{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:4FSB426ZSS6QVC2D7USKOXPX2J","short_pith_number":"pith:4FSB426Z","schema_version":"1.0","canonical_sha256":"e1641e6bd994bd0a8b43fd24a75df7d27a2bce7fad66fe13d504853d4af1031e","source":{"kind":"arxiv","id":"2411.18729","version":2},"attestation_state":"computed","paper":{"title":"Multi-Task Model Merging via Adaptive Weight Disentanglement","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CV"],"primary_cat":"cs.LG","authors_text":"Chun Yuan, Feng Xiong, Ruifeng Xu, Runxi Cheng, Wang Chen, Yiwen Guo, Zhanqiu Zhang","submitted_at":"2024-11-27T20:08:55Z","abstract_excerpt":"Model merging has recently gained attention as an economical and scalable approach to incorporate task-specific weights from various tasks into a unified multi-task model. For example, in Task Arithmetic (TA), adding the fine-tuned weights of different tasks can enhance the model's performance on those tasks, while subtracting them leads to task forgetting. Although TA is highly effective, interference among task still hampers the performance of the merged model. Existing methods for handling conflicts between task generally rely on empirical selection, resulting in suboptimal performance. In "},"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":"2411.18729","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-11-27T20:08:55Z","cross_cats_sorted":["cs.CL","cs.CV"],"title_canon_sha256":"081d77bbea9f175b810036e3ec86234edf0f69e9c38f6502ace66a69c6986dee","abstract_canon_sha256":"ac0bac4b60bb366dc2fc708a03e12f821eaa8777d012bc8a2ac2b5662cfe3f63"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:58:43.894150Z","signature_b64":"Xmuos2cRhtvowitzSGjADC/cZAyazVAV9GvJKd3aP9QKAjjCWvkZuKzbfP8NdsCsWipQcgsMktwuS9P+9iZbDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e1641e6bd994bd0a8b43fd24a75df7d27a2bce7fad66fe13d504853d4af1031e","last_reissued_at":"2026-07-05T09:58:43.893715Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:58:43.893715Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Multi-Task Model Merging via Adaptive Weight Disentanglement","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CV"],"primary_cat":"cs.LG","authors_text":"Chun Yuan, Feng Xiong, Ruifeng Xu, Runxi Cheng, Wang Chen, Yiwen Guo, Zhanqiu Zhang","submitted_at":"2024-11-27T20:08:55Z","abstract_excerpt":"Model merging has recently gained attention as an economical and scalable approach to incorporate task-specific weights from various tasks into a unified multi-task model. For example, in Task Arithmetic (TA), adding the fine-tuned weights of different tasks can enhance the model's performance on those tasks, while subtracting them leads to task forgetting. Although TA is highly effective, interference among task still hampers the performance of the merged model. Existing methods for handling conflicts between task generally rely on empirical selection, resulting in suboptimal performance. In "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.18729","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/2411.18729/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":"2411.18729","created_at":"2026-07-05T09:58:43.893771+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.18729v2","created_at":"2026-07-05T09:58:43.893771+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.18729","created_at":"2026-07-05T09:58:43.893771+00:00"},{"alias_kind":"pith_short_12","alias_value":"4FSB426ZSS6Q","created_at":"2026-07-05T09:58:43.893771+00:00"},{"alias_kind":"pith_short_16","alias_value":"4FSB426ZSS6QVC2D","created_at":"2026-07-05T09:58:43.893771+00:00"},{"alias_kind":"pith_short_8","alias_value":"4FSB426Z","created_at":"2026-07-05T09:58:43.893771+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26902","citing_title":"Learning to Recover Task Experts from a Multi-Task Merged Model","ref_index":97,"is_internal_anchor":false},{"citing_arxiv_id":"2606.18627","citing_title":"PACT: Preserving Anchored Cores in Task-vectors for Model Merging","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03391","citing_title":"When Model Merging Breaks Routing: Training-Free Calibration for MoE","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20948","citing_title":"Memory Grafting: Scaling Language Model Pre-training via Offline Conditional Memory","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2408.07666","citing_title":"Model Merging in LLMs, MLLMs, and Beyond: Methods, Theories, Applications and Opportunities","ref_index":260,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17078","citing_title":"Understanding and Enforcing Weight Disentanglement in Task Arithmetic","ref_index":45,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4FSB426ZSS6QVC2D7USKOXPX2J","json":"https://pith.science/pith/4FSB426ZSS6QVC2D7USKOXPX2J.json","graph_json":"https://pith.science/api/pith-number/4FSB426ZSS6QVC2D7USKOXPX2J/graph.json","events_json":"https://pith.science/api/pith-number/4FSB426ZSS6QVC2D7USKOXPX2J/events.json","paper":"https://pith.science/paper/4FSB426Z"},"agent_actions":{"view_html":"https://pith.science/pith/4FSB426ZSS6QVC2D7USKOXPX2J","download_json":"https://pith.science/pith/4FSB426ZSS6QVC2D7USKOXPX2J.json","view_paper":"https://pith.science/paper/4FSB426Z","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.18729&json=true","fetch_graph":"https://pith.science/api/pith-number/4FSB426ZSS6QVC2D7USKOXPX2J/graph.json","fetch_events":"https://pith.science/api/pith-number/4FSB426ZSS6QVC2D7USKOXPX2J/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4FSB426ZSS6QVC2D7USKOXPX2J/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4FSB426ZSS6QVC2D7USKOXPX2J/action/storage_attestation","attest_author":"https://pith.science/pith/4FSB426ZSS6QVC2D7USKOXPX2J/action/author_attestation","sign_citation":"https://pith.science/pith/4FSB426ZSS6QVC2D7USKOXPX2J/action/citation_signature","submit_replication":"https://pith.science/pith/4FSB426ZSS6QVC2D7USKOXPX2J/action/replication_record"}},"created_at":"2026-07-05T09:58:43.893771+00:00","updated_at":"2026-07-05T09:58:43.893771+00:00"}