{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:KZP6QXMOL6BP5NMJT6KJGZ2OMP","short_pith_number":"pith:KZP6QXMO","schema_version":"1.0","canonical_sha256":"565fe85d8e5f82feb5899f9493674e63e25dae51365f8c43763050fb2dec95f3","source":{"kind":"arxiv","id":"2502.00258","version":2},"attestation_state":"computed","paper":{"title":"ProxSparse: Regularized Learning of Semi-Structured Sparsity Masks for Pretrained LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"George Karypis, Hongyi Liu, Jiaji Huang, Rajarshi Saha, Shoham Sabach, Youngsuk Park, Yu-Xiang Wang, Zhen Jia","submitted_at":"2025-02-01T01:35:23Z","abstract_excerpt":"Large Language Models (LLMs) have demonstrated exceptional performance in natural language processing tasks, yet their massive size makes serving them inefficient and costly. Semi-structured pruning has emerged as an effective method for model acceleration, but existing approaches are suboptimal because they focus on local, layer-wise optimizations using heuristic rules, failing to leverage global feedback. We present ProxSparse, a learning-based framework for mask selection enabled by regularized optimization. ProxSparse transforms the rigid, non-differentiable mask selection process into a s"},"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":"2502.00258","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-01T01:35:23Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"78699494ba9bafeab07c8acdb2b9a6c52a91ecf4595f90fa7cc1b39063721ba6","abstract_canon_sha256":"0ad83a71241f4a60f5f4fada7d04d4894bc82b191ce62c18aace05da9c10df15"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:26:02.806609Z","signature_b64":"gz9ks8Ow8Pmx8fO27OaFQUCxX8QIwK82DVlt+Ctkbw8UWf2n3mkr26zMpgY+BIUgFVL38QupA8W7IaVnYrG3Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"565fe85d8e5f82feb5899f9493674e63e25dae51365f8c43763050fb2dec95f3","last_reissued_at":"2026-07-05T11:26:02.805968Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:26:02.805968Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ProxSparse: Regularized Learning of Semi-Structured Sparsity Masks for Pretrained LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"George Karypis, Hongyi Liu, Jiaji Huang, Rajarshi Saha, Shoham Sabach, Youngsuk Park, Yu-Xiang Wang, Zhen Jia","submitted_at":"2025-02-01T01:35:23Z","abstract_excerpt":"Large Language Models (LLMs) have demonstrated exceptional performance in natural language processing tasks, yet their massive size makes serving them inefficient and costly. Semi-structured pruning has emerged as an effective method for model acceleration, but existing approaches are suboptimal because they focus on local, layer-wise optimizations using heuristic rules, failing to leverage global feedback. We present ProxSparse, a learning-based framework for mask selection enabled by regularized optimization. ProxSparse transforms the rigid, non-differentiable mask selection process into a s"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.00258","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/2502.00258/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":"2502.00258","created_at":"2026-07-05T11:26:02.806053+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.00258v2","created_at":"2026-07-05T11:26:02.806053+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.00258","created_at":"2026-07-05T11:26:02.806053+00:00"},{"alias_kind":"pith_short_12","alias_value":"KZP6QXMOL6BP","created_at":"2026-07-05T11:26:02.806053+00:00"},{"alias_kind":"pith_short_16","alias_value":"KZP6QXMOL6BP5NMJ","created_at":"2026-07-05T11:26:02.806053+00:00"},{"alias_kind":"pith_short_8","alias_value":"KZP6QXMO","created_at":"2026-07-05T11:26:02.806053+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.26632","citing_title":"RT-Lynx: Putting the GEMM Sparsity In a Right Way for Diffusion Models","ref_index":34,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KZP6QXMOL6BP5NMJT6KJGZ2OMP","json":"https://pith.science/pith/KZP6QXMOL6BP5NMJT6KJGZ2OMP.json","graph_json":"https://pith.science/api/pith-number/KZP6QXMOL6BP5NMJT6KJGZ2OMP/graph.json","events_json":"https://pith.science/api/pith-number/KZP6QXMOL6BP5NMJT6KJGZ2OMP/events.json","paper":"https://pith.science/paper/KZP6QXMO"},"agent_actions":{"view_html":"https://pith.science/pith/KZP6QXMOL6BP5NMJT6KJGZ2OMP","download_json":"https://pith.science/pith/KZP6QXMOL6BP5NMJT6KJGZ2OMP.json","view_paper":"https://pith.science/paper/KZP6QXMO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.00258&json=true","fetch_graph":"https://pith.science/api/pith-number/KZP6QXMOL6BP5NMJT6KJGZ2OMP/graph.json","fetch_events":"https://pith.science/api/pith-number/KZP6QXMOL6BP5NMJT6KJGZ2OMP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KZP6QXMOL6BP5NMJT6KJGZ2OMP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KZP6QXMOL6BP5NMJT6KJGZ2OMP/action/storage_attestation","attest_author":"https://pith.science/pith/KZP6QXMOL6BP5NMJT6KJGZ2OMP/action/author_attestation","sign_citation":"https://pith.science/pith/KZP6QXMOL6BP5NMJT6KJGZ2OMP/action/citation_signature","submit_replication":"https://pith.science/pith/KZP6QXMOL6BP5NMJT6KJGZ2OMP/action/replication_record"}},"created_at":"2026-07-05T11:26:02.806053+00:00","updated_at":"2026-07-05T11:26:02.806053+00:00"}