{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:DRDBSOJUZP7HJ5RCM54YV5E62W","short_pith_number":"pith:DRDBSOJU","canonical_record":{"source":{"id":"2605.08464","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-08T20:30:22Z","cross_cats_sorted":[],"title_canon_sha256":"77607c69e021cb880fe63eec12b930157c1c8fc22ab1213560a67c3184249fa3","abstract_canon_sha256":"05a5e5d4b0216f5dbb4313e92ce1a8b57ea3950b29b5bf62dd042a4e57e6ae97"},"schema_version":"1.0"},"canonical_sha256":"1c46193934cbfe74f62267798af49ed5a521499f2dafb7129c80ef81d8bff2d0","source":{"kind":"arxiv","id":"2605.08464","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.08464","created_at":"2026-05-20T00:01:43Z"},{"alias_kind":"arxiv_version","alias_value":"2605.08464v2","created_at":"2026-05-20T00:01:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.08464","created_at":"2026-05-20T00:01:43Z"},{"alias_kind":"pith_short_12","alias_value":"DRDBSOJUZP7H","created_at":"2026-05-20T00:01:43Z"},{"alias_kind":"pith_short_16","alias_value":"DRDBSOJUZP7HJ5RC","created_at":"2026-05-20T00:01:43Z"},{"alias_kind":"pith_short_8","alias_value":"DRDBSOJU","created_at":"2026-05-20T00:01:43Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:DRDBSOJUZP7HJ5RCM54YV5E62W","target":"record","payload":{"canonical_record":{"source":{"id":"2605.08464","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-08T20:30:22Z","cross_cats_sorted":[],"title_canon_sha256":"77607c69e021cb880fe63eec12b930157c1c8fc22ab1213560a67c3184249fa3","abstract_canon_sha256":"05a5e5d4b0216f5dbb4313e92ce1a8b57ea3950b29b5bf62dd042a4e57e6ae97"},"schema_version":"1.0"},"canonical_sha256":"1c46193934cbfe74f62267798af49ed5a521499f2dafb7129c80ef81d8bff2d0","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:01:43.020944Z","signature_b64":"6ug7BVrAT+gP5CJWw89VmThVl9jr7gN/p1FYarrZV8tI8cnM4D+SfI1/8lDqvE5O4wGt11hrnpVFNM+vdiX8BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1c46193934cbfe74f62267798af49ed5a521499f2dafb7129c80ef81d8bff2d0","last_reissued_at":"2026-05-20T00:01:43.020288Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:01:43.020288Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.08464","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":"2026-05-20T00:01:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1tHpGVCOoPm3uPz0Blnxlr/yrkxLmOcgQQtNpLV5JEQdcI2qGgYdaTiJLFXbaYyjodoIdWRy9B7Fj60pMVkUAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-31T11:23:05.802070Z"},"content_sha256":"d680a58a6ddfa8498f1e33874e5ada1c41c8eba8a60081f8aa4aaa5a918eae4d","schema_version":"1.0","event_id":"sha256:d680a58a6ddfa8498f1e33874e5ada1c41c8eba8a60081f8aa4aaa5a918eae4d"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:DRDBSOJUZP7HJ5RCM54YV5E62W","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"The Geometric Structure of Models Learning Sparse Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Models succeed on sparse data by making their input-output Jacobians rank-one and perfectly aligned with each training point.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Ahmed Imtiaz Humayun, Randall Balestriero, Richard Baraniuk, Thomas Walker, T. Mitchell Roddenberry","submitted_at":"2026-05-08T20:30:22Z","abstract_excerpt":"The manifold hypothesis (MH) is often used to explain how machine learning can overcome the curse of dimensionality. However, the MH is only applicable in regimes where the training data provides a sufficiently dense sample of the underlying low-dimensional data manifold, or where such a low-dimensional manifold is conceivably present. We describe the regimes where the MH is not applicable as sparse. In this paper, we demonstrate that models succeed in the sparse regime by exploiting a highly structured local geometry, a property we formalize as normal alignment. We prove that normal-aligned c"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"normal-aligned classifiers -- whose input-output Jacobians are rank-one and align perfectly with the training data -- minimize the training objective under norm constraints and achieve maximal local robustness under a non-zero Jacobian constraint","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that success in the sparse regime is explained by normal alignment rather than other mechanisms, and that this alignment arises specifically from the feature-learning regime in continuous piecewise-affine networks (as described in the abstract when discussing power-diagram partitions).","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Normal alignment is the rank-one Jacobian structure that lets classifiers minimize loss and maximize local robustness in sparse regimes; the paper proves its optimality and uses it to create GrokAlign and RFAMs.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Models succeed on sparse data by making their input-output Jacobians rank-one and perfectly aligned with each training point.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"f309b0fc6a67e224939f246169de880d14dfefdd7dfbbcd9e17a30eedcd7ef57"},"source":{"id":"2605.08464","kind":"arxiv","version":2},"verdict":{"id":"67b2ff44-b3c0-4e74-835d-9a0b1772d5a4","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T17:56:29.297098Z","strongest_claim":"normal-aligned classifiers -- whose input-output Jacobians are rank-one and align perfectly with the training data -- minimize the training objective under norm constraints and achieve maximal local robustness under a non-zero Jacobian constraint","one_line_summary":"Normal alignment is the rank-one Jacobian structure that lets classifiers minimize loss and maximize local robustness in sparse regimes; the paper proves its optimality and uses it to create GrokAlign and RFAMs.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that success in the sparse regime is explained by normal alignment rather than other mechanisms, and that this alignment arises specifically from the feature-learning regime in continuous piecewise-affine networks (as described in the abstract when discussing power-diagram partitions).","pith_extraction_headline":"Models succeed on sparse data by making their input-output Jacobians rank-one and perfectly aligned with each training point."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.08464/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T15:01:17.790576Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T11:04:31.266533Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"83d621ee3c9123b3fe62b65a989400e7aad285eaa19e6c414e347ddef459c69f"},"references":{"count":46,"sample":[{"doi":"","year":2000,"title":"Tenenbaum, Vin de Silva, and John C","work_id":"5a469211-854a-42bd-9bdb-9ae3b6a90c2d","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2000,"title":"Roweis and Lawrence K","work_id":"8beed03d-4aea-49dd-aad3-2ac0f4b1eaf9","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"Dauphin, and David Lopez-Paz","work_id":"eddc5252-cf0b-47e6-be05-0151d26825be","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features","work_id":"71760d81-100c-4b37-81f3-ec00607e0354","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Data Augmentation Using Random Image Cropping and Patching for Deep CNNs.IEEE Trans","work_id":"31c1b904-c1f8-47f4-a532-3eedf4857393","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":46,"snapshot_sha256":"f694b9fb32aeea1809eb50c906fbc14956f2a0e2167167bd0f2cc6c5398439d6","internal_anchors":2},"formal_canon":{"evidence_count":3,"snapshot_sha256":"79b2252d9ba4fc4b26bab114156e1b0fd4f840b655305a387fa9faf4b1b6d949"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"67b2ff44-b3c0-4e74-835d-9a0b1772d5a4"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:01:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6kCXaU2+Yk2N1vAuYY5V39TEU7QZoz0i6PuZ3Z4Njmh1l+7UUukMfFPJ1lm4JoYoYuQ2v3BprlHAjwrj7BsNDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-31T11:23:05.803020Z"},"content_sha256":"ee13a0eb971773ca493eb649bc04a5dcece74a50b45c42762f51542b841584a7","schema_version":"1.0","event_id":"sha256:ee13a0eb971773ca493eb649bc04a5dcece74a50b45c42762f51542b841584a7"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/DRDBSOJUZP7HJ5RCM54YV5E62W/bundle.json","state_url":"https://pith.science/pith/DRDBSOJUZP7HJ5RCM54YV5E62W/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/DRDBSOJUZP7HJ5RCM54YV5E62W/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":"2026-05-31T11:23:05Z","links":{"resolver":"https://pith.science/pith/DRDBSOJUZP7HJ5RCM54YV5E62W","bundle":"https://pith.science/pith/DRDBSOJUZP7HJ5RCM54YV5E62W/bundle.json","state":"https://pith.science/pith/DRDBSOJUZP7HJ5RCM54YV5E62W/state.json","well_known_bundle":"https://pith.science/.well-known/pith/DRDBSOJUZP7HJ5RCM54YV5E62W/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:DRDBSOJUZP7HJ5RCM54YV5E62W","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":"05a5e5d4b0216f5dbb4313e92ce1a8b57ea3950b29b5bf62dd042a4e57e6ae97","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-08T20:30:22Z","title_canon_sha256":"77607c69e021cb880fe63eec12b930157c1c8fc22ab1213560a67c3184249fa3"},"schema_version":"1.0","source":{"id":"2605.08464","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.08464","created_at":"2026-05-20T00:01:43Z"},{"alias_kind":"arxiv_version","alias_value":"2605.08464v2","created_at":"2026-05-20T00:01:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.08464","created_at":"2026-05-20T00:01:43Z"},{"alias_kind":"pith_short_12","alias_value":"DRDBSOJUZP7H","created_at":"2026-05-20T00:01:43Z"},{"alias_kind":"pith_short_16","alias_value":"DRDBSOJUZP7HJ5RC","created_at":"2026-05-20T00:01:43Z"},{"alias_kind":"pith_short_8","alias_value":"DRDBSOJU","created_at":"2026-05-20T00:01:43Z"}],"graph_snapshots":[{"event_id":"sha256:ee13a0eb971773ca493eb649bc04a5dcece74a50b45c42762f51542b841584a7","target":"graph","created_at":"2026-05-20T00:01:43Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"normal-aligned classifiers -- whose input-output Jacobians are rank-one and align perfectly with the training data -- minimize the training objective under norm constraints and achieve maximal local robustness under a non-zero Jacobian constraint"},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The assumption that success in the sparse regime is explained by normal alignment rather than other mechanisms, and that this alignment arises specifically from the feature-learning regime in continuous piecewise-affine networks (as described in the abstract when discussing power-diagram partitions)."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Normal alignment is the rank-one Jacobian structure that lets classifiers minimize loss and maximize local robustness in sparse regimes; the paper proves its optimality and uses it to create GrokAlign and RFAMs."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Models succeed on sparse data by making their input-output Jacobians rank-one and perfectly aligned with each training point."}],"snapshot_sha256":"f309b0fc6a67e224939f246169de880d14dfefdd7dfbbcd9e17a30eedcd7ef57"},"formal_canon":{"evidence_count":3,"snapshot_sha256":"79b2252d9ba4fc4b26bab114156e1b0fd4f840b655305a387fa9faf4b1b6d949"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"doi_title_agreement","ran_at":"2026-05-19T15:01:17.790576Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-19T11:04:31.266533Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2605.08464/integrity.json","findings":[],"snapshot_sha256":"83d621ee3c9123b3fe62b65a989400e7aad285eaa19e6c414e347ddef459c69f","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"The manifold hypothesis (MH) is often used to explain how machine learning can overcome the curse of dimensionality. However, the MH is only applicable in regimes where the training data provides a sufficiently dense sample of the underlying low-dimensional data manifold, or where such a low-dimensional manifold is conceivably present. We describe the regimes where the MH is not applicable as sparse. In this paper, we demonstrate that models succeed in the sparse regime by exploiting a highly structured local geometry, a property we formalize as normal alignment. We prove that normal-aligned c","authors_text":"Ahmed Imtiaz Humayun, Randall Balestriero, Richard Baraniuk, Thomas Walker, T. Mitchell Roddenberry","cross_cats":[],"headline":"Models succeed on sparse data by making their input-output Jacobians rank-one and perfectly aligned with each training point.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-08T20:30:22Z","title":"The Geometric Structure of Models Learning Sparse Data"},"references":{"count":46,"internal_anchors":2,"resolved_work":46,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Tenenbaum, Vin de Silva, and John C","work_id":"5a469211-854a-42bd-9bdb-9ae3b6a90c2d","year":2000},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Roweis and Lawrence K","work_id":"8beed03d-4aea-49dd-aad3-2ac0f4b1eaf9","year":2000},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Dauphin, and David Lopez-Paz","work_id":"eddc5252-cf0b-47e6-be05-0151d26825be","year":2018},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features","work_id":"71760d81-100c-4b37-81f3-ec00607e0354","year":2019},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Data Augmentation Using Random Image Cropping and Patching for Deep CNNs.IEEE Trans","work_id":"31c1b904-c1f8-47f4-a532-3eedf4857393","year":2020}],"snapshot_sha256":"f694b9fb32aeea1809eb50c906fbc14956f2a0e2167167bd0f2cc6c5398439d6"},"source":{"id":"2605.08464","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-19T17:56:29.297098Z","id":"67b2ff44-b3c0-4e74-835d-9a0b1772d5a4","model_set":{"reader":"grok-4.3"},"one_line_summary":"Normal alignment is the rank-one Jacobian structure that lets classifiers minimize loss and maximize local robustness in sparse regimes; the paper proves its optimality and uses it to create GrokAlign and RFAMs.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Models succeed on sparse data by making their input-output Jacobians rank-one and perfectly aligned with each training point.","strongest_claim":"normal-aligned classifiers -- whose input-output Jacobians are rank-one and align perfectly with the training data -- minimize the training objective under norm constraints and achieve maximal local robustness under a non-zero Jacobian constraint","weakest_assumption":"The assumption that success in the sparse regime is explained by normal alignment rather than other mechanisms, and that this alignment arises specifically from the feature-learning regime in continuous piecewise-affine networks (as described in the abstract when discussing power-diagram partitions)."}},"verdict_id":"67b2ff44-b3c0-4e74-835d-9a0b1772d5a4"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:d680a58a6ddfa8498f1e33874e5ada1c41c8eba8a60081f8aa4aaa5a918eae4d","target":"record","created_at":"2026-05-20T00:01:43Z","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":"05a5e5d4b0216f5dbb4313e92ce1a8b57ea3950b29b5bf62dd042a4e57e6ae97","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-08T20:30:22Z","title_canon_sha256":"77607c69e021cb880fe63eec12b930157c1c8fc22ab1213560a67c3184249fa3"},"schema_version":"1.0","source":{"id":"2605.08464","kind":"arxiv","version":2}},"canonical_sha256":"1c46193934cbfe74f62267798af49ed5a521499f2dafb7129c80ef81d8bff2d0","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"1c46193934cbfe74f62267798af49ed5a521499f2dafb7129c80ef81d8bff2d0","first_computed_at":"2026-05-20T00:01:43.020288Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:01:43.020288Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"6ug7BVrAT+gP5CJWw89VmThVl9jr7gN/p1FYarrZV8tI8cnM4D+SfI1/8lDqvE5O4wGt11hrnpVFNM+vdiX8BQ==","signature_status":"signed_v1","signed_at":"2026-05-20T00:01:43.020944Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.08464","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:d680a58a6ddfa8498f1e33874e5ada1c41c8eba8a60081f8aa4aaa5a918eae4d","sha256:ee13a0eb971773ca493eb649bc04a5dcece74a50b45c42762f51542b841584a7"],"state_sha256":"71aac913ad03f3d35c6b52be5c5ffa5b0ba857e8981c7d302b1063dc657b581f"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ndtDpjo8Ll64FbCuzS50eIxEGALD5FBistEOaWCxxnr6OJHgM5VRVLAYzftyJjnZXxqAh0u6LNUIVrf35C1jBg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-31T11:23:05.807083Z","bundle_sha256":"cd9eeb280ff8c1b6d67053fe2427bcffa8881560266a0a020a6db454285020cf"}}