{"id":"https://openalex.org/W7203923645","doi":"https://doi.org/10.48550/arxiv.2608.20281","title":"Inject, Align, Recover: Staged Post-Training for Retrieval-Free Document Knowledge Internalization","display_name":"Inject, Align, Recover: Staged Post-Training for Retrieval-Free Document Knowledge Internalization","publication_year":2026,"publication_date":"2026-08-20","ids":{"openalex":"https://openalex.org/W7203923645","doi":"https://doi.org/10.48550/arxiv.2608.20281"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.20281","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.20281","pdf_url":"https://arxiv.org/pdf/2608.20281","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2608.20281","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137629515","display_name":"Qian Kou","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210100255","display_name":"Beijing Academy of Artificial Intelligence","ror":"https://ror.org/016a74861","country_code":"CN","type":"other","lineage":["https://openalex.org/I4210100255"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kou, Qian","raw_affiliation_strings":["Beijing Academy of Artificial Intelligence (BAAI)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beijing Academy of Artificial Intelligence (BAAI)","institution_ids":["https://openalex.org/I4210100255"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137668165","display_name":"Xiaofeng Shi","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210100255","display_name":"Beijing Academy of Artificial Intelligence","ror":"https://ror.org/016a74861","country_code":"CN","type":"other","lineage":["https://openalex.org/I4210100255"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shi, Xiaofeng","raw_affiliation_strings":["Beijing Academy of Artificial Intelligence (BAAI)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beijing Academy of Artificial Intelligence (BAAI)","institution_ids":["https://openalex.org/I4210100255"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137583774","display_name":"Xiaosong Qiu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210100255","display_name":"Beijing Academy of Artificial Intelligence","ror":"https://ror.org/016a74861","country_code":"CN","type":"other","lineage":["https://openalex.org/I4210100255"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Qiu, Xiaosong","raw_affiliation_strings":["Beijing Academy of Artificial Intelligence (BAAI)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beijing Academy of Artificial Intelligence (BAAI)","institution_ids":["https://openalex.org/I4210100255"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5124888193","display_name":"Hua Zhou","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210100255","display_name":"Beijing Academy of Artificial Intelligence","ror":"https://ror.org/016a74861","country_code":"CN","type":"other","lineage":["https://openalex.org/I4210100255"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhou, Hua","raw_affiliation_strings":["Beijing Academy of Artificial Intelligence (BAAI)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beijing Academy of Artificial Intelligence (BAAI)","institution_ids":["https://openalex.org/I4210100255"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I4210100255"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"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":0.9663000106811523,"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":0.9663000106811523,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.011599999852478504,"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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.006599999964237213,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/post-training","display_name":"post-training","score":0.7210000157356262},{"id":"https://openalex.org/keywords/knowledge-injection","display_name":"knowledge injection","score":0.718999981880188},{"id":"https://openalex.org/keywords/domain-adaptation","display_name":"domain adaptation","score":0.4819999933242798},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.6480000019073486},{"id":"https://openalex.org/keywords/instruction-tuning","display_name":"instruction tuning","score":0.34200000762939453},{"id":"https://openalex.org/keywords/lora","display_name":"LoRa","score":0.48500001430511475},{"id":"https://openalex.org/keywords/fine-tuning","display_name":"fine-tuning","score":0.546999990940094}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6916999816894531},{"id":"https://openalex.org/C2780615836","wikidata":"https://www.wikidata.org/wiki/Q2471869","display_name":"USable","level":2,"score":0.6470999717712402},{"id":"https://openalex.org/C4554734","wikidata":"https://www.wikidata.org/wiki/Q593744","display_name":"Knowledge base","level":2,"score":0.640999972820282},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5633000135421753},{"id":"https://openalex.org/C49929091","wikidata":"https://www.wikidata.org/wiki/Q1930471","display_name":"General knowledge","level":2,"score":0.5135999917984009},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.47769999504089355},{"id":"https://openalex.org/C117251300","wikidata":"https://www.wikidata.org/wiki/Q1849855","display_name":"Parametric statistics","level":2,"score":0.4691999852657318},{"id":"https://openalex.org/C36503486","wikidata":"https://www.wikidata.org/wiki/Q11235244","display_name":"Domain (mathematical analysis)","level":2,"score":0.4577000141143799},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.4449000060558319},{"id":"https://openalex.org/C207685749","wikidata":"https://www.wikidata.org/wiki/Q2088941","display_name":"Domain knowledge","level":2,"score":0.4447999894618988},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.40139999985694885},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.39340001344680786},{"id":"https://openalex.org/C44291984","wikidata":"https://www.wikidata.org/wiki/Q1074173","display_name":"Question answering","level":2,"score":0.367000013589859},{"id":"https://openalex.org/C115925183","wikidata":"https://www.wikidata.org/wiki/Q1412694","display_name":"Knowledge-based systems","level":2,"score":0.3328000009059906},{"id":"https://openalex.org/C34388435","wikidata":"https://www.wikidata.org/wiki/Q2267362","display_name":"Bounded function","level":2,"score":0.3244999945163727},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2939000129699707},{"id":"https://openalex.org/C105888452","wikidata":"https://www.wikidata.org/wiki/Q7565148","display_name":"Source document","level":2,"score":0.2863999903202057},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2847000062465668},{"id":"https://openalex.org/C2982832238","wikidata":"https://www.wikidata.org/wiki/Q5531640","display_name":"General purpose","level":2,"score":0.2833000123500824},{"id":"https://openalex.org/C120567893","wikidata":"https://www.wikidata.org/wiki/Q1582085","display_name":"Knowledge extraction","level":2,"score":0.27480000257492065},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2718000113964081},{"id":"https://openalex.org/C2777220311","wikidata":"https://www.wikidata.org/wiki/Q6423340","display_name":"Knowledge acquisition","level":2,"score":0.2660999894142151},{"id":"https://openalex.org/C24574437","wikidata":"https://www.wikidata.org/wiki/Q7135228","display_name":"Parametric model","level":3,"score":0.2603999972343445},{"id":"https://openalex.org/C2474386","wikidata":"https://www.wikidata.org/wiki/Q461183","display_name":"Text corpus","level":2,"score":0.258899986743927},{"id":"https://openalex.org/C84685590","wikidata":"https://www.wikidata.org/wiki/Q1540472","display_name":"Knowledge engineering","level":2,"score":0.25839999318122864},{"id":"https://openalex.org/C171686336","wikidata":"https://www.wikidata.org/wiki/Q3532085","display_name":"Topic model","level":2,"score":0.251800000667572},{"id":"https://openalex.org/C133462117","wikidata":"https://www.wikidata.org/wiki/Q4929239","display_name":"Data collection","level":2,"score":0.25060001015663147}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.20281","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.20281","pdf_url":"https://arxiv.org/pdf/2608.20281","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.20281","doi":"https://doi.org/10.48550/arxiv.2608.20281","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.20281","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.20281","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.20281","pdf_url":"https://arxiv.org/pdf/2608.20281","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7203923645.pdf","grobid_xml":"https://content.openalex.org/works/W7203923645.grobid-xml"},"referenced_works_count":4,"referenced_works":["https://openalex.org/W3001279689","https://openalex.org/W4388482013","https://openalex.org/W4388718089","https://openalex.org/W4417006854"],"related_works":["https://openalex.org/W3007672467","https://openalex.org/W4392930030","https://openalex.org/W4416034764","https://openalex.org/W7165530975","https://openalex.org/W4385570798","https://openalex.org/W4385573021","https://openalex.org/W4386113969","https://openalex.org/W4404784426","https://openalex.org/W4402699347","https://openalex.org/W7172431420"],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"often":[3],"fail":[4],"to":[5,103],"answer":[6],"questions":[7],"about":[8],"a":[9,32,50],"bounded":[10],"document":[11,28,57,130],"collection":[12],"when":[13],"the":[14,85,95,99,124,133,204],"source":[15,73],"documents":[16,74],"are":[17],"not":[18],"retrieved":[19],"at":[20],"inference":[21],"time.":[22],"We":[23,43],"study":[24],"this":[25],"setting":[26],"as":[27],"knowledge":[29,38,58],"internalization:":[30],"converting":[31],"fixed":[33],"corpus":[34],"into":[35,75],"usable":[36],"parametric":[37],"for":[39,128],"retrieval-free":[40,129],"question":[41],"answering.":[42],"propose":[44],"IAR":[45,122,136,200],"(Inject,":[46],"Align,":[47],"and":[48,63,78,111,113,118,163,174,182],"Recover),":[49],"three-stage":[51],"post-training":[52],"framework":[53],"that":[54,180,192],"separates":[55],"structured":[56],"injection,":[59],"QA":[60,90,161],"behavior":[61],"alignment,":[62],"general":[64,105,169,187,206],"ability":[65],"recovery.":[66],"Unlike":[67],"conventional":[68],"continued":[69],"pretraining,":[70],"Inject":[71],"converts":[72],"continuation,":[76],"rewrite,":[77],"instruction-conditioned":[79],"reconstruction":[80],"objectives.":[81],"Align":[82],"then":[83],"adapts":[84],"injected":[86],"model":[87,97,102,120],"with":[88,98,152],"answer-only":[89],"supervision,":[91],"while":[92],"Recover":[93],"merges":[94],"domain-adapted":[96],"base":[100],"instruction":[101],"recover":[104],"capabilities.":[106],"Across":[107],"Common":[108],"Corpus":[109],"(CC)":[110],"CCI,":[112],"across":[114,171],"Llama,":[115],"Phi,":[116],"Qwen,":[117],"SmolLM":[119],"families,":[121],"improves":[123,137],"domain-primary":[125],"domain-general":[126],"frontier":[127],"internalization.":[131],"In":[132],"main":[134],"comparison,":[135],"over":[138],"Vanilla":[139],"SFT":[140],"on":[141],"all":[142],"four":[143],"reported":[144],"metrics":[145],"in":[146,159,167],"7":[147],"of":[148,155,203],"8":[149],"dataset-model":[150],"settings,":[151],"average":[153],"gains":[154],"3.6":[156],"percentage":[157,165],"points":[158,166],"domain":[160,198],"accuracy":[162],"12.1":[164],"mean":[168],"performance":[170],"IFEval,":[172],"MMLU,":[173],"MSBench.":[175],"Extended":[176],"CC":[177],"baselines":[178],"show":[179],"LoRA":[181],"FAPM":[183],"can":[184],"win":[185],"individual":[186],"metrics,":[188],"but":[189],"among":[190],"methods":[191],"also":[193],"reach":[194],"leading":[195],"or":[196],"near-leading":[197],"internalization,":[199],"retains":[201],"one":[202],"strongest":[205],"profiles.":[207]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-08-22T00:00:00"}