{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:KFKTDL5H6ZCIZBOTZJ73ZEO5UE","short_pith_number":"pith:KFKTDL5H","canonical_record":{"source":{"id":"2604.08302","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-09T14:35:42Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"bb2f046bd6663f117409a9fd6365c2e74dc4bd6fa03d41f06e4e26c85543b725","abstract_canon_sha256":"b2fbbf793f6564d8d824c34f8ff318f7b8e9dcc70d0a8b7f72eb494ac28afed4"},"schema_version":"1.0"},"canonical_sha256":"515531afa7f6448c85d3ca7fbc91dda137e8fdf76d1b98e180ad1c4ef4b31d6c","source":{"kind":"arxiv","id":"2604.08302","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.08302","created_at":"2026-05-20T00:00:37Z"},{"alias_kind":"arxiv_version","alias_value":"2604.08302v3","created_at":"2026-05-20T00:00:37Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.08302","created_at":"2026-05-20T00:00:37Z"},{"alias_kind":"pith_short_12","alias_value":"KFKTDL5H6ZCI","created_at":"2026-05-20T00:00:37Z"},{"alias_kind":"pith_short_16","alias_value":"KFKTDL5H6ZCIZBOT","created_at":"2026-05-20T00:00:37Z"},{"alias_kind":"pith_short_8","alias_value":"KFKTDL5H","created_at":"2026-05-20T00:00:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:KFKTDL5H6ZCIZBOTZJ73ZEO5UE","target":"record","payload":{"canonical_record":{"source":{"id":"2604.08302","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-09T14:35:42Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"bb2f046bd6663f117409a9fd6365c2e74dc4bd6fa03d41f06e4e26c85543b725","abstract_canon_sha256":"b2fbbf793f6564d8d824c34f8ff318f7b8e9dcc70d0a8b7f72eb494ac28afed4"},"schema_version":"1.0"},"canonical_sha256":"515531afa7f6448c85d3ca7fbc91dda137e8fdf76d1b98e180ad1c4ef4b31d6c","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:00:37.788663Z","signature_b64":"pdsEG/IjDKeyALxBtGtqcdVSMk6tW0eL9Joh2uujvsRuYiUop2Cv5t0HDNMFypKZKK+n/CRMCE09H1KYQ+61DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"515531afa7f6448c85d3ca7fbc91dda137e8fdf76d1b98e180ad1c4ef4b31d6c","last_reissued_at":"2026-05-20T00:00:37.787996Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:00:37.787996Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2604.08302","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":"2026-05-20T00:00:37Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"QVtihXvOw50Ho+toUrkXopto7OO+yi//ZRmdR/agqEipQQ1s2qKyYUhgxGyWBskzTNU2oasZ/tkWX7O+z+g8Cg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-30T08:54:08.322825Z"},"content_sha256":"90fabf01e2f33083010b87100a89b022031c62e81276b98d3a4f9283bdc2ff9f","schema_version":"1.0","event_id":"sha256:90fabf01e2f33083010b87100a89b022031c62e81276b98d3a4f9283bdc2ff9f"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:KFKTDL5H6ZCIZBOTZJ73ZEO5UE","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"DMax: Aggressive Parallel Decoding for dLLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"DMax reformulates parallel decoding for diffusion language models as progressive self-refinement from mask to token embeddings.","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Gongfan Fang, Ruonan Yu, Xinchao Wang, Xinyin Ma, Zigeng Chen","submitted_at":"2026-04-09T14:35:42Z","abstract_excerpt":"We present DMax, a new paradigm for efficient diffusion language models (dLLMs). It mitigates error accumulation in parallel decoding, enabling aggressive decoding parallelism while preserving generation quality. Unlike conventional masked dLLMs that decode through a binary mask-to-token transition, DMax reformulates decoding as a progressive self-refinement from mask embeddings to token embeddings. At the core of our approach is On-Policy Uniform Training, a novel training strategy that efficiently unifies masked and uniform dLLMs, equipping the model to recover clean tokens from both masked "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Compared with the original LLaDA-2.0-mini, our method improves TPF on GSM8K from 2.04 to 5.47 while preserving accuracy. On MBPP, it increases TPF from 2.71 to 5.86 while maintaining comparable performance.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That representing each intermediate decoding state as an interpolation between the predicted token embedding and the mask embedding enables effective iterative self-revising in embedding space without accumulating new errors or requiring additional post-processing.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"DMax uses On-Policy Uniform Training and Soft Parallel Decoding to enable aggressive parallelism in dLLMs, raising TPF on GSM8K from 2.04 to 5.47 and on MBPP from 2.71 to 5.86 while preserving accuracy.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"DMax reformulates parallel decoding for diffusion language models as progressive self-refinement from mask to token embeddings.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"88a1417ccbf03f881f079de4f7d9e6a79155ba529afa6b780e822d8e044d8aac"},"source":{"id":"2604.08302","kind":"arxiv","version":3},"verdict":{"id":"d96f9437-f937-4ee8-bf67-17e4ff74376c","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T16:42:17.239764Z","strongest_claim":"Compared with the original LLaDA-2.0-mini, our method improves TPF on GSM8K from 2.04 to 5.47 while preserving accuracy. On MBPP, it increases TPF from 2.71 to 5.86 while maintaining comparable performance.","one_line_summary":"DMax uses On-Policy Uniform Training and Soft Parallel Decoding to enable aggressive parallelism in dLLMs, raising TPF on GSM8K from 2.04 to 5.47 and on MBPP from 2.71 to 5.86 while preserving accuracy.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That representing each intermediate decoding state as an interpolation between the predicted token embedding and the mask embedding enables effective iterative self-revising in embedding space without accumulating new errors or requiring additional post-processing.","pith_extraction_headline":"DMax reformulates parallel decoding for diffusion language models as progressive self-refinement from mask to token embeddings."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.08302/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":110,"sample":[{"doi":"","year":2023,"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","ref_index":1,"cited_arxiv_id":"2303.08774","is_internal_anchor":true},{"doi":"","year":2025,"title":"Opencodeinstruct: A large-scale instruction tuning dataset for code llms","work_id":"136700a7-11f6-48b3-8b28-1be2840954e9","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Models","work_id":"b34ab928-6ffb-4028-b13c-395a8924d76b","ref_index":3,"cited_arxiv_id":"2503.09573","is_internal_anchor":true},{"doi":"","year":2021,"title":"Structured denoising diffusion models in discrete state-spaces.Advances in neural information processing systems, 34:17981–17993","work_id":"b3c340e2-b557-4352-95ba-6967124a292d","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"Program Synthesis with Large Language Models","work_id":"fd241a05-03b9-4de2-9588-9d77ce176125","ref_index":5,"cited_arxiv_id":"2108.07732","is_internal_anchor":true}],"resolved_work":110,"snapshot_sha256":"2bcf7381ee7cc3574e455f924fe4b6a1de39ccc2999c2a7d90fa0bbf20468b79","internal_anchors":24},"formal_canon":{"evidence_count":2,"snapshot_sha256":"ba9ca5ca9f94de360a5275a5428aa7f69840320bff9e0bfcad24bea93087f76d"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"d96f9437-f937-4ee8-bf67-17e4ff74376c"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:00:37Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"YlfC/CnZ3+o62dwR5mk1JmEiFwS5uhFAbNmxMtiZvozYukpTMrn1dHdEaXuyOsUOZ7/P+3fUseLJiqPXGrwxCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-30T08:54:08.323536Z"},"content_sha256":"702660b2121a860c5ea8f4e67de2d15757d883c2c85efa9b85fd6b7c149b2afb","schema_version":"1.0","event_id":"sha256:702660b2121a860c5ea8f4e67de2d15757d883c2c85efa9b85fd6b7c149b2afb"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/KFKTDL5H6ZCIZBOTZJ73ZEO5UE/bundle.json","state_url":"https://pith.science/pith/KFKTDL5H6ZCIZBOTZJ73ZEO5UE/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/KFKTDL5H6ZCIZBOTZJ73ZEO5UE/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":"2026-05-30T08:54:08Z","links":{"resolver":"https://pith.science/pith/KFKTDL5H6ZCIZBOTZJ73ZEO5UE","bundle":"https://pith.science/pith/KFKTDL5H6ZCIZBOTZJ73ZEO5UE/bundle.json","state":"https://pith.science/pith/KFKTDL5H6ZCIZBOTZJ73ZEO5UE/state.json","well_known_bundle":"https://pith.science/.well-known/pith/KFKTDL5H6ZCIZBOTZJ73ZEO5UE/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:KFKTDL5H6ZCIZBOTZJ73ZEO5UE","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":"b2fbbf793f6564d8d824c34f8ff318f7b8e9dcc70d0a8b7f72eb494ac28afed4","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-09T14:35:42Z","title_canon_sha256":"bb2f046bd6663f117409a9fd6365c2e74dc4bd6fa03d41f06e4e26c85543b725"},"schema_version":"1.0","source":{"id":"2604.08302","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.08302","created_at":"2026-05-20T00:00:37Z"},{"alias_kind":"arxiv_version","alias_value":"2604.08302v3","created_at":"2026-05-20T00:00:37Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.08302","created_at":"2026-05-20T00:00:37Z"},{"alias_kind":"pith_short_12","alias_value":"KFKTDL5H6ZCI","created_at":"2026-05-20T00:00:37Z"},{"alias_kind":"pith_short_16","alias_value":"KFKTDL5H6ZCIZBOT","created_at":"2026-05-20T00:00:37Z"},{"alias_kind":"pith_short_8","alias_value":"KFKTDL5H","created_at":"2026-05-20T00:00:37Z"}],"graph_snapshots":[{"event_id":"sha256:702660b2121a860c5ea8f4e67de2d15757d883c2c85efa9b85fd6b7c149b2afb","target":"graph","created_at":"2026-05-20T00:00:37Z","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":"Compared with the original LLaDA-2.0-mini, our method improves TPF on GSM8K from 2.04 to 5.47 while preserving accuracy. On MBPP, it increases TPF from 2.71 to 5.86 while maintaining comparable performance."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That representing each intermediate decoding state as an interpolation between the predicted token embedding and the mask embedding enables effective iterative self-revising in embedding space without accumulating new errors or requiring additional post-processing."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"DMax uses On-Policy Uniform Training and Soft Parallel Decoding to enable aggressive parallelism in dLLMs, raising TPF on GSM8K from 2.04 to 5.47 and on MBPP from 2.71 to 5.86 while preserving accuracy."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"DMax reformulates parallel decoding for diffusion language models as progressive self-refinement from mask to token embeddings."}],"snapshot_sha256":"88a1417ccbf03f881f079de4f7d9e6a79155ba529afa6b780e822d8e044d8aac"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"ba9ca5ca9f94de360a5275a5428aa7f69840320bff9e0bfcad24bea93087f76d"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2604.08302/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We present DMax, a new paradigm for efficient diffusion language models (dLLMs). It mitigates error accumulation in parallel decoding, enabling aggressive decoding parallelism while preserving generation quality. Unlike conventional masked dLLMs that decode through a binary mask-to-token transition, DMax reformulates decoding as a progressive self-refinement from mask embeddings to token embeddings. At the core of our approach is On-Policy Uniform Training, a novel training strategy that efficiently unifies masked and uniform dLLMs, equipping the model to recover clean tokens from both masked ","authors_text":"Gongfan Fang, Ruonan Yu, Xinchao Wang, Xinyin Ma, Zigeng Chen","cross_cats":["cs.AI"],"headline":"DMax reformulates parallel decoding for diffusion language models as progressive self-refinement from mask to token embeddings.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-09T14:35:42Z","title":"DMax: Aggressive Parallel Decoding for dLLMs"},"references":{"count":110,"internal_anchors":24,"resolved_work":110,"sample":[{"cited_arxiv_id":"2303.08774","doi":"","is_internal_anchor":true,"ref_index":1,"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","year":2023},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Opencodeinstruct: A large-scale instruction tuning dataset for code llms","work_id":"136700a7-11f6-48b3-8b28-1be2840954e9","year":2025},{"cited_arxiv_id":"2503.09573","doi":"","is_internal_anchor":true,"ref_index":3,"title":"Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Models","work_id":"b34ab928-6ffb-4028-b13c-395a8924d76b","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Structured denoising diffusion models in discrete state-spaces.Advances in neural information processing systems, 34:17981–17993","work_id":"b3c340e2-b557-4352-95ba-6967124a292d","year":2021},{"cited_arxiv_id":"2108.07732","doi":"","is_internal_anchor":true,"ref_index":5,"title":"Program Synthesis with Large Language Models","work_id":"fd241a05-03b9-4de2-9588-9d77ce176125","year":2021}],"snapshot_sha256":"2bcf7381ee7cc3574e455f924fe4b6a1de39ccc2999c2a7d90fa0bbf20468b79"},"source":{"id":"2604.08302","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-19T16:42:17.239764Z","id":"d96f9437-f937-4ee8-bf67-17e4ff74376c","model_set":{"reader":"grok-4.3"},"one_line_summary":"DMax uses On-Policy Uniform Training and Soft Parallel Decoding to enable aggressive parallelism in dLLMs, raising TPF on GSM8K from 2.04 to 5.47 and on MBPP from 2.71 to 5.86 while preserving accuracy.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"DMax reformulates parallel decoding for diffusion language models as progressive self-refinement from mask to token embeddings.","strongest_claim":"Compared with the original LLaDA-2.0-mini, our method improves TPF on GSM8K from 2.04 to 5.47 while preserving accuracy. On MBPP, it increases TPF from 2.71 to 5.86 while maintaining comparable performance.","weakest_assumption":"That representing each intermediate decoding state as an interpolation between the predicted token embedding and the mask embedding enables effective iterative self-revising in embedding space without accumulating new errors or requiring additional post-processing."}},"verdict_id":"d96f9437-f937-4ee8-bf67-17e4ff74376c"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:90fabf01e2f33083010b87100a89b022031c62e81276b98d3a4f9283bdc2ff9f","target":"record","created_at":"2026-05-20T00:00:37Z","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":"b2fbbf793f6564d8d824c34f8ff318f7b8e9dcc70d0a8b7f72eb494ac28afed4","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-09T14:35:42Z","title_canon_sha256":"bb2f046bd6663f117409a9fd6365c2e74dc4bd6fa03d41f06e4e26c85543b725"},"schema_version":"1.0","source":{"id":"2604.08302","kind":"arxiv","version":3}},"canonical_sha256":"515531afa7f6448c85d3ca7fbc91dda137e8fdf76d1b98e180ad1c4ef4b31d6c","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"515531afa7f6448c85d3ca7fbc91dda137e8fdf76d1b98e180ad1c4ef4b31d6c","first_computed_at":"2026-05-20T00:00:37.787996Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:00:37.787996Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"pdsEG/IjDKeyALxBtGtqcdVSMk6tW0eL9Joh2uujvsRuYiUop2Cv5t0HDNMFypKZKK+n/CRMCE09H1KYQ+61DA==","signature_status":"signed_v1","signed_at":"2026-05-20T00:00:37.788663Z","signed_message":"canonical_sha256_bytes"},"source_id":"2604.08302","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:90fabf01e2f33083010b87100a89b022031c62e81276b98d3a4f9283bdc2ff9f","sha256:702660b2121a860c5ea8f4e67de2d15757d883c2c85efa9b85fd6b7c149b2afb"],"state_sha256":"4755a4220f5257d5e24f555032e56751942c8b316366d031efdddd8f63a02473"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"xOqhmI8c1A6iZWWtZ7TEXd+uWUl2G7HrTBqmXwz5UO0GaEA9HxUR3mB7vM9uY9S3gzPJABNfUKV4okP+G+pgDA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-30T08:54:08.326298Z","bundle_sha256":"cd1ef77873df9e8abc9ca8ffc4075d82241018ac8bd08788f55da315271a6ef7"}}