{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:K6EHR2L4CQYD32HL7ON5NJXBM2","short_pith_number":"pith:K6EHR2L4","canonical_record":{"source":{"id":"2511.08544","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-11-11T18:21:55Z","cross_cats_sorted":["cs.AI","cs.CV","stat.ML"],"title_canon_sha256":"8864f231b3df8d62e7247350818cb3ec58832e957fc58576e3c0681bb84d44ef","abstract_canon_sha256":"1865d810e53231cfd50cd36d8342752f7036d7e986858f9c3bc5b802b0b80e58"},"schema_version":"1.0"},"canonical_sha256":"578878e97c14303de8ebfb9bd6a6e166979eace01d53a484ce3341e3b40cb4c9","source":{"kind":"arxiv","id":"2511.08544","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2511.08544","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"arxiv_version","alias_value":"2511.08544v3","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2511.08544","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"pith_short_12","alias_value":"K6EHR2L4CQYD","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"K6EHR2L4CQYD32HL","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"K6EHR2L4","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:K6EHR2L4CQYD32HL7ON5NJXBM2","target":"record","payload":{"canonical_record":{"source":{"id":"2511.08544","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-11-11T18:21:55Z","cross_cats_sorted":["cs.AI","cs.CV","stat.ML"],"title_canon_sha256":"8864f231b3df8d62e7247350818cb3ec58832e957fc58576e3c0681bb84d44ef","abstract_canon_sha256":"1865d810e53231cfd50cd36d8342752f7036d7e986858f9c3bc5b802b0b80e58"},"schema_version":"1.0"},"canonical_sha256":"578878e97c14303de8ebfb9bd6a6e166979eace01d53a484ce3341e3b40cb4c9","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:48.705163Z","signature_b64":"AF9ZbSARLNIrzeJW5iqtmEJk8p5ybsJ6kM6cmzf8kkqAEUK8r/aflpAs9N0Gg3JtrmYwoNh2qPFAx+h01B+IDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"578878e97c14303de8ebfb9bd6a6e166979eace01d53a484ce3341e3b40cb4c9","last_reissued_at":"2026-05-17T23:38:48.704724Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:48.704724Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2511.08544","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"mqaXYDQlQILNKC13zJpJ0D9Fb+e1vnA7me2JOPUZIDy0zzlXAdXhygz76p0mPrXXJPrHpHHqfKfbWzmnvXKiDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-05T22:25:07.429039Z"},"content_sha256":"8bc42e5da3f39a389c3cbf60389e710171c0bb38a18c2a5afabba63a1509fca1","schema_version":"1.0","event_id":"sha256:8bc42e5da3f39a389c3cbf60389e710171c0bb38a18c2a5afabba63a1509fca1"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:K6EHR2L4CQYD32HL7ON5NJXBM2","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristics","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"LeJEPA shows that self-supervised embeddings reach minimal downstream prediction risk when constrained to an isotropic Gaussian distribution via sketched regularization added to the JEPA loss.","cross_cats":["cs.AI","cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Randall Balestriero, Yann LeCun","submitted_at":"2025-11-11T18:21:55Z","abstract_excerpt":"Learning manipulable representations of the world and its dynamics is central to AI. Joint-Embedding Predictive Architectures (JEPAs) offer a promising blueprint, but lack of practical guidance and theory has led to ad-hoc R&D. We present a comprehensive theory of JEPAs and instantiate it in {\\bf LeJEPA}, a lean, scalable, and theoretically grounded training objective. First, we identify the isotropic Gaussian as the optimal distribution that JEPAs' embeddings should follow to minimize downstream prediction risk. Second, we introduce a novel objective--{\\bf Sketched Isotropic Gaussian Regulari"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Combining the JEPA predictive loss with SIGReg yields LeJEPA with single trade-off hyperparameter, linear time/memory, stability across architectures and domains, and heuristics-free training; on ImageNet-1k pretraining with linear evaluation, LeJEPA reaches 79% with ViT-H/14.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The derivation that the isotropic Gaussian is the distribution minimizing downstream prediction risk holds under the paper's stated conditions on the embedding space and loss; if this optimality proof contains gaps or unstated assumptions about the data distribution, the regularization target and claimed benefits would not follow.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"LeJEPA derives an optimal isotropic Gaussian target for embeddings and enforces it via sketched regularization to deliver scalable, heuristics-free self-supervised pretraining with 79% ImageNet linear accuracy on ViT-H/14.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"LeJEPA shows that self-supervised embeddings reach minimal downstream prediction risk when constrained to an isotropic Gaussian distribution via sketched regularization added to the JEPA loss.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"9a9fd7648fa3f5ff0d4ecb0365f29a0e6d5f691b4f0b859e6bed4c6eab0119d2"},"source":{"id":"2511.08544","kind":"arxiv","version":3},"verdict":{"id":"43ed7a62-561c-4631-a8ec-6a66afb5431b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T07:18:53.098609Z","strongest_claim":"Combining the JEPA predictive loss with SIGReg yields LeJEPA with single trade-off hyperparameter, linear time/memory, stability across architectures and domains, and heuristics-free training; on ImageNet-1k pretraining with linear evaluation, LeJEPA reaches 79% with ViT-H/14.","one_line_summary":"LeJEPA derives an optimal isotropic Gaussian target for embeddings and enforces it via sketched regularization to deliver scalable, heuristics-free self-supervised pretraining with 79% ImageNet linear accuracy on ViT-H/14.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The derivation that the isotropic Gaussian is the distribution minimizing downstream prediction risk holds under the paper's stated conditions on the embedding space and loss; if this optimality proof contains gaps or unstated assumptions about the data distribution, the regularization target and claimed benefits would not follow.","pith_extraction_headline":"LeJEPA shows that self-supervised embeddings reach minimal downstream prediction risk when constrained to an isotropic Gaussian distribution via sketched regularization added to the JEPA loss."},"references":{"count":144,"sample":[{"doi":"","year":1965,"title":"An analysis of variance test for normality (complete samples) , author=. Biometrika , volume=. 1965 , publisher=","work_id":"0151ffb6-7db8-4053-9a0c-943b5a8e8fe3","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1961,"title":"Goodness-of-fit tests on a circle , author=. Biometrika , volume=. 1961 , publisher=","work_id":"6070f507-1302-4db3-9eb6-4ebbf589c455","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2010,"title":"Digital nets and sequences: discrepancy theory and quasi--Monte Carlo integration , author=. 2010 , publisher=","work_id":"9f67bedc-52a6-4695-9428-c8dcb748044b","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Characteristic functions , author=. NA , year=","work_id":"0053b0da-84ac-4b43-99cb-c60b82ee2b30","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2004,"title":"Econometric reviews , volume=","work_id":"a0c6cdbb-c529-46f3-8583-34ce1bbb1b6b","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":144,"snapshot_sha256":"606fd85a5ad83012d4d8fdf6c8e5d1410d11200dc28da6f41be161f758791e78","internal_anchors":11},"formal_canon":{"evidence_count":3,"snapshot_sha256":"a4c82e9b7a69010c1ce3893e89b3e844e6e02764901f145e67d18a84a16ff61b"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"43ed7a62-561c-4631-a8ec-6a66afb5431b"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"meGmLyT4M8JBtixYNGPfUZb7PL47iJvsSFJKOvO7ZqGFkYON39tIRkmjCBv83Rgnu5QV32Blz79lK8lBqdiWDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-05T22:25:07.430016Z"},"content_sha256":"87fcd00c70d8879981206c146a1b5d0a6fdf61b00a105d27a1fbf7fdc9b46d9c","schema_version":"1.0","event_id":"sha256:87fcd00c70d8879981206c146a1b5d0a6fdf61b00a105d27a1fbf7fdc9b46d9c"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/K6EHR2L4CQYD32HL7ON5NJXBM2/bundle.json","state_url":"https://pith.science/pith/K6EHR2L4CQYD32HL7ON5NJXBM2/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/K6EHR2L4CQYD32HL7ON5NJXBM2/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":"2026-06-05T22:25:07Z","links":{"resolver":"https://pith.science/pith/K6EHR2L4CQYD32HL7ON5NJXBM2","bundle":"https://pith.science/pith/K6EHR2L4CQYD32HL7ON5NJXBM2/bundle.json","state":"https://pith.science/pith/K6EHR2L4CQYD32HL7ON5NJXBM2/state.json","well_known_bundle":"https://pith.science/.well-known/pith/K6EHR2L4CQYD32HL7ON5NJXBM2/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:K6EHR2L4CQYD32HL7ON5NJXBM2","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":"1865d810e53231cfd50cd36d8342752f7036d7e986858f9c3bc5b802b0b80e58","cross_cats_sorted":["cs.AI","cs.CV","stat.ML"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-11-11T18:21:55Z","title_canon_sha256":"8864f231b3df8d62e7247350818cb3ec58832e957fc58576e3c0681bb84d44ef"},"schema_version":"1.0","source":{"id":"2511.08544","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2511.08544","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"arxiv_version","alias_value":"2511.08544v3","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2511.08544","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"pith_short_12","alias_value":"K6EHR2L4CQYD","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"K6EHR2L4CQYD32HL","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"K6EHR2L4","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:87fcd00c70d8879981206c146a1b5d0a6fdf61b00a105d27a1fbf7fdc9b46d9c","target":"graph","created_at":"2026-05-17T23:38:48Z","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":"Combining the JEPA predictive loss with SIGReg yields LeJEPA with single trade-off hyperparameter, linear time/memory, stability across architectures and domains, and heuristics-free training; on ImageNet-1k pretraining with linear evaluation, LeJEPA reaches 79% with ViT-H/14."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The derivation that the isotropic Gaussian is the distribution minimizing downstream prediction risk holds under the paper's stated conditions on the embedding space and loss; if this optimality proof contains gaps or unstated assumptions about the data distribution, the regularization target and claimed benefits would not follow."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"LeJEPA derives an optimal isotropic Gaussian target for embeddings and enforces it via sketched regularization to deliver scalable, heuristics-free self-supervised pretraining with 79% ImageNet linear accuracy on ViT-H/14."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"LeJEPA shows that self-supervised embeddings reach minimal downstream prediction risk when constrained to an isotropic Gaussian distribution via sketched regularization added to the JEPA loss."}],"snapshot_sha256":"9a9fd7648fa3f5ff0d4ecb0365f29a0e6d5f691b4f0b859e6bed4c6eab0119d2"},"formal_canon":{"evidence_count":3,"snapshot_sha256":"a4c82e9b7a69010c1ce3893e89b3e844e6e02764901f145e67d18a84a16ff61b"},"paper":{"abstract_excerpt":"Learning manipulable representations of the world and its dynamics is central to AI. Joint-Embedding Predictive Architectures (JEPAs) offer a promising blueprint, but lack of practical guidance and theory has led to ad-hoc R&D. We present a comprehensive theory of JEPAs and instantiate it in {\\bf LeJEPA}, a lean, scalable, and theoretically grounded training objective. First, we identify the isotropic Gaussian as the optimal distribution that JEPAs' embeddings should follow to minimize downstream prediction risk. Second, we introduce a novel objective--{\\bf Sketched Isotropic Gaussian Regulari","authors_text":"Randall Balestriero, Yann LeCun","cross_cats":["cs.AI","cs.CV","stat.ML"],"headline":"LeJEPA shows that self-supervised embeddings reach minimal downstream prediction risk when constrained to an isotropic Gaussian distribution via sketched regularization added to the JEPA loss.","license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-11-11T18:21:55Z","title":"LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristics"},"references":{"count":144,"internal_anchors":11,"resolved_work":144,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"An analysis of variance test for normality (complete samples) , author=. Biometrika , volume=. 1965 , publisher=","work_id":"0151ffb6-7db8-4053-9a0c-943b5a8e8fe3","year":1965},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Goodness-of-fit tests on a circle , author=. Biometrika , volume=. 1961 , publisher=","work_id":"6070f507-1302-4db3-9eb6-4ebbf589c455","year":1961},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Digital nets and sequences: discrepancy theory and quasi--Monte Carlo integration , author=. 2010 , publisher=","work_id":"9f67bedc-52a6-4695-9428-c8dcb748044b","year":2010},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Characteristic functions , author=. NA , year=","work_id":"0053b0da-84ac-4b43-99cb-c60b82ee2b30","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Econometric reviews , volume=","work_id":"a0c6cdbb-c529-46f3-8583-34ce1bbb1b6b","year":2004}],"snapshot_sha256":"606fd85a5ad83012d4d8fdf6c8e5d1410d11200dc28da6f41be161f758791e78"},"source":{"id":"2511.08544","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-16T07:18:53.098609Z","id":"43ed7a62-561c-4631-a8ec-6a66afb5431b","model_set":{"reader":"grok-4.3"},"one_line_summary":"LeJEPA derives an optimal isotropic Gaussian target for embeddings and enforces it via sketched regularization to deliver scalable, heuristics-free self-supervised pretraining with 79% ImageNet linear accuracy on ViT-H/14.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"LeJEPA shows that self-supervised embeddings reach minimal downstream prediction risk when constrained to an isotropic Gaussian distribution via sketched regularization added to the JEPA loss.","strongest_claim":"Combining the JEPA predictive loss with SIGReg yields LeJEPA with single trade-off hyperparameter, linear time/memory, stability across architectures and domains, and heuristics-free training; on ImageNet-1k pretraining with linear evaluation, LeJEPA reaches 79% with ViT-H/14.","weakest_assumption":"The derivation that the isotropic Gaussian is the distribution minimizing downstream prediction risk holds under the paper's stated conditions on the embedding space and loss; if this optimality proof contains gaps or unstated assumptions about the data distribution, the regularization target and claimed benefits would not follow."}},"verdict_id":"43ed7a62-561c-4631-a8ec-6a66afb5431b"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:8bc42e5da3f39a389c3cbf60389e710171c0bb38a18c2a5afabba63a1509fca1","target":"record","created_at":"2026-05-17T23:38:48Z","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":"1865d810e53231cfd50cd36d8342752f7036d7e986858f9c3bc5b802b0b80e58","cross_cats_sorted":["cs.AI","cs.CV","stat.ML"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-11-11T18:21:55Z","title_canon_sha256":"8864f231b3df8d62e7247350818cb3ec58832e957fc58576e3c0681bb84d44ef"},"schema_version":"1.0","source":{"id":"2511.08544","kind":"arxiv","version":3}},"canonical_sha256":"578878e97c14303de8ebfb9bd6a6e166979eace01d53a484ce3341e3b40cb4c9","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"578878e97c14303de8ebfb9bd6a6e166979eace01d53a484ce3341e3b40cb4c9","first_computed_at":"2026-05-17T23:38:48.704724Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:48.704724Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"AF9ZbSARLNIrzeJW5iqtmEJk8p5ybsJ6kM6cmzf8kkqAEUK8r/aflpAs9N0Gg3JtrmYwoNh2qPFAx+h01B+IDA==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:48.705163Z","signed_message":"canonical_sha256_bytes"},"source_id":"2511.08544","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:8bc42e5da3f39a389c3cbf60389e710171c0bb38a18c2a5afabba63a1509fca1","sha256:87fcd00c70d8879981206c146a1b5d0a6fdf61b00a105d27a1fbf7fdc9b46d9c"],"state_sha256":"4a576063656b87928e76277d9fada18fd99c1fdc1fae086f1f178675253909cb"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"fzawgG9bxXNNi7nmi5VxwT2vfAiqU9ytcf+rHVeFDMILtsaIyMEFWTNkPLYxU/vdqHH5aS0rn/z0x4DZUsREBg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-05T22:25:07.435383Z","bundle_sha256":"ee79fc7e82bd770da7084c2a7bdfdb6129f6c1e8c9a65f8781dbf856cfe18458"}}