{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:GOJYYRJTGSHN5C2XRGWBOUROBI","short_pith_number":"pith:GOJYYRJT","schema_version":"1.0","canonical_sha256":"33938c4533348ede8b5789ac17522e0a2e914039ac3e33290f8720654c15b8ca","source":{"kind":"arxiv","id":"2410.06809","version":3},"attestation_state":"computed","paper":{"title":"Root Defence Strategies: Ensuring Safety of LLM at the Decoding Level","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.CL","authors_text":"Jiawei Chen, Jingyuan Zhang, Xinyi Zeng, Yu Tian, Yuying Shang","submitted_at":"2024-10-09T12:09:30Z","abstract_excerpt":"Large language models (LLMs) have demonstrated immense utility across various industries. However, as LLMs advance, the risk of harmful outputs increases due to incorrect or malicious instruction prompts. While current methods effectively address jailbreak risks, they share common limitations: 1) Judging harmful responses from the prefill-level lacks utilization of the model's decoding outputs, leading to relatively lower effectiveness and robustness. 2) Rejecting potentially harmful responses based on a single evaluation can significantly impair the model's helpfulness.This paper examines the"},"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":"2410.06809","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-10-09T12:09:30Z","cross_cats_sorted":["cs.CR"],"title_canon_sha256":"984db862dc7f8c3a1c8241a34fbe725e4bd1d983b60d7715967f7a00127a1c20","abstract_canon_sha256":"80a4d1156d36c547ab1be25cd964f4e6758b9bfe643b907161f8baa8b01ebeff"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:23:18.708467Z","signature_b64":"utmeJQac/RugtuQWthdSdn+lixtIG74mk0rJeyr9ZB6QDDC9UptGgl0KXRvbKLQqCZ+8dnLPfKwo56M1+kwaDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"33938c4533348ede8b5789ac17522e0a2e914039ac3e33290f8720654c15b8ca","last_reissued_at":"2026-07-05T11:23:18.707866Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:23:18.707866Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Root Defence Strategies: Ensuring Safety of LLM at the Decoding Level","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR"],"primary_cat":"cs.CL","authors_text":"Jiawei Chen, Jingyuan Zhang, Xinyi Zeng, Yu Tian, Yuying Shang","submitted_at":"2024-10-09T12:09:30Z","abstract_excerpt":"Large language models (LLMs) have demonstrated immense utility across various industries. However, as LLMs advance, the risk of harmful outputs increases due to incorrect or malicious instruction prompts. While current methods effectively address jailbreak risks, they share common limitations: 1) Judging harmful responses from the prefill-level lacks utilization of the model's decoding outputs, leading to relatively lower effectiveness and robustness. 2) Rejecting potentially harmful responses based on a single evaluation can significantly impair the model's helpfulness.This paper examines the"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.06809","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/2410.06809/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":"2410.06809","created_at":"2026-07-05T11:23:18.707933+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.06809v3","created_at":"2026-07-05T11:23:18.707933+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.06809","created_at":"2026-07-05T11:23:18.707933+00:00"},{"alias_kind":"pith_short_12","alias_value":"GOJYYRJTGSHN","created_at":"2026-07-05T11:23:18.707933+00:00"},{"alias_kind":"pith_short_16","alias_value":"GOJYYRJTGSHN5C2X","created_at":"2026-07-05T11:23:18.707933+00:00"},{"alias_kind":"pith_short_8","alias_value":"GOJYYRJT","created_at":"2026-07-05T11:23:18.707933+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2509.03871","citing_title":"A Comprehensive Survey on Trustworthiness in Reasoning with Large Language Models","ref_index":266,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GOJYYRJTGSHN5C2XRGWBOUROBI","json":"https://pith.science/pith/GOJYYRJTGSHN5C2XRGWBOUROBI.json","graph_json":"https://pith.science/api/pith-number/GOJYYRJTGSHN5C2XRGWBOUROBI/graph.json","events_json":"https://pith.science/api/pith-number/GOJYYRJTGSHN5C2XRGWBOUROBI/events.json","paper":"https://pith.science/paper/GOJYYRJT"},"agent_actions":{"view_html":"https://pith.science/pith/GOJYYRJTGSHN5C2XRGWBOUROBI","download_json":"https://pith.science/pith/GOJYYRJTGSHN5C2XRGWBOUROBI.json","view_paper":"https://pith.science/paper/GOJYYRJT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.06809&json=true","fetch_graph":"https://pith.science/api/pith-number/GOJYYRJTGSHN5C2XRGWBOUROBI/graph.json","fetch_events":"https://pith.science/api/pith-number/GOJYYRJTGSHN5C2XRGWBOUROBI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GOJYYRJTGSHN5C2XRGWBOUROBI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GOJYYRJTGSHN5C2XRGWBOUROBI/action/storage_attestation","attest_author":"https://pith.science/pith/GOJYYRJTGSHN5C2XRGWBOUROBI/action/author_attestation","sign_citation":"https://pith.science/pith/GOJYYRJTGSHN5C2XRGWBOUROBI/action/citation_signature","submit_replication":"https://pith.science/pith/GOJYYRJTGSHN5C2XRGWBOUROBI/action/replication_record"}},"created_at":"2026-07-05T11:23:18.707933+00:00","updated_at":"2026-07-05T11:23:18.707933+00:00"}