{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2019:H2V73KGYMGOAIPHIU4U3EM722P","short_pith_number":"pith:H2V73KGY","canonical_record":{"source":{"id":"1911.05507","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-11-13T14:36:01Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"9c91268b8020966108bc25936156fe7dd205c38221cd3b9e1fbc2596cd27ee2f","abstract_canon_sha256":"4f4c0020b10318538db1c245e758a633e4e5b232f2d0f095f78456b5d2445bf9"},"schema_version":"1.0"},"canonical_sha256":"3eabfda8d8619c043ce8a729b233fad3ffff44259e5aa5bd72c8a17d519df321","source":{"kind":"arxiv","id":"1911.05507","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1911.05507","created_at":"2026-07-05T00:18:57Z"},{"alias_kind":"arxiv_version","alias_value":"1911.05507v1","created_at":"2026-07-05T00:18:57Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1911.05507","created_at":"2026-07-05T00:18:57Z"},{"alias_kind":"pith_short_12","alias_value":"H2V73KGYMGOA","created_at":"2026-07-05T00:18:57Z"},{"alias_kind":"pith_short_16","alias_value":"H2V73KGYMGOAIPHI","created_at":"2026-07-05T00:18:57Z"},{"alias_kind":"pith_short_8","alias_value":"H2V73KGY","created_at":"2026-07-05T00:18:57Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2019:H2V73KGYMGOAIPHIU4U3EM722P","target":"record","payload":{"canonical_record":{"source":{"id":"1911.05507","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-11-13T14:36:01Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"9c91268b8020966108bc25936156fe7dd205c38221cd3b9e1fbc2596cd27ee2f","abstract_canon_sha256":"4f4c0020b10318538db1c245e758a633e4e5b232f2d0f095f78456b5d2445bf9"},"schema_version":"1.0"},"canonical_sha256":"3eabfda8d8619c043ce8a729b233fad3ffff44259e5aa5bd72c8a17d519df321","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:18:57.920797Z","signature_b64":"jCmuF/kLlFaSzxsUZDw7d4aYTTm4Nh81CF9fbOT6sjRxdSPYfSNSZm68ocd9sZCxgOOhoSyPHWw+FIrfmGg1Ag==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3eabfda8d8619c043ce8a729b233fad3ffff44259e5aa5bd72c8a17d519df321","last_reissued_at":"2026-07-05T00:18:57.920367Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:18:57.920367Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"1911.05507","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":"2026-07-05T00:18:57Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"geBV6HHdIvpI83f41aGNUp2APE9Z56PjJzZmdm4dxt6aE8yYyV9F2CU72Fp3s1jMOXzff8QjMTlKHDqhBjP6Dg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T04:05:41.034670Z"},"content_sha256":"924425353251e444a76e606858548e8603c0ab966c9287d11ee5bfb24d5f4506","schema_version":"1.0","event_id":"sha256:924425353251e444a76e606858548e8603c0ab966c9287d11ee5bfb24d5f4506"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2019:H2V73KGYMGOAIPHIU4U3EM722P","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Compressive Transformers for Long-Range Sequence Modelling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"The Compressive Transformer compresses past memories to achieve state-of-the-art results on long-range language modeling benchmarks.","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Anna Potapenko, Jack W. Rae, Siddhant M. Jayakumar, Timothy P. Lillicrap","submitted_at":"2019-11-13T14:36:01Z","abstract_excerpt":"We present the Compressive Transformer, an attentive sequence model which compresses past memories for long-range sequence learning. We find the Compressive Transformer obtains state-of-the-art language modelling results in the WikiText-103 and Enwik8 benchmarks, achieving 17.1 ppl and 0.97 bpc respectively. We also find it can model high-frequency speech effectively and can be used as a memory mechanism for RL, demonstrated on an object matching task. To promote the domain of long-range sequence learning, we propose a new open-vocabulary language modelling benchmark derived from books, PG-19."},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We find the Compressive Transformer obtains state-of-the-art language modelling results in the WikiText-103 and Enwik8 benchmarks, achieving 17.1 ppl and 0.97 bpc respectively.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"Compressing past memories retains the information necessary for accurate long-range predictions on the reported benchmarks without introducing unacceptable information loss.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Compressive Transformer sets new records on WikiText-103 (17.1 ppl) and Enwik8 (0.97 bpc) via memory compression and introduces the PG-19 long-range language benchmark.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"The Compressive Transformer compresses past memories to achieve state-of-the-art results on long-range language modeling benchmarks.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"02f2c603de0e20dc39bf40ba3bf49601178458216a36aaa58b9e1a848e46f68d"},"source":{"id":"1911.05507","kind":"arxiv","version":1},"verdict":{"id":"39efda88-77bc-41d7-80f4-de1296821bb8","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-18T10:41:35.199769Z","strongest_claim":"We find the Compressive Transformer obtains state-of-the-art language modelling results in the WikiText-103 and Enwik8 benchmarks, achieving 17.1 ppl and 0.97 bpc respectively.","one_line_summary":"Compressive Transformer sets new records on WikiText-103 (17.1 ppl) and Enwik8 (0.97 bpc) via memory compression and introduces the PG-19 long-range language benchmark.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"Compressing past memories retains the information necessary for accurate long-range predictions on the reported benchmarks without introducing unacceptable information loss.","pith_extraction_headline":"The Compressive Transformer compresses past memories to achieve state-of-the-art results on long-range language modeling benchmarks."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/1911.05507/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":125,"sample":[{"doi":"","year":2019,"title":"R. Al-Rfou, D. Choe, N. Constant, M. Guo, and L. Jones. Character-level language modeling with deeper self-attention. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages","work_id":"a44d68e5-77fc-447a-9411-a6ecaa323cb1","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"S. Bai, J. Z. Kolter, and V. Koltun. Convolutional sequence modeling revisited, 2018 a . URL https://openreview.net/forum?id=rk8wKk-R-","work_id":"240767ac-2879-4ebd-82f9-06e11f530dba","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2016,"title":"DeepMind Lab","work_id":"8a8d827f-5377-4733-bfe8-bc66c011d458","ref_index":6,"cited_arxiv_id":"1612.03801","is_internal_anchor":true},{"doi":"","year":2003,"title":"D. M. Blei, A. Y. Ng, and M. I. Jordan. Latent dirichlet allocation. J. Mach. Learn. Res., 3: 0 993--1022, Mar. 2003. ISSN 1532-4435","work_id":"a0cb2852-a473-45a6-adab-0fa427c3ef4c","ref_index":7,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"L. Espeholt, H. Soyer, R. Munos, K. Simonyan, V. Mnih, T. Ward, Y. Doron, V. Firoiu, T. Harley, I. Dunning, et al. Impala: Scalable distributed deep-rl with importance weighted actor-learner architect","work_id":"a5ca2a29-ba14-460b-bfe2-ffeaa367077b","ref_index":15,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":125,"snapshot_sha256":"7a73f28639549e599f92a0d30254a7e1fbac8809ba49f4e748ab61bbb7610361","internal_anchors":44},"formal_canon":{"evidence_count":2,"snapshot_sha256":"f9c6c909304c6f0766824af81938fda464ec40be90ef59fa8fcdb3cae295a210"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"39efda88-77bc-41d7-80f4-de1296821bb8"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T00:18:57Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"a4zQMF2eEObCg2I8e8sHMo0nhSl9jAlslijQ307BYrF5U6nET1UXvqmr8BCGpSVwTvCVTerucwRxMUHZBZOhDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T04:05:41.035519Z"},"content_sha256":"a96e1eff9009d648c3ed695c8edf21281ea703ac0a704e3c4f35dd88b7113324","schema_version":"1.0","event_id":"sha256:a96e1eff9009d648c3ed695c8edf21281ea703ac0a704e3c4f35dd88b7113324"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/H2V73KGYMGOAIPHIU4U3EM722P/bundle.json","state_url":"https://pith.science/pith/H2V73KGYMGOAIPHIU4U3EM722P/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/H2V73KGYMGOAIPHIU4U3EM722P/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":"2026-08-06T04:05:41Z","links":{"resolver":"https://pith.science/pith/H2V73KGYMGOAIPHIU4U3EM722P","bundle":"https://pith.science/pith/H2V73KGYMGOAIPHIU4U3EM722P/bundle.json","state":"https://pith.science/pith/H2V73KGYMGOAIPHIU4U3EM722P/state.json","well_known_bundle":"https://pith.science/.well-known/pith/H2V73KGYMGOAIPHIU4U3EM722P/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2019:H2V73KGYMGOAIPHIU4U3EM722P","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":"4f4c0020b10318538db1c245e758a633e4e5b232f2d0f095f78456b5d2445bf9","cross_cats_sorted":["stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-11-13T14:36:01Z","title_canon_sha256":"9c91268b8020966108bc25936156fe7dd205c38221cd3b9e1fbc2596cd27ee2f"},"schema_version":"1.0","source":{"id":"1911.05507","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1911.05507","created_at":"2026-07-05T00:18:57Z"},{"alias_kind":"arxiv_version","alias_value":"1911.05507v1","created_at":"2026-07-05T00:18:57Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1911.05507","created_at":"2026-07-05T00:18:57Z"},{"alias_kind":"pith_short_12","alias_value":"H2V73KGYMGOA","created_at":"2026-07-05T00:18:57Z"},{"alias_kind":"pith_short_16","alias_value":"H2V73KGYMGOAIPHI","created_at":"2026-07-05T00:18:57Z"},{"alias_kind":"pith_short_8","alias_value":"H2V73KGY","created_at":"2026-07-05T00:18:57Z"}],"graph_snapshots":[{"event_id":"sha256:a96e1eff9009d648c3ed695c8edf21281ea703ac0a704e3c4f35dd88b7113324","target":"graph","created_at":"2026-07-05T00:18:57Z","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":"We find the Compressive Transformer obtains state-of-the-art language modelling results in the WikiText-103 and Enwik8 benchmarks, achieving 17.1 ppl and 0.97 bpc respectively."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"Compressing past memories retains the information necessary for accurate long-range predictions on the reported benchmarks without introducing unacceptable information loss."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Compressive Transformer sets new records on WikiText-103 (17.1 ppl) and Enwik8 (0.97 bpc) via memory compression and introduces the PG-19 long-range language benchmark."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"The Compressive Transformer compresses past memories to achieve state-of-the-art results on long-range language modeling benchmarks."}],"snapshot_sha256":"02f2c603de0e20dc39bf40ba3bf49601178458216a36aaa58b9e1a848e46f68d"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"f9c6c909304c6f0766824af81938fda464ec40be90ef59fa8fcdb3cae295a210"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/1911.05507/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We present the Compressive Transformer, an attentive sequence model which compresses past memories for long-range sequence learning. We find the Compressive Transformer obtains state-of-the-art language modelling results in the WikiText-103 and Enwik8 benchmarks, achieving 17.1 ppl and 0.97 bpc respectively. We also find it can model high-frequency speech effectively and can be used as a memory mechanism for RL, demonstrated on an object matching task. To promote the domain of long-range sequence learning, we propose a new open-vocabulary language modelling benchmark derived from books, PG-19.","authors_text":"Anna Potapenko, Jack W. Rae, Siddhant M. Jayakumar, Timothy P. Lillicrap","cross_cats":["stat.ML"],"headline":"The Compressive Transformer compresses past memories to achieve state-of-the-art results on long-range language modeling benchmarks.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-11-13T14:36:01Z","title":"Compressive Transformers for Long-Range Sequence Modelling"},"references":{"count":125,"internal_anchors":44,"resolved_work":125,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"R. Al-Rfou, D. Choe, N. Constant, M. Guo, and L. Jones. Character-level language modeling with deeper self-attention. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages","work_id":"a44d68e5-77fc-447a-9411-a6ecaa323cb1","year":2019},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"S. Bai, J. Z. Kolter, and V. Koltun. Convolutional sequence modeling revisited, 2018 a . URL https://openreview.net/forum?id=rk8wKk-R-","work_id":"240767ac-2879-4ebd-82f9-06e11f530dba","year":2018},{"cited_arxiv_id":"1612.03801","doi":"","is_internal_anchor":true,"ref_index":6,"title":"DeepMind Lab","work_id":"8a8d827f-5377-4733-bfe8-bc66c011d458","year":2016},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":7,"title":"D. M. Blei, A. Y. Ng, and M. I. Jordan. Latent dirichlet allocation. J. Mach. Learn. Res., 3: 0 993--1022, Mar. 2003. ISSN 1532-4435","work_id":"a0cb2852-a473-45a6-adab-0fa427c3ef4c","year":2003},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":15,"title":"L. Espeholt, H. Soyer, R. Munos, K. Simonyan, V. Mnih, T. Ward, Y. Doron, V. Firoiu, T. Harley, I. Dunning, et al. Impala: Scalable distributed deep-rl with importance weighted actor-learner architect","work_id":"a5ca2a29-ba14-460b-bfe2-ffeaa367077b","year":2018}],"snapshot_sha256":"7a73f28639549e599f92a0d30254a7e1fbac8809ba49f4e748ab61bbb7610361"},"source":{"id":"1911.05507","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-18T10:41:35.199769Z","id":"39efda88-77bc-41d7-80f4-de1296821bb8","model_set":{"reader":"grok-4.3"},"one_line_summary":"Compressive Transformer sets new records on WikiText-103 (17.1 ppl) and Enwik8 (0.97 bpc) via memory compression and introduces the PG-19 long-range language benchmark.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"The Compressive Transformer compresses past memories to achieve state-of-the-art results on long-range language modeling benchmarks.","strongest_claim":"We find the Compressive Transformer obtains state-of-the-art language modelling results in the WikiText-103 and Enwik8 benchmarks, achieving 17.1 ppl and 0.97 bpc respectively.","weakest_assumption":"Compressing past memories retains the information necessary for accurate long-range predictions on the reported benchmarks without introducing unacceptable information loss."}},"verdict_id":"39efda88-77bc-41d7-80f4-de1296821bb8"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:924425353251e444a76e606858548e8603c0ab966c9287d11ee5bfb24d5f4506","target":"record","created_at":"2026-07-05T00:18:57Z","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":"4f4c0020b10318538db1c245e758a633e4e5b232f2d0f095f78456b5d2445bf9","cross_cats_sorted":["stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-11-13T14:36:01Z","title_canon_sha256":"9c91268b8020966108bc25936156fe7dd205c38221cd3b9e1fbc2596cd27ee2f"},"schema_version":"1.0","source":{"id":"1911.05507","kind":"arxiv","version":1}},"canonical_sha256":"3eabfda8d8619c043ce8a729b233fad3ffff44259e5aa5bd72c8a17d519df321","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3eabfda8d8619c043ce8a729b233fad3ffff44259e5aa5bd72c8a17d519df321","first_computed_at":"2026-07-05T00:18:57.920367Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T00:18:57.920367Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"jCmuF/kLlFaSzxsUZDw7d4aYTTm4Nh81CF9fbOT6sjRxdSPYfSNSZm68ocd9sZCxgOOhoSyPHWw+FIrfmGg1Ag==","signature_status":"signed_v1","signed_at":"2026-07-05T00:18:57.920797Z","signed_message":"canonical_sha256_bytes"},"source_id":"1911.05507","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:924425353251e444a76e606858548e8603c0ab966c9287d11ee5bfb24d5f4506","sha256:a96e1eff9009d648c3ed695c8edf21281ea703ac0a704e3c4f35dd88b7113324"],"state_sha256":"ba104ec60882da499ca20af5fae080829b2aaf0b5d2248a3f39b1a3002bb04f4"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"sqmwt2dB6Qbse8uK0Ps1ySMD4ab3+eYnXlagE6T9q7r67JBSZHVSIPqCoZKQTCYMfJFOF0OgP3h5M2R84tUpCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T04:05:41.040065Z","bundle_sha256":"3ce6ec3e07e4ddd284251318c5f83eee900f2438c6c665870b58120cbcbd74d3"}}