{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:G5PHVFIDGAADWC23K6NVBP63LN","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"12d878280646b316ab674d039cd7d2ac0928643b7de47e413c5b730ca4cfa0ba","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-01-29T20:00:26Z","title_canon_sha256":"f7b423489f9f9ad0a232ce96a629dcadcaf3bb5155bc709c0dc487ee8cd977f3"},"schema_version":"1.0","source":{"id":"2601.22285","kind":"arxiv","version":7}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2601.22285","created_at":"2026-05-20T01:05:07Z"},{"alias_kind":"arxiv_version","alias_value":"2601.22285v7","created_at":"2026-05-20T01:05:07Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2601.22285","created_at":"2026-05-20T01:05:07Z"},{"alias_kind":"pith_short_12","alias_value":"G5PHVFIDGAAD","created_at":"2026-05-20T01:05:07Z"},{"alias_kind":"pith_short_16","alias_value":"G5PHVFIDGAADWC23","created_at":"2026-05-20T01:05:07Z"},{"alias_kind":"pith_short_8","alias_value":"G5PHVFID","created_at":"2026-05-20T01:05:07Z"}],"graph_snapshots":[{"event_id":"sha256:b8ecb2ef49a989d209e9a6ff541565b6ee2eb2d41f00c0c16acc8fc8da981635","target":"graph","created_at":"2026-05-20T01:05:07Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Crucially, however, gradient alignment metrics consistently emerge as the most fundamental signals of compatibility."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the chosen set of interpretable pairwise metrics (such as gradient L2 distance) plus L1-regularized linear optimization is sufficient to identify the true drivers of merge success without missing key unmeasured factors."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Mergeability is not intrinsic to models but depends on merging method and tasks, with gradient alignment as the key predictive signal uncovered via interpretable metrics."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Gradient alignment between fine-tuned models is the strongest predictor of successful merging across methods and tasks."}],"snapshot_sha256":"00cb05c533c0e179389859adb011782384da3ab79ca3fe96afc3fd518f657d7d"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2601.22285/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Model merging combines knowledge from separately fine-tuned models, yet the factors driving its success remain poorly understood. While recent work treats mergeability as an intrinsic property of the models, we show with an architecture-agnostic framework that it fundamentally depends on both the merging method and the partner tasks. Using L1-regularized linear optimization over a set of interpretable pairwise metrics (e.g., gradient L_2 distance), we uncover properties correlating with post-merge normalized accuracy across five merging methods. We find architecture- and method-specific variat","authors_text":"Bo Zhao, Emanuele Rodol\\`a, Luca Zhou, Rose Yu","cross_cats":[],"headline":"Gradient alignment between fine-tuned models is the strongest predictor of successful merging across methods and tasks.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-01-29T20:00:26Z","title":"Demystifying Mergeability: Interpretable Properties to Predict Model Merging Success"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2601.22285","kind":"arxiv","version":7},"verdict":{"created_at":"2026-05-16T09:47:05.919474Z","id":"dbf9c059-828e-416f-a712-6c617498264b","model_set":{"reader":"grok-4.3"},"one_line_summary":"Mergeability is not intrinsic to models but depends on merging method and tasks, with gradient alignment as the key predictive signal uncovered via interpretable metrics.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Gradient alignment between fine-tuned models is the strongest predictor of successful merging across methods and tasks.","strongest_claim":"Crucially, however, gradient alignment metrics consistently emerge as the most fundamental signals of compatibility.","weakest_assumption":"That the chosen set of interpretable pairwise metrics (such as gradient L2 distance) plus L1-regularized linear optimization is sufficient to identify the true drivers of merge success without missing key unmeasured factors."}},"verdict_id":"dbf9c059-828e-416f-a712-6c617498264b"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:37ef8bc7589c96f4640767a50c24236831db3816a43241684ca95538dc82d65a","target":"record","created_at":"2026-05-20T01:05:07Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"12d878280646b316ab674d039cd7d2ac0928643b7de47e413c5b730ca4cfa0ba","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-01-29T20:00:26Z","title_canon_sha256":"f7b423489f9f9ad0a232ce96a629dcadcaf3bb5155bc709c0dc487ee8cd977f3"},"schema_version":"1.0","source":{"id":"2601.22285","kind":"arxiv","version":7}},"canonical_sha256":"375e7a950330003b0b5b579b50bfdb5b67b3a269208585bf5dd3c0aa84c7e4d7","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"375e7a950330003b0b5b579b50bfdb5b67b3a269208585bf5dd3c0aa84c7e4d7","first_computed_at":"2026-05-20T01:05:07.281625Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T01:05:07.281625Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"JIywg1QJMsZH8Vwh082XMDgm0ALJgxyNgA9PeRfKuerIQaW4uKwgqAWR/Q6rq8xAnuDk+ROM0vMY5tE3GAn2Dg==","signature_status":"signed_v1","signed_at":"2026-05-20T01:05:07.282667Z","signed_message":"canonical_sha256_bytes"},"source_id":"2601.22285","source_kind":"arxiv","source_version":7}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:37ef8bc7589c96f4640767a50c24236831db3816a43241684ca95538dc82d65a","sha256:b8ecb2ef49a989d209e9a6ff541565b6ee2eb2d41f00c0c16acc8fc8da981635"],"state_sha256":"bcd8b5c807d8adf0297de08882cca6d01c51fdee3717205ae995874613f2711b"}