{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:CMQP2XMJU4XRMGZLDAJHWF6W3V","short_pith_number":"pith:CMQP2XMJ","schema_version":"1.0","canonical_sha256":"1320fd5d89a72f161b2b18127b17d6dd40f4287005ab79c4d3bb2a9ad971f4d7","source":{"kind":"arxiv","id":"2412.21023","version":2},"attestation_state":"computed","paper":{"title":"EdgeRAG: Online-Indexed RAG for Edge Devices","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Korakit Seemakhupt, Samira Khan, Sihang Liu","submitted_at":"2024-12-30T15:46:53Z","abstract_excerpt":"Deploying Retrieval Augmented Generation (RAG) on resource-constrained edge devices is challenging due to limited memory and processing power. In this work, we propose EdgeRAG which addresses the memory constraint by pruning embeddings within clusters and generating embeddings on-demand during retrieval. To avoid the latency of generating embeddings for large tail clusters, EdgeRAG pre-computes and stores embeddings for these clusters, while adaptively caching remaining embeddings to minimize redundant computations and further optimize latency. The result from BEIR suite shows that EdgeRAG off"},"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":"2412.21023","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-12-30T15:46:53Z","cross_cats_sorted":[],"title_canon_sha256":"15094c23c5dbc8d8a90d9f13501caf5c03733f7a975e76d6bb821758891657ff","abstract_canon_sha256":"ceaf3d9a09189b06911e78dd8c998387d0396614283a855d2e0d71876677f52a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:56:05.047794Z","signature_b64":"/eJ6ibqka2m6gjfHZ5Lsbxhp0ekdxAyYpmCse33/2SIStr9jeoHYi4P4l4MYBd24nRn4j7VNRYXuw0psTV+DBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1320fd5d89a72f161b2b18127b17d6dd40f4287005ab79c4d3bb2a9ad971f4d7","last_reissued_at":"2026-07-05T09:56:05.047382Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:56:05.047382Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"EdgeRAG: Online-Indexed RAG for Edge Devices","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Korakit Seemakhupt, Samira Khan, Sihang Liu","submitted_at":"2024-12-30T15:46:53Z","abstract_excerpt":"Deploying Retrieval Augmented Generation (RAG) on resource-constrained edge devices is challenging due to limited memory and processing power. In this work, we propose EdgeRAG which addresses the memory constraint by pruning embeddings within clusters and generating embeddings on-demand during retrieval. To avoid the latency of generating embeddings for large tail clusters, EdgeRAG pre-computes and stores embeddings for these clusters, while adaptively caching remaining embeddings to minimize redundant computations and further optimize latency. The result from BEIR suite shows that EdgeRAG off"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.21023","kind":"arxiv","version":2},"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/2412.21023/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":"2412.21023","created_at":"2026-07-05T09:56:05.047443+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.21023v2","created_at":"2026-07-05T09:56:05.047443+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.21023","created_at":"2026-07-05T09:56:05.047443+00:00"},{"alias_kind":"pith_short_12","alias_value":"CMQP2XMJU4XR","created_at":"2026-07-05T09:56:05.047443+00:00"},{"alias_kind":"pith_short_16","alias_value":"CMQP2XMJU4XRMGZL","created_at":"2026-07-05T09:56:05.047443+00:00"},{"alias_kind":"pith_short_8","alias_value":"CMQP2XMJ","created_at":"2026-07-05T09:56:05.047443+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.09493","citing_title":"Policy-Aware Edge LLM-RAG Framework for Internet of Battlefield Things Mission Orchestration","ref_index":16,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CMQP2XMJU4XRMGZLDAJHWF6W3V","json":"https://pith.science/pith/CMQP2XMJU4XRMGZLDAJHWF6W3V.json","graph_json":"https://pith.science/api/pith-number/CMQP2XMJU4XRMGZLDAJHWF6W3V/graph.json","events_json":"https://pith.science/api/pith-number/CMQP2XMJU4XRMGZLDAJHWF6W3V/events.json","paper":"https://pith.science/paper/CMQP2XMJ"},"agent_actions":{"view_html":"https://pith.science/pith/CMQP2XMJU4XRMGZLDAJHWF6W3V","download_json":"https://pith.science/pith/CMQP2XMJU4XRMGZLDAJHWF6W3V.json","view_paper":"https://pith.science/paper/CMQP2XMJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.21023&json=true","fetch_graph":"https://pith.science/api/pith-number/CMQP2XMJU4XRMGZLDAJHWF6W3V/graph.json","fetch_events":"https://pith.science/api/pith-number/CMQP2XMJU4XRMGZLDAJHWF6W3V/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CMQP2XMJU4XRMGZLDAJHWF6W3V/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CMQP2XMJU4XRMGZLDAJHWF6W3V/action/storage_attestation","attest_author":"https://pith.science/pith/CMQP2XMJU4XRMGZLDAJHWF6W3V/action/author_attestation","sign_citation":"https://pith.science/pith/CMQP2XMJU4XRMGZLDAJHWF6W3V/action/citation_signature","submit_replication":"https://pith.science/pith/CMQP2XMJU4XRMGZLDAJHWF6W3V/action/replication_record"}},"created_at":"2026-07-05T09:56:05.047443+00:00","updated_at":"2026-07-05T09:56:05.047443+00:00"}