{"id":"https://openalex.org/W7166147437","doi":"https://doi.org/10.48550/arxiv.2606.26671","title":"NebulaExp-8B: An Empirical Post-Training Pipeline via Full-Scale Ablation Research","display_name":"NebulaExp-8B: An Empirical Post-Training Pipeline via Full-Scale Ablation Research","publication_year":2026,"publication_date":"2026-06-25","ids":{"openalex":"https://openalex.org/W7166147437","doi":"https://doi.org/10.48550/arxiv.2606.26671"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.26671","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.26671","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.2606.26671","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5024679777","display_name":"Qiaobo Hao","orcid":"https://orcid.org/0000-0002-0177-7263"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hao, Qiaobo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5010505918","display_name":"Yangqian Wu","orcid":"https://orcid.org/0000-0003-3914-7395"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Yangqian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139419315","display_name":"Shunyi Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Shunyi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139425221","display_name":"Zhongjian Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Zhongjian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139453165","display_name":"Ziqun Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Ziqun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5121999607","display_name":"Y He","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"He, Yayin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139428218","display_name":"Muqing Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Muqing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139454047","display_name":"Chen Zhong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhong, Chen","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":0.39750000834465027,"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":0.39750000834465027,"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":0.20600000023841858,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.10729999840259552,"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":0.6945000290870667},{"id":"https://openalex.org/keywords/baseline","display_name":"Baseline (sea)","score":0.6413999795913696},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.6355999708175659},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.6315000057220459},{"id":"https://openalex.org/keywords/dependency","display_name":"Dependency (UML)","score":0.51419997215271},{"id":"https://openalex.org/keywords/raw-data","display_name":"Raw data","score":0.43970000743865967},{"id":"https://openalex.org/keywords/relevance","display_name":"Relevance (law)","score":0.4383000135421753},{"id":"https://openalex.org/keywords/point","display_name":"Point (geometry)","score":0.4334000051021576},{"id":"https://openalex.org/keywords/f1-score","display_name":"F1 score","score":0.3686000108718872},{"id":"https://openalex.org/keywords/stack","display_name":"Stack (abstract data type)","score":0.36169999837875366}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7455000281333923},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.6945000290870667},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.6413999795913696},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.6355999708175659},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.6315000057220459},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5713000297546387},{"id":"https://openalex.org/C19768560","wikidata":"https://www.wikidata.org/wiki/Q320727","display_name":"Dependency (UML)","level":2,"score":0.51419997215271},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4772999882698059},{"id":"https://openalex.org/C132964779","wikidata":"https://www.wikidata.org/wiki/Q2110223","display_name":"Raw data","level":2,"score":0.43970000743865967},{"id":"https://openalex.org/C158154518","wikidata":"https://www.wikidata.org/wiki/Q7310970","display_name":"Relevance (law)","level":2,"score":0.4383000135421753},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.4334000051021576},{"id":"https://openalex.org/C148524875","wikidata":"https://www.wikidata.org/wiki/Q6975395","display_name":"F1 score","level":2,"score":0.3686000108718872},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3619000017642975},{"id":"https://openalex.org/C9395851","wikidata":"https://www.wikidata.org/wiki/Q177929","display_name":"Stack (abstract data type)","level":2,"score":0.36169999837875366},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.3528999984264374},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.3409999907016754},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.3402000069618225},{"id":"https://openalex.org/C4554734","wikidata":"https://www.wikidata.org/wiki/Q593744","display_name":"Knowledge base","level":2,"score":0.3330000042915344},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.31369999051094055},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.3082999885082245},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3052000105381012},{"id":"https://openalex.org/C175309249","wikidata":"https://www.wikidata.org/wiki/Q725864","display_name":"Pipeline transport","level":2,"score":0.29580000042915344},{"id":"https://openalex.org/C168065819","wikidata":"https://www.wikidata.org/wiki/Q845566","display_name":"Debugging","level":2,"score":0.29109999537467957},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.2856999933719635},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.2827000021934509},{"id":"https://openalex.org/C124304363","wikidata":"https://www.wikidata.org/wiki/Q673661","display_name":"Abstraction","level":2,"score":0.28220000863075256},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.28049999475479126},{"id":"https://openalex.org/C127162648","wikidata":"https://www.wikidata.org/wiki/Q16858953","display_name":"Channel (broadcasting)","level":2,"score":0.2793999910354614},{"id":"https://openalex.org/C49289754","wikidata":"https://www.wikidata.org/wiki/Q2267081","display_name":"Side channel attack","level":3,"score":0.26930001378059387},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.26589998602867126},{"id":"https://openalex.org/C165838908","wikidata":"https://www.wikidata.org/wiki/Q736777","display_name":"Calibration","level":2,"score":0.26179999113082886},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.26030001044273376},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.25600001215934753},{"id":"https://openalex.org/C21080849","wikidata":"https://www.wikidata.org/wiki/Q13611879","display_name":"Data point","level":2,"score":0.2540000081062317},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.25200000405311584}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.26671","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.26671","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.2606.26671","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.26671","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":[{"score":0.5127196311950684,"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Post-training":[0],"alignment":[1],"determines":[2],"the":[3,99,110,115,128,134,197,215],"reasoning":[4,142],"and":[5,24,31,56,70,77,95,159,168,212,225],"human":[6],"preference":[7],"following":[8],"capabilities":[9],"of":[10,65,118],"large":[11],"language":[12],"models,":[13],"yet":[14],"most":[15],"existing":[16],"works":[17],"withhold":[18],"detailed":[19],"data":[20,81],"construction,":[21],"filtering":[22],"rules":[23],"training":[25],"recipes,":[26],"which":[27],"hinders":[28],"community":[29],"reproducibility":[30],"lightweight":[32],"model":[33,51,55],"optimization.":[34],"This":[35,200],"work":[36],"presents":[37],"NebulaExp,":[38],"a":[39,62,71,203],"fully":[40,204],"transparent,":[41],"ablation-driven":[42],"post-training":[43,207],"pipeline":[44],"built":[45],"on":[46,152,175],"Qwen3-8B-base,":[47],"covering":[48],"two":[49],"orthogonal":[50],"branches:":[52],"general":[53,226],"instruct":[54],"complex":[57],"reasoning-specialized":[58],"model.":[59,199],"We":[60],"curate":[61],"raw":[63],"corpus":[64],"3.84M":[66],"multi-source":[67],"SFT":[68],"samples":[69,167],"200K":[72],"verifiable":[73],"RL":[74,139,170],"candidate":[75],"pool,":[76],"design":[78],"an":[79],"end-to-end":[80],"processing":[82],"stack":[83],"including":[84],"response":[85],"distillation,":[86],"multi-dimensional":[87],"cross-verification":[88],"filtering,":[89],"fine-grained":[90],"difficulty":[91],"grading,":[92],"task":[93,153],"classification":[94],"diversity-aware":[96],"sampling.":[97],"For":[98,133],"Instruct":[100],"branch,":[101,136],"our":[102],"three-stage":[103],"optimized":[104],"supervised":[105],"fine-tuning":[106],"approach":[107],"NebulaExp-Ins-SFT":[108],"improves":[109,140],"average":[111,129,141,179,192],"benchmark":[112],"score":[113,130,143],"from":[114,144],"55.01":[116],"baseline":[117,171],"Qwen3-8B-nothink":[119],"to":[120,131,146],"60.99.":[121],"GRPO":[122,138],"reinforcement":[123],"learning":[124],"then":[125],"further":[126],"elevates":[127],"61.85.":[132],"Reasoning":[135],"medium-difficulty":[137],"73.88":[145],"75.17.":[147],"To":[148],"address":[149],"RL's":[150],"dependency":[151],"verifiers,":[154],"we":[155],"systematically":[156],"investigate":[157],"single-teacher":[158],"multi-teacher":[160],"OPD":[161],"(MOPD):":[162],"utilizing":[163],"merely":[164,188],"4K":[165],"instruction-following":[166],"outperforms":[169],"by":[172,194],"3.26":[173],"points":[174],"IFEval":[176],"with":[177,187],"+4.43":[178],"overall":[180],"gain;":[181],"MOPD":[182],"fuses":[183],"four":[184],"domain-specialist":[185],"teachers":[186],"10K":[189],"samples,":[190],"lifting":[191],"performance":[193],"4.18":[195],"over":[196],"base":[198],"report":[201],"provides":[202],"reproducible":[205],"empirical":[206],"recipe":[208],"for":[209],"8B-scale":[210],"LLMs,":[211],"comprehensively":[213],"dissects":[214],"capability":[216],"trade-offs":[217],"among":[218],"instruction":[219],"adherence,":[220],"mathematical":[221],"reasoning,":[222],"code":[223],"generation":[224],"knowledge.":[227]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-27T00:00:00"}