{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:RWZ2P3RZOQWXC7C7Z4EMTYMR2B","short_pith_number":"pith:RWZ2P3RZ","schema_version":"1.0","canonical_sha256":"8db3a7ee39742d717c5fcf08c9e191d05f27ecc66ceccdebc15589e3fa091de9","source":{"kind":"arxiv","id":"2508.06163","version":1},"attestation_state":"computed","paper":{"title":"One Size Does Not Fit All: A Distribution-Aware Sparsification for More Precise Model Merging","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Anxiang Ma, Bei Li, Dingyang Lin, Jingbo Zhu, Junxin Wang, Kaiyan Chang, Tong Xiao, Tong Zheng, yingfeng luo, Zhengtao Yu, Ziqiang Xu","submitted_at":"2025-08-08T09:33:08Z","abstract_excerpt":"Model merging has emerged as a compelling data-free paradigm for multi-task learning, enabling the fusion of multiple fine-tuned models into a single, powerful entity. A key technique in merging methods is sparsification, which prunes redundant parameters from task vectors to mitigate interference. However, prevailing approaches employ a ``one-size-fits-all'' strategy, applying a uniform sparsity ratio that overlooks the inherent structural and statistical heterogeneity of model parameters. This often leads to a suboptimal trade-off, where critical parameters are inadvertently pruned while les"},"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":"2508.06163","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-08-08T09:33:08Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"cabcbf4eda948644d7b6791248559e52118f973a54bf5896fcef4af4dbd49eb6","abstract_canon_sha256":"6de8647113d17af3e8caa9ba7583430a1e66ec562cb05fae216e31c8325bc32e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:50:51.086744Z","signature_b64":"0t/Us4+wCeW1boIMDXdogFTgCVvhaGOhUaDUNI1Hdi/oOFxW7/n00WEDZ5zOnAwWZKSaeNY7HWPvUlxMjc0pAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8db3a7ee39742d717c5fcf08c9e191d05f27ecc66ceccdebc15589e3fa091de9","last_reissued_at":"2026-07-05T11:50:51.086229Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:50:51.086229Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"One Size Does Not Fit All: A Distribution-Aware Sparsification for More Precise Model Merging","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Anxiang Ma, Bei Li, Dingyang Lin, Jingbo Zhu, Junxin Wang, Kaiyan Chang, Tong Xiao, Tong Zheng, yingfeng luo, Zhengtao Yu, Ziqiang Xu","submitted_at":"2025-08-08T09:33:08Z","abstract_excerpt":"Model merging has emerged as a compelling data-free paradigm for multi-task learning, enabling the fusion of multiple fine-tuned models into a single, powerful entity. A key technique in merging methods is sparsification, which prunes redundant parameters from task vectors to mitigate interference. However, prevailing approaches employ a ``one-size-fits-all'' strategy, applying a uniform sparsity ratio that overlooks the inherent structural and statistical heterogeneity of model parameters. This often leads to a suboptimal trade-off, where critical parameters are inadvertently pruned while les"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.06163","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/2508.06163/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":"2508.06163","created_at":"2026-07-05T11:50:51.086297+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.06163v1","created_at":"2026-07-05T11:50:51.086297+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.06163","created_at":"2026-07-05T11:50:51.086297+00:00"},{"alias_kind":"pith_short_12","alias_value":"RWZ2P3RZOQWX","created_at":"2026-07-05T11:50:51.086297+00:00"},{"alias_kind":"pith_short_16","alias_value":"RWZ2P3RZOQWXC7C7","created_at":"2026-07-05T11:50:51.086297+00:00"},{"alias_kind":"pith_short_8","alias_value":"RWZ2P3RZ","created_at":"2026-07-05T11:50:51.086297+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2602.00585","citing_title":"Exploring Information Seeking Agent Consolidation","ref_index":2025,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RWZ2P3RZOQWXC7C7Z4EMTYMR2B","json":"https://pith.science/pith/RWZ2P3RZOQWXC7C7Z4EMTYMR2B.json","graph_json":"https://pith.science/api/pith-number/RWZ2P3RZOQWXC7C7Z4EMTYMR2B/graph.json","events_json":"https://pith.science/api/pith-number/RWZ2P3RZOQWXC7C7Z4EMTYMR2B/events.json","paper":"https://pith.science/paper/RWZ2P3RZ"},"agent_actions":{"view_html":"https://pith.science/pith/RWZ2P3RZOQWXC7C7Z4EMTYMR2B","download_json":"https://pith.science/pith/RWZ2P3RZOQWXC7C7Z4EMTYMR2B.json","view_paper":"https://pith.science/paper/RWZ2P3RZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.06163&json=true","fetch_graph":"https://pith.science/api/pith-number/RWZ2P3RZOQWXC7C7Z4EMTYMR2B/graph.json","fetch_events":"https://pith.science/api/pith-number/RWZ2P3RZOQWXC7C7Z4EMTYMR2B/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RWZ2P3RZOQWXC7C7Z4EMTYMR2B/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RWZ2P3RZOQWXC7C7Z4EMTYMR2B/action/storage_attestation","attest_author":"https://pith.science/pith/RWZ2P3RZOQWXC7C7Z4EMTYMR2B/action/author_attestation","sign_citation":"https://pith.science/pith/RWZ2P3RZOQWXC7C7Z4EMTYMR2B/action/citation_signature","submit_replication":"https://pith.science/pith/RWZ2P3RZOQWXC7C7Z4EMTYMR2B/action/replication_record"}},"created_at":"2026-07-05T11:50:51.086297+00:00","updated_at":"2026-07-05T11:50:51.086297+00:00"}