{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:TZ2XLSKAVCDKCSPEGHDGG35W22","short_pith_number":"pith:TZ2XLSKA","canonical_record":{"source":{"id":"2403.04894","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-03-07T20:58:04Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c65ef2d1d1a7074102c8ca00b1eaf5d12448df852e1907ca0f0c861b103043b3","abstract_canon_sha256":"4718ac660274509ac1414e7d0468a6a3f2efabce0880314c6fbb62d8bb945801"},"schema_version":"1.0"},"canonical_sha256":"9e7575c940a886a149e431c6636fb6d6a5074ce2af4e00f7e129ce70c5a99a6f","source":{"kind":"arxiv","id":"2403.04894","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.04894","created_at":"2026-07-05T07:53:35Z"},{"alias_kind":"arxiv_version","alias_value":"2403.04894v1","created_at":"2026-07-05T07:53:35Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.04894","created_at":"2026-07-05T07:53:35Z"},{"alias_kind":"pith_short_12","alias_value":"TZ2XLSKAVCDK","created_at":"2026-07-05T07:53:35Z"},{"alias_kind":"pith_short_16","alias_value":"TZ2XLSKAVCDKCSPE","created_at":"2026-07-05T07:53:35Z"},{"alias_kind":"pith_short_8","alias_value":"TZ2XLSKA","created_at":"2026-07-05T07:53:35Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:TZ2XLSKAVCDKCSPEGHDGG35W22","target":"record","payload":{"canonical_record":{"source":{"id":"2403.04894","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-03-07T20:58:04Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c65ef2d1d1a7074102c8ca00b1eaf5d12448df852e1907ca0f0c861b103043b3","abstract_canon_sha256":"4718ac660274509ac1414e7d0468a6a3f2efabce0880314c6fbb62d8bb945801"},"schema_version":"1.0"},"canonical_sha256":"9e7575c940a886a149e431c6636fb6d6a5074ce2af4e00f7e129ce70c5a99a6f","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:53:35.554012Z","signature_b64":"jxapmkgeiUpco/DgWNnItdN/9P08vs2CoTQnoqKzxPnAg6lrIazjYE+CeylYZo4MuyOInm7t/XtbL+7dNThbBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9e7575c940a886a149e431c6636fb6d6a5074ce2af4e00f7e129ce70c5a99a6f","last_reissued_at":"2026-07-05T07:53:35.553511Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:53:35.553511Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2403.04894","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":"2026-07-05T07:53:35Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"hee2tw+gEWn/Ypf3/ZA35xZPOy1IVHopM8ibxE426zj1g+dAh3ZJzRm29YRKPryg3u4ZQV9CMM/5gc79cndIDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T15:32:26.713783Z"},"content_sha256":"cbb4fdbd05c11db5961d325e9e7f434f6628a2133ce3e36e64b060b29bab5f3a","schema_version":"1.0","event_id":"sha256:cbb4fdbd05c11db5961d325e9e7f434f6628a2133ce3e36e64b060b29bab5f3a"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:TZ2XLSKAVCDKCSPEGHDGG35W22","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"ConstitutionalExperts: Training a Mixture of Principle-based Prompts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ann Yuan, Ben Wedin, James Wexler, Nithum Thain, Savvas Petridis","submitted_at":"2024-03-07T20:58:04Z","abstract_excerpt":"Large language models (LLMs) are highly capable at a variety of tasks given the right prompt, but writing one is still a difficult and tedious process. In this work, we introduce ConstitutionalExperts, a method for learning a prompt consisting of constitutional principles (i.e. rules), given a training dataset. Unlike prior methods that optimize the prompt as a single entity, our method incrementally improves the prompt by surgically editing individual principles. We also show that we can improve overall performance by learning unique prompts for different semantic regions of the training data"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.04894","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/2403.04894/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":"2026-07-05T07:53:35Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"IIsJejmmx1UA4paLP5ESV2UGgrDE2k3u3arF8qzZ5PeD+DGmrUmm+CJq9Zw967YCNb9ZolgfzA2eakVcXDqwAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-07T15:32:26.714200Z"},"content_sha256":"902462a5e9b89ca327125d0fc5dac22a74df6aa85cb909038d356b95431b05a1","schema_version":"1.0","event_id":"sha256:902462a5e9b89ca327125d0fc5dac22a74df6aa85cb909038d356b95431b05a1"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/TZ2XLSKAVCDKCSPEGHDGG35W22/bundle.json","state_url":"https://pith.science/pith/TZ2XLSKAVCDKCSPEGHDGG35W22/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/TZ2XLSKAVCDKCSPEGHDGG35W22/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":"2026-07-07T15:32:26Z","links":{"resolver":"https://pith.science/pith/TZ2XLSKAVCDKCSPEGHDGG35W22","bundle":"https://pith.science/pith/TZ2XLSKAVCDKCSPEGHDGG35W22/bundle.json","state":"https://pith.science/pith/TZ2XLSKAVCDKCSPEGHDGG35W22/state.json","well_known_bundle":"https://pith.science/.well-known/pith/TZ2XLSKAVCDKCSPEGHDGG35W22/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:TZ2XLSKAVCDKCSPEGHDGG35W22","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":"4718ac660274509ac1414e7d0468a6a3f2efabce0880314c6fbb62d8bb945801","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-03-07T20:58:04Z","title_canon_sha256":"c65ef2d1d1a7074102c8ca00b1eaf5d12448df852e1907ca0f0c861b103043b3"},"schema_version":"1.0","source":{"id":"2403.04894","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.04894","created_at":"2026-07-05T07:53:35Z"},{"alias_kind":"arxiv_version","alias_value":"2403.04894v1","created_at":"2026-07-05T07:53:35Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.04894","created_at":"2026-07-05T07:53:35Z"},{"alias_kind":"pith_short_12","alias_value":"TZ2XLSKAVCDK","created_at":"2026-07-05T07:53:35Z"},{"alias_kind":"pith_short_16","alias_value":"TZ2XLSKAVCDKCSPE","created_at":"2026-07-05T07:53:35Z"},{"alias_kind":"pith_short_8","alias_value":"TZ2XLSKA","created_at":"2026-07-05T07:53:35Z"}],"graph_snapshots":[{"event_id":"sha256:902462a5e9b89ca327125d0fc5dac22a74df6aa85cb909038d356b95431b05a1","target":"graph","created_at":"2026-07-05T07:53:35Z","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/2403.04894/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) are highly capable at a variety of tasks given the right prompt, but writing one is still a difficult and tedious process. In this work, we introduce ConstitutionalExperts, a method for learning a prompt consisting of constitutional principles (i.e. rules), given a training dataset. Unlike prior methods that optimize the prompt as a single entity, our method incrementally improves the prompt by surgically editing individual principles. We also show that we can improve overall performance by learning unique prompts for different semantic regions of the training data","authors_text":"Ann Yuan, Ben Wedin, James Wexler, Nithum Thain, Savvas Petridis","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-03-07T20:58:04Z","title":"ConstitutionalExperts: Training a Mixture of Principle-based Prompts"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.04894","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:cbb4fdbd05c11db5961d325e9e7f434f6628a2133ce3e36e64b060b29bab5f3a","target":"record","created_at":"2026-07-05T07:53:35Z","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":"4718ac660274509ac1414e7d0468a6a3f2efabce0880314c6fbb62d8bb945801","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-03-07T20:58:04Z","title_canon_sha256":"c65ef2d1d1a7074102c8ca00b1eaf5d12448df852e1907ca0f0c861b103043b3"},"schema_version":"1.0","source":{"id":"2403.04894","kind":"arxiv","version":1}},"canonical_sha256":"9e7575c940a886a149e431c6636fb6d6a5074ce2af4e00f7e129ce70c5a99a6f","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9e7575c940a886a149e431c6636fb6d6a5074ce2af4e00f7e129ce70c5a99a6f","first_computed_at":"2026-07-05T07:53:35.553511Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:53:35.553511Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"jxapmkgeiUpco/DgWNnItdN/9P08vs2CoTQnoqKzxPnAg6lrIazjYE+CeylYZo4MuyOInm7t/XtbL+7dNThbBA==","signature_status":"signed_v1","signed_at":"2026-07-05T07:53:35.554012Z","signed_message":"canonical_sha256_bytes"},"source_id":"2403.04894","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:cbb4fdbd05c11db5961d325e9e7f434f6628a2133ce3e36e64b060b29bab5f3a","sha256:902462a5e9b89ca327125d0fc5dac22a74df6aa85cb909038d356b95431b05a1"],"state_sha256":"14eaebf30bf8b79940db608e0d2565d5043945a003e9560e7018bcc9ec0e1e77"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"U53DMqc3T3a/EUT33c81G+M3trvZf4zPWcJ6G87DuXHb3+2hIaMQO/wlYkaQr6hTf6Cu5pGayb00q5Tv+70pDg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-07T15:32:26.716120Z","bundle_sha256":"e639a4b68918712cebf568ebbaaf5bcc3bdde6aa9a91c41d3d8d8304ef7fb944"}}