{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:IEAI7A3NQWGUD5PTJBC2AXJBBB","short_pith_number":"pith:IEAI7A3N","schema_version":"1.0","canonical_sha256":"41008f836d858d41f5f34845a05d21084e9e13760063d08b7e5fd0a5ac74e06e","source":{"kind":"arxiv","id":"2112.07899","version":1},"attestation_state":"computed","paper":{"title":"Large Dual Encoders Are Generalizable Retrievers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.IR","authors_text":"Chen Qu, Gustavo Hern\\'andez \\'Abrego, Jianmo Ni, Ji Ma, Jing Lu, Keith B. Hall, Ming-Wei Chang, Vincent Y. Zhao, Yi Luan, Yinfei Yang, Zhuyun Dai","submitted_at":"2021-12-15T05:33:27Z","abstract_excerpt":"It has been shown that dual encoders trained on one domain often fail to generalize to other domains for retrieval tasks. One widespread belief is that the bottleneck layer of a dual encoder, where the final score is simply a dot-product between a query vector and a passage vector, is too limited to make dual encoders an effective retrieval model for out-of-domain generalization. In this paper, we challenge this belief by scaling up the size of the dual encoder model {\\em while keeping the bottleneck embedding size fixed.} With multi-stage training, surprisingly, scaling up the model size brin"},"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":"2112.07899","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.IR","submitted_at":"2021-12-15T05:33:27Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"4f2bbe5dcd37d2340b78a6ed8554b90c9f00bb82b01df020ba5b4e71bacf052a","abstract_canon_sha256":"2e9b04d4ba56455421591fda61c0e3818d0663a0fc17c5552675c0380fb3a2d1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:41:08.206092Z","signature_b64":"r0YYQu2tJnc+LkqcfqM0VaF3l9UQ3py1NG1iig5Vqu8wFLSpSmhkOkHBbw+Qmbzyzo7Tfrad8FOxD+JDPR3fCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"41008f836d858d41f5f34845a05d21084e9e13760063d08b7e5fd0a5ac74e06e","last_reissued_at":"2026-07-05T03:41:08.205701Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:41:08.205701Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Large Dual Encoders Are Generalizable Retrievers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.IR","authors_text":"Chen Qu, Gustavo Hern\\'andez \\'Abrego, Jianmo Ni, Ji Ma, Jing Lu, Keith B. Hall, Ming-Wei Chang, Vincent Y. Zhao, Yi Luan, Yinfei Yang, Zhuyun Dai","submitted_at":"2021-12-15T05:33:27Z","abstract_excerpt":"It has been shown that dual encoders trained on one domain often fail to generalize to other domains for retrieval tasks. One widespread belief is that the bottleneck layer of a dual encoder, where the final score is simply a dot-product between a query vector and a passage vector, is too limited to make dual encoders an effective retrieval model for out-of-domain generalization. In this paper, we challenge this belief by scaling up the size of the dual encoder model {\\em while keeping the bottleneck embedding size fixed.} With multi-stage training, surprisingly, scaling up the model size brin"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2112.07899","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/2112.07899/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":"2112.07899","created_at":"2026-07-05T03:41:08.205759+00:00"},{"alias_kind":"arxiv_version","alias_value":"2112.07899v1","created_at":"2026-07-05T03:41:08.205759+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2112.07899","created_at":"2026-07-05T03:41:08.205759+00:00"},{"alias_kind":"pith_short_12","alias_value":"IEAI7A3NQWGU","created_at":"2026-07-05T03:41:08.205759+00:00"},{"alias_kind":"pith_short_16","alias_value":"IEAI7A3NQWGUD5PT","created_at":"2026-07-05T03:41:08.205759+00:00"},{"alias_kind":"pith_short_8","alias_value":"IEAI7A3N","created_at":"2026-07-05T03:41:08.205759+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":13,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22807","citing_title":"KaLM-Reranker-V1: Fast but Not Late Interaction for Compressed Document Reranking","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03728","citing_title":"Re-Ranking Through an Attribution Lens for Citation Quality in Legal QA","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01504","citing_title":"Semantic Retrieval for Product Search in E-Commerce","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.24530","citing_title":"Unveil: Unified Visual-Textual Integration and Distillation for Multi-modal Document Retrieval","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2405.13068","citing_title":"Uncovering Logit Suppression Vulnerabilities in LLM Safety Alignment","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19366","citing_title":"Accurate, Efficient, and Explainable Deep Learning Approaches for Environmental Science Problems","ref_index":178,"is_internal_anchor":false},{"citing_arxiv_id":"2603.04816","citing_title":"Scaling Laws for Cross-Encoder Reranking","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2509.20354","citing_title":"EmbeddingGemma: Powerful and Lightweight Text Representations","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2405.17428","citing_title":"NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models","ref_index":112,"is_internal_anchor":false},{"citing_arxiv_id":"2309.07597","citing_title":"C-Pack: Packed Resources For General Chinese Embeddings","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2402.03216","citing_title":"M3-Embedding: Multi-Linguality, Multi-Functionality, Multi-Granularity Text Embeddings Through Self-Knowledge Distillation","ref_index":55,"is_internal_anchor":false},{"citing_arxiv_id":"2212.03533","citing_title":"Text Embeddings by Weakly-Supervised Contrastive Pre-training","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15109","citing_title":"IUQ: Interrogative Uncertainty Quantification for Long-Form Large Language Model Generation","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IEAI7A3NQWGUD5PTJBC2AXJBBB","json":"https://pith.science/pith/IEAI7A3NQWGUD5PTJBC2AXJBBB.json","graph_json":"https://pith.science/api/pith-number/IEAI7A3NQWGUD5PTJBC2AXJBBB/graph.json","events_json":"https://pith.science/api/pith-number/IEAI7A3NQWGUD5PTJBC2AXJBBB/events.json","paper":"https://pith.science/paper/IEAI7A3N"},"agent_actions":{"view_html":"https://pith.science/pith/IEAI7A3NQWGUD5PTJBC2AXJBBB","download_json":"https://pith.science/pith/IEAI7A3NQWGUD5PTJBC2AXJBBB.json","view_paper":"https://pith.science/paper/IEAI7A3N","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2112.07899&json=true","fetch_graph":"https://pith.science/api/pith-number/IEAI7A3NQWGUD5PTJBC2AXJBBB/graph.json","fetch_events":"https://pith.science/api/pith-number/IEAI7A3NQWGUD5PTJBC2AXJBBB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IEAI7A3NQWGUD5PTJBC2AXJBBB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IEAI7A3NQWGUD5PTJBC2AXJBBB/action/storage_attestation","attest_author":"https://pith.science/pith/IEAI7A3NQWGUD5PTJBC2AXJBBB/action/author_attestation","sign_citation":"https://pith.science/pith/IEAI7A3NQWGUD5PTJBC2AXJBBB/action/citation_signature","submit_replication":"https://pith.science/pith/IEAI7A3NQWGUD5PTJBC2AXJBBB/action/replication_record"}},"created_at":"2026-07-05T03:41:08.205759+00:00","updated_at":"2026-07-05T03:41:08.205759+00:00"}