{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:FKZDUEPIY5LRLHYLISX3XDLD3I","short_pith_number":"pith:FKZDUEPI","schema_version":"1.0","canonical_sha256":"2ab23a11e8c757159f0b44afbb8d63da3fbae53170b03e886b00fd4ddf19b92f","source":{"kind":"arxiv","id":"2210.17167","version":1},"attestation_state":"computed","paper":{"title":"Reduce Catastrophic Forgetting of Dense Retrieval Training with Teleportation Negatives","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Arnold Overwijk, Chenyan Xiong, Jie Bao, Si Sun, Yue Yu, Zhiyuan Liu","submitted_at":"2022-10-31T09:25:42Z","abstract_excerpt":"In this paper, we investigate the instability in the standard dense retrieval training, which iterates between model training and hard negative selection using the being-trained model. We show the catastrophic forgetting phenomena behind the training instability, where models learn and forget different negative groups during training iterations. We then propose ANCE-Tele, which accumulates momentum negatives from past iterations and approximates future iterations using lookahead negatives, as \"teleportations\" along the time axis to smooth the learning process. On web search and OpenQA, ANCE-Te"},"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":"2210.17167","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-10-31T09:25:42Z","cross_cats_sorted":[],"title_canon_sha256":"3b3d700d69a3a4caf014f640de8a88740fc5ffb7d03415df9038a893c4e76e86","abstract_canon_sha256":"f2f17e567a98d0864a65d1212d0def4c1d935a61ebedd3ea3129d35805abb703"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:11:55.734400Z","signature_b64":"9RQMOLM6spIuWREKRgXBgINBySZKnrclwCDSLJRzpNhDdVqI7ldm5PzLodke02rQXMDMfw5+H+Y+CU/EHHzdAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2ab23a11e8c757159f0b44afbb8d63da3fbae53170b03e886b00fd4ddf19b92f","last_reissued_at":"2026-07-05T05:11:55.733982Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:11:55.733982Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Reduce Catastrophic Forgetting of Dense Retrieval Training with Teleportation Negatives","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Arnold Overwijk, Chenyan Xiong, Jie Bao, Si Sun, Yue Yu, Zhiyuan Liu","submitted_at":"2022-10-31T09:25:42Z","abstract_excerpt":"In this paper, we investigate the instability in the standard dense retrieval training, which iterates between model training and hard negative selection using the being-trained model. We show the catastrophic forgetting phenomena behind the training instability, where models learn and forget different negative groups during training iterations. We then propose ANCE-Tele, which accumulates momentum negatives from past iterations and approximates future iterations using lookahead negatives, as \"teleportations\" along the time axis to smooth the learning process. On web search and OpenQA, ANCE-Te"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2210.17167","kind":"arxiv","version":1},"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/2210.17167/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":"2210.17167","created_at":"2026-07-05T05:11:55.734049+00:00"},{"alias_kind":"arxiv_version","alias_value":"2210.17167v1","created_at":"2026-07-05T05:11:55.734049+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2210.17167","created_at":"2026-07-05T05:11:55.734049+00:00"},{"alias_kind":"pith_short_12","alias_value":"FKZDUEPIY5LR","created_at":"2026-07-05T05:11:55.734049+00:00"},{"alias_kind":"pith_short_16","alias_value":"FKZDUEPIY5LRLHYL","created_at":"2026-07-05T05:11:55.734049+00:00"},{"alias_kind":"pith_short_8","alias_value":"FKZDUEPI","created_at":"2026-07-05T05:11:55.734049+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2412.17250","citing_title":"SyNeg: LLM-Driven Synthetic Hard-Negatives for Dense Retrieval","ref_index":30,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FKZDUEPIY5LRLHYLISX3XDLD3I","json":"https://pith.science/pith/FKZDUEPIY5LRLHYLISX3XDLD3I.json","graph_json":"https://pith.science/api/pith-number/FKZDUEPIY5LRLHYLISX3XDLD3I/graph.json","events_json":"https://pith.science/api/pith-number/FKZDUEPIY5LRLHYLISX3XDLD3I/events.json","paper":"https://pith.science/paper/FKZDUEPI"},"agent_actions":{"view_html":"https://pith.science/pith/FKZDUEPIY5LRLHYLISX3XDLD3I","download_json":"https://pith.science/pith/FKZDUEPIY5LRLHYLISX3XDLD3I.json","view_paper":"https://pith.science/paper/FKZDUEPI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2210.17167&json=true","fetch_graph":"https://pith.science/api/pith-number/FKZDUEPIY5LRLHYLISX3XDLD3I/graph.json","fetch_events":"https://pith.science/api/pith-number/FKZDUEPIY5LRLHYLISX3XDLD3I/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FKZDUEPIY5LRLHYLISX3XDLD3I/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FKZDUEPIY5LRLHYLISX3XDLD3I/action/storage_attestation","attest_author":"https://pith.science/pith/FKZDUEPIY5LRLHYLISX3XDLD3I/action/author_attestation","sign_citation":"https://pith.science/pith/FKZDUEPIY5LRLHYLISX3XDLD3I/action/citation_signature","submit_replication":"https://pith.science/pith/FKZDUEPIY5LRLHYLISX3XDLD3I/action/replication_record"}},"created_at":"2026-07-05T05:11:55.734049+00:00","updated_at":"2026-07-05T05:11:55.734049+00:00"}