{"id":"https://openalex.org/W4414460334","doi":"https://doi.org/10.48550/arxiv.2508.12116","title":"DynamixSFT: Dynamic Mixture Optimization of Instruction Tuning Collections","display_name":"DynamixSFT: Dynamic Mixture Optimization of Instruction Tuning Collections","publication_year":2025,"publication_date":"2025-08-16","ids":{"openalex":"https://openalex.org/W4414460334","doi":"https://doi.org/10.48550/arxiv.2508.12116"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2508.12116","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2508.12116","pdf_url":"https://arxiv.org/pdf/2508.12116","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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"},"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2508.12116","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5023169667","display_name":"Haebin Shin","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Shin, Haebin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119713564","display_name":"Lei Ji","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ji, Lei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119713565","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/A5119713566","display_name":"Zhiwei Yu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu, Zhiwei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119713567","display_name":"Qi Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Qi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5119713568","display_name":"Yeyun Gong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gong, Yeyun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5023169667"],"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/T11338","display_name":"Advancements in Photolithography Techniques","score":0.9592000246047974,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11338","display_name":"Advancements in Photolithography Techniques","score":0.9592000246047974,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/adaptive-sampling","display_name":"Adaptive sampling","score":0.616100013256073},{"id":"https://openalex.org/keywords/mixture-model","display_name":"Mixture model","score":0.5824000239372253},{"id":"https://openalex.org/keywords/sampling","display_name":"Sampling (signal processing)","score":0.5527999997138977},{"id":"https://openalex.org/keywords/data-collection","display_name":"Data collection","score":0.3089999854564667},{"id":"https://openalex.org/keywords/current","display_name":"Current (fluid)","score":0.30219998955726624},{"id":"https://openalex.org/keywords/importance-sampling","display_name":"Importance sampling","score":0.30090001225471497}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7972999811172485},{"id":"https://openalex.org/C2781395549","wikidata":"https://www.wikidata.org/wiki/Q4680762","display_name":"Adaptive sampling","level":3,"score":0.616100013256073},{"id":"https://openalex.org/C61224824","wikidata":"https://www.wikidata.org/wiki/Q2260434","display_name":"Mixture model","level":2,"score":0.5824000239372253},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.5527999997138977},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.45669999718666077},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.38940000534057617},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3562000095844269},{"id":"https://openalex.org/C133462117","wikidata":"https://www.wikidata.org/wiki/Q4929239","display_name":"Data collection","level":2,"score":0.3089999854564667},{"id":"https://openalex.org/C148043351","wikidata":"https://www.wikidata.org/wiki/Q4456944","display_name":"Current (fluid)","level":2,"score":0.30219998955726624},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.30090001225471497},{"id":"https://openalex.org/C2778915421","wikidata":"https://www.wikidata.org/wiki/Q3643177","display_name":"Performance improvement","level":2,"score":0.2759000062942505},{"id":"https://openalex.org/C75684735","wikidata":"https://www.wikidata.org/wiki/Q858810","display_name":"Big data","level":2,"score":0.2581999897956848},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.2574000060558319}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2508.12116","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2508.12116","pdf_url":"https://arxiv.org/pdf/2508.12116","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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.2508.12116","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2508.12116","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":"article"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2508.12116","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2508.12116","pdf_url":"https://arxiv.org/pdf/2508.12116","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"As":[0],"numerous":[1],"instruction-tuning":[2,34,108],"datasets":[3],"continue":[4],"to":[5,60,91,102,113,129],"emerge":[6],"during":[7],"the":[8,40,56,61,67,73,88,93,103,134],"post-training":[9],"stage,":[10],"dynamically":[11],"balancing":[12],"and":[13,30,47,70,127],"optimizing":[14],"their":[15],"mixtures":[16],"has":[17],"become":[18],"a":[19,28,43,49,80,114,124],"critical":[20],"challenge.":[21],"To":[22],"address":[23],"this,":[24],"we":[25,122],"propose":[26],"DynamixSFT,":[27],"dynamic":[29],"automated":[31],"method":[32],"for":[33],"dataset":[35,63,89],"mixture":[36],"optimization.":[37],"We":[38],"formulate":[39],"problem":[41],"as":[42],"multi-armed":[44],"bandit":[45],"setup":[46],"introduce":[48],"Prior-scaled":[50],"Boltzmann":[51],"Exploration":[52],"that":[53],"softly":[54],"anchors":[55],"updated":[57,78],"sampling":[58],"distribution":[59],"original":[62],"proportions,":[64],"thereby":[65],"preserving":[66],"inherent":[68],"diversity":[69],"coverage":[71],"of":[72,137],"collection.":[74],"Sampling":[75],"probabilities":[76],"are":[77],"using":[79],"lightweight":[81],"1-Step":[82],"Look-ahead":[83],"Reward,":[84],"reflecting":[85],"how":[86],"much":[87],"contributes":[90],"improving":[92],"model's":[94],"performance":[95,116],"at":[96],"its":[97],"current":[98],"state.":[99],"When":[100],"applied":[101],"Tulu-v2-mixture":[104],"collection":[105],"comprising":[106],"16":[107],"datasets,":[109],"DynamixSFT":[110],"achieves":[111],"up":[112],"2.2%":[115],"improvement":[117],"across":[118],"10":[119],"benchmarks.":[120],"Furthermore,":[121],"provide":[123],"comprehensive":[125],"analysis":[126],"visualizations":[128],"offer":[130],"deeper":[131],"insights":[132],"into":[133],"adaptive":[135],"dynamics":[136],"our":[138],"method.":[139]},"counts_by_year":[],"updated_date":"2026-05-05T08:41:31.759640","created_date":"2025-10-10T00:00:00"}