{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:7V3FX5744UJF27KO4H7MIDNZFP","short_pith_number":"pith:7V3FX574","canonical_record":{"source":{"id":"2501.17112","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-28T17:59:56Z","cross_cats_sorted":[],"title_canon_sha256":"c1e9bb3f851550a28557c1dc627ab503d816bf4d9ee861c3f346c29606a1639b","abstract_canon_sha256":"26f7071d8d7a3a823b6b3a2cbef54fca040e0caf80eeec5cfcbdf51c67633baa"},"schema_version":"1.0"},"canonical_sha256":"fd765bf7fce5125d7d4ee1fec40db92bc8364a8e3c57e1f7a066d32e0ca13d22","source":{"kind":"arxiv","id":"2501.17112","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.17112","created_at":"2026-07-05T10:41:19Z"},{"alias_kind":"arxiv_version","alias_value":"2501.17112v2","created_at":"2026-07-05T10:41:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.17112","created_at":"2026-07-05T10:41:19Z"},{"alias_kind":"pith_short_12","alias_value":"7V3FX5744UJF","created_at":"2026-07-05T10:41:19Z"},{"alias_kind":"pith_short_16","alias_value":"7V3FX5744UJF27KO","created_at":"2026-07-05T10:41:19Z"},{"alias_kind":"pith_short_8","alias_value":"7V3FX574","created_at":"2026-07-05T10:41:19Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:7V3FX5744UJF27KO4H7MIDNZFP","target":"record","payload":{"canonical_record":{"source":{"id":"2501.17112","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-28T17:59:56Z","cross_cats_sorted":[],"title_canon_sha256":"c1e9bb3f851550a28557c1dc627ab503d816bf4d9ee861c3f346c29606a1639b","abstract_canon_sha256":"26f7071d8d7a3a823b6b3a2cbef54fca040e0caf80eeec5cfcbdf51c67633baa"},"schema_version":"1.0"},"canonical_sha256":"fd765bf7fce5125d7d4ee1fec40db92bc8364a8e3c57e1f7a066d32e0ca13d22","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:41:19.970044Z","signature_b64":"L3XysQpyka74k9xy8sYmu/Mb/yN1p5fbRx6ZTklRGCRxXKwQTZB7O9L/AZX8Sg/xo1oNk/3msZ5B+3Q6rEKuDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fd765bf7fce5125d7d4ee1fec40db92bc8364a8e3c57e1f7a066d32e0ca13d22","last_reissued_at":"2026-07-05T10:41:19.969586Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:41:19.969586Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2501.17112","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:41:19Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"pyjMpSfFrinCEqCXXx0fX+oTZkMjGQT7pscU98JZ7VnteG0OcRnRdWEnn4b7zFMvxBxk8g6F2H/svB7qSE7QCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-24T10:36:19.007054Z"},"content_sha256":"51dc00a1d6e8232621b5f3669350aa0147600e0fa8fe6d831783c26c06088afa","schema_version":"1.0","event_id":"sha256:51dc00a1d6e8232621b5f3669350aa0147600e0fa8fe6d831783c26c06088afa"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:7V3FX5744UJF27KO4H7MIDNZFP","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Decoding Human Preferences in Alignment: An Improved Approach to Inverse Constitutional AI","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Carl-Leander Henneking, Claas Beger","submitted_at":"2025-01-28T17:59:56Z","abstract_excerpt":"Traditional methods for aligning Large Language Models (LLMs), such as Reinforcement Learning from Human Feedback (RLHF) and Direct Preference Optimization (DPO), rely on implicit principles, limiting interpretability. Constitutional AI (CAI) offers an explicit, rule-based framework for guiding LLM alignment. Building on this, we refine the Inverse Constitutional AI (ICAI) algorithm, which extracts constitutions from preference datasets. By improving principle generation, clustering, and embedding processes, our approach enhances the accuracy and generalizability of extracted principles across"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.17112","kind":"arxiv","version":2},"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/2501.17112/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":"2026-07-05T10:41:19Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"DEM4yqkPwr1MkQSQG8Tsk/9RYRncKZRqW1+Yd146wbk1sqdykJLSQLxZL+Os32mqaTlV6mpP2NiFNwEzfECzBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-24T10:36:19.007446Z"},"content_sha256":"b77d03f85b19348315393493b9e9e32e4ee0fc4455547874868dfd87764109d5","schema_version":"1.0","event_id":"sha256:b77d03f85b19348315393493b9e9e32e4ee0fc4455547874868dfd87764109d5"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/7V3FX5744UJF27KO4H7MIDNZFP/bundle.json","state_url":"https://pith.science/pith/7V3FX5744UJF27KO4H7MIDNZFP/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/7V3FX5744UJF27KO4H7MIDNZFP/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":"2026-07-24T10:36:19Z","links":{"resolver":"https://pith.science/pith/7V3FX5744UJF27KO4H7MIDNZFP","bundle":"https://pith.science/pith/7V3FX5744UJF27KO4H7MIDNZFP/bundle.json","state":"https://pith.science/pith/7V3FX5744UJF27KO4H7MIDNZFP/state.json","well_known_bundle":"https://pith.science/.well-known/pith/7V3FX5744UJF27KO4H7MIDNZFP/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:7V3FX5744UJF27KO4H7MIDNZFP","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":"26f7071d8d7a3a823b6b3a2cbef54fca040e0caf80eeec5cfcbdf51c67633baa","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-28T17:59:56Z","title_canon_sha256":"c1e9bb3f851550a28557c1dc627ab503d816bf4d9ee861c3f346c29606a1639b"},"schema_version":"1.0","source":{"id":"2501.17112","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.17112","created_at":"2026-07-05T10:41:19Z"},{"alias_kind":"arxiv_version","alias_value":"2501.17112v2","created_at":"2026-07-05T10:41:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.17112","created_at":"2026-07-05T10:41:19Z"},{"alias_kind":"pith_short_12","alias_value":"7V3FX5744UJF","created_at":"2026-07-05T10:41:19Z"},{"alias_kind":"pith_short_16","alias_value":"7V3FX5744UJF27KO","created_at":"2026-07-05T10:41:19Z"},{"alias_kind":"pith_short_8","alias_value":"7V3FX574","created_at":"2026-07-05T10:41:19Z"}],"graph_snapshots":[{"event_id":"sha256:b77d03f85b19348315393493b9e9e32e4ee0fc4455547874868dfd87764109d5","target":"graph","created_at":"2026-07-05T10:41:19Z","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/2501.17112/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Traditional methods for aligning Large Language Models (LLMs), such as Reinforcement Learning from Human Feedback (RLHF) and Direct Preference Optimization (DPO), rely on implicit principles, limiting interpretability. Constitutional AI (CAI) offers an explicit, rule-based framework for guiding LLM alignment. Building on this, we refine the Inverse Constitutional AI (ICAI) algorithm, which extracts constitutions from preference datasets. By improving principle generation, clustering, and embedding processes, our approach enhances the accuracy and generalizability of extracted principles across","authors_text":"Carl-Leander Henneking, Claas Beger","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-28T17:59:56Z","title":"Decoding Human Preferences in Alignment: An Improved Approach to Inverse Constitutional AI"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.17112","kind":"arxiv","version":2},"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":"sha256:51dc00a1d6e8232621b5f3669350aa0147600e0fa8fe6d831783c26c06088afa","target":"record","created_at":"2026-07-05T10:41:19Z","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":"26f7071d8d7a3a823b6b3a2cbef54fca040e0caf80eeec5cfcbdf51c67633baa","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-28T17:59:56Z","title_canon_sha256":"c1e9bb3f851550a28557c1dc627ab503d816bf4d9ee861c3f346c29606a1639b"},"schema_version":"1.0","source":{"id":"2501.17112","kind":"arxiv","version":2}},"canonical_sha256":"fd765bf7fce5125d7d4ee1fec40db92bc8364a8e3c57e1f7a066d32e0ca13d22","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"fd765bf7fce5125d7d4ee1fec40db92bc8364a8e3c57e1f7a066d32e0ca13d22","first_computed_at":"2026-07-05T10:41:19.969586Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:41:19.969586Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"L3XysQpyka74k9xy8sYmu/Mb/yN1p5fbRx6ZTklRGCRxXKwQTZB7O9L/AZX8Sg/xo1oNk/3msZ5B+3Q6rEKuDA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:41:19.970044Z","signed_message":"canonical_sha256_bytes"},"source_id":"2501.17112","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:51dc00a1d6e8232621b5f3669350aa0147600e0fa8fe6d831783c26c06088afa","sha256:b77d03f85b19348315393493b9e9e32e4ee0fc4455547874868dfd87764109d5"],"state_sha256":"50f99b4155033551a122267b0d7fbf7aa227d5df67720957c4d12cc7d1f6ce25"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"OIy8pWxdGORmwRsHMqQ+Iv9nTiuIgdMqKSqaEunrqTG3VAebkNB8Iib3wuSKAY9mLU9Zr1aE95khNw/XVy8aCA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-24T10:36:19.009976Z","bundle_sha256":"33fc7977d9509403b7107f48629e6f9141a43fcbc6c9e3140e699bd9ebe5d25a"}}