{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:OF3N5QAQZKHR5PQPZGDFZXF4X5","short_pith_number":"pith:OF3N5QAQ","schema_version":"1.0","canonical_sha256":"7176dec010ca8f1ebe0fc9865cdcbcbf484747d0661c4f093ba591b36ae13d41","source":{"kind":"arxiv","id":"2406.11837","version":1},"attestation_state":"computed","paper":{"title":"Scaling the Codebook Size of VQGAN to 100,000 with a Utilization Rate of 99%","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Dong Chen, Fangyun Wei, Lei Zhu, Yanye Lu","submitted_at":"2024-06-17T17:59:57Z","abstract_excerpt":"In the realm of image quantization exemplified by VQGAN, the process encodes images into discrete tokens drawn from a codebook with a predefined size. Recent advancements, particularly with LLAMA 3, reveal that enlarging the codebook significantly enhances model performance. However, VQGAN and its derivatives, such as VQGAN-FC (Factorized Codes) and VQGAN-EMA, continue to grapple with challenges related to expanding the codebook size and enhancing codebook utilization. For instance, VQGAN-FC is restricted to learning a codebook with a maximum size of 16,384, maintaining a typically low utiliza"},"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":"2406.11837","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-06-17T17:59:57Z","cross_cats_sorted":[],"title_canon_sha256":"2cf09ad20b3b8002c26590d38a0d3a5e74c8f6050a686c0e3b4c3ca4b966cba3","abstract_canon_sha256":"94ff2662832f04a3051de478e7b3b5f3e84f0fbfc7955f9681f75ae3b91aa7ee"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:32:58.482734Z","signature_b64":"qLpW0w4GCOJqUIt8rKLGCWeP+89JMlBfd+SM2SAg8Vn87U4rYmzUKmF9KLtPp2ULt3E2YaegmxQJuv96HFGFDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7176dec010ca8f1ebe0fc9865cdcbcbf484747d0661c4f093ba591b36ae13d41","last_reissued_at":"2026-07-05T08:32:58.482269Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:32:58.482269Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scaling the Codebook Size of VQGAN to 100,000 with a Utilization Rate of 99%","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Dong Chen, Fangyun Wei, Lei Zhu, Yanye Lu","submitted_at":"2024-06-17T17:59:57Z","abstract_excerpt":"In the realm of image quantization exemplified by VQGAN, the process encodes images into discrete tokens drawn from a codebook with a predefined size. Recent advancements, particularly with LLAMA 3, reveal that enlarging the codebook significantly enhances model performance. However, VQGAN and its derivatives, such as VQGAN-FC (Factorized Codes) and VQGAN-EMA, continue to grapple with challenges related to expanding the codebook size and enhancing codebook utilization. For instance, VQGAN-FC is restricted to learning a codebook with a maximum size of 16,384, maintaining a typically low utiliza"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.11837","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/2406.11837/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":"2406.11837","created_at":"2026-07-05T08:32:58.482325+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.11837v1","created_at":"2026-07-05T08:32:58.482325+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.11837","created_at":"2026-07-05T08:32:58.482325+00:00"},{"alias_kind":"pith_short_12","alias_value":"OF3N5QAQZKHR","created_at":"2026-07-05T08:32:58.482325+00:00"},{"alias_kind":"pith_short_16","alias_value":"OF3N5QAQZKHR5PQP","created_at":"2026-07-05T08:32:58.482325+00:00"},{"alias_kind":"pith_short_8","alias_value":"OF3N5QAQ","created_at":"2026-07-05T08:32:58.482325+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.12940","citing_title":"Self-Guidance: Enhancing Neural Codecs via Decoder Manifold Alignment","ref_index":143,"is_internal_anchor":false},{"citing_arxiv_id":"2606.11096","citing_title":"IDEAL: In-DEpth ALignment Makes A Discrete Representation AutoEncoder","ref_index":74,"is_internal_anchor":false},{"citing_arxiv_id":"2606.11363","citing_title":"NSVQ: Mitigating Codebook Collapse by Stabilizing Encoder Drift in Vector Quantization","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01157","citing_title":"HiTokSR: A Coarse-to-Fine Tokenizer with Hierarchical Codebooks for High-Fidelity Real-World Image Super-Resolution","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16384","citing_title":"Mutual Enhancement Between Global Tokens and Patch Tokens: From Theory to Practice","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18390","citing_title":"Vision Foundation Models as Generalist Tokenizers for Image Generation","ref_index":103,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OF3N5QAQZKHR5PQPZGDFZXF4X5","json":"https://pith.science/pith/OF3N5QAQZKHR5PQPZGDFZXF4X5.json","graph_json":"https://pith.science/api/pith-number/OF3N5QAQZKHR5PQPZGDFZXF4X5/graph.json","events_json":"https://pith.science/api/pith-number/OF3N5QAQZKHR5PQPZGDFZXF4X5/events.json","paper":"https://pith.science/paper/OF3N5QAQ"},"agent_actions":{"view_html":"https://pith.science/pith/OF3N5QAQZKHR5PQPZGDFZXF4X5","download_json":"https://pith.science/pith/OF3N5QAQZKHR5PQPZGDFZXF4X5.json","view_paper":"https://pith.science/paper/OF3N5QAQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.11837&json=true","fetch_graph":"https://pith.science/api/pith-number/OF3N5QAQZKHR5PQPZGDFZXF4X5/graph.json","fetch_events":"https://pith.science/api/pith-number/OF3N5QAQZKHR5PQPZGDFZXF4X5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OF3N5QAQZKHR5PQPZGDFZXF4X5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OF3N5QAQZKHR5PQPZGDFZXF4X5/action/storage_attestation","attest_author":"https://pith.science/pith/OF3N5QAQZKHR5PQPZGDFZXF4X5/action/author_attestation","sign_citation":"https://pith.science/pith/OF3N5QAQZKHR5PQPZGDFZXF4X5/action/citation_signature","submit_replication":"https://pith.science/pith/OF3N5QAQZKHR5PQPZGDFZXF4X5/action/replication_record"}},"created_at":"2026-07-05T08:32:58.482325+00:00","updated_at":"2026-07-05T08:32:58.482325+00:00"}