{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:BDGBLRAUPSC7TOO2W3CZCNBMSG","short_pith_number":"pith:BDGBLRAU","schema_version":"1.0","canonical_sha256":"08cc15c4147c85f9b9dab6c591342c91ad65c304655b4bb33c340fadb2067252","source":{"kind":"arxiv","id":"2407.03641","version":2},"attestation_state":"computed","paper":{"title":"Learning Scalable Model Soup on a Single GPU: An Efficient Subspace Training Strategy","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Fanghui Liu, James T. Kwok, Tao Li, Weisen Jiang, Xiaolin Huang","submitted_at":"2024-07-04T05:23:22Z","abstract_excerpt":"Pre-training followed by fine-tuning is widely adopted among practitioners. The performance can be improved by \"model soups\"~\\cite{wortsman2022model} via exploring various hyperparameter configurations.The Learned-Soup, a variant of model soups, significantly improves the performance but suffers from substantial memory and time costs due to the requirements of (i) having to load all fine-tuned models simultaneously, and (ii) a large computational graph encompassing all fine-tuned models. In this paper, we propose Memory Efficient Hyperplane Learned Soup (MEHL-Soup) to tackle this issue by form"},"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":"2407.03641","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-07-04T05:23:22Z","cross_cats_sorted":[],"title_canon_sha256":"595c3cf376e5744b207edc6895f7649114b0e7b1f16b76d1c89c116509e11fc2","abstract_canon_sha256":"8726840d01e4a2807f1171bd15ba004ac5aec9756cd070f68bc84c348ec1d249"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:47:21.875187Z","signature_b64":"7YEBVjiESRy6RjxMQk5l6FW9Vn+NiZVUBM+GCMuUi6O7mpxSgkUYnBFovAdp+Dgv0Lkpu0D7ryAfuOZCsz1LDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"08cc15c4147c85f9b9dab6c591342c91ad65c304655b4bb33c340fadb2067252","last_reissued_at":"2026-07-05T08:47:21.874662Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:47:21.874662Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning Scalable Model Soup on a Single GPU: An Efficient Subspace Training Strategy","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Fanghui Liu, James T. Kwok, Tao Li, Weisen Jiang, Xiaolin Huang","submitted_at":"2024-07-04T05:23:22Z","abstract_excerpt":"Pre-training followed by fine-tuning is widely adopted among practitioners. The performance can be improved by \"model soups\"~\\cite{wortsman2022model} via exploring various hyperparameter configurations.The Learned-Soup, a variant of model soups, significantly improves the performance but suffers from substantial memory and time costs due to the requirements of (i) having to load all fine-tuned models simultaneously, and (ii) a large computational graph encompassing all fine-tuned models. In this paper, we propose Memory Efficient Hyperplane Learned Soup (MEHL-Soup) to tackle this issue by form"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.03641","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/2407.03641/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":"2407.03641","created_at":"2026-07-05T08:47:21.874726+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.03641v2","created_at":"2026-07-05T08:47:21.874726+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.03641","created_at":"2026-07-05T08:47:21.874726+00:00"},{"alias_kind":"pith_short_12","alias_value":"BDGBLRAUPSC7","created_at":"2026-07-05T08:47:21.874726+00:00"},{"alias_kind":"pith_short_16","alias_value":"BDGBLRAUPSC7TOO2","created_at":"2026-07-05T08:47:21.874726+00:00"},{"alias_kind":"pith_short_8","alias_value":"BDGBLRAU","created_at":"2026-07-05T08:47:21.874726+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":124,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BDGBLRAUPSC7TOO2W3CZCNBMSG","json":"https://pith.science/pith/BDGBLRAUPSC7TOO2W3CZCNBMSG.json","graph_json":"https://pith.science/api/pith-number/BDGBLRAUPSC7TOO2W3CZCNBMSG/graph.json","events_json":"https://pith.science/api/pith-number/BDGBLRAUPSC7TOO2W3CZCNBMSG/events.json","paper":"https://pith.science/paper/BDGBLRAU"},"agent_actions":{"view_html":"https://pith.science/pith/BDGBLRAUPSC7TOO2W3CZCNBMSG","download_json":"https://pith.science/pith/BDGBLRAUPSC7TOO2W3CZCNBMSG.json","view_paper":"https://pith.science/paper/BDGBLRAU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.03641&json=true","fetch_graph":"https://pith.science/api/pith-number/BDGBLRAUPSC7TOO2W3CZCNBMSG/graph.json","fetch_events":"https://pith.science/api/pith-number/BDGBLRAUPSC7TOO2W3CZCNBMSG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BDGBLRAUPSC7TOO2W3CZCNBMSG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BDGBLRAUPSC7TOO2W3CZCNBMSG/action/storage_attestation","attest_author":"https://pith.science/pith/BDGBLRAUPSC7TOO2W3CZCNBMSG/action/author_attestation","sign_citation":"https://pith.science/pith/BDGBLRAUPSC7TOO2W3CZCNBMSG/action/citation_signature","submit_replication":"https://pith.science/pith/BDGBLRAUPSC7TOO2W3CZCNBMSG/action/replication_record"}},"created_at":"2026-07-05T08:47:21.874726+00:00","updated_at":"2026-07-05T08:47:21.874726+00:00"}