{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:V4G67KJH5S2GQBC5XCYR35LW4M","short_pith_number":"pith:V4G67KJH","schema_version":"1.0","canonical_sha256":"af0defa927ecb468045db8b11df576e31c6bacd0962b167923af8e5608012887","source":{"kind":"arxiv","id":"2002.01808","version":5},"attestation_state":"computed","paper":{"title":"K-Adapter: Infusing Knowledge into Pre-Trained Models with Adapters","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Daxin Jiang, Duyu Tang, Guihong Cao, Jianshu Ji, Ming Zhou, Nan Duan, Ruize Wang, Xuanjing Huang, Zhongyu Wei","submitted_at":"2020-02-05T14:30:49Z","abstract_excerpt":"We study the problem of injecting knowledge into large pre-trained models like BERT and RoBERTa. Existing methods typically update the original parameters of pre-trained models when injecting knowledge. However, when multiple kinds of knowledge are injected, the historically injected knowledge would be flushed away. To address this, we propose K-Adapter, a framework that retains the original parameters of the pre-trained model fixed and supports the development of versatile knowledge-infused model. Taking RoBERTa as the backbone model, K-Adapter has a neural adapter for each kind of infused kn"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2002.01808","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-02-05T14:30:49Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"a68a5f433b2d5c0d358341f7aaa76ac50b70d7504e0e79c72f194ca09f972658","abstract_canon_sha256":"1048cd275d8561c4f3d3bff5322993beb3c767484c5f8c6f7470c32253639b5c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:02:02.234568Z","signature_b64":"v9L4QZUb98RS1Pwkvgu6KF3P+aeMgNd8FZ9lhQgsd5sqcn7ncLw2/80j5r01hl93B7bAO7X+t4ySC0uOpFvMDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"af0defa927ecb468045db8b11df576e31c6bacd0962b167923af8e5608012887","last_reissued_at":"2026-07-05T02:02:02.234108Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:02:02.234108Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"K-Adapter: Infusing Knowledge into Pre-Trained Models with Adapters","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Daxin Jiang, Duyu Tang, Guihong Cao, Jianshu Ji, Ming Zhou, Nan Duan, Ruize Wang, Xuanjing Huang, Zhongyu Wei","submitted_at":"2020-02-05T14:30:49Z","abstract_excerpt":"We study the problem of injecting knowledge into large pre-trained models like BERT and RoBERTa. Existing methods typically update the original parameters of pre-trained models when injecting knowledge. However, when multiple kinds of knowledge are injected, the historically injected knowledge would be flushed away. To address this, we propose K-Adapter, a framework that retains the original parameters of the pre-trained model fixed and supports the development of versatile knowledge-infused model. Taking RoBERTa as the backbone model, K-Adapter has a neural adapter for each kind of infused kn"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2002.01808","kind":"arxiv","version":5},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2002.01808/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2002.01808","created_at":"2026-07-05T02:02:02.234167+00:00"},{"alias_kind":"arxiv_version","alias_value":"2002.01808v5","created_at":"2026-07-05T02:02:02.234167+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2002.01808","created_at":"2026-07-05T02:02:02.234167+00:00"},{"alias_kind":"pith_short_12","alias_value":"V4G67KJH5S2G","created_at":"2026-07-05T02:02:02.234167+00:00"},{"alias_kind":"pith_short_16","alias_value":"V4G67KJH5S2GQBC5","created_at":"2026-07-05T02:02:02.234167+00:00"},{"alias_kind":"pith_short_8","alias_value":"V4G67KJH","created_at":"2026-07-05T02:02:02.234167+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.20170","citing_title":"KoRe: Compact Knowledge Representations for Large Language Models","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02452","citing_title":"Position: How can Graphs Help Large Language Models?","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/V4G67KJH5S2GQBC5XCYR35LW4M","json":"https://pith.science/pith/V4G67KJH5S2GQBC5XCYR35LW4M.json","graph_json":"https://pith.science/api/pith-number/V4G67KJH5S2GQBC5XCYR35LW4M/graph.json","events_json":"https://pith.science/api/pith-number/V4G67KJH5S2GQBC5XCYR35LW4M/events.json","paper":"https://pith.science/paper/V4G67KJH"},"agent_actions":{"view_html":"https://pith.science/pith/V4G67KJH5S2GQBC5XCYR35LW4M","download_json":"https://pith.science/pith/V4G67KJH5S2GQBC5XCYR35LW4M.json","view_paper":"https://pith.science/paper/V4G67KJH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2002.01808&json=true","fetch_graph":"https://pith.science/api/pith-number/V4G67KJH5S2GQBC5XCYR35LW4M/graph.json","fetch_events":"https://pith.science/api/pith-number/V4G67KJH5S2GQBC5XCYR35LW4M/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/V4G67KJH5S2GQBC5XCYR35LW4M/action/timestamp_anchor","attest_storage":"https://pith.science/pith/V4G67KJH5S2GQBC5XCYR35LW4M/action/storage_attestation","attest_author":"https://pith.science/pith/V4G67KJH5S2GQBC5XCYR35LW4M/action/author_attestation","sign_citation":"https://pith.science/pith/V4G67KJH5S2GQBC5XCYR35LW4M/action/citation_signature","submit_replication":"https://pith.science/pith/V4G67KJH5S2GQBC5XCYR35LW4M/action/replication_record"}},"created_at":"2026-07-05T02:02:02.234167+00:00","updated_at":"2026-07-05T02:02:02.234167+00:00"}