{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:LP23F3ALY6EP37REPK6TXFWUGJ","short_pith_number":"pith:LP23F3AL","schema_version":"1.0","canonical_sha256":"5bf5b2ec0bc788fdfe247abd3b96d432692d7b892070dd2b7fe60932570f6f6f","source":{"kind":"arxiv","id":"2409.14683","version":1},"attestation_state":"computed","paper":{"title":"Reducing the Footprint of Multi-Vector Retrieval with Minimal Performance Impact via Token Pooling","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.IR","authors_text":"Antoine Chaffin, Benjamin Clavi\\'e, Griffin Adams","submitted_at":"2024-09-23T03:12:43Z","abstract_excerpt":"Over the last few years, multi-vector retrieval methods, spearheaded by ColBERT, have become an increasingly popular approach to Neural IR. By storing representations at the token level rather than at the document level, these methods have demonstrated very strong retrieval performance, especially in out-of-domain settings. However, the storage and memory requirements necessary to store the large number of associated vectors remain an important drawback, hindering practical adoption. In this paper, we introduce a simple clustering-based token pooling approach to aggressively reduce the number "},"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":"2409.14683","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.IR","submitted_at":"2024-09-23T03:12:43Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"2b0d2d3e1d823ff93a0bcec6c9e2a2b2f4e7585118969c5ab37184b22e18e75d","abstract_canon_sha256":"5dc3bf9d790bd94d4454cf3f4ffdacf418384edf48c87395427993586632794e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:10:20.204830Z","signature_b64":"ea+VQN9C9khWj0fv58H3RE7SWKyxlRe89SzR7VvgI062yr+0r26S5/vVsJQy4RnBNg/ScVuTL+Y4jm6N3TLKCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5bf5b2ec0bc788fdfe247abd3b96d432692d7b892070dd2b7fe60932570f6f6f","last_reissued_at":"2026-07-05T09:10:20.204426Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:10:20.204426Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Reducing the Footprint of Multi-Vector Retrieval with Minimal Performance Impact via Token Pooling","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.IR","authors_text":"Antoine Chaffin, Benjamin Clavi\\'e, Griffin Adams","submitted_at":"2024-09-23T03:12:43Z","abstract_excerpt":"Over the last few years, multi-vector retrieval methods, spearheaded by ColBERT, have become an increasingly popular approach to Neural IR. By storing representations at the token level rather than at the document level, these methods have demonstrated very strong retrieval performance, especially in out-of-domain settings. However, the storage and memory requirements necessary to store the large number of associated vectors remain an important drawback, hindering practical adoption. In this paper, we introduce a simple clustering-based token pooling approach to aggressively reduce the number "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.14683","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/2409.14683/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":"2409.14683","created_at":"2026-07-05T09:10:20.204485+00:00"},{"alias_kind":"arxiv_version","alias_value":"2409.14683v1","created_at":"2026-07-05T09:10:20.204485+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.14683","created_at":"2026-07-05T09:10:20.204485+00:00"},{"alias_kind":"pith_short_12","alias_value":"LP23F3ALY6EP","created_at":"2026-07-05T09:10:20.204485+00:00"},{"alias_kind":"pith_short_16","alias_value":"LP23F3ALY6EP37RE","created_at":"2026-07-05T09:10:20.204485+00:00"},{"alias_kind":"pith_short_8","alias_value":"LP23F3AL","created_at":"2026-07-05T09:10:20.204485+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2607.05927","citing_title":"CMDR: Contextual Multimodal Document Retrieval","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2607.06036","citing_title":"Learn to Pool: Lightweight Fine-Tuning for Flexible Multi-Vector Compression","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2601.21853","citing_title":"LEMUR: Learned Multi-Vector Retrieval","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2602.19549","citing_title":"Sculpting the Vector Space: Towards Efficient Multi-Vector Visual Document Retrieval via Prune-then-Merge Framework","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2603.09933","citing_title":"A Voronoi Cell Formulation for Principled Token Pruning in Late-Interaction Retrieval Models","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00646","citing_title":"A Replicability Study of XTR","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10167","citing_title":"Visual Late Chunking: An Empirical Study of Contextual Chunking for Efficient Visual Document Retrieval","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LP23F3ALY6EP37REPK6TXFWUGJ","json":"https://pith.science/pith/LP23F3ALY6EP37REPK6TXFWUGJ.json","graph_json":"https://pith.science/api/pith-number/LP23F3ALY6EP37REPK6TXFWUGJ/graph.json","events_json":"https://pith.science/api/pith-number/LP23F3ALY6EP37REPK6TXFWUGJ/events.json","paper":"https://pith.science/paper/LP23F3AL"},"agent_actions":{"view_html":"https://pith.science/pith/LP23F3ALY6EP37REPK6TXFWUGJ","download_json":"https://pith.science/pith/LP23F3ALY6EP37REPK6TXFWUGJ.json","view_paper":"https://pith.science/paper/LP23F3AL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2409.14683&json=true","fetch_graph":"https://pith.science/api/pith-number/LP23F3ALY6EP37REPK6TXFWUGJ/graph.json","fetch_events":"https://pith.science/api/pith-number/LP23F3ALY6EP37REPK6TXFWUGJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LP23F3ALY6EP37REPK6TXFWUGJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LP23F3ALY6EP37REPK6TXFWUGJ/action/storage_attestation","attest_author":"https://pith.science/pith/LP23F3ALY6EP37REPK6TXFWUGJ/action/author_attestation","sign_citation":"https://pith.science/pith/LP23F3ALY6EP37REPK6TXFWUGJ/action/citation_signature","submit_replication":"https://pith.science/pith/LP23F3ALY6EP37REPK6TXFWUGJ/action/replication_record"}},"created_at":"2026-07-05T09:10:20.204485+00:00","updated_at":"2026-07-05T09:10:20.204485+00:00"}