{"id":"https://openalex.org/W7153186253","doi":"https://doi.org/10.48550/arxiv.2604.08423","title":"Synthetic Data for any Differentiable Target","display_name":"Synthetic Data for any Differentiable Target","publication_year":2026,"publication_date":"2026-04-09","ids":{"openalex":"https://openalex.org/W7153186253","doi":"https://doi.org/10.48550/arxiv.2604.08423"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.08423","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.08423","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.2604.08423","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5026811404","display_name":"Tristan Thrush","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Thrush, Tristan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133354292","display_name":"Sung Keun Park","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Park, Sung Min","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133327805","display_name":"Herman Brunborg","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Brunborg, Herman","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133356816","display_name":"Luke Bailey","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bailey, Luke","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5023564562","display_name":"Marcel Roed","orcid":"https://orcid.org/0000-0002-1369-8933"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Roed, Marcel","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030721728","display_name":"Neil Band","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Band, Neil","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133349927","display_name":"Christopher Potts","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Potts, Christopher","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5133380807","display_name":"Tatsunori Hashimoto","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hashimoto, Tatsunori","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":0.2542000114917755,"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":0.2542000114917755,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.08720000088214874,"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/T12072","display_name":"Machine Learning and Algorithms","score":0.0869000032544136,"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/generator","display_name":"Generator (circuit theory)","score":0.724399983882904},{"id":"https://openalex.org/keywords/synthetic-data","display_name":"Synthetic data","score":0.6732000112533569},{"id":"https://openalex.org/keywords/metric","display_name":"Metric (unit)","score":0.6626999974250793},{"id":"https://openalex.org/keywords/differentiable-function","display_name":"Differentiable function","score":0.5934000015258789},{"id":"https://openalex.org/keywords/gradient-method","display_name":"Gradient method","score":0.4316999912261963},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.4311999976634979},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.38190001249313354}],"concepts":[{"id":"https://openalex.org/C2780992000","wikidata":"https://www.wikidata.org/wiki/Q17016113","display_name":"Generator (circuit theory)","level":3,"score":0.724399983882904},{"id":"https://openalex.org/C160920958","wikidata":"https://www.wikidata.org/wiki/Q7662746","display_name":"Synthetic data","level":2,"score":0.6732000112533569},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.6626999974250793},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6022999882698059},{"id":"https://openalex.org/C202615002","wikidata":"https://www.wikidata.org/wiki/Q783507","display_name":"Differentiable function","level":2,"score":0.5934000015258789},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.529699981212616},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.45339998602867126},{"id":"https://openalex.org/C115680565","wikidata":"https://www.wikidata.org/wiki/Q5977448","display_name":"Gradient method","level":2,"score":0.4316999912261963},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.4311999976634979},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.38190001249313354},{"id":"https://openalex.org/C117896860","wikidata":"https://www.wikidata.org/wiki/Q11376","display_name":"Acceleration","level":2,"score":0.3537999987602234},{"id":"https://openalex.org/C2781170535","wikidata":"https://www.wikidata.org/wiki/Q30587856","display_name":"Noisy data","level":2,"score":0.34529998898506165},{"id":"https://openalex.org/C149364088","wikidata":"https://www.wikidata.org/wiki/Q185917","display_name":"Translation (biology)","level":4,"score":0.2978000044822693},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2904999852180481},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.28529998660087585},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.2761000096797943},{"id":"https://openalex.org/C21080849","wikidata":"https://www.wikidata.org/wiki/Q13611879","display_name":"Data point","level":2,"score":0.2687000036239624},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.2551000118255615}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.08423","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.08423","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.2604.08423","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.08423","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":{"What":[0],"are":[1],"the":[2,19,51,91,96,102,118,132,148,173],"limits":[3],"of":[4,35,44,61,104,167],"controlling":[5],"language":[6,157],"models":[7],"via":[8,73],"synthetic":[9,28,97,194],"training":[10,195],"data?":[11],"We":[12,84,141],"develop":[13],"a":[14,33,45,58,127,155,161,183],"reinforcement":[15],"learning":[16],"(RL)":[17],"primitive,":[18],"Dataset":[20],"Policy":[21],"Gradient":[22],"(DPG),":[23],"which":[24],"can":[25,116,146],"precisely":[26],"optimize":[27],"data":[29,71,98],"generators":[30],"to":[31,54,124,150],"produce":[32,160],"dataset":[34],"targeted":[36],"examples.":[37,196],"When":[38],"used":[39],"for":[40,95,188],"supervised":[41],"fine-tuning":[42],"(SFT)":[43],"target":[46,52,119],"model,":[47],"these":[48,168],"examples":[49],"cause":[50,117,147],"model":[53,190],"do":[55],"well":[56],"on":[57,112],"differentiable":[59],"metric":[60],"our":[62],"choice.":[63],"Our":[64],"approach":[65],"achieves":[66],"this":[67,87],"by":[68],"taking":[69],"exact":[70],"attribution":[72],"higher-order":[74],"gradients":[75],"and":[76,135,158,185],"using":[77,109,192],"those":[78],"scores":[79],"as":[80],"policy":[81],"gradient":[82,94],"rewards.":[83],"prove":[85],"that":[86,144,180],"procedure":[88],"closely":[89],"approximates":[90],"true,":[92],"intractable":[93],"generator.":[99],"To":[100],"illustrate":[101],"potential":[103],"DPG,":[105],"we":[106,115,145],"show":[107,143],"that,":[108],"only":[110,193],"SFT":[111],"generated":[113],"examples,":[114],"model's":[120],"LM":[121],"head":[122],"weights":[123],"(1)":[125],"embed":[126,131],"QR":[128],"code,":[129],"(2)":[130],"pattern":[133],"$\\texttt{67}$,":[134],"(3)":[136],"have":[137],"lower":[138],"$\\ell^2$":[139],"norm.":[140],"additionally":[142],"generator":[149],"(4)":[151],"rephrase":[152],"inputs":[153],"in":[154,172],"new":[156],"(5)":[159],"specific":[162],"UUID,":[163],"even":[164],"though":[165],"neither":[166],"objectives":[169],"is":[170,182],"conveyed":[171],"generator's":[174],"input":[175],"prompts.":[176],"These":[177],"findings":[178],"suggest":[179],"DPG":[181],"powerful":[184],"flexible":[186],"technique":[187],"shaping":[189],"properties":[191]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-04-11T00:00:00"}