{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:JWGPNU354KXCEDWSZLVCDE5IMR","short_pith_number":"pith:JWGPNU35","canonical_record":{"source":{"id":"2310.09573","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2023-10-14T12:51:38Z","cross_cats_sorted":[],"title_canon_sha256":"f6acd2cfb03fe13e33fc335d22599cd853aa1f1057429cd819c1da5eb7de94d2","abstract_canon_sha256":"07ea86e724df754a3b758f1263d660b446960c17e27f4e1a4c9d2b8aab5d768d"},"schema_version":"1.0"},"canonical_sha256":"4d8cf6d37de2ae220ed2caea2193a8645a52343e7e7407128ed0bb4897abe02b","source":{"kind":"arxiv","id":"2310.09573","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.09573","created_at":"2026-07-05T07:01:08Z"},{"alias_kind":"arxiv_version","alias_value":"2310.09573v1","created_at":"2026-07-05T07:01:08Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.09573","created_at":"2026-07-05T07:01:08Z"},{"alias_kind":"pith_short_12","alias_value":"JWGPNU354KXC","created_at":"2026-07-05T07:01:08Z"},{"alias_kind":"pith_short_16","alias_value":"JWGPNU354KXCEDWS","created_at":"2026-07-05T07:01:08Z"},{"alias_kind":"pith_short_8","alias_value":"JWGPNU35","created_at":"2026-07-05T07:01:08Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:JWGPNU354KXCEDWSZLVCDE5IMR","target":"record","payload":{"canonical_record":{"source":{"id":"2310.09573","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2023-10-14T12:51:38Z","cross_cats_sorted":[],"title_canon_sha256":"f6acd2cfb03fe13e33fc335d22599cd853aa1f1057429cd819c1da5eb7de94d2","abstract_canon_sha256":"07ea86e724df754a3b758f1263d660b446960c17e27f4e1a4c9d2b8aab5d768d"},"schema_version":"1.0"},"canonical_sha256":"4d8cf6d37de2ae220ed2caea2193a8645a52343e7e7407128ed0bb4897abe02b","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:01:08.078106Z","signature_b64":"PGYYY7rtAXbQ8jMv7VpAsY7VYa4zraaAPrSEkjgOYXa1sA7Xj9Ni2vYGb/cMXgxw+x7NKnEilRHVMKEGjis2Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4d8cf6d37de2ae220ed2caea2193a8645a52343e7e7407128ed0bb4897abe02b","last_reissued_at":"2026-07-05T07:01:08.077601Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:01:08.077601Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2310.09573","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T07:01:08Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"z4r6JNvYVH1+M9DMephqCa8POa+EEVFZjmJ48tk9FvbRea9AgC9lMOpsDSE6ky0RMgBz4Yw3+R8AEe+uQNarBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T07:09:56.418592Z"},"content_sha256":"0a687185dcb145cf220f57aa322be40f2def433b6b0e59d9f421c306c89a140f","schema_version":"1.0","event_id":"sha256:0a687185dcb145cf220f57aa322be40f2def433b6b0e59d9f421c306c89a140f"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:JWGPNU354KXCEDWSZLVCDE5IMR","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Self-Detoxifying Language Models via Toxification Reversal","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chak Tou Leong, Jian Wang, Jiashuo Wang, Wenjie Li, Yi Cheng","submitted_at":"2023-10-14T12:51:38Z","abstract_excerpt":"Language model detoxification aims to minimize the risk of generating offensive or harmful content in pretrained language models (PLMs) for safer deployment. Existing methods can be roughly categorized as finetuning-based and decoding-based. However, the former is often resource-intensive, while the latter relies on additional components and potentially compromises the generation fluency. In this paper, we propose a more lightweight approach that enables the PLM itself to achieve \"self-detoxification\". Our method is built upon the observation that prepending a negative steering prompt can effe"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.09573","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/2310.09573/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":"2026-07-05T07:01:08Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"efrp+BplmMsljc+Igg3YYX/t/YO7IVIYTgFLjMIFw8p89kvYukuaHgw3l39IQBYct8Yklg7kw71DYwprfdawBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T07:09:56.418971Z"},"content_sha256":"fb4f522c9a480cf8b5a2199e11b62879e5cdaa5d4407b712bec3b99fbdab2478","schema_version":"1.0","event_id":"sha256:fb4f522c9a480cf8b5a2199e11b62879e5cdaa5d4407b712bec3b99fbdab2478"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/JWGPNU354KXCEDWSZLVCDE5IMR/bundle.json","state_url":"https://pith.science/pith/JWGPNU354KXCEDWSZLVCDE5IMR/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/JWGPNU354KXCEDWSZLVCDE5IMR/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":"2026-07-07T07:09:56Z","links":{"resolver":"https://pith.science/pith/JWGPNU354KXCEDWSZLVCDE5IMR","bundle":"https://pith.science/pith/JWGPNU354KXCEDWSZLVCDE5IMR/bundle.json","state":"https://pith.science/pith/JWGPNU354KXCEDWSZLVCDE5IMR/state.json","well_known_bundle":"https://pith.science/.well-known/pith/JWGPNU354KXCEDWSZLVCDE5IMR/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:JWGPNU354KXCEDWSZLVCDE5IMR","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":"07ea86e724df754a3b758f1263d660b446960c17e27f4e1a4c9d2b8aab5d768d","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2023-10-14T12:51:38Z","title_canon_sha256":"f6acd2cfb03fe13e33fc335d22599cd853aa1f1057429cd819c1da5eb7de94d2"},"schema_version":"1.0","source":{"id":"2310.09573","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2310.09573","created_at":"2026-07-05T07:01:08Z"},{"alias_kind":"arxiv_version","alias_value":"2310.09573v1","created_at":"2026-07-05T07:01:08Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.09573","created_at":"2026-07-05T07:01:08Z"},{"alias_kind":"pith_short_12","alias_value":"JWGPNU354KXC","created_at":"2026-07-05T07:01:08Z"},{"alias_kind":"pith_short_16","alias_value":"JWGPNU354KXCEDWS","created_at":"2026-07-05T07:01:08Z"},{"alias_kind":"pith_short_8","alias_value":"JWGPNU35","created_at":"2026-07-05T07:01:08Z"}],"graph_snapshots":[{"event_id":"sha256:fb4f522c9a480cf8b5a2199e11b62879e5cdaa5d4407b712bec3b99fbdab2478","target":"graph","created_at":"2026-07-05T07:01:08Z","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/2310.09573/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Language model detoxification aims to minimize the risk of generating offensive or harmful content in pretrained language models (PLMs) for safer deployment. Existing methods can be roughly categorized as finetuning-based and decoding-based. However, the former is often resource-intensive, while the latter relies on additional components and potentially compromises the generation fluency. In this paper, we propose a more lightweight approach that enables the PLM itself to achieve \"self-detoxification\". Our method is built upon the observation that prepending a negative steering prompt can effe","authors_text":"Chak Tou Leong, Jian Wang, Jiashuo Wang, Wenjie Li, Yi Cheng","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2023-10-14T12:51:38Z","title":"Self-Detoxifying Language Models via Toxification Reversal"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.09573","kind":"arxiv","version":1},"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":"sha256:0a687185dcb145cf220f57aa322be40f2def433b6b0e59d9f421c306c89a140f","target":"record","created_at":"2026-07-05T07:01:08Z","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":"07ea86e724df754a3b758f1263d660b446960c17e27f4e1a4c9d2b8aab5d768d","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2023-10-14T12:51:38Z","title_canon_sha256":"f6acd2cfb03fe13e33fc335d22599cd853aa1f1057429cd819c1da5eb7de94d2"},"schema_version":"1.0","source":{"id":"2310.09573","kind":"arxiv","version":1}},"canonical_sha256":"4d8cf6d37de2ae220ed2caea2193a8645a52343e7e7407128ed0bb4897abe02b","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"4d8cf6d37de2ae220ed2caea2193a8645a52343e7e7407128ed0bb4897abe02b","first_computed_at":"2026-07-05T07:01:08.077601Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:01:08.077601Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"PGYYY7rtAXbQ8jMv7VpAsY7VYa4zraaAPrSEkjgOYXa1sA7Xj9Ni2vYGb/cMXgxw+x7NKnEilRHVMKEGjis2Bw==","signature_status":"signed_v1","signed_at":"2026-07-05T07:01:08.078106Z","signed_message":"canonical_sha256_bytes"},"source_id":"2310.09573","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:0a687185dcb145cf220f57aa322be40f2def433b6b0e59d9f421c306c89a140f","sha256:fb4f522c9a480cf8b5a2199e11b62879e5cdaa5d4407b712bec3b99fbdab2478"],"state_sha256":"b03a4b5c2462c54af791008b990fc7b3b9e74595219bb2868f962a16dc6baa2c"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"uGMul6eh9zXPha3LSYsRO7SuNaORtouXK3B4QgoBUVOxXzUeVLWwNqgKarjF1A5r8x/me4mYjulXvEXdTKBNDw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-07T07:09:56.420951Z","bundle_sha256":"5da1d499777560ecc504c41774948aae321f3c9d1523f1bf2f9e42df09e31e2a"}}