{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2IEYPBOPY6HTKQXPMQJTDW7V4K","short_pith_number":"pith:2IEYPBOP","schema_version":"1.0","canonical_sha256":"d2098785cfc78f3542ef641331dbf5e2a94fb621ab935816ea51334057262f2b","source":{"kind":"arxiv","id":"2503.18986","version":1},"attestation_state":"computed","paper":{"title":"SplitFrozen: Split Learning with Device-side Model Frozen for Fine-Tuning LLM on Heterogeneous Resource-Constrained Devices","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Haipeng Yao, Jian Ma, Jun Jiang, Qimei Cui, Xiaofeng Tao, Xinchen Lyu","submitted_at":"2025-03-23T08:03:44Z","abstract_excerpt":"Fine-tuning large language models (LLMs) on private, on-device data can empower tailored personalized AI agents. However, fine-tuning LLMs on resource-constrained edge devices faces significant challenges, including excessive computation overhead, device heterogeneity, and data imbalance. This paper proposes SplitFrozen, a split learning framework that enables efficient LLM fine-tuning by strategically freezing device-side model layers while centralizing parameter-efficient fine-tuning on the server. Our framework partitions LLMs into device-side frozen layers and server-side fine-tuning layer"},"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":"2503.18986","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-03-23T08:03:44Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"a5ee72eb1a370f070a5e6f17234c936b72f600b4af9a63f7519f978b698b6d7d","abstract_canon_sha256":"672a03fd66da7b84b20775d8fddc5b4e82fcc1a5c6c725ad7552c1e810eb8dc1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:38:47.621693Z","signature_b64":"DjmaIZxgewpcOz6WCMfwtCda+sXDPRqDyrvX7kEg40F6u2+ApdO4PDac44ANfPMmX/FfxWeUSCCgMOzKWqfLBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d2098785cfc78f3542ef641331dbf5e2a94fb621ab935816ea51334057262f2b","last_reissued_at":"2026-07-05T10:38:47.621219Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:38:47.621219Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SplitFrozen: Split Learning with Device-side Model Frozen for Fine-Tuning LLM on Heterogeneous Resource-Constrained Devices","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Haipeng Yao, Jian Ma, Jun Jiang, Qimei Cui, Xiaofeng Tao, Xinchen Lyu","submitted_at":"2025-03-23T08:03:44Z","abstract_excerpt":"Fine-tuning large language models (LLMs) on private, on-device data can empower tailored personalized AI agents. However, fine-tuning LLMs on resource-constrained edge devices faces significant challenges, including excessive computation overhead, device heterogeneity, and data imbalance. This paper proposes SplitFrozen, a split learning framework that enables efficient LLM fine-tuning by strategically freezing device-side model layers while centralizing parameter-efficient fine-tuning on the server. Our framework partitions LLMs into device-side frozen layers and server-side fine-tuning layer"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.18986","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/2503.18986/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":"2503.18986","created_at":"2026-07-05T10:38:47.621274+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.18986v1","created_at":"2026-07-05T10:38:47.621274+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.18986","created_at":"2026-07-05T10:38:47.621274+00:00"},{"alias_kind":"pith_short_12","alias_value":"2IEYPBOPY6HT","created_at":"2026-07-05T10:38:47.621274+00:00"},{"alias_kind":"pith_short_16","alias_value":"2IEYPBOPY6HTKQXP","created_at":"2026-07-05T10:38:47.621274+00:00"},{"alias_kind":"pith_short_8","alias_value":"2IEYPBOP","created_at":"2026-07-05T10:38:47.621274+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.26120","citing_title":"Semantic-aware Token Selection and Resource Optimization for Communication-efficient Split Federated Fine-tuning in Edge Intelligence","ref_index":21,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2IEYPBOPY6HTKQXPMQJTDW7V4K","json":"https://pith.science/pith/2IEYPBOPY6HTKQXPMQJTDW7V4K.json","graph_json":"https://pith.science/api/pith-number/2IEYPBOPY6HTKQXPMQJTDW7V4K/graph.json","events_json":"https://pith.science/api/pith-number/2IEYPBOPY6HTKQXPMQJTDW7V4K/events.json","paper":"https://pith.science/paper/2IEYPBOP"},"agent_actions":{"view_html":"https://pith.science/pith/2IEYPBOPY6HTKQXPMQJTDW7V4K","download_json":"https://pith.science/pith/2IEYPBOPY6HTKQXPMQJTDW7V4K.json","view_paper":"https://pith.science/paper/2IEYPBOP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.18986&json=true","fetch_graph":"https://pith.science/api/pith-number/2IEYPBOPY6HTKQXPMQJTDW7V4K/graph.json","fetch_events":"https://pith.science/api/pith-number/2IEYPBOPY6HTKQXPMQJTDW7V4K/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2IEYPBOPY6HTKQXPMQJTDW7V4K/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2IEYPBOPY6HTKQXPMQJTDW7V4K/action/storage_attestation","attest_author":"https://pith.science/pith/2IEYPBOPY6HTKQXPMQJTDW7V4K/action/author_attestation","sign_citation":"https://pith.science/pith/2IEYPBOPY6HTKQXPMQJTDW7V4K/action/citation_signature","submit_replication":"https://pith.science/pith/2IEYPBOPY6HTKQXPMQJTDW7V4K/action/replication_record"}},"created_at":"2026-07-05T10:38:47.621274+00:00","updated_at":"2026-07-05T10:38:47.621274+00:00"}