{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:3GM4RFGAGFGAC5TWUXX5BHCWO4","short_pith_number":"pith:3GM4RFGA","canonical_record":{"source":{"id":"2604.27147","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-04-29T19:56:53Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"3e507dbcf04cafc1041733300149f26b66da165b267a75b89b379adfd41b4525","abstract_canon_sha256":"e5d90b16e5401c58e65b5807a6d45f1d2793580e11b4b4859aab0e6d9bceeefb"},"schema_version":"1.0"},"canonical_sha256":"d999c894c0314c017676a5efd09c56770643adcf68f4025a59e700f8b8a70935","source":{"kind":"arxiv","id":"2604.27147","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.27147","created_at":"2026-05-21T01:04:26Z"},{"alias_kind":"arxiv_version","alias_value":"2604.27147v2","created_at":"2026-05-21T01:04:26Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.27147","created_at":"2026-05-21T01:04:26Z"},{"alias_kind":"pith_short_12","alias_value":"3GM4RFGAGFGA","created_at":"2026-05-21T01:04:26Z"},{"alias_kind":"pith_short_16","alias_value":"3GM4RFGAGFGAC5TW","created_at":"2026-05-21T01:04:26Z"},{"alias_kind":"pith_short_8","alias_value":"3GM4RFGA","created_at":"2026-05-21T01:04:26Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:3GM4RFGAGFGAC5TWUXX5BHCWO4","target":"record","payload":{"canonical_record":{"source":{"id":"2604.27147","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-04-29T19:56:53Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"3e507dbcf04cafc1041733300149f26b66da165b267a75b89b379adfd41b4525","abstract_canon_sha256":"e5d90b16e5401c58e65b5807a6d45f1d2793580e11b4b4859aab0e6d9bceeefb"},"schema_version":"1.0"},"canonical_sha256":"d999c894c0314c017676a5efd09c56770643adcf68f4025a59e700f8b8a70935","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T01:04:26.516501Z","signature_b64":"6Y2zctFMklYfoGZc3D8BY36TQOva8aZxMzSkTdAZuxDS7Zf5B2ty4f0kQKENwiYrq6HIkLe62WID/dooYjYDDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d999c894c0314c017676a5efd09c56770643adcf68f4025a59e700f8b8a70935","last_reissued_at":"2026-05-21T01:04:26.515890Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T01:04:26.515890Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2604.27147","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":"2026-05-21T01:04:26Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5iKTcQ1trWgYx9HkTB1G7CXtRMt4foQPhgNjWHvIUx1muvRIKRMGMRTHJkcJ+LGvaYvExIU2CSso6s3btrhMBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-29T20:42:39.741584Z"},"content_sha256":"a5c83d0a6f33db8fddc550ba2d93edf051b1644419768a81c9aa789523048ec4","schema_version":"1.0","event_id":"sha256:a5c83d0a6f33db8fddc550ba2d93edf051b1644419768a81c9aa789523048ec4"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:3GM4RFGAGFGAC5TWUXX5BHCWO4","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"How to Guide Your Flow: Few-Step Alignment via Flow Map Reward Guidance","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Reformulating guidance as deterministic optimal control lets the flow map steer trajectories to rewards in a single pass.","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Jerry Y. Huang, Justin Lin, Kartik Nair, Nicholas M. Boffi, Sheel Shah","submitted_at":"2026-04-29T19:56:53Z","abstract_excerpt":"In generative modeling, we often wish to produce samples that maximize a user-specified reward such as aesthetic quality or alignment with human preferences, a problem known as \\textit{guidance}. Despite their widespread use, existing guidance methods either require expensive multi-particle, many-step schemes or rely on poorly understood approximations. We reformulate guidance as a \\textit{deterministic optimal control problem}, yielding a hierarchy of algorithms that subsumes existing approaches at the coarsest level. We show that the \\textit{flow map}, an object of significant recent interes"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"FMRG matches or surpasses baselines across inverse problems, style transfer, human preferences, and VLM rewards with as few as 3 NFEs, giving at least an order-of-magnitude speedup in comparison to prior state of the art.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The deterministic optimal control reformulation of the guidance problem remains accurate and useful when discretized to very few steps and when the flow map is applied directly without additional approximation error or training.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"FMRG is a training-free, single-trajectory guidance method for flow models derived from optimal control that achieves strong reward alignment with only 3 NFEs.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Reformulating guidance as deterministic optimal control lets the flow map steer trajectories to rewards in a single pass.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"e7c6ec9a6189f4c33b23f7a4e2905ca7ff06216bc4ceb74471fdc274b4d47d99"},"source":{"id":"2604.27147","kind":"arxiv","version":2},"verdict":{"id":"ab95eaf6-6ef7-45f4-9403-12eaa69abd0b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-07T09:19:02.890713Z","strongest_claim":"FMRG matches or surpasses baselines across inverse problems, style transfer, human preferences, and VLM rewards with as few as 3 NFEs, giving at least an order-of-magnitude speedup in comparison to prior state of the art.","one_line_summary":"FMRG is a training-free, single-trajectory guidance method for flow models derived from optimal control that achieves strong reward alignment with only 3 NFEs.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The deterministic optimal control reformulation of the guidance problem remains accurate and useful when discretized to very few steps and when the flow map is applied directly without additional approximation error or training.","pith_extraction_headline":"Reformulating guidance as deterministic optimal control lets the flow map steer trajectories to rewards in a single pass."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.27147/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-20T22:43:29.101797Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T19:32:54.631214Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"870f7aa9afe299441c6d7b716908e6aa8ed9eac568009969fedb00d0c7d962f6"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"ab95eaf6-6ef7-45f4-9403-12eaa69abd0b"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-21T01:04:26Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"cfb2Hx5/U8op+CW0JjZBBqJ6Uu8Lye2/Tys0i3N1+S+1nDuybFUphegukqGcgoC7jFH9ojPDWAtrzcv+ccMwBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-29T20:42:39.742465Z"},"content_sha256":"8e2a0834afb46f12b13fd15e18b321fac6ef53a7512c9108259a479c4959753b","schema_version":"1.0","event_id":"sha256:8e2a0834afb46f12b13fd15e18b321fac6ef53a7512c9108259a479c4959753b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/3GM4RFGAGFGAC5TWUXX5BHCWO4/bundle.json","state_url":"https://pith.science/pith/3GM4RFGAGFGAC5TWUXX5BHCWO4/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/3GM4RFGAGFGAC5TWUXX5BHCWO4/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":"2026-05-29T20:42:39Z","links":{"resolver":"https://pith.science/pith/3GM4RFGAGFGAC5TWUXX5BHCWO4","bundle":"https://pith.science/pith/3GM4RFGAGFGAC5TWUXX5BHCWO4/bundle.json","state":"https://pith.science/pith/3GM4RFGAGFGAC5TWUXX5BHCWO4/state.json","well_known_bundle":"https://pith.science/.well-known/pith/3GM4RFGAGFGAC5TWUXX5BHCWO4/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:3GM4RFGAGFGAC5TWUXX5BHCWO4","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":"e5d90b16e5401c58e65b5807a6d45f1d2793580e11b4b4859aab0e6d9bceeefb","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-04-29T19:56:53Z","title_canon_sha256":"3e507dbcf04cafc1041733300149f26b66da165b267a75b89b379adfd41b4525"},"schema_version":"1.0","source":{"id":"2604.27147","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.27147","created_at":"2026-05-21T01:04:26Z"},{"alias_kind":"arxiv_version","alias_value":"2604.27147v2","created_at":"2026-05-21T01:04:26Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.27147","created_at":"2026-05-21T01:04:26Z"},{"alias_kind":"pith_short_12","alias_value":"3GM4RFGAGFGA","created_at":"2026-05-21T01:04:26Z"},{"alias_kind":"pith_short_16","alias_value":"3GM4RFGAGFGAC5TW","created_at":"2026-05-21T01:04:26Z"},{"alias_kind":"pith_short_8","alias_value":"3GM4RFGA","created_at":"2026-05-21T01:04:26Z"}],"graph_snapshots":[{"event_id":"sha256:8e2a0834afb46f12b13fd15e18b321fac6ef53a7512c9108259a479c4959753b","target":"graph","created_at":"2026-05-21T01:04:26Z","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":"FMRG matches or surpasses baselines across inverse problems, style transfer, human preferences, and VLM rewards with as few as 3 NFEs, giving at least an order-of-magnitude speedup in comparison to prior state of the art."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The deterministic optimal control reformulation of the guidance problem remains accurate and useful when discretized to very few steps and when the flow map is applied directly without additional approximation error or training."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"FMRG is a training-free, single-trajectory guidance method for flow models derived from optimal control that achieves strong reward alignment with only 3 NFEs."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Reformulating guidance as deterministic optimal control lets the flow map steer trajectories to rewards in a single pass."}],"snapshot_sha256":"e7c6ec9a6189f4c33b23f7a4e2905ca7ff06216bc4ceb74471fdc274b4d47d99"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"ai_meta_artifact","ran_at":"2026-05-20T22:43:29.101797Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-19T19:32:54.631214Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2604.27147/integrity.json","findings":[],"snapshot_sha256":"870f7aa9afe299441c6d7b716908e6aa8ed9eac568009969fedb00d0c7d962f6","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"In generative modeling, we often wish to produce samples that maximize a user-specified reward such as aesthetic quality or alignment with human preferences, a problem known as \\textit{guidance}. Despite their widespread use, existing guidance methods either require expensive multi-particle, many-step schemes or rely on poorly understood approximations. We reformulate guidance as a \\textit{deterministic optimal control problem}, yielding a hierarchy of algorithms that subsumes existing approaches at the coarsest level. We show that the \\textit{flow map}, an object of significant recent interes","authors_text":"Jerry Y. Huang, Justin Lin, Kartik Nair, Nicholas M. Boffi, Sheel Shah","cross_cats":["cs.AI"],"headline":"Reformulating guidance as deterministic optimal control lets the flow map steer trajectories to rewards in a single pass.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-04-29T19:56:53Z","title":"How to Guide Your Flow: Few-Step Alignment via Flow Map Reward Guidance"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2604.27147","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-07T09:19:02.890713Z","id":"ab95eaf6-6ef7-45f4-9403-12eaa69abd0b","model_set":{"reader":"grok-4.3"},"one_line_summary":"FMRG is a training-free, single-trajectory guidance method for flow models derived from optimal control that achieves strong reward alignment with only 3 NFEs.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Reformulating guidance as deterministic optimal control lets the flow map steer trajectories to rewards in a single pass.","strongest_claim":"FMRG matches or surpasses baselines across inverse problems, style transfer, human preferences, and VLM rewards with as few as 3 NFEs, giving at least an order-of-magnitude speedup in comparison to prior state of the art.","weakest_assumption":"The deterministic optimal control reformulation of the guidance problem remains accurate and useful when discretized to very few steps and when the flow map is applied directly without additional approximation error or training."}},"verdict_id":"ab95eaf6-6ef7-45f4-9403-12eaa69abd0b"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:a5c83d0a6f33db8fddc550ba2d93edf051b1644419768a81c9aa789523048ec4","target":"record","created_at":"2026-05-21T01:04:26Z","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":"e5d90b16e5401c58e65b5807a6d45f1d2793580e11b4b4859aab0e6d9bceeefb","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-04-29T19:56:53Z","title_canon_sha256":"3e507dbcf04cafc1041733300149f26b66da165b267a75b89b379adfd41b4525"},"schema_version":"1.0","source":{"id":"2604.27147","kind":"arxiv","version":2}},"canonical_sha256":"d999c894c0314c017676a5efd09c56770643adcf68f4025a59e700f8b8a70935","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d999c894c0314c017676a5efd09c56770643adcf68f4025a59e700f8b8a70935","first_computed_at":"2026-05-21T01:04:26.515890Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-21T01:04:26.515890Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"6Y2zctFMklYfoGZc3D8BY36TQOva8aZxMzSkTdAZuxDS7Zf5B2ty4f0kQKENwiYrq6HIkLe62WID/dooYjYDDQ==","signature_status":"signed_v1","signed_at":"2026-05-21T01:04:26.516501Z","signed_message":"canonical_sha256_bytes"},"source_id":"2604.27147","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:a5c83d0a6f33db8fddc550ba2d93edf051b1644419768a81c9aa789523048ec4","sha256:8e2a0834afb46f12b13fd15e18b321fac6ef53a7512c9108259a479c4959753b"],"state_sha256":"e0267f264517979fff8b9895ce4d64d3dcfd6b8600434e8c18d261562d7b38cb"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lQMIeVjvu24DlvEt8M+z8qE/34CSHCrI0OI8gKGPuRsieI2wPr9Wi1VGby3ViGUerupiXNCs+BBqCefzHi+3Ag==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-29T20:42:39.746720Z","bundle_sha256":"d16f76cbc517af3c1bed32fa17f6c9d88eb58e8bd19dedcdbcedda0eb3a68c52"}}