{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:D5IJXTCOOSNFURMIHEUUOZBBAP","short_pith_number":"pith:D5IJXTCO","canonical_record":{"source":{"id":"2303.15389","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-03-27T17:02:21Z","cross_cats_sorted":[],"title_canon_sha256":"2f74502de0eed85b21d45e4952242da7b637c6c5302a8f6b11330eab6507cd44","abstract_canon_sha256":"f790cd3f9582e1afd0f8ee958881604f311c8278fa6821ce93dade246b871e37"},"schema_version":"1.0"},"canonical_sha256":"1f509bcc4e749a5a4588392947642103c300f546aa4dcf3367c92d60d02810c5","source":{"kind":"arxiv","id":"2303.15389","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2303.15389","created_at":"2026-07-05T05:55:04Z"},{"alias_kind":"arxiv_version","alias_value":"2303.15389v1","created_at":"2026-07-05T05:55:04Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2303.15389","created_at":"2026-07-05T05:55:04Z"},{"alias_kind":"pith_short_12","alias_value":"D5IJXTCOOSNF","created_at":"2026-07-05T05:55:04Z"},{"alias_kind":"pith_short_16","alias_value":"D5IJXTCOOSNFURMI","created_at":"2026-07-05T05:55:04Z"},{"alias_kind":"pith_short_8","alias_value":"D5IJXTCO","created_at":"2026-07-05T05:55:04Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:D5IJXTCOOSNFURMIHEUUOZBBAP","target":"record","payload":{"canonical_record":{"source":{"id":"2303.15389","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-03-27T17:02:21Z","cross_cats_sorted":[],"title_canon_sha256":"2f74502de0eed85b21d45e4952242da7b637c6c5302a8f6b11330eab6507cd44","abstract_canon_sha256":"f790cd3f9582e1afd0f8ee958881604f311c8278fa6821ce93dade246b871e37"},"schema_version":"1.0"},"canonical_sha256":"1f509bcc4e749a5a4588392947642103c300f546aa4dcf3367c92d60d02810c5","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:55:04.629034Z","signature_b64":"/Y4rfYl9XndEiwMSB+NmNCvIdkm4ulxzbKqnn0aI5d8cEvzNw5n7UHyOphqXXk4uj8QUqGvNqf7p0n07fVeCAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1f509bcc4e749a5a4588392947642103c300f546aa4dcf3367c92d60d02810c5","last_reissued_at":"2026-07-05T05:55:04.628558Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:55:04.628558Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2303.15389","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":"2026-07-05T05:55:04Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"9CfosqIQlytU0QHsj0uadHFVFnnnh7NPcQcqAlRVZdMXzX3n4O7JbPNcRck7BLOT+v37d536MALoYQpQUi8WDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T07:20:05.981524Z"},"content_sha256":"129a524a54ef882cff331d0f2646d959f24e2e87ef9a7ebad1378292675bddf9","schema_version":"1.0","event_id":"sha256:129a524a54ef882cff331d0f2646d959f24e2e87ef9a7ebad1378292675bddf9"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:D5IJXTCOOSNFURMIHEUUOZBBAP","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"EVA-CLIP: Improved Training Techniques for CLIP at Scale","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"EVA-CLIP applies new techniques in representation learning, optimization, and augmentation to train CLIP models that reach higher zero-shot ImageNet accuracy with far fewer samples than prior versions.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Ledell Wu, Quan Sun, Xinlong Wang, Yue Cao, Yuxin Fang","submitted_at":"2023-03-27T17:02:21Z","abstract_excerpt":"Contrastive language-image pre-training, CLIP for short, has gained increasing attention for its potential in various scenarios. In this paper, we propose EVA-CLIP, a series of models that significantly improve the efficiency and effectiveness of CLIP training. Our approach incorporates new techniques for representation learning, optimization, and augmentation, enabling EVA-CLIP to achieve superior performance compared to previous CLIP models with the same number of parameters but significantly smaller training costs. Notably, our largest 5.0B-parameter EVA-02-CLIP-E/14+ with only 9 billion se"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our largest 5.0B-parameter EVA-02-CLIP-E/14+ with only 9 billion seen samples achieves 82.0 zero-shot top-1 accuracy on ImageNet-1K val. A smaller EVA-02-CLIP-L/14+ with only 430 million parameters and 6 billion seen samples achieves 80.4 zero-shot top-1 accuracy on ImageNet-1K val.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the reported gains stem primarily from the proposed representation, optimization, and augmentation techniques rather than from undisclosed differences in training data scale, curation, or hardware-specific optimizations.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"EVA-CLIP delivers improved CLIP training recipes that yield 82.0% zero-shot ImageNet-1K accuracy for a 5B-parameter model after only 9 billion samples.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"EVA-CLIP applies new techniques in representation learning, optimization, and augmentation to train CLIP models that reach higher zero-shot ImageNet accuracy with far fewer samples than prior versions.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"30e53d50fbd873204ca0dc6a1e0c1703fd8d08e1e94b4e65f7c330b3a5fe8e9f"},"source":{"id":"2303.15389","kind":"arxiv","version":1},"verdict":{"id":"0114d8d0-de8b-4273-b21c-f0bf7a4a8bc1","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T01:48:55.298479Z","strongest_claim":"Our largest 5.0B-parameter EVA-02-CLIP-E/14+ with only 9 billion seen samples achieves 82.0 zero-shot top-1 accuracy on ImageNet-1K val. A smaller EVA-02-CLIP-L/14+ with only 430 million parameters and 6 billion seen samples achieves 80.4 zero-shot top-1 accuracy on ImageNet-1K val.","one_line_summary":"EVA-CLIP delivers improved CLIP training recipes that yield 82.0% zero-shot ImageNet-1K accuracy for a 5B-parameter model after only 9 billion samples.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the reported gains stem primarily from the proposed representation, optimization, and augmentation techniques rather than from undisclosed differences in training data scale, curation, or hardware-specific optimizations.","pith_extraction_headline":"EVA-CLIP applies new techniques in representation learning, optimization, and augmentation to train CLIP models that reach higher zero-shot ImageNet accuracy with far fewer samples than prior versions."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2303.15389/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":53,"sample":[{"doi":"","year":null,"title":"https://laion.ai/blog/giant-openclip/","work_id":"397cdab5-c78c-4464-947d-a8015ccddb8c","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"BEiT: BERT Pre-Training of Image Transformers","work_id":"d74eda3c-bf7e-45f1-a8f1-a0137ecca3f4","ref_index":2,"cited_arxiv_id":"2106.08254","is_internal_anchor":true},{"doi":"","year":2019,"title":"Ob- jectnet: A large-scale bias-controlled dataset for pushing the limits of object recognition models","work_id":"39996506-9c46-4fc3-88ad-f2f46f2f25a5","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2014,"title":"Birdsnap: Large- 5 config EV A-01-CLIP-g / EV A-02-CLIP-g+ image enc","work_id":"f1c2a878-b555-4592-9b4c-5480522ff5ee","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2014,"title":"Food- 101–mining discriminative components with random forests","work_id":"4691f693-8753-4bb2-bf05-c9f571b59cc0","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":53,"snapshot_sha256":"925539f49a8d0d2301c5eca8de8900023b8d928e50c883a63bfc0cdda16d43e4","internal_anchors":10},"formal_canon":{"evidence_count":2,"snapshot_sha256":"a36ccffafc48d3e61a8af64d8d5c02a4556a795fe3757e24be8d4155e4839d5e"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"0114d8d0-de8b-4273-b21c-f0bf7a4a8bc1"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T05:55:04Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"fYvn5O2Wh1Tz2ekoHqJLUVF7ZX1QddmRvlEgJtGyNYQSUN3xFAtMHLUnM+bn2318dcUbsbugJfFmhP5PS8OCCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T07:20:05.982551Z"},"content_sha256":"229eeef733e9905238a5f7dc64355248ae33e9e4d7121b0e91e1809654946570","schema_version":"1.0","event_id":"sha256:229eeef733e9905238a5f7dc64355248ae33e9e4d7121b0e91e1809654946570"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/D5IJXTCOOSNFURMIHEUUOZBBAP/bundle.json","state_url":"https://pith.science/pith/D5IJXTCOOSNFURMIHEUUOZBBAP/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/D5IJXTCOOSNFURMIHEUUOZBBAP/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":"2026-08-06T07:20:05Z","links":{"resolver":"https://pith.science/pith/D5IJXTCOOSNFURMIHEUUOZBBAP","bundle":"https://pith.science/pith/D5IJXTCOOSNFURMIHEUUOZBBAP/bundle.json","state":"https://pith.science/pith/D5IJXTCOOSNFURMIHEUUOZBBAP/state.json","well_known_bundle":"https://pith.science/.well-known/pith/D5IJXTCOOSNFURMIHEUUOZBBAP/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:D5IJXTCOOSNFURMIHEUUOZBBAP","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":"f790cd3f9582e1afd0f8ee958881604f311c8278fa6821ce93dade246b871e37","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-03-27T17:02:21Z","title_canon_sha256":"2f74502de0eed85b21d45e4952242da7b637c6c5302a8f6b11330eab6507cd44"},"schema_version":"1.0","source":{"id":"2303.15389","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2303.15389","created_at":"2026-07-05T05:55:04Z"},{"alias_kind":"arxiv_version","alias_value":"2303.15389v1","created_at":"2026-07-05T05:55:04Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2303.15389","created_at":"2026-07-05T05:55:04Z"},{"alias_kind":"pith_short_12","alias_value":"D5IJXTCOOSNF","created_at":"2026-07-05T05:55:04Z"},{"alias_kind":"pith_short_16","alias_value":"D5IJXTCOOSNFURMI","created_at":"2026-07-05T05:55:04Z"},{"alias_kind":"pith_short_8","alias_value":"D5IJXTCO","created_at":"2026-07-05T05:55:04Z"}],"graph_snapshots":[{"event_id":"sha256:229eeef733e9905238a5f7dc64355248ae33e9e4d7121b0e91e1809654946570","target":"graph","created_at":"2026-07-05T05:55:04Z","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":"Our largest 5.0B-parameter EVA-02-CLIP-E/14+ with only 9 billion seen samples achieves 82.0 zero-shot top-1 accuracy on ImageNet-1K val. A smaller EVA-02-CLIP-L/14+ with only 430 million parameters and 6 billion seen samples achieves 80.4 zero-shot top-1 accuracy on ImageNet-1K val."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the reported gains stem primarily from the proposed representation, optimization, and augmentation techniques rather than from undisclosed differences in training data scale, curation, or hardware-specific optimizations."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"EVA-CLIP delivers improved CLIP training recipes that yield 82.0% zero-shot ImageNet-1K accuracy for a 5B-parameter model after only 9 billion samples."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"EVA-CLIP applies new techniques in representation learning, optimization, and augmentation to train CLIP models that reach higher zero-shot ImageNet accuracy with far fewer samples than prior versions."}],"snapshot_sha256":"30e53d50fbd873204ca0dc6a1e0c1703fd8d08e1e94b4e65f7c330b3a5fe8e9f"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"a36ccffafc48d3e61a8af64d8d5c02a4556a795fe3757e24be8d4155e4839d5e"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2303.15389/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Contrastive language-image pre-training, CLIP for short, has gained increasing attention for its potential in various scenarios. In this paper, we propose EVA-CLIP, a series of models that significantly improve the efficiency and effectiveness of CLIP training. Our approach incorporates new techniques for representation learning, optimization, and augmentation, enabling EVA-CLIP to achieve superior performance compared to previous CLIP models with the same number of parameters but significantly smaller training costs. Notably, our largest 5.0B-parameter EVA-02-CLIP-E/14+ with only 9 billion se","authors_text":"Ledell Wu, Quan Sun, Xinlong Wang, Yue Cao, Yuxin Fang","cross_cats":[],"headline":"EVA-CLIP applies new techniques in representation learning, optimization, and augmentation to train CLIP models that reach higher zero-shot ImageNet accuracy with far fewer samples than prior versions.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-03-27T17:02:21Z","title":"EVA-CLIP: Improved Training Techniques for CLIP at Scale"},"references":{"count":53,"internal_anchors":10,"resolved_work":53,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"https://laion.ai/blog/giant-openclip/","work_id":"397cdab5-c78c-4464-947d-a8015ccddb8c","year":null},{"cited_arxiv_id":"2106.08254","doi":"","is_internal_anchor":true,"ref_index":2,"title":"BEiT: BERT Pre-Training of Image Transformers","work_id":"d74eda3c-bf7e-45f1-a8f1-a0137ecca3f4","year":2021},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Ob- jectnet: A large-scale bias-controlled dataset for pushing the limits of object recognition models","work_id":"39996506-9c46-4fc3-88ad-f2f46f2f25a5","year":2019},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Birdsnap: Large- 5 config EV A-01-CLIP-g / EV A-02-CLIP-g+ image enc","work_id":"f1c2a878-b555-4592-9b4c-5480522ff5ee","year":2014},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Food- 101–mining discriminative components with random forests","work_id":"4691f693-8753-4bb2-bf05-c9f571b59cc0","year":2014}],"snapshot_sha256":"925539f49a8d0d2301c5eca8de8900023b8d928e50c883a63bfc0cdda16d43e4"},"source":{"id":"2303.15389","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-13T01:48:55.298479Z","id":"0114d8d0-de8b-4273-b21c-f0bf7a4a8bc1","model_set":{"reader":"grok-4.3"},"one_line_summary":"EVA-CLIP delivers improved CLIP training recipes that yield 82.0% zero-shot ImageNet-1K accuracy for a 5B-parameter model after only 9 billion samples.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"EVA-CLIP applies new techniques in representation learning, optimization, and augmentation to train CLIP models that reach higher zero-shot ImageNet accuracy with far fewer samples than prior versions.","strongest_claim":"Our largest 5.0B-parameter EVA-02-CLIP-E/14+ with only 9 billion seen samples achieves 82.0 zero-shot top-1 accuracy on ImageNet-1K val. A smaller EVA-02-CLIP-L/14+ with only 430 million parameters and 6 billion seen samples achieves 80.4 zero-shot top-1 accuracy on ImageNet-1K val.","weakest_assumption":"That the reported gains stem primarily from the proposed representation, optimization, and augmentation techniques rather than from undisclosed differences in training data scale, curation, or hardware-specific optimizations."}},"verdict_id":"0114d8d0-de8b-4273-b21c-f0bf7a4a8bc1"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:129a524a54ef882cff331d0f2646d959f24e2e87ef9a7ebad1378292675bddf9","target":"record","created_at":"2026-07-05T05:55:04Z","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":"f790cd3f9582e1afd0f8ee958881604f311c8278fa6821ce93dade246b871e37","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-03-27T17:02:21Z","title_canon_sha256":"2f74502de0eed85b21d45e4952242da7b637c6c5302a8f6b11330eab6507cd44"},"schema_version":"1.0","source":{"id":"2303.15389","kind":"arxiv","version":1}},"canonical_sha256":"1f509bcc4e749a5a4588392947642103c300f546aa4dcf3367c92d60d02810c5","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"1f509bcc4e749a5a4588392947642103c300f546aa4dcf3367c92d60d02810c5","first_computed_at":"2026-07-05T05:55:04.628558Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T05:55:04.628558Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"/Y4rfYl9XndEiwMSB+NmNCvIdkm4ulxzbKqnn0aI5d8cEvzNw5n7UHyOphqXXk4uj8QUqGvNqf7p0n07fVeCAQ==","signature_status":"signed_v1","signed_at":"2026-07-05T05:55:04.629034Z","signed_message":"canonical_sha256_bytes"},"source_id":"2303.15389","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:129a524a54ef882cff331d0f2646d959f24e2e87ef9a7ebad1378292675bddf9","sha256:229eeef733e9905238a5f7dc64355248ae33e9e4d7121b0e91e1809654946570"],"state_sha256":"42d9f2610d83371ad2e7e3d460908c34087add8523933397e2bc958d4482d3c7"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"EAf2aA4a2Qx1vgORORkqKvnXwiC4xvGc03zQZ3qgUCXqL8B3VnqE7AzJWvEIvlz52OT6AvChRrzcwq0/2hunCQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T07:20:05.990396Z","bundle_sha256":"11c9005329766f4e5d7dc28065b131b563cc981f2e7f094c7dd090477a528d5c"}}