{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:Q7ASRHN6SCKAWFSKM2XTKE43EL","short_pith_number":"pith:Q7ASRHN6","schema_version":"1.0","canonical_sha256":"87c1289dbe90940b164a66af35139b22eb037a47630948cdb0bd9f968d26af09","source":{"kind":"arxiv","id":"2509.22075","version":6},"attestation_state":"computed","paper":{"title":"CoSpaDi: Compressing LLMs via Calibration-Guided Sparse Dictionary Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"CoSpaDi replaces low-rank factorization with a sparse dictionary model that better preserves LLM accuracy at 20-40 percent compression.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ammar Ali, Denis Makhov, Dmitriy Shopkhoev, Magauiya Zhussip, Stamatios Lefkimmiatis","submitted_at":"2025-09-26T08:55:09Z","abstract_excerpt":"Post-training LLM compression often relies on low-rank approximations, which force all columns of a projection matrix to share a single low-dimensional subspace. We propose CoSpaDi, a training-free compression framework that replaces this single-subspace assumption with a union-of-subspaces model via sparse dictionary learning. CoSpaDi factorizes each weight matrix into a dense dictionary and column-sparse coefficients, allowing different columns to select different subsets of dictionary atoms at the same storage budget. To preserve model behavior, we use calibration activations to transform f"},"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":"2509.22075","kind":"arxiv","version":6},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-09-26T08:55:09Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"d6e17c26284705680c617022401148f8074164ae5ff3afd6c70469945f8f3af8","abstract_canon_sha256":"bc82722b17729c3ca381508862d7ffaa450d2d34c2fcc91ee4412e4685255600"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T01:11:57.156794Z","signature_b64":"MSPKpzwjX3rDf60mHlWMh13+Z/O8cdTkSd//EEqG5HU0zq69OH/1bCJO5NZ9zaCPWudNdI55gCRQrzYb3CWGDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"87c1289dbe90940b164a66af35139b22eb037a47630948cdb0bd9f968d26af09","last_reissued_at":"2026-06-23T01:11:57.156174Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T01:11:57.156174Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CoSpaDi: Compressing LLMs via Calibration-Guided Sparse Dictionary Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"CoSpaDi replaces low-rank factorization with a sparse dictionary model that better preserves LLM accuracy at 20-40 percent compression.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ammar Ali, Denis Makhov, Dmitriy Shopkhoev, Magauiya Zhussip, Stamatios Lefkimmiatis","submitted_at":"2025-09-26T08:55:09Z","abstract_excerpt":"Post-training LLM compression often relies on low-rank approximations, which force all columns of a projection matrix to share a single low-dimensional subspace. We propose CoSpaDi, a training-free compression framework that replaces this single-subspace assumption with a union-of-subspaces model via sparse dictionary learning. CoSpaDi factorizes each weight matrix into a dense dictionary and column-sparse coefficients, allowing different columns to select different subsets of dictionary atoms at the same storage budget. To preserve model behavior, we use calibration activations to transform f"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Across Llama and Qwen model families, CoSpaDi consistently improves the accuracy-compression and perplexity-compression trade-offs over state-of-the-art SVD-based baselines and strong structured pruning baselines at 20-40% compression ratios.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that minimizing functional reconstruction error on a small calibration set will produce a factorization whose downstream task performance remains close to the original model without any fine-tuning or further adaptation.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"CoSpaDi introduces a training-free sparse dictionary learning framework for post-training LLM compression that optimizes functional reconstruction error via activation-derived orthonormalization and achieves improved accuracy-compression trade-offs over SVD and pruning baselines.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"CoSpaDi replaces low-rank factorization with a sparse dictionary model that better preserves LLM accuracy at 20-40 percent compression.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"f7e2648c9a5527e0cfc6f00a30dade4184bacec6357a9e3215bb0202c79939f6"},"source":{"id":"2509.22075","kind":"arxiv","version":6},"verdict":{"id":"5e9533e4-fa16-4a29-9e32-8f1fd8b16999","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-18T13:14:45.608935Z","strongest_claim":"Across Llama and Qwen model families, CoSpaDi consistently improves the accuracy-compression and perplexity-compression trade-offs over state-of-the-art SVD-based baselines and strong structured pruning baselines at 20-40% compression ratios.","one_line_summary":"CoSpaDi introduces a training-free sparse dictionary learning framework for post-training LLM compression that optimizes functional reconstruction error via activation-derived orthonormalization and achieves improved accuracy-compression trade-offs over SVD and pruning baselines.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that minimizing functional reconstruction error on a small calibration set will produce a factorization whose downstream task performance remains close to the original model without any fine-tuning or further adaptation.","pith_extraction_headline":"CoSpaDi replaces low-rank factorization with a sparse dictionary model that better preserves LLM accuracy at 20-40 percent compression."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2509.22075/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":"2509.22075","created_at":"2026-06-23T01:11:57.156238+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.22075v6","created_at":"2026-06-23T01:11:57.156238+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.22075","created_at":"2026-06-23T01:11:57.156238+00:00"},{"alias_kind":"pith_short_12","alias_value":"Q7ASRHN6SCKA","created_at":"2026-06-23T01:11:57.156238+00:00"},{"alias_kind":"pith_short_16","alias_value":"Q7ASRHN6SCKAWFSK","created_at":"2026-06-23T01:11:57.156238+00:00"},{"alias_kind":"pith_short_8","alias_value":"Q7ASRHN6","created_at":"2026-06-23T01:11:57.156238+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.24754","citing_title":"Motion-Compensated Weight Compression","ref_index":38,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Q7ASRHN6SCKAWFSKM2XTKE43EL","json":"https://pith.science/pith/Q7ASRHN6SCKAWFSKM2XTKE43EL.json","graph_json":"https://pith.science/api/pith-number/Q7ASRHN6SCKAWFSKM2XTKE43EL/graph.json","events_json":"https://pith.science/api/pith-number/Q7ASRHN6SCKAWFSKM2XTKE43EL/events.json","paper":"https://pith.science/paper/Q7ASRHN6"},"agent_actions":{"view_html":"https://pith.science/pith/Q7ASRHN6SCKAWFSKM2XTKE43EL","download_json":"https://pith.science/pith/Q7ASRHN6SCKAWFSKM2XTKE43EL.json","view_paper":"https://pith.science/paper/Q7ASRHN6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.22075&json=true","fetch_graph":"https://pith.science/api/pith-number/Q7ASRHN6SCKAWFSKM2XTKE43EL/graph.json","fetch_events":"https://pith.science/api/pith-number/Q7ASRHN6SCKAWFSKM2XTKE43EL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Q7ASRHN6SCKAWFSKM2XTKE43EL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Q7ASRHN6SCKAWFSKM2XTKE43EL/action/storage_attestation","attest_author":"https://pith.science/pith/Q7ASRHN6SCKAWFSKM2XTKE43EL/action/author_attestation","sign_citation":"https://pith.science/pith/Q7ASRHN6SCKAWFSKM2XTKE43EL/action/citation_signature","submit_replication":"https://pith.science/pith/Q7ASRHN6SCKAWFSKM2XTKE43EL/action/replication_record"}},"created_at":"2026-06-23T01:11:57.156238+00:00","updated_at":"2026-06-23T01:11:57.156238+00:00"}