidata":"https://www.wikidata.org/wiki/Q2555318","display_name":"Automated reasoning","level":2,"score":0.2533999979496002}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.16517","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.16517","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.2606.16517","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.16517","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":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Scientific":[0],"reasoning":[1,32,62,156],"models":[2,7,10,22,63,110,129],"for":[3],"biology":[4],"combine":[5],"language":[6],"with":[8,111,141,161],"foundation":[9],"trained":[11],"on":[12,169],"multimodal":[13],"biological":[14,61,112,155],"data,":[15],"including":[16],"DNA,":[17],"RNA,":[18],"and":[19,33,44,52,56,75,83,126,147,190],"proteins.":[20],"These":[21,151],"are":[23,173],"built":[24],"through":[25],"post-training,":[26],"yet":[27],"how":[28,170],"each":[29,90],"stage":[30,92],"shapes":[31],"generalization":[34,94],"remains":[35],"poorly":[36],"understood.":[37],"We":[38,87],"study":[39],"when":[40,45,135],"post-training":[41,91,177],"improves":[42,105,144],"performance":[43,107,118,122,146,167],"it":[46],"induces":[47],"over-specialization.":[48],"Across":[49],"genomics,":[50],"transcriptomics,":[51],"proteins,":[53],"we":[54],"train":[55],"evaluate":[57],"more":[58],"than":[59,100],"100":[60],"under":[64],"controlled":[65],"variation":[66],"in":[67,95],"backbone,":[68],"continued":[69],"pre-training":[70],"(CPT),":[71],"supervised":[72],"fine-tuning":[73],"(SFT),":[74],"reinforcement":[76],"learning":[77],"(RL),":[78],"measuring":[79],"both":[80],"in-domain":[81],"(ID)":[82],"out-of-domain":[84],"(OOD)":[85],"performance.":[86],"find":[88],"that":[89,154],"reshapes":[93],"a":[96],"distinct":[97],"way":[98],"rather":[99],"contributing":[101],"uniform":[102],"gains.":[103],"CPT":[104],"downstream":[106],"by":[108],"aligning":[109],"language.":[113],"SFT":[114,139],"consistently":[115],"increases":[116],"ID":[117],"but":[119],"causes":[120],"OOD":[121,145],"to":[123,137],"peak":[124],"early":[125],"decline":[127],"as":[128],"fit":[130],"the":[131,179],"training":[132,171],"distribution.":[133],"RL,":[134],"applied":[136],"strong":[138],"checkpoints":[140],"aligned":[142],"rewards,":[143],"partially":[148],"recovers":[149],"generalization.":[150],"results":[152],"show":[153],"does":[157],"not":[158],"improve":[159],"monotonically":[160],"additional":[162],"supervision":[163],"or":[164],"compute.":[165],"Instead,":[166],"depends":[168],"stages":[172],"composed.":[174],"Under":[175],"fixed":[176],"budgets,":[178],"strongest":[180],"ID-OOD":[181],"trade-off":[182],"comes":[183],"from":[184],"brief":[185],"SFT,":[186],"larger":[187],"RL":[188],"allocations,":[189],"asymmetric":[191],"adaptation":[192],"capacity":[193],"across":[194],"stages.":[195]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-17T00:00:00"}