{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:ZEOQ7X6BZXG5BHSA4V76JMKGOL","short_pith_number":"pith:ZEOQ7X6B","canonical_record":{"source":{"id":"2603.12893","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-03-13T10:54:09Z","cross_cats_sorted":["cs.AI","cs.LG","cs.NE","stat.ML"],"title_canon_sha256":"a79b0306a1b3f92285d4716dd0f5599fa6d65f3bdf6f55521bf4681779c5c738","abstract_canon_sha256":"a4d647f94d53b3512cd5bf9396d03e7c42856bf03ae703501c0ab5011f290f6e"},"schema_version":"1.0"},"canonical_sha256":"c91d0fdfc1cdcdd09e40e57fe4b14672ef8049051cee5ac1a11a50c9aeba48a4","source":{"kind":"arxiv","id":"2603.12893","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2603.12893","created_at":"2026-07-01T01:17:48Z"},{"alias_kind":"arxiv_version","alias_value":"2603.12893v2","created_at":"2026-07-01T01:17:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.12893","created_at":"2026-07-01T01:17:48Z"},{"alias_kind":"pith_short_12","alias_value":"ZEOQ7X6BZXG5","created_at":"2026-07-01T01:17:48Z"},{"alias_kind":"pith_short_16","alias_value":"ZEOQ7X6BZXG5BHSA","created_at":"2026-07-01T01:17:48Z"},{"alias_kind":"pith_short_8","alias_value":"ZEOQ7X6B","created_at":"2026-07-01T01:17:48Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:ZEOQ7X6BZXG5BHSA4V76JMKGOL","target":"record","payload":{"canonical_record":{"source":{"id":"2603.12893","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-03-13T10:54:09Z","cross_cats_sorted":["cs.AI","cs.LG","cs.NE","stat.ML"],"title_canon_sha256":"a79b0306a1b3f92285d4716dd0f5599fa6d65f3bdf6f55521bf4681779c5c738","abstract_canon_sha256":"a4d647f94d53b3512cd5bf9396d03e7c42856bf03ae703501c0ab5011f290f6e"},"schema_version":"1.0"},"canonical_sha256":"c91d0fdfc1cdcdd09e40e57fe4b14672ef8049051cee5ac1a11a50c9aeba48a4","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-01T01:17:48.891877Z","signature_b64":"QXQp/2T9zRc0FX/Widvx4M3xyMoiQ4bN4ylCvVaJ+DaSpDlEnvxKU9ZWM/H7+4TxBRSFgEIIrE8aaOTqoqTICA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c91d0fdfc1cdcdd09e40e57fe4b14672ef8049051cee5ac1a11a50c9aeba48a4","last_reissued_at":"2026-07-01T01:17:48.891388Z","signature_status":"signed_v1","first_computed_at":"2026-07-01T01:17:48.891388Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2603.12893","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":"2026-07-01T01:17:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"C2zYqhupAQgiE9Qh62a6UIKUsakLfYyP0j2SWWYFPU4uIbzbdisiGTRrf5KjUawzAnXlB+pk/8XE7VoUgGj9CA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-03T16:48:01.843082Z"},"content_sha256":"e9093bb9d79d54dac123817ac21f8b4f807ecae2806b870cd74613135b28826f","schema_version":"1.0","event_id":"sha256:e9093bb9d79d54dac123817ac21f8b4f807ecae2806b870cd74613135b28826f"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:ZEOQ7X6BZXG5BHSA4V76JMKGOL","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Finite Difference Flow Optimization for RL Post-Training of Text-to-Image Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG","cs.NE","stat.ML"],"primary_cat":"cs.CV","authors_text":"Angjoo Kanazawa, David McAllister, Janne Hellsten, Miika Aittala, Samuli Laine, Tero Karras, Timo Aila","submitted_at":"2026-03-13T10:54:09Z","abstract_excerpt":"Reinforcement learning (RL) has become a standard technique for post-training diffusion-based image synthesis models, as it enables learning from reward signals to explicitly improve desirable aspects such as image quality and prompt alignment. In this paper, we propose an online RL variant that reduces the variance in the model updates by sampling paired trajectories and pulling the flow velocity in the direction of the more favorable image. Unlike existing methods that treat each sampling step as a separate policy action, we consider the entire sampling process as a single action. We experim"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.12893","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2603.12893/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"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":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-01T01:17:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"7YxHTBb/1Xq2DxeKAcmTlE38cHg4ngZ/KpxrC9RsiC4tfSJZJo1V/ax3c62McYtqeMD6GsqzAsI1fpXhBBjFBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-03T16:48:01.843490Z"},"content_sha256":"84e7990659490079cc0e9b5217abef9364bf94ab67721a562ec8a07fc63ca5d8","schema_version":"1.0","event_id":"sha256:84e7990659490079cc0e9b5217abef9364bf94ab67721a562ec8a07fc63ca5d8"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/ZEOQ7X6BZXG5BHSA4V76JMKGOL/bundle.json","state_url":"https://pith.science/pith/ZEOQ7X6BZXG5BHSA4V76JMKGOL/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/ZEOQ7X6BZXG5BHSA4V76JMKGOL/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":"2026-07-03T16:48:01Z","links":{"resolver":"https://pith.science/pith/ZEOQ7X6BZXG5BHSA4V76JMKGOL","bundle":"https://pith.science/pith/ZEOQ7X6BZXG5BHSA4V76JMKGOL/bundle.json","state":"https://pith.science/pith/ZEOQ7X6BZXG5BHSA4V76JMKGOL/state.json","well_known_bundle":"https://pith.science/.well-known/pith/ZEOQ7X6BZXG5BHSA4V76JMKGOL/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:ZEOQ7X6BZXG5BHSA4V76JMKGOL","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":"a4d647f94d53b3512cd5bf9396d03e7c42856bf03ae703501c0ab5011f290f6e","cross_cats_sorted":["cs.AI","cs.LG","cs.NE","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-03-13T10:54:09Z","title_canon_sha256":"a79b0306a1b3f92285d4716dd0f5599fa6d65f3bdf6f55521bf4681779c5c738"},"schema_version":"1.0","source":{"id":"2603.12893","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2603.12893","created_at":"2026-07-01T01:17:48Z"},{"alias_kind":"arxiv_version","alias_value":"2603.12893v2","created_at":"2026-07-01T01:17:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.12893","created_at":"2026-07-01T01:17:48Z"},{"alias_kind":"pith_short_12","alias_value":"ZEOQ7X6BZXG5","created_at":"2026-07-01T01:17:48Z"},{"alias_kind":"pith_short_16","alias_value":"ZEOQ7X6BZXG5BHSA","created_at":"2026-07-01T01:17:48Z"},{"alias_kind":"pith_short_8","alias_value":"ZEOQ7X6B","created_at":"2026-07-01T01:17:48Z"}],"graph_snapshots":[{"event_id":"sha256:84e7990659490079cc0e9b5217abef9364bf94ab67721a562ec8a07fc63ca5d8","target":"graph","created_at":"2026-07-01T01:17:48Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2603.12893/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reinforcement learning (RL) has become a standard technique for post-training diffusion-based image synthesis models, as it enables learning from reward signals to explicitly improve desirable aspects such as image quality and prompt alignment. In this paper, we propose an online RL variant that reduces the variance in the model updates by sampling paired trajectories and pulling the flow velocity in the direction of the more favorable image. Unlike existing methods that treat each sampling step as a separate policy action, we consider the entire sampling process as a single action. We experim","authors_text":"Angjoo Kanazawa, David McAllister, Janne Hellsten, Miika Aittala, Samuli Laine, Tero Karras, Timo Aila","cross_cats":["cs.AI","cs.LG","cs.NE","stat.ML"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-03-13T10:54:09Z","title":"Finite Difference Flow Optimization for RL Post-Training of Text-to-Image Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.12893","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:e9093bb9d79d54dac123817ac21f8b4f807ecae2806b870cd74613135b28826f","target":"record","created_at":"2026-07-01T01:17:48Z","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":"a4d647f94d53b3512cd5bf9396d03e7c42856bf03ae703501c0ab5011f290f6e","cross_cats_sorted":["cs.AI","cs.LG","cs.NE","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-03-13T10:54:09Z","title_canon_sha256":"a79b0306a1b3f92285d4716dd0f5599fa6d65f3bdf6f55521bf4681779c5c738"},"schema_version":"1.0","source":{"id":"2603.12893","kind":"arxiv","version":2}},"canonical_sha256":"c91d0fdfc1cdcdd09e40e57fe4b14672ef8049051cee5ac1a11a50c9aeba48a4","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"c91d0fdfc1cdcdd09e40e57fe4b14672ef8049051cee5ac1a11a50c9aeba48a4","first_computed_at":"2026-07-01T01:17:48.891388Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-01T01:17:48.891388Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"QXQp/2T9zRc0FX/Widvx4M3xyMoiQ4bN4ylCvVaJ+DaSpDlEnvxKU9ZWM/H7+4TxBRSFgEIIrE8aaOTqoqTICA==","signature_status":"signed_v1","signed_at":"2026-07-01T01:17:48.891877Z","signed_message":"canonical_sha256_bytes"},"source_id":"2603.12893","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e9093bb9d79d54dac123817ac21f8b4f807ecae2806b870cd74613135b28826f","sha256:84e7990659490079cc0e9b5217abef9364bf94ab67721a562ec8a07fc63ca5d8"],"state_sha256":"4f20368e392df35e119a344948b06b90db12e0a5f9901cf0c9746788bedac633"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"jMwQD/Li4TFATil6SCv92v5xr20ZmeDf0LZxPSOUKZvR2kx/bZJTB1kaGRvGztY99W8iRiPpM3WILcNV88+nAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-03T16:48:01.845646Z","bundle_sha256":"435431742e2aa594905bdda35ce0f73a13968b4bde1d56fcd7df686b996107b3"}}