{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:RUF6BGSYN3XGX4ZLX652ZGAVDR","short_pith_number":"pith:RUF6BGSY","schema_version":"1.0","canonical_sha256":"8d0be09a586eee6bf32bbfbbac98151c4380fa522de1a2b85aa268a9517c59dc","source":{"kind":"arxiv","id":"2507.04455","version":1},"attestation_state":"computed","paper":{"title":"GradOT: Training-free Gradient-preserving Offsite-tuning for Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jianke Zhu, Kaixin Wu, Kai Yao, Lichun Li, Penglei Gao, Wei Wang, Yinggui Wang, Yixin Ji, Yuan Zhao, Zhaorui Tan","submitted_at":"2025-07-06T16:27:27Z","abstract_excerpt":"The rapid growth of large language models (LLMs) with traditional centralized fine-tuning emerges as a key technique for adapting these models to domain-specific challenges, yielding privacy risks for both model and data owners. One promising solution, called offsite-tuning (OT), is proposed to address these challenges, where a weaker emulator is compressed from the original model and further fine-tuned with adapter to enhance privacy. However, the existing OT-based methods require high computational costs and lack theoretical analysis. This paper introduces a novel OT approach based on gradie"},"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":"2507.04455","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-07-06T16:27:27Z","cross_cats_sorted":[],"title_canon_sha256":"88e7a9e063184be55c16f3d1c12db5144ca43106c1ad23c86bb3d307e32d9db6","abstract_canon_sha256":"45d6ea0e1aba8158b91c555a247c89b27b654ccea81b79f43aa750c388e8e83d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:32:48.387612Z","signature_b64":"P3jHFFzFWiLC7L8HcFyLy3tbIiQYbQhee8/N0jUlFYdSgm3e2B9Aye0X+MQElXQe55rb9eJnaCTcidqWBC1vBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8d0be09a586eee6bf32bbfbbac98151c4380fa522de1a2b85aa268a9517c59dc","last_reissued_at":"2026-07-05T11:32:48.387147Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:32:48.387147Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GradOT: Training-free Gradient-preserving Offsite-tuning for Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jianke Zhu, Kaixin Wu, Kai Yao, Lichun Li, Penglei Gao, Wei Wang, Yinggui Wang, Yixin Ji, Yuan Zhao, Zhaorui Tan","submitted_at":"2025-07-06T16:27:27Z","abstract_excerpt":"The rapid growth of large language models (LLMs) with traditional centralized fine-tuning emerges as a key technique for adapting these models to domain-specific challenges, yielding privacy risks for both model and data owners. One promising solution, called offsite-tuning (OT), is proposed to address these challenges, where a weaker emulator is compressed from the original model and further fine-tuned with adapter to enhance privacy. However, the existing OT-based methods require high computational costs and lack theoretical analysis. This paper introduces a novel OT approach based on gradie"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.04455","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/2507.04455/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":"2507.04455","created_at":"2026-07-05T11:32:48.387203+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.04455v1","created_at":"2026-07-05T11:32:48.387203+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.04455","created_at":"2026-07-05T11:32:48.387203+00:00"},{"alias_kind":"pith_short_12","alias_value":"RUF6BGSYN3XG","created_at":"2026-07-05T11:32:48.387203+00:00"},{"alias_kind":"pith_short_16","alias_value":"RUF6BGSYN3XGX4ZL","created_at":"2026-07-05T11:32:48.387203+00:00"},{"alias_kind":"pith_short_8","alias_value":"RUF6BGSY","created_at":"2026-07-05T11:32:48.387203+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RUF6BGSYN3XGX4ZLX652ZGAVDR","json":"https://pith.science/pith/RUF6BGSYN3XGX4ZLX652ZGAVDR.json","graph_json":"https://pith.science/api/pith-number/RUF6BGSYN3XGX4ZLX652ZGAVDR/graph.json","events_json":"https://pith.science/api/pith-number/RUF6BGSYN3XGX4ZLX652ZGAVDR/events.json","paper":"https://pith.science/paper/RUF6BGSY"},"agent_actions":{"view_html":"https://pith.science/pith/RUF6BGSYN3XGX4ZLX652ZGAVDR","download_json":"https://pith.science/pith/RUF6BGSYN3XGX4ZLX652ZGAVDR.json","view_paper":"https://pith.science/paper/RUF6BGSY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.04455&json=true","fetch_graph":"https://pith.science/api/pith-number/RUF6BGSYN3XGX4ZLX652ZGAVDR/graph.json","fetch_events":"https://pith.science/api/pith-number/RUF6BGSYN3XGX4ZLX652ZGAVDR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RUF6BGSYN3XGX4ZLX652ZGAVDR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RUF6BGSYN3XGX4ZLX652ZGAVDR/action/storage_attestation","attest_author":"https://pith.science/pith/RUF6BGSYN3XGX4ZLX652ZGAVDR/action/author_attestation","sign_citation":"https://pith.science/pith/RUF6BGSYN3XGX4ZLX652ZGAVDR/action/citation_signature","submit_replication":"https://pith.science/pith/RUF6BGSYN3XGX4ZLX652ZGAVDR/action/replication_record"}},"created_at":"2026-07-05T11:32:48.387203+00:00","updated_at":"2026-07-05T11:32:48.387203+00:00"}