{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:5GDYLWOL7NGIY3SRLTSKHFM63X","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":"c31255ea55fe59d514aec810f8bb685f645b85e58b45ffc7ddd80805d91a6a11","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T00:36:47Z","title_canon_sha256":"7d511b61ff1cd74c70bf56c33d21e3963dd51d07833af05b906bc12ee40a9b8d"},"schema_version":"1.0","source":{"id":"2605.12843","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12843","created_at":"2026-05-18T03:09:11Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12843v1","created_at":"2026-05-18T03:09:11Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12843","created_at":"2026-05-18T03:09:11Z"},{"alias_kind":"pith_short_12","alias_value":"5GDYLWOL7NGI","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"5GDYLWOL7NGIY3SR","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"5GDYLWOL","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:de1a3e60595e17ecabb9ae8e4ee938affb1a55b38b2880ac6eb5655aaa5285c1","target":"graph","created_at":"2026-05-18T03:09:11Z","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":"Across extensive benchmarks, including up to 20-task merging in vision and 5-task merging in language, BMM consistently outperforms all plug-and-play anchor baselines (e.g., TA, WUDI-Merging, and TSV). In particular, on the ViT-L/14 benchmark for 8-task merging, a single merged model reaches 95.1, closely matching the average performance of eight task-specific experts (95.8)."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The claimed alignment between activation statistics and task vectors that enables the data-free Gram-matrix estimation, together with the assumption that the inner-level Bayesian regression under the anchor prior produces a solution that generalizes without hidden post-hoc adjustments."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Bayesian Model Merging introduces a bi-level optimization framework that merges task-specific models via closed-form Bayesian regression with an anchor prior and global hyperparameter search, outperforming baselines and nearly matching expert averages on up to 20-task vision and 5-task language Merg"},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Bayesian Model Merging fuses task-specific models into one via inner Bayesian regression under anchor priors and outer Bayesian optimization of per-module hyperparameters."}],"snapshot_sha256":"ab7c2ad3c1596973df5628ea20ea381dffce5e00899de90a989efa72921045f4"},"formal_canon":{"evidence_count":1,"snapshot_sha256":"c3a047cd8172fed4c4a0b52f3b10ba64430131dd805daceefaa009725d2019bf"},"paper":{"abstract_excerpt":"Model merging aims to combine multiple task-specific expert models into a single model without joint retraining, offering a practical alternative to multi-task learning when data access or computational budget is limited. Existing methods, however, face two key limitations: (1) they overlook the valuable inductive bias of strong anchor models and estimate the merged weights from scratch, and (2) they rely on a shared hyperparameter setting across different modules of the network, lacking a global optimization strategy. This paper introduces Bayesian Model Merging (BMM), a plug-and-play bi-leve","authors_text":"Kaiyang Li, Qing Su, Shaobo Han, Shihao Ji","cross_cats":["cs.AI"],"headline":"Bayesian Model Merging fuses task-specific models into one via inner Bayesian regression under anchor priors and outer Bayesian optimization of per-module hyperparameters.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T00:36:47Z","title":"Bayesian Model Merging"},"references":{"count":66,"internal_anchors":11,"resolved_work":66,"sample":[{"cited_arxiv_id":"1706.05098","doi":"","is_internal_anchor":true,"ref_index":1,"title":"An Overview of Multi-Task Learning in Deep Neural Networks","work_id":"2c849184-06ce-40df-8143-ec29eb925f39","year":2017},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Matena and Colin A","work_id":"f85bb7a2-7dff-4a41-aa1a-a2013fc51ef4","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Editing models with task arithmetic","work_id":"a4f3855c-b66f-4915-8b5b-581e494a8cb5","year":2023},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Raffel, and Mohit Bansal","work_id":"ac02f012-a7c9-4546-a950-364c16bc8719","year":2023},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"The hugging face hub","work_id":"051718a0-462f-4455-962a-aa2930577451","year":2026}],"snapshot_sha256":"507badcd941774027e9060eec0a645288c2ed7293cb430fdf90147a4c0ac252a"},"source":{"id":"2605.12843","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T20:38:04.112438Z","id":"755c71a3-bfc8-40b5-9114-45252ba5e03b","model_set":{"reader":"grok-4.3"},"one_line_summary":"Bayesian Model Merging introduces a bi-level optimization framework that merges task-specific models via closed-form Bayesian regression with an anchor prior and global hyperparameter search, outperforming baselines and nearly matching expert averages on up to 20-task vision and 5-task language Merg","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Bayesian Model Merging fuses task-specific models into one via inner Bayesian regression under anchor priors and outer Bayesian optimization of per-module hyperparameters.","strongest_claim":"Across extensive benchmarks, including up to 20-task merging in vision and 5-task merging in language, BMM consistently outperforms all plug-and-play anchor baselines (e.g., TA, WUDI-Merging, and TSV). In particular, on the ViT-L/14 benchmark for 8-task merging, a single merged model reaches 95.1, closely matching the average performance of eight task-specific experts (95.8).","weakest_assumption":"The claimed alignment between activation statistics and task vectors that enables the data-free Gram-matrix estimation, together with the assumption that the inner-level Bayesian regression under the anchor prior produces a solution that generalizes without hidden post-hoc adjustments."}},"verdict_id":"755c71a3-bfc8-40b5-9114-45252ba5e03b"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:03d6206c5bf19387d88f98cb09162211dca0a96d8bb9cfe2a4484d0255133a15","target":"record","created_at":"2026-05-18T03:09:11Z","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":"c31255ea55fe59d514aec810f8bb685f645b85e58b45ffc7ddd80805d91a6a11","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T00:36:47Z","title_canon_sha256":"7d511b61ff1cd74c70bf56c33d21e3963dd51d07833af05b906bc12ee40a9b8d"},"schema_version":"1.0","source":{"id":"2605.12843","kind":"arxiv","version":1}},"canonical_sha256":"e98785d9cbfb4c8c6e515ce4a3959eddff2175548d26e17b1e05dfdf0e48e916","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e98785d9cbfb4c8c6e515ce4a3959eddff2175548d26e17b1e05dfdf0e48e916","first_computed_at":"2026-05-18T03:09:11.934066Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T03:09:11.934066Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"2tKE5AB/YqjJ1TVADBoc4/sYutUCWoeGpfnokQY+czKQi8QZNXkbw8kjUMR1q8iKiwPBUrSkl/SWdjYe5memCA==","signature_status":"signed_v1","signed_at":"2026-05-18T03:09:11.934739Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.12843","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:03d6206c5bf19387d88f98cb09162211dca0a96d8bb9cfe2a4484d0255133a15","sha256:de1a3e60595e17ecabb9ae8e4ee938affb1a55b38b2880ac6eb5655aaa5285c1"],"state_sha256":"4cdeee3465d79c910ae62d10daf33c39604a9e34c03bd37d3dc46abf1f94dd49"}