{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:GZF7TYGSBDD4JMW2XECO5K4TKD","short_pith_number":"pith:GZF7TYGS","canonical_record":{"source":{"id":"2602.05319","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-05T05:37:14Z","cross_cats_sorted":[],"title_canon_sha256":"a8001fc97401e7bca5073ceec612bf3a45ca58182fa2315015b70928d854bb05","abstract_canon_sha256":"8dbe0bc461f3f81409dccebedb35379f9bcf12f25494dcef10de331fb7c28606"},"schema_version":"1.0"},"canonical_sha256":"364bf9e0d208c7c4b2dab904eeab9350ef3251e760a701d13fd335c8761efe5b","source":{"kind":"arxiv","id":"2602.05319","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2602.05319","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"arxiv_version","alias_value":"2602.05319v3","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.05319","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"pith_short_12","alias_value":"GZF7TYGSBDD4","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"GZF7TYGSBDD4JMW2","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"GZF7TYGS","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:GZF7TYGSBDD4JMW2XECO5K4TKD","target":"record","payload":{"canonical_record":{"source":{"id":"2602.05319","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-05T05:37:14Z","cross_cats_sorted":[],"title_canon_sha256":"a8001fc97401e7bca5073ceec612bf3a45ca58182fa2315015b70928d854bb05","abstract_canon_sha256":"8dbe0bc461f3f81409dccebedb35379f9bcf12f25494dcef10de331fb7c28606"},"schema_version":"1.0"},"canonical_sha256":"364bf9e0d208c7c4b2dab904eeab9350ef3251e760a701d13fd335c8761efe5b","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:16.305677Z","signature_b64":"am9ACMJii7Q5csnwzVSa96RXHRfC4SVFlIuPRU2A4FRUw0w6BhdaPmRZXruQSdznjHcK/ZQ6biLt9PWMfCcPDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"364bf9e0d208c7c4b2dab904eeab9350ef3251e760a701d13fd335c8761efe5b","last_reissued_at":"2026-05-17T23:39:16.305129Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:16.305129Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2602.05319","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":"2026-05-17T23:39:16Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"tBIfzwlnrWTNhf2fvXVHkEuihKJecu0niZ77z94r9KtmNAspoi08yDIRi6NV94UILsIYtdHEg2z5NAZk0LTICw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-23T00:04:38.419045Z"},"content_sha256":"411b32b12af299545e8276e145b18b1a75f424e13e6ed8bd2f1cc2af1fda9da3","schema_version":"1.0","event_id":"sha256:411b32b12af299545e8276e145b18b1a75f424e13e6ed8bd2f1cc2af1fda9da3"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:GZF7TYGSBDD4JMW2XECO5K4TKD","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Accelerated Sequential Flow Matching: A Bayesian Filtering Perspective","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Sequential Bayesian Flow Matching reuses the previous posterior as a source distribution to accelerate sampling from streaming observations.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bo Dai, Hans Hao-Hsun Hsu, Junran Wang, Pan Li, Yinan Huang","submitted_at":"2026-02-05T05:37:14Z","abstract_excerpt":"Sequential probabilistic inference from streaming observations requires modeling distributions over future trajectories as new observations arrive. Although diffusion and flow-matching models are effective at capturing high-dimensional, multimodal distributions, their deployment in real-time streaming settings typically relies on repeatedly sampling from a non-informative initial distribution. This results in substantial inference latency, particularly when multiple samples are needed to characterize the predictive distribution. In this work, we introduce Sequential Bayesian Flow Matching, a f"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"By using the previous belief as an informative source distribution, it enables substantially faster sampling than naive resampling from scratch while achieving performance competitive with full-step diffusion on distributional metrics.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That a learned flow can reliably transport the full posterior distribution (including multimodality) from one time step to the next without accumulating approximation error or requiring retraining when the observation model changes.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Sequential Bayesian Flow Matching accelerates flow-based sampling for streaming probabilistic inference by transporting posteriors recursively like Bayesian filters rather than restarting from noise each step.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Sequential Bayesian Flow Matching reuses the previous posterior as a source distribution to accelerate sampling from streaming observations.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"53da81f087669c8aae81a6b6547a851b1f4c67752db5974c61163d6af138f386"},"source":{"id":"2602.05319","kind":"arxiv","version":3},"verdict":{"id":"4a48b38d-2752-4db3-b5e7-fc56265d0e48","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T07:28:04.950920Z","strongest_claim":"By using the previous belief as an informative source distribution, it enables substantially faster sampling than naive resampling from scratch while achieving performance competitive with full-step diffusion on distributional metrics.","one_line_summary":"Sequential Bayesian Flow Matching accelerates flow-based sampling for streaming probabilistic inference by transporting posteriors recursively like Bayesian filters rather than restarting from noise each step.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That a learned flow can reliably transport the full posterior distribution (including multimodality) from one time step to the next without accumulating approximation error or requiring retraining when the observation model changes.","pith_extraction_headline":"Sequential Bayesian Flow Matching reuses the previous posterior as a source distribution to accelerate sampling from streaming observations."},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"27e09ef0999223a651f718af491f5ee6070a22877db8e628c758b516da5c7352"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"4a48b38d-2752-4db3-b5e7-fc56265d0e48"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:16Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"33HhUSPge0J6GuHQDOtDDrKgeehSYWKdovEUipxeI+7g36vb9kJxauL6LKxSpyd6j+Y/KfQt2PKWEGzYJor5AQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-23T00:04:38.419644Z"},"content_sha256":"67795b4ede3c14124f042358410fe2f1fe299121ec044ed26863611ab9ee5dc4","schema_version":"1.0","event_id":"sha256:67795b4ede3c14124f042358410fe2f1fe299121ec044ed26863611ab9ee5dc4"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/GZF7TYGSBDD4JMW2XECO5K4TKD/bundle.json","state_url":"https://pith.science/pith/GZF7TYGSBDD4JMW2XECO5K4TKD/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/GZF7TYGSBDD4JMW2XECO5K4TKD/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":"2026-05-23T00:04:38Z","links":{"resolver":"https://pith.science/pith/GZF7TYGSBDD4JMW2XECO5K4TKD","bundle":"https://pith.science/pith/GZF7TYGSBDD4JMW2XECO5K4TKD/bundle.json","state":"https://pith.science/pith/GZF7TYGSBDD4JMW2XECO5K4TKD/state.json","well_known_bundle":"https://pith.science/.well-known/pith/GZF7TYGSBDD4JMW2XECO5K4TKD/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:GZF7TYGSBDD4JMW2XECO5K4TKD","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":"8dbe0bc461f3f81409dccebedb35379f9bcf12f25494dcef10de331fb7c28606","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-05T05:37:14Z","title_canon_sha256":"a8001fc97401e7bca5073ceec612bf3a45ca58182fa2315015b70928d854bb05"},"schema_version":"1.0","source":{"id":"2602.05319","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2602.05319","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"arxiv_version","alias_value":"2602.05319v3","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.05319","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"pith_short_12","alias_value":"GZF7TYGSBDD4","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"GZF7TYGSBDD4JMW2","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"GZF7TYGS","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:67795b4ede3c14124f042358410fe2f1fe299121ec044ed26863611ab9ee5dc4","target":"graph","created_at":"2026-05-17T23:39:16Z","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":"By using the previous belief as an informative source distribution, it enables substantially faster sampling than naive resampling from scratch while achieving performance competitive with full-step diffusion on distributional metrics."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That a learned flow can reliably transport the full posterior distribution (including multimodality) from one time step to the next without accumulating approximation error or requiring retraining when the observation model changes."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Sequential Bayesian Flow Matching accelerates flow-based sampling for streaming probabilistic inference by transporting posteriors recursively like Bayesian filters rather than restarting from noise each step."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Sequential Bayesian Flow Matching reuses the previous posterior as a source distribution to accelerate sampling from streaming observations."}],"snapshot_sha256":"53da81f087669c8aae81a6b6547a851b1f4c67752db5974c61163d6af138f386"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"27e09ef0999223a651f718af491f5ee6070a22877db8e628c758b516da5c7352"},"paper":{"abstract_excerpt":"Sequential probabilistic inference from streaming observations requires modeling distributions over future trajectories as new observations arrive. Although diffusion and flow-matching models are effective at capturing high-dimensional, multimodal distributions, their deployment in real-time streaming settings typically relies on repeatedly sampling from a non-informative initial distribution. This results in substantial inference latency, particularly when multiple samples are needed to characterize the predictive distribution. In this work, we introduce Sequential Bayesian Flow Matching, a f","authors_text":"Bo Dai, Hans Hao-Hsun Hsu, Junran Wang, Pan Li, Yinan Huang","cross_cats":[],"headline":"Sequential Bayesian Flow Matching reuses the previous posterior as a source distribution to accelerate sampling from streaming observations.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-05T05:37:14Z","title":"Accelerated Sequential Flow Matching: A Bayesian Filtering Perspective"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.05319","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-16T07:28:04.950920Z","id":"4a48b38d-2752-4db3-b5e7-fc56265d0e48","model_set":{"reader":"grok-4.3"},"one_line_summary":"Sequential Bayesian Flow Matching accelerates flow-based sampling for streaming probabilistic inference by transporting posteriors recursively like Bayesian filters rather than restarting from noise each step.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Sequential Bayesian Flow Matching reuses the previous posterior as a source distribution to accelerate sampling from streaming observations.","strongest_claim":"By using the previous belief as an informative source distribution, it enables substantially faster sampling than naive resampling from scratch while achieving performance competitive with full-step diffusion on distributional metrics.","weakest_assumption":"That a learned flow can reliably transport the full posterior distribution (including multimodality) from one time step to the next without accumulating approximation error or requiring retraining when the observation model changes."}},"verdict_id":"4a48b38d-2752-4db3-b5e7-fc56265d0e48"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:411b32b12af299545e8276e145b18b1a75f424e13e6ed8bd2f1cc2af1fda9da3","target":"record","created_at":"2026-05-17T23:39:16Z","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":"8dbe0bc461f3f81409dccebedb35379f9bcf12f25494dcef10de331fb7c28606","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-05T05:37:14Z","title_canon_sha256":"a8001fc97401e7bca5073ceec612bf3a45ca58182fa2315015b70928d854bb05"},"schema_version":"1.0","source":{"id":"2602.05319","kind":"arxiv","version":3}},"canonical_sha256":"364bf9e0d208c7c4b2dab904eeab9350ef3251e760a701d13fd335c8761efe5b","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"364bf9e0d208c7c4b2dab904eeab9350ef3251e760a701d13fd335c8761efe5b","first_computed_at":"2026-05-17T23:39:16.305129Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:16.305129Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"am9ACMJii7Q5csnwzVSa96RXHRfC4SVFlIuPRU2A4FRUw0w6BhdaPmRZXruQSdznjHcK/ZQ6biLt9PWMfCcPDw==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:16.305677Z","signed_message":"canonical_sha256_bytes"},"source_id":"2602.05319","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:411b32b12af299545e8276e145b18b1a75f424e13e6ed8bd2f1cc2af1fda9da3","sha256:67795b4ede3c14124f042358410fe2f1fe299121ec044ed26863611ab9ee5dc4"],"state_sha256":"7f7862a979e9defdc04ef8f41e570228f64826b04a44bf930983ccf314ff17f0"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"mUl+7HKf/iJ/VANonm7TB1cd5VzPG3hHWI6fwBQrRfj7vDmn1hlT+UCuvNxCv8shHQAhjQFJ4AtOBoNImBeBBQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-23T00:04:38.423609Z","bundle_sha256":"50194afcd60f234209c005c8660ba3a142649946e2932713e348094ea04d1a55"}}