{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:ETCXPF3EIBIFUSCE6JIEV2FTGN","short_pith_number":"pith:ETCXPF3E","canonical_record":{"source":{"id":"2406.01514","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-03T16:46:18Z","cross_cats_sorted":["cs.AI","cs.CR"],"title_canon_sha256":"201cdaf18eb5cc9ef5cdb04170fcc0385b8fa331e3567c2017d21e80db4bcdc1","abstract_canon_sha256":"4ce7d3d0070b0b9d87920a775d3f2f23cf8afeefc325f61f98ac8b871e301e9e"},"schema_version":"1.0"},"canonical_sha256":"24c577976440505a4844f2504ae8b33356edc51ee358ea9a229e04eb5a1b0912","source":{"kind":"arxiv","id":"2406.01514","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.01514","created_at":"2026-07-05T08:28:11Z"},{"alias_kind":"arxiv_version","alias_value":"2406.01514v3","created_at":"2026-07-05T08:28:11Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.01514","created_at":"2026-07-05T08:28:11Z"},{"alias_kind":"pith_short_12","alias_value":"ETCXPF3EIBIF","created_at":"2026-07-05T08:28:11Z"},{"alias_kind":"pith_short_16","alias_value":"ETCXPF3EIBIFUSCE","created_at":"2026-07-05T08:28:11Z"},{"alias_kind":"pith_short_8","alias_value":"ETCXPF3E","created_at":"2026-07-05T08:28:11Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:ETCXPF3EIBIFUSCE6JIEV2FTGN","target":"record","payload":{"canonical_record":{"source":{"id":"2406.01514","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-03T16:46:18Z","cross_cats_sorted":["cs.AI","cs.CR"],"title_canon_sha256":"201cdaf18eb5cc9ef5cdb04170fcc0385b8fa331e3567c2017d21e80db4bcdc1","abstract_canon_sha256":"4ce7d3d0070b0b9d87920a775d3f2f23cf8afeefc325f61f98ac8b871e301e9e"},"schema_version":"1.0"},"canonical_sha256":"24c577976440505a4844f2504ae8b33356edc51ee358ea9a229e04eb5a1b0912","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:28:11.089407Z","signature_b64":"iL/I97QdyteVC1HbGcJTK98CFfuU7XHjAMkoBYR0MOKe4cUyQ9rFpgy6Vy4Fi2k8rDJs4DVZE4PGXgBlt7NfDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"24c577976440505a4844f2504ae8b33356edc51ee358ea9a229e04eb5a1b0912","last_reissued_at":"2026-07-05T08:28:11.088936Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:28:11.088936Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2406.01514","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:28:11Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"oAxDi7Uaj232QYxJpnseUsQ4BweTIT4gQGkYnZkwdbJ8VnawlInGhGoFLimZlQoRt1AL6BSU1rGSt8vXuR8uAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T16:33:22.642192Z"},"content_sha256":"544573ebaaa44208de9ecb3d0d61e88d7c4ff7726a1ecc78056774cdc301fad6","schema_version":"1.0","event_id":"sha256:544573ebaaa44208de9ecb3d0d61e88d7c4ff7726a1ecc78056774cdc301fad6"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:ETCXPF3EIBIFUSCE6JIEV2FTGN","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Decoupled Alignment for Robust Plug-and-Play Adaptation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CR"],"primary_cat":"cs.CL","authors_text":"Han Liu, Haozheng Luo, Jerry Yao-Chieh Hu, Jiahao Yu, Jialong Li, Wenxin Zhang, Xinyu Xing","submitted_at":"2024-06-03T16:46:18Z","abstract_excerpt":"We introduce a low-resource safety enhancement method for aligning large language models (LLMs) without the need for supervised fine-tuning (SFT) or reinforcement learning from human feedback (RLHF). Our main idea is to exploit knowledge distillation to extract the alignment information from existing well-aligned LLMs and integrate it into unaligned LLMs in a plug-and-play fashion. Methodology, we employ delta debugging to identify the critical components of knowledge necessary for effective distillation. On the harmful question dataset, our method significantly enhances the average defense su"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.01514","kind":"arxiv","version":3},"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/2406.01514/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:28:11Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"jTM9x5+uNeiPqiwpaTQzsVKpL34PME/DV2b5E2jV4KFCoJnZj+MfTRmUz83hNsVlpV91s8R+Z9M9Hz4v2zGNBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T16:33:22.642753Z"},"content_sha256":"966a8966b60e1a0ace0685c1870b6e46fd5cdb6f4fde75a99b8a984be6180a46","schema_version":"1.0","event_id":"sha256:966a8966b60e1a0ace0685c1870b6e46fd5cdb6f4fde75a99b8a984be6180a46"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/ETCXPF3EIBIFUSCE6JIEV2FTGN/bundle.json","state_url":"https://pith.science/pith/ETCXPF3EIBIFUSCE6JIEV2FTGN/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/ETCXPF3EIBIFUSCE6JIEV2FTGN/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-05T16:33:22Z","links":{"resolver":"https://pith.science/pith/ETCXPF3EIBIFUSCE6JIEV2FTGN","bundle":"https://pith.science/pith/ETCXPF3EIBIFUSCE6JIEV2FTGN/bundle.json","state":"https://pith.science/pith/ETCXPF3EIBIFUSCE6JIEV2FTGN/state.json","well_known_bundle":"https://pith.science/.well-known/pith/ETCXPF3EIBIFUSCE6JIEV2FTGN/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:ETCXPF3EIBIFUSCE6JIEV2FTGN","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":"4ce7d3d0070b0b9d87920a775d3f2f23cf8afeefc325f61f98ac8b871e301e9e","cross_cats_sorted":["cs.AI","cs.CR"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-03T16:46:18Z","title_canon_sha256":"201cdaf18eb5cc9ef5cdb04170fcc0385b8fa331e3567c2017d21e80db4bcdc1"},"schema_version":"1.0","source":{"id":"2406.01514","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.01514","created_at":"2026-07-05T08:28:11Z"},{"alias_kind":"arxiv_version","alias_value":"2406.01514v3","created_at":"2026-07-05T08:28:11Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.01514","created_at":"2026-07-05T08:28:11Z"},{"alias_kind":"pith_short_12","alias_value":"ETCXPF3EIBIF","created_at":"2026-07-05T08:28:11Z"},{"alias_kind":"pith_short_16","alias_value":"ETCXPF3EIBIFUSCE","created_at":"2026-07-05T08:28:11Z"},{"alias_kind":"pith_short_8","alias_value":"ETCXPF3E","created_at":"2026-07-05T08:28:11Z"}],"graph_snapshots":[{"event_id":"sha256:966a8966b60e1a0ace0685c1870b6e46fd5cdb6f4fde75a99b8a984be6180a46","target":"graph","created_at":"2026-07-05T08:28:11Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2406.01514/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We introduce a low-resource safety enhancement method for aligning large language models (LLMs) without the need for supervised fine-tuning (SFT) or reinforcement learning from human feedback (RLHF). Our main idea is to exploit knowledge distillation to extract the alignment information from existing well-aligned LLMs and integrate it into unaligned LLMs in a plug-and-play fashion. Methodology, we employ delta debugging to identify the critical components of knowledge necessary for effective distillation. On the harmful question dataset, our method significantly enhances the average defense su","authors_text":"Han Liu, Haozheng Luo, Jerry Yao-Chieh Hu, Jiahao Yu, Jialong Li, Wenxin Zhang, Xinyu Xing","cross_cats":["cs.AI","cs.CR"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-03T16:46:18Z","title":"Decoupled Alignment for Robust Plug-and-Play Adaptation"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.01514","kind":"arxiv","version":3},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:544573ebaaa44208de9ecb3d0d61e88d7c4ff7726a1ecc78056774cdc301fad6","target":"record","created_at":"2026-07-05T08:28:11Z","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":"4ce7d3d0070b0b9d87920a775d3f2f23cf8afeefc325f61f98ac8b871e301e9e","cross_cats_sorted":["cs.AI","cs.CR"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-03T16:46:18Z","title_canon_sha256":"201cdaf18eb5cc9ef5cdb04170fcc0385b8fa331e3567c2017d21e80db4bcdc1"},"schema_version":"1.0","source":{"id":"2406.01514","kind":"arxiv","version":3}},"canonical_sha256":"24c577976440505a4844f2504ae8b33356edc51ee358ea9a229e04eb5a1b0912","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"24c577976440505a4844f2504ae8b33356edc51ee358ea9a229e04eb5a1b0912","first_computed_at":"2026-07-05T08:28:11.088936Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:28:11.088936Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"iL/I97QdyteVC1HbGcJTK98CFfuU7XHjAMkoBYR0MOKe4cUyQ9rFpgy6Vy4Fi2k8rDJs4DVZE4PGXgBlt7NfDA==","signature_status":"signed_v1","signed_at":"2026-07-05T08:28:11.089407Z","signed_message":"canonical_sha256_bytes"},"source_id":"2406.01514","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:544573ebaaa44208de9ecb3d0d61e88d7c4ff7726a1ecc78056774cdc301fad6","sha256:966a8966b60e1a0ace0685c1870b6e46fd5cdb6f4fde75a99b8a984be6180a46"],"state_sha256":"00b99c3dbca2a7e224b64ec76ebb687b4b441b3bd962ab1bc05f6c2d475982c2"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"y4wvSOD/Y3i74xKkMY/O9ePH1EEWsQDDyJToezFo0jRy0rPcfSefdFuvcgaXxfSXpVhnL8CCZWJrx5iNSCf+Bw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-05T16:33:22.649015Z","bundle_sha256":"e7895262f8e3b5aad1c45fc6e66359f3319f4260f3b2805ba9eabf7b8e9819e1"}}