{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:L575ERTKOBJAZ25BPUKFBYVVQG","short_pith_number":"pith:L575ERTK","schema_version":"1.0","canonical_sha256":"5f7fd2466a70520ceba17d1450e2b5818ad83372590a8a260a8f5282c9917423","source":{"kind":"arxiv","id":"2503.07493","version":1},"attestation_state":"computed","paper":{"title":"V2Flow: Unifying Visual Tokenization and Large Language Model Vocabularies for Autoregressive Image Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Biye Li, Guiwei Zhang, Mohan Zhou, Tianyu Zhang, Yalong Bai","submitted_at":"2025-03-10T16:12:50Z","abstract_excerpt":"We propose V2Flow, a novel tokenizer that produces discrete visual tokens capable of high-fidelity reconstruction, while ensuring structural and latent distribution alignment with the vocabulary space of large language models (LLMs). Leveraging this tight visual-vocabulary coupling, V2Flow enables autoregressive visual generation on top of existing LLMs. Our approach formulates visual tokenization as a flow-matching problem, aiming to learn a mapping from a standard normal prior to the continuous image distribution, conditioned on token sequences embedded within the LLMs vocabulary space. The "},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2503.07493","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-03-10T16:12:50Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"397e5cf7550ccda652b003747657d81812138abed249a389a7fac6793033d572","abstract_canon_sha256":"458e933f547f22d56c3c5954b2d3950866b96b2877ac04de571a55162f3a463a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:27:57.594163Z","signature_b64":"oO/h2/ziMuOAm8q8k/57+ifLz3O+xIHzv/2wIjEPR7Qp/XqfPSP7DJmtbv1cKlltV28jNUxg5jDNmlAIOibyCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5f7fd2466a70520ceba17d1450e2b5818ad83372590a8a260a8f5282c9917423","last_reissued_at":"2026-07-05T10:27:57.593626Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:27:57.593626Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"V2Flow: Unifying Visual Tokenization and Large Language Model Vocabularies for Autoregressive Image Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Biye Li, Guiwei Zhang, Mohan Zhou, Tianyu Zhang, Yalong Bai","submitted_at":"2025-03-10T16:12:50Z","abstract_excerpt":"We propose V2Flow, a novel tokenizer that produces discrete visual tokens capable of high-fidelity reconstruction, while ensuring structural and latent distribution alignment with the vocabulary space of large language models (LLMs). Leveraging this tight visual-vocabulary coupling, V2Flow enables autoregressive visual generation on top of existing LLMs. Our approach formulates visual tokenization as a flow-matching problem, aiming to learn a mapping from a standard normal prior to the continuous image distribution, conditioned on token sequences embedded within the LLMs vocabulary space. The "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.07493","kind":"arxiv","version":1},"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/2503.07493/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2503.07493","created_at":"2026-07-05T10:27:57.593697+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.07493v1","created_at":"2026-07-05T10:27:57.593697+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.07493","created_at":"2026-07-05T10:27:57.593697+00:00"},{"alias_kind":"pith_short_12","alias_value":"L575ERTKOBJA","created_at":"2026-07-05T10:27:57.593697+00:00"},{"alias_kind":"pith_short_16","alias_value":"L575ERTKOBJAZ25B","created_at":"2026-07-05T10:27:57.593697+00:00"},{"alias_kind":"pith_short_8","alias_value":"L575ERTK","created_at":"2026-07-05T10:27:57.593697+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.30904","citing_title":"MergeTok: Unified Continuous and Discrete Visual Tokenization via Token Merging","ref_index":42,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/L575ERTKOBJAZ25BPUKFBYVVQG","json":"https://pith.science/pith/L575ERTKOBJAZ25BPUKFBYVVQG.json","graph_json":"https://pith.science/api/pith-number/L575ERTKOBJAZ25BPUKFBYVVQG/graph.json","events_json":"https://pith.science/api/pith-number/L575ERTKOBJAZ25BPUKFBYVVQG/events.json","paper":"https://pith.science/paper/L575ERTK"},"agent_actions":{"view_html":"https://pith.science/pith/L575ERTKOBJAZ25BPUKFBYVVQG","download_json":"https://pith.science/pith/L575ERTKOBJAZ25BPUKFBYVVQG.json","view_paper":"https://pith.science/paper/L575ERTK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.07493&json=true","fetch_graph":"https://pith.science/api/pith-number/L575ERTKOBJAZ25BPUKFBYVVQG/graph.json","fetch_events":"https://pith.science/api/pith-number/L575ERTKOBJAZ25BPUKFBYVVQG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/L575ERTKOBJAZ25BPUKFBYVVQG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/L575ERTKOBJAZ25BPUKFBYVVQG/action/storage_attestation","attest_author":"https://pith.science/pith/L575ERTKOBJAZ25BPUKFBYVVQG/action/author_attestation","sign_citation":"https://pith.science/pith/L575ERTKOBJAZ25BPUKFBYVVQG/action/citation_signature","submit_replication":"https://pith.science/pith/L575ERTKOBJAZ25BPUKFBYVVQG/action/replication_record"}},"created_at":"2026-07-05T10:27:57.593697+00:00","updated_at":"2026-07-05T10:27:57.593697+00:00"}