{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:FCMRYIMQ3K7E2T635YNZCF7EAL","short_pith_number":"pith:FCMRYIMQ","schema_version":"1.0","canonical_sha256":"28991c2190dabe4d4fdbee1b9117e402f07cea0ee120d9348c1cc3fd0869554d","source":{"kind":"arxiv","id":"2312.04339","version":2},"attestation_state":"computed","paper":{"title":"Merging by Matching Models in Task Parameter Subspaces","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Colin Raffel, Derek Tam, Mohit Bansal","submitted_at":"2023-12-07T14:59:15Z","abstract_excerpt":"Model merging aims to cheaply combine individual task-specific models into a single multitask model. In this work, we view past merging methods as leveraging different notions of a ''task parameter subspace'' in which models are matched before being merged. We connect the task parameter subspace of a given model to its loss landscape and formalize how this approach to model merging can be seen as solving a linear system of equations. While past work has generally been limited to linear systems that have a closed-form solution, we consider using the conjugate gradient method to find a solution."},"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":"2312.04339","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-12-07T14:59:15Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"8865ac7906df5cb6faa2aaedb6c0de2877a27df1ecbb84d518c1efc9ef26a797","abstract_canon_sha256":"dafcd81642f8d2e82cb1189d6c5dc34b6ae31f3f67caa89f0e937661ab64aa81"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:07:31.423063Z","signature_b64":"GzRZqTWyQ8ITEyXl2TYQP0gkK08oPQXS6BJVEd85CiRUtcUBhTXZUb4EgQjBqLNIPi7lH1PerqNw28bZ320uBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"28991c2190dabe4d4fdbee1b9117e402f07cea0ee120d9348c1cc3fd0869554d","last_reissued_at":"2026-07-05T08:07:31.422569Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:07:31.422569Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Merging by Matching Models in Task Parameter Subspaces","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Colin Raffel, Derek Tam, Mohit Bansal","submitted_at":"2023-12-07T14:59:15Z","abstract_excerpt":"Model merging aims to cheaply combine individual task-specific models into a single multitask model. In this work, we view past merging methods as leveraging different notions of a ''task parameter subspace'' in which models are matched before being merged. We connect the task parameter subspace of a given model to its loss landscape and formalize how this approach to model merging can be seen as solving a linear system of equations. While past work has generally been limited to linear systems that have a closed-form solution, we consider using the conjugate gradient method to find a solution."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.04339","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/2312.04339/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":"2312.04339","created_at":"2026-07-05T08:07:31.422629+00:00"},{"alias_kind":"arxiv_version","alias_value":"2312.04339v2","created_at":"2026-07-05T08:07:31.422629+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.04339","created_at":"2026-07-05T08:07:31.422629+00:00"},{"alias_kind":"pith_short_12","alias_value":"FCMRYIMQ3K7E","created_at":"2026-07-05T08:07:31.422629+00:00"},{"alias_kind":"pith_short_16","alias_value":"FCMRYIMQ3K7E2T63","created_at":"2026-07-05T08:07:31.422629+00:00"},{"alias_kind":"pith_short_8","alias_value":"FCMRYIMQ","created_at":"2026-07-05T08:07:31.422629+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2408.07666","citing_title":"Model Merging in LLMs, MLLMs, and Beyond: Methods, Theories, Applications and Opportunities","ref_index":204,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FCMRYIMQ3K7E2T635YNZCF7EAL","json":"https://pith.science/pith/FCMRYIMQ3K7E2T635YNZCF7EAL.json","graph_json":"https://pith.science/api/pith-number/FCMRYIMQ3K7E2T635YNZCF7EAL/graph.json","events_json":"https://pith.science/api/pith-number/FCMRYIMQ3K7E2T635YNZCF7EAL/events.json","paper":"https://pith.science/paper/FCMRYIMQ"},"agent_actions":{"view_html":"https://pith.science/pith/FCMRYIMQ3K7E2T635YNZCF7EAL","download_json":"https://pith.science/pith/FCMRYIMQ3K7E2T635YNZCF7EAL.json","view_paper":"https://pith.science/paper/FCMRYIMQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2312.04339&json=true","fetch_graph":"https://pith.science/api/pith-number/FCMRYIMQ3K7E2T635YNZCF7EAL/graph.json","fetch_events":"https://pith.science/api/pith-number/FCMRYIMQ3K7E2T635YNZCF7EAL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FCMRYIMQ3K7E2T635YNZCF7EAL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FCMRYIMQ3K7E2T635YNZCF7EAL/action/storage_attestation","attest_author":"https://pith.science/pith/FCMRYIMQ3K7E2T635YNZCF7EAL/action/author_attestation","sign_citation":"https://pith.science/pith/FCMRYIMQ3K7E2T635YNZCF7EAL/action/citation_signature","submit_replication":"https://pith.science/pith/FCMRYIMQ3K7E2T635YNZCF7EAL/action/replication_record"}},"created_at":"2026-07-05T08:07:31.422629+00:00","updated_at":"2026-07-05T08:07:31.422629+00:00"}