{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:SYPRT7MHZKDXNZKC7BZFYD6SBC","short_pith_number":"pith:SYPRT7MH","schema_version":"1.0","canonical_sha256":"961f19fd87ca8776e542f8725c0fd208a0d38070dabd41d4aec36c4e6613445b","source":{"kind":"arxiv","id":"2605.26693","version":1},"attestation_state":"computed","paper":{"title":"Model Merging on Loss Landscape: A Geometry Perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Anand Bhaskar, Brian Axelrod, Ekaterina Tolstaya, Juanwu Lu, Tristan Emrich","submitted_at":"2026-05-26T08:33:30Z","abstract_excerpt":"Model merging offers a promising avenue for knowledge integration and parallel development without retraining. Yet, existing methods either ignore the geometry of the loss landscape or rely on intractable full-space Hessian approximations. We propose EpiMer, a framework that casts model merging as solving the Fr\\'echet mean on a Riemannian manifold and restricts the computation to a low-rank subspace spanned by the task vectors. With the expected Hessian as the metric, we reveal a connection between local curvature and epistemic uncertainty of the parameters. Our theoretical analysis decompose"},"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":"2605.26693","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-26T08:33:30Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"ba57d673620176c47a3bf1c45cee033bcf99f75e6b6a03a08f574283f5aae1c3","abstract_canon_sha256":"dca5337b7825e2762e448cf6f14d486b47ae090bcaf630dd113405a73ed99d93"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T01:06:06.420488Z","signature_b64":"bTPD5cXHb8kbdvmtQO4XoYgHQaP3yZmPcJXeUhWQ7njo4GyH/sa5XaVNQcp5cxkTdk5uvbHXt/fweQPwKfCYCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"961f19fd87ca8776e542f8725c0fd208a0d38070dabd41d4aec36c4e6613445b","last_reissued_at":"2026-05-27T01:06:06.419692Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T01:06:06.419692Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Model Merging on Loss Landscape: A Geometry Perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Anand Bhaskar, Brian Axelrod, Ekaterina Tolstaya, Juanwu Lu, Tristan Emrich","submitted_at":"2026-05-26T08:33:30Z","abstract_excerpt":"Model merging offers a promising avenue for knowledge integration and parallel development without retraining. Yet, existing methods either ignore the geometry of the loss landscape or rely on intractable full-space Hessian approximations. We propose EpiMer, a framework that casts model merging as solving the Fr\\'echet mean on a Riemannian manifold and restricts the computation to a low-rank subspace spanned by the task vectors. With the expected Hessian as the metric, we reveal a connection between local curvature and epistemic uncertainty of the parameters. Our theoretical analysis decompose"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.26693","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/2605.26693/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":"2605.26693","created_at":"2026-05-27T01:06:06.419814+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.26693v1","created_at":"2026-05-27T01:06:06.419814+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.26693","created_at":"2026-05-27T01:06:06.419814+00:00"},{"alias_kind":"pith_short_12","alias_value":"SYPRT7MHZKDX","created_at":"2026-05-27T01:06:06.419814+00:00"},{"alias_kind":"pith_short_16","alias_value":"SYPRT7MHZKDXNZKC","created_at":"2026-05-27T01:06:06.419814+00:00"},{"alias_kind":"pith_short_8","alias_value":"SYPRT7MH","created_at":"2026-05-27T01:06:06.419814+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SYPRT7MHZKDXNZKC7BZFYD6SBC","json":"https://pith.science/pith/SYPRT7MHZKDXNZKC7BZFYD6SBC.json","graph_json":"https://pith.science/api/pith-number/SYPRT7MHZKDXNZKC7BZFYD6SBC/graph.json","events_json":"https://pith.science/api/pith-number/SYPRT7MHZKDXNZKC7BZFYD6SBC/events.json","paper":"https://pith.science/paper/SYPRT7MH"},"agent_actions":{"view_html":"https://pith.science/pith/SYPRT7MHZKDXNZKC7BZFYD6SBC","download_json":"https://pith.science/pith/SYPRT7MHZKDXNZKC7BZFYD6SBC.json","view_paper":"https://pith.science/paper/SYPRT7MH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.26693&json=true","fetch_graph":"https://pith.science/api/pith-number/SYPRT7MHZKDXNZKC7BZFYD6SBC/graph.json","fetch_events":"https://pith.science/api/pith-number/SYPRT7MHZKDXNZKC7BZFYD6SBC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SYPRT7MHZKDXNZKC7BZFYD6SBC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SYPRT7MHZKDXNZKC7BZFYD6SBC/action/storage_attestation","attest_author":"https://pith.science/pith/SYPRT7MHZKDXNZKC7BZFYD6SBC/action/author_attestation","sign_citation":"https://pith.science/pith/SYPRT7MHZKDXNZKC7BZFYD6SBC/action/citation_signature","submit_replication":"https://pith.science/pith/SYPRT7MHZKDXNZKC7BZFYD6SBC/action/replication_record"}},"created_at":"2026-05-27T01:06:06.419814+00:00","updated_at":"2026-05-27T01:06:06.419814+00:00"}