{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:NHFUK5FQZ6ROWOJKM3CAN5SYFW","short_pith_number":"pith:NHFUK5FQ","canonical_record":{"source":{"id":"2605.09994","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2026-05-11T05:10:16Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"95e337d1c75af465b6532b3b6aeda0777402c2554a5fe9e9aa6d6c4f1ba36531","abstract_canon_sha256":"6d2f71694bc4c4a94be716719986f5e3e1078dae4a4a464e5773ee95df9c9284"},"schema_version":"1.0"},"canonical_sha256":"69cb4574b0cfa2eb392a66c406f6582dae2021bb86f2c6ea7f42c89450822884","source":{"kind":"arxiv","id":"2605.09994","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.09994","created_at":"2026-05-20T00:00:42Z"},{"alias_kind":"arxiv_version","alias_value":"2605.09994v2","created_at":"2026-05-20T00:00:42Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.09994","created_at":"2026-05-20T00:00:42Z"},{"alias_kind":"pith_short_12","alias_value":"NHFUK5FQZ6RO","created_at":"2026-05-20T00:00:42Z"},{"alias_kind":"pith_short_16","alias_value":"NHFUK5FQZ6ROWOJK","created_at":"2026-05-20T00:00:42Z"},{"alias_kind":"pith_short_8","alias_value":"NHFUK5FQ","created_at":"2026-05-20T00:00:42Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:NHFUK5FQZ6ROWOJKM3CAN5SYFW","target":"record","payload":{"canonical_record":{"source":{"id":"2605.09994","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2026-05-11T05:10:16Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"95e337d1c75af465b6532b3b6aeda0777402c2554a5fe9e9aa6d6c4f1ba36531","abstract_canon_sha256":"6d2f71694bc4c4a94be716719986f5e3e1078dae4a4a464e5773ee95df9c9284"},"schema_version":"1.0"},"canonical_sha256":"69cb4574b0cfa2eb392a66c406f6582dae2021bb86f2c6ea7f42c89450822884","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:00:42.205832Z","signature_b64":"kVbuSt0kSDj2MQ8WrshPOkmCroI7u6YFZ5u0PeLiZduI60S1ROWZmC/2N5M8Nxowug5s2fTdTA+KBierdey+Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"69cb4574b0cfa2eb392a66c406f6582dae2021bb86f2c6ea7f42c89450822884","last_reissued_at":"2026-05-20T00:00:42.205104Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:00:42.205104Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.09994","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":"2026-05-20T00:00:42Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ZbjQiZsWVH80BF7Mmcr7eAHHWsC71KuFhje6J4dpJLE5XcAwHM76r3iOVDW2evjmCgFAx6zgu3IrMIcd+B8nCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T23:47:50.428299Z"},"content_sha256":"84a0b2610acfb12473ed9e70700caaf661b0a9ceda7a5e087d1165848c264f1a","schema_version":"1.0","event_id":"sha256:84a0b2610acfb12473ed9e70700caaf661b0a9ceda7a5e087d1165848c264f1a"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:NHFUK5FQZ6ROWOJKM3CAN5SYFW","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"BatchWeave: A Consistent Object-Store-Native Data Plane for Large Foundation Model Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"BatchWeave builds a consistent object-store-native data plane that delivers atomic all-rank batch visibility and exactly-once recovery for distributed foundation model training.","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Bingyi Jing, Jiaxing Zhang, Jingyi Xi, Junjie Zhang, Songxin Zhang, Ting Sun, Xiao Yan, Zejian Xie, Zhuoyang Song, Zunyao Mao","submitted_at":"2026-05-11T05:10:16Z","abstract_excerpt":"Modern Large Foundation Model (LFM) training has transformed the data pipeline from a static ingestion layer into a dynamic component that must co-evolve with the training process. Existing systems are ill-equipped: colocated dataloaders offer no failure isolation, while message queue-based disaggregated dataloaders operate on a record/offset abstraction that cannot express the batch-level semantics required by distributed training. We present BatchWeave, an object-store-native training data plane for distributed LFM training. BatchWeave uses versioned manifests and conditional object writes t"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Evaluations on large-scale multimodal pre-training and SFT workloads using 64 GPUs show that BatchWeave outperforms colocated dataloader throughput while providing full failure isolation, outperforms Apache Kafka in ingestion throughput, and achieves lower consumer read latency than Kafka.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"Object stores can deliver the versioned-manifest ACID semantics and conditional-write performance needed for atomic all-rank batch visibility and checkpoint-aligned lifecycle management without introducing latency or throughput penalties that would erase the reported gains over colocated and Kafka baselines.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"BatchWeave delivers an object-store-native data plane for distributed large foundation model training via transactional global batches and a decentralized adaptive commit algorithm.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"BatchWeave builds a consistent object-store-native data plane that delivers atomic all-rank batch visibility and exactly-once recovery for distributed foundation model training.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"ac97c3594df3ca218e5ce806eddba61d46d048aacaac407c88b9dbafabe103be"},"source":{"id":"2605.09994","kind":"arxiv","version":2},"verdict":{"id":"e9970b4d-eb96-49b8-89a4-d848c2ac7d69","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T17:51:06.961186Z","strongest_claim":"Evaluations on large-scale multimodal pre-training and SFT workloads using 64 GPUs show that BatchWeave outperforms colocated dataloader throughput while providing full failure isolation, outperforms Apache Kafka in ingestion throughput, and achieves lower consumer read latency than Kafka.","one_line_summary":"BatchWeave delivers an object-store-native data plane for distributed large foundation model training via transactional global batches and a decentralized adaptive commit algorithm.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"Object stores can deliver the versioned-manifest ACID semantics and conditional-write performance needed for atomic all-rank batch visibility and checkpoint-aligned lifecycle management without introducing latency or throughput penalties that would erase the reported gains over colocated and Kafka baselines.","pith_extraction_headline":"BatchWeave builds a consistent object-store-native data plane that delivers atomic all-rank batch visibility and exactly-once recovery for distributed foundation model training."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.09994/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-19T15:42:56.345285Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T12:01:18.261085Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T09:44:55.726346Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"a35c4b9e0ca5b430be91d9f83b353e89c4f22de22b1a75e14ad374db1bb1cad7"},"references":{"count":42,"sample":[{"doi":"","year":2020,"title":"Alex Aizman, Gavin Maltby, and Thomas Breuel. 2020. High Performance I/O For Large Scale Deep Learning. https://arxiv.org/abs/2001.01858. doi:10.48550/ arXiv.2001.01858 arXiv:2001.01858","work_id":"0e1f92fc-97f3-47f6-9fc5-3e5750b0d18e","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2015,"title":"Tyler Akidau, Robert Bradshaw, Craig Chambers, Slava Chernyak, Rafael J Dagum, Sam Knight, Frances Perry, Reiner Schmidt, and Sam Whittle. 2015. The dataflow model: a practical approach to balancing c","work_id":"8fd6f7fe-750d-4216-b987-e3bc78c14439","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Michael Armbrust, Tathagata Das, Liwen Sun, Burak Yavuz, Shixiong Zhu, Mukul Murthy, Joseph Torres, Herman van Hovell, Adrian Ionescu, Bogdan Ghit, Mad- hukara Bhat, Reynold Xin, Ali Ghodsi, Ion Stoic","work_id":"855dacad-91e9-44e2-99f9-368de4d0f63c","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"Michael Armbrust, Tathagata Das, Joseph Torres, Burak Yavuz, Shixiong Liao, Yin Huai, Hossein Hosseini, Matei Zaharia, and Reynold Xin. 2018. Structured streaming: A declarative api for real-time appl","work_id":"95c29681-53b8-45e5-810c-284bbf8cc33a","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Andrew Audibert, Yang Chen, Dan Graur, Ana Klimovic, Jiri Simsa, and Chan- dramohan A. Thekkath. 2023. tf.data service: A Case for Disaggregating ML Input Data Processing. InProceedings of the 2023 AC","work_id":"99ca57d8-dad1-4ccb-b319-5599d2883688","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":42,"snapshot_sha256":"6c8257e4d3f24c4949a5d1360c199ba12a3aeeee1385ac3ec0c7749f2f08c170","internal_anchors":5},"formal_canon":{"evidence_count":2,"snapshot_sha256":"f250df4380c4554db2087519e206eef6e961167b27546c500c9e354e412d7f29"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"e9970b4d-eb96-49b8-89a4-d848c2ac7d69"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:00:42Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ssOr2ub0cZZ6q1PgdbNwTSm/AbwpVHWZdK22a+c/+7Qf3F7n4zICdY7wmjeJs4RMuXAN0nj1zeqoKuhDVHy/CQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T23:47:50.429432Z"},"content_sha256":"f8a1b87cd50f4a30cf39a9fba6785f947e09ae2b161a51a25b0e0f9dcb58ece8","schema_version":"1.0","event_id":"sha256:f8a1b87cd50f4a30cf39a9fba6785f947e09ae2b161a51a25b0e0f9dcb58ece8"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/NHFUK5FQZ6ROWOJKM3CAN5SYFW/bundle.json","state_url":"https://pith.science/pith/NHFUK5FQZ6ROWOJKM3CAN5SYFW/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/NHFUK5FQZ6ROWOJKM3CAN5SYFW/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":"2026-05-25T23:47:50Z","links":{"resolver":"https://pith.science/pith/NHFUK5FQZ6ROWOJKM3CAN5SYFW","bundle":"https://pith.science/pith/NHFUK5FQZ6ROWOJKM3CAN5SYFW/bundle.json","state":"https://pith.science/pith/NHFUK5FQZ6ROWOJKM3CAN5SYFW/state.json","well_known_bundle":"https://pith.science/.well-known/pith/NHFUK5FQZ6ROWOJKM3CAN5SYFW/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:NHFUK5FQZ6ROWOJKM3CAN5SYFW","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":"6d2f71694bc4c4a94be716719986f5e3e1078dae4a4a464e5773ee95df9c9284","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2026-05-11T05:10:16Z","title_canon_sha256":"95e337d1c75af465b6532b3b6aeda0777402c2554a5fe9e9aa6d6c4f1ba36531"},"schema_version":"1.0","source":{"id":"2605.09994","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.09994","created_at":"2026-05-20T00:00:42Z"},{"alias_kind":"arxiv_version","alias_value":"2605.09994v2","created_at":"2026-05-20T00:00:42Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.09994","created_at":"2026-05-20T00:00:42Z"},{"alias_kind":"pith_short_12","alias_value":"NHFUK5FQZ6RO","created_at":"2026-05-20T00:00:42Z"},{"alias_kind":"pith_short_16","alias_value":"NHFUK5FQZ6ROWOJK","created_at":"2026-05-20T00:00:42Z"},{"alias_kind":"pith_short_8","alias_value":"NHFUK5FQ","created_at":"2026-05-20T00:00:42Z"}],"graph_snapshots":[{"event_id":"sha256:f8a1b87cd50f4a30cf39a9fba6785f947e09ae2b161a51a25b0e0f9dcb58ece8","target":"graph","created_at":"2026-05-20T00:00:42Z","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":"Evaluations on large-scale multimodal pre-training and SFT workloads using 64 GPUs show that BatchWeave outperforms colocated dataloader throughput while providing full failure isolation, outperforms Apache Kafka in ingestion throughput, and achieves lower consumer read latency than Kafka."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"Object stores can deliver the versioned-manifest ACID semantics and conditional-write performance needed for atomic all-rank batch visibility and checkpoint-aligned lifecycle management without introducing latency or throughput penalties that would erase the reported gains over colocated and Kafka baselines."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"BatchWeave delivers an object-store-native data plane for distributed large foundation model training via transactional global batches and a decentralized adaptive commit algorithm."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"BatchWeave builds a consistent object-store-native data plane that delivers atomic all-rank batch visibility and exactly-once recovery for distributed foundation model training."}],"snapshot_sha256":"ac97c3594df3ca218e5ce806eddba61d46d048aacaac407c88b9dbafabe103be"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"f250df4380c4554db2087519e206eef6e961167b27546c500c9e354e412d7f29"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"ai_meta_artifact","ran_at":"2026-05-19T15:42:56.345285Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_title_agreement","ran_at":"2026-05-19T12:01:18.261085Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-19T09:44:55.726346Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2605.09994/integrity.json","findings":[],"snapshot_sha256":"a35c4b9e0ca5b430be91d9f83b353e89c4f22de22b1a75e14ad374db1bb1cad7","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Modern Large Foundation Model (LFM) training has transformed the data pipeline from a static ingestion layer into a dynamic component that must co-evolve with the training process. Existing systems are ill-equipped: colocated dataloaders offer no failure isolation, while message queue-based disaggregated dataloaders operate on a record/offset abstraction that cannot express the batch-level semantics required by distributed training. We present BatchWeave, an object-store-native training data plane for distributed LFM training. BatchWeave uses versioned manifests and conditional object writes t","authors_text":"Bingyi Jing, Jiaxing Zhang, Jingyi Xi, Junjie Zhang, Songxin Zhang, Ting Sun, Xiao Yan, Zejian Xie, Zhuoyang Song, Zunyao Mao","cross_cats":["cs.LG"],"headline":"BatchWeave builds a consistent object-store-native data plane that delivers atomic all-rank batch visibility and exactly-once recovery for distributed foundation model training.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2026-05-11T05:10:16Z","title":"BatchWeave: A Consistent Object-Store-Native Data Plane for Large Foundation Model Training"},"references":{"count":42,"internal_anchors":5,"resolved_work":42,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Alex Aizman, Gavin Maltby, and Thomas Breuel. 2020. High Performance I/O For Large Scale Deep Learning. https://arxiv.org/abs/2001.01858. doi:10.48550/ arXiv.2001.01858 arXiv:2001.01858","work_id":"0e1f92fc-97f3-47f6-9fc5-3e5750b0d18e","year":2020},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Tyler Akidau, Robert Bradshaw, Craig Chambers, Slava Chernyak, Rafael J Dagum, Sam Knight, Frances Perry, Reiner Schmidt, and Sam Whittle. 2015. The dataflow model: a practical approach to balancing c","work_id":"8fd6f7fe-750d-4216-b987-e3bc78c14439","year":2015},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Michael Armbrust, Tathagata Das, Liwen Sun, Burak Yavuz, Shixiong Zhu, Mukul Murthy, Joseph Torres, Herman van Hovell, Adrian Ionescu, Bogdan Ghit, Mad- hukara Bhat, Reynold Xin, Ali Ghodsi, Ion Stoic","work_id":"855dacad-91e9-44e2-99f9-368de4d0f63c","year":2020},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Michael Armbrust, Tathagata Das, Joseph Torres, Burak Yavuz, Shixiong Liao, Yin Huai, Hossein Hosseini, Matei Zaharia, and Reynold Xin. 2018. Structured streaming: A declarative api for real-time appl","work_id":"95c29681-53b8-45e5-810c-284bbf8cc33a","year":2018},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Andrew Audibert, Yang Chen, Dan Graur, Ana Klimovic, Jiri Simsa, and Chan- dramohan A. Thekkath. 2023. tf.data service: A Case for Disaggregating ML Input Data Processing. InProceedings of the 2023 AC","work_id":"99ca57d8-dad1-4ccb-b319-5599d2883688","year":2023}],"snapshot_sha256":"6c8257e4d3f24c4949a5d1360c199ba12a3aeeee1385ac3ec0c7749f2f08c170"},"source":{"id":"2605.09994","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-19T17:51:06.961186Z","id":"e9970b4d-eb96-49b8-89a4-d848c2ac7d69","model_set":{"reader":"grok-4.3"},"one_line_summary":"BatchWeave delivers an object-store-native data plane for distributed large foundation model training via transactional global batches and a decentralized adaptive commit algorithm.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"BatchWeave builds a consistent object-store-native data plane that delivers atomic all-rank batch visibility and exactly-once recovery for distributed foundation model training.","strongest_claim":"Evaluations on large-scale multimodal pre-training and SFT workloads using 64 GPUs show that BatchWeave outperforms colocated dataloader throughput while providing full failure isolation, outperforms Apache Kafka in ingestion throughput, and achieves lower consumer read latency than Kafka.","weakest_assumption":"Object stores can deliver the versioned-manifest ACID semantics and conditional-write performance needed for atomic all-rank batch visibility and checkpoint-aligned lifecycle management without introducing latency or throughput penalties that would erase the reported gains over colocated and Kafka baselines."}},"verdict_id":"e9970b4d-eb96-49b8-89a4-d848c2ac7d69"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:84a0b2610acfb12473ed9e70700caaf661b0a9ceda7a5e087d1165848c264f1a","target":"record","created_at":"2026-05-20T00:00:42Z","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":"6d2f71694bc4c4a94be716719986f5e3e1078dae4a4a464e5773ee95df9c9284","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2026-05-11T05:10:16Z","title_canon_sha256":"95e337d1c75af465b6532b3b6aeda0777402c2554a5fe9e9aa6d6c4f1ba36531"},"schema_version":"1.0","source":{"id":"2605.09994","kind":"arxiv","version":2}},"canonical_sha256":"69cb4574b0cfa2eb392a66c406f6582dae2021bb86f2c6ea7f42c89450822884","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"69cb4574b0cfa2eb392a66c406f6582dae2021bb86f2c6ea7f42c89450822884","first_computed_at":"2026-05-20T00:00:42.205104Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:00:42.205104Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"kVbuSt0kSDj2MQ8WrshPOkmCroI7u6YFZ5u0PeLiZduI60S1ROWZmC/2N5M8Nxowug5s2fTdTA+KBierdey+Dg==","signature_status":"signed_v1","signed_at":"2026-05-20T00:00:42.205832Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.09994","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:84a0b2610acfb12473ed9e70700caaf661b0a9ceda7a5e087d1165848c264f1a","sha256:f8a1b87cd50f4a30cf39a9fba6785f947e09ae2b161a51a25b0e0f9dcb58ece8"],"state_sha256":"af5ca871496d011bfde55cdfff3d5d45b6bfb46237f195687217a961bfcfa031"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"DUWK9crL5WwU86pmq/8U47+LIZ5VMdAn1VSuorIWlftPFQ8bffepJrxKMD78q/0GaHCxxXlc4C5JvvLkbh3OCA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-25T23:47:50.434409Z","bundle_sha256":"3a332c0d03deb8c46406271309fe27bc6f7a9c924400d2749c1f275f0490e803"}}