{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:4W5EPPKRVJKMVP6WO6RSRBV7RU","short_pith_number":"pith:4W5EPPKR","schema_version":"1.0","canonical_sha256":"e5ba47bd51aa54cabfd677a32886bf8d100a3d4c5a6c8aa796c07fc0ceb2bafe","source":{"kind":"arxiv","id":"2405.18886","version":2},"attestation_state":"computed","paper":{"title":"Compressing Large Language Models using Low Rank and Low Precision Decomposition","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Andrea J. Goldsmith, Mert Pilanci, Naomi Sagan, Rajarshi Saha, Varun Srivastava","submitted_at":"2024-05-29T08:42:30Z","abstract_excerpt":"The prohibitive sizes of Large Language Models (LLMs) today make it difficult to deploy them on memory-constrained edge devices. This work introduces $\\rm CALDERA$ -- a new post-training LLM compression algorithm that harnesses the inherent low-rank structure of a weight matrix $\\mathbf{W}$ by approximating it via a low-rank, low-precision decomposition as $\\mathbf{W} \\approx \\mathbf{Q} + \\mathbf{L}\\mathbf{R}$. Here, $\\mathbf{L}$ and $\\mathbf{R}$ are low rank factors, and the entries of $\\mathbf{Q}$, $\\mathbf{L}$ and $\\mathbf{R}$ are quantized. The model is compressed by substituting each laye"},"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":"2405.18886","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-05-29T08:42:30Z","cross_cats_sorted":["cs.AI","math.OC","stat.ML"],"title_canon_sha256":"8d77466e9b708ccd6cecc586690d004e1e4661237c3410366423b51e45ad0f01","abstract_canon_sha256":"3afeb79aac793c21f28cf3fb4d6384e0f49ea79706a2528374b5c91e08d3d47a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:30:15.528616Z","signature_b64":"IMBEuBqNQh/Kw9dRQeJqHwi/8AG+6aHw838XCtcznCIkiLG2AOGGIIvb5Hg5B7T+8CZfutjOFahYs03vJiceDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e5ba47bd51aa54cabfd677a32886bf8d100a3d4c5a6c8aa796c07fc0ceb2bafe","last_reissued_at":"2026-07-05T09:30:15.528069Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:30:15.528069Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Compressing Large Language Models using Low Rank and Low Precision Decomposition","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Andrea J. Goldsmith, Mert Pilanci, Naomi Sagan, Rajarshi Saha, Varun Srivastava","submitted_at":"2024-05-29T08:42:30Z","abstract_excerpt":"The prohibitive sizes of Large Language Models (LLMs) today make it difficult to deploy them on memory-constrained edge devices. This work introduces $\\rm CALDERA$ -- a new post-training LLM compression algorithm that harnesses the inherent low-rank structure of a weight matrix $\\mathbf{W}$ by approximating it via a low-rank, low-precision decomposition as $\\mathbf{W} \\approx \\mathbf{Q} + \\mathbf{L}\\mathbf{R}$. Here, $\\mathbf{L}$ and $\\mathbf{R}$ are low rank factors, and the entries of $\\mathbf{Q}$, $\\mathbf{L}$ and $\\mathbf{R}$ are quantized. The model is compressed by substituting each laye"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.18886","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/2405.18886/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":"2405.18886","created_at":"2026-07-05T09:30:15.528129+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.18886v2","created_at":"2026-07-05T09:30:15.528129+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.18886","created_at":"2026-07-05T09:30:15.528129+00:00"},{"alias_kind":"pith_short_12","alias_value":"4W5EPPKRVJKM","created_at":"2026-07-05T09:30:15.528129+00:00"},{"alias_kind":"pith_short_16","alias_value":"4W5EPPKRVJKMVP6W","created_at":"2026-07-05T09:30:15.528129+00:00"},{"alias_kind":"pith_short_8","alias_value":"4W5EPPKR","created_at":"2026-07-05T09:30:15.528129+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.07098","citing_title":"SigmaScale: LLM Compression with SVD-based Low-Rank Decomposition and Learned Scaling Matrices","ref_index":40,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4W5EPPKRVJKMVP6WO6RSRBV7RU","json":"https://pith.science/pith/4W5EPPKRVJKMVP6WO6RSRBV7RU.json","graph_json":"https://pith.science/api/pith-number/4W5EPPKRVJKMVP6WO6RSRBV7RU/graph.json","events_json":"https://pith.science/api/pith-number/4W5EPPKRVJKMVP6WO6RSRBV7RU/events.json","paper":"https://pith.science/paper/4W5EPPKR"},"agent_actions":{"view_html":"https://pith.science/pith/4W5EPPKRVJKMVP6WO6RSRBV7RU","download_json":"https://pith.science/pith/4W5EPPKRVJKMVP6WO6RSRBV7RU.json","view_paper":"https://pith.science/paper/4W5EPPKR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.18886&json=true","fetch_graph":"https://pith.science/api/pith-number/4W5EPPKRVJKMVP6WO6RSRBV7RU/graph.json","fetch_events":"https://pith.science/api/pith-number/4W5EPPKRVJKMVP6WO6RSRBV7RU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4W5EPPKRVJKMVP6WO6RSRBV7RU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4W5EPPKRVJKMVP6WO6RSRBV7RU/action/storage_attestation","attest_author":"https://pith.science/pith/4W5EPPKRVJKMVP6WO6RSRBV7RU/action/author_attestation","sign_citation":"https://pith.science/pith/4W5EPPKRVJKMVP6WO6RSRBV7RU/action/citation_signature","submit_replication":"https://pith.science/pith/4W5EPPKRVJKMVP6WO6RSRBV7RU/action/replication_record"}},"created_at":"2026-07-05T09:30:15.528129+00:00","updated_at":"2026-07-05T09:30:15.528129+00:00"}