{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/arxiv-2605-26693","title":"Model Merging on Loss Landscape: A Geometry Perspective","arxiv_id":"2605.26693","date":"2026-05-26","proceeding":null,"authors":["Juanwu Lu","Anand Bhaskar","Brian Axelrod","Ekaterina Tolstaya","Tristan Emrich"],"abstract":"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 Fréchet 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 decomposes the merging error bound into the subspace Fréchet variance and the residual energy, and provides a closed-form characterization of when curvature-aware merging provably outperforms flat-geometry methods. In addition, our framework unifies both curvature-aware methods and recent spectral methods as special cases of the subspace Fréchet mean with different geometric metrics. Merging fine-tuned CLIP-ViT models on eight image classification tasks, Epistemic Merging strictly outperforms the baselines on all three CLIP-ViT backbones at matched rank, improving the across-task average accuracy and worst-task accuracy on every backbone.","url_abs":"https://arxiv.org/abs/2605.26693","url_pdf":"https://arxiv.org/pdf/2605.26693","source":{"archive":null,"snapshot":"2025-07-28","note":"not in the Papers with Code archive (frozen at the snapshot)","row_kind":"graph","title_abstract_authors_date":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)"},"code_links":[],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2605.26693","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2605.26693"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"mentioned_in_github":null,"is_official":null,"provenance":"deterministic:regex_extraction","mentioned_in_paper":null,"url":"https://github.com/mlfoundations/task_vectors","reach":null}],"summary":{"ran":3,"ran_draft_wrong":1,"unverified":4},"by_repo_kind":{"found_in_text":{"samples":8,"ran":4,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":8,"samples":[{"code_sha256_prefix":"6d77ee840479c381","entry":"convert","repo":"mlfoundations/task_vectors","repo_kind":"found_in_text","path":"src/datasets/cifar10.py","file_url":"https://github.com/mlfoundations/task_vectors/blob/HEAD/src/datasets/cifar10.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"6d77ee840479c381"}},{"code_sha256_prefix":"2d5d7a12d8de02b7","entry":"cosine_lr","repo":"mlfoundations/task_vectors","repo_kind":"found_in_text","path":"src/utils.py","file_url":"https://github.com/mlfoundations/task_vectors/blob/HEAD/src/utils.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"2d5d7a12d8de02b7"}},{"code_sha256_prefix":"b1e9e0d3d7d7e615","entry":"maybe_dictionarize","repo":"mlfoundations/task_vectors","repo_kind":"found_in_text","path":"src/datasets/common.py","file_url":"https://github.com/mlfoundations/task_vectors/blob/HEAD/src/datasets/common.py","link_basis":"plan_row","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"b1e9e0d3d7d7e615"}},{"code_sha256_prefix":"241a2a00ba0c307b","entry":"pretify_classname","repo":"mlfoundations/task_vectors","repo_kind":"found_in_text","path":"src/datasets/eurosat.py","file_url":"https://github.com/mlfoundations/task_vectors/blob/HEAD/src/datasets/eurosat.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"241a2a00ba0c307b"}},{"code_sha256_prefix":"aafcff3ad1ccef6e","entry":"accuracy","repo":"mlfoundations/task_vectors","repo_kind":"found_in_text","path":"src/utils.py","file_url":"https://github.com/mlfoundations/task_vectors/blob/HEAD/src/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"aafcff3ad1ccef6e"}},{"code_sha256_prefix":"3b97c12751c0cf5b","entry":"get_features","repo":"mlfoundations/task_vectors","repo_kind":"found_in_text","path":"src/datasets/common.py","file_url":"https://github.com/mlfoundations/task_vectors/blob/HEAD/src/datasets/common.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"3b97c12751c0cf5b"}},{"code_sha256_prefix":"238ff32fcbf410d6","entry":"get_features_helper","repo":"mlfoundations/task_vectors","repo_kind":"found_in_text","path":"src/datasets/common.py","file_url":"https://github.com/mlfoundations/task_vectors/blob/HEAD/src/datasets/common.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"238ff32fcbf410d6"}},{"code_sha256_prefix":"7d051a7adefedb6c","entry":"torch_load_old","repo":"mlfoundations/task_vectors","repo_kind":"found_in_text","path":"src/utils.py","file_url":"https://github.com/mlfoundations/task_vectors/blob/HEAD/src/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"7d051a7adefedb6c"}}]},"arxiv_metadata":{"licence":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)","fields":["title","abstract","authors","date"],"primary_category":"cs.LG","source":"arxiv_2026.jsonl"},"syntology_extracted_results":null}