{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:JUSI6Y6RDIT2FHAUJ5QEUZYQ3V","short_pith_number":"pith:JUSI6Y6R","schema_version":"1.0","canonical_sha256":"4d248f63d11a27a29c144f604a6710dd5bffc2c742cb93387c51a6ddb172457c","source":{"kind":"arxiv","id":"2406.11838","version":3},"attestation_state":"computed","paper":{"title":"Autoregressive Image Generation without Vector Quantization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"He Li, Kaiming He, Mingyang Deng, Tianhong Li, Yonglong Tian","submitted_at":"2024-06-17T17:59:58Z","abstract_excerpt":"Conventional wisdom holds that autoregressive models for image generation are typically accompanied by vector-quantized tokens. We observe that while a discrete-valued space can facilitate representing a categorical distribution, it is not a necessity for autoregressive modeling. In this work, we propose to model the per-token probability distribution using a diffusion procedure, which allows us to apply autoregressive models in a continuous-valued space. Rather than using categorical cross-entropy loss, we define a Diffusion Loss function to model the per-token probability. This approach elim"},"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":"2406.11838","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-06-17T17:59:58Z","cross_cats_sorted":[],"title_canon_sha256":"b93f97b9d3496115529171079e444776050239685fb750c283948096f00066fb","abstract_canon_sha256":"48c8141839b054d18a2c9d03afacf335e705ddb6148d73f265f3ea5e37a25dab"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:29:30.550456Z","signature_b64":"8tRwLlHKOX0ICDktlrkqSPbhlR6vPVWBN8Bnz4F/DG15SaRJ099AM1NHw2I9YaiwTlN53H6Do8pjKFANfRuZCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4d248f63d11a27a29c144f604a6710dd5bffc2c742cb93387c51a6ddb172457c","last_reissued_at":"2026-07-05T09:29:30.549889Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:29:30.549889Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Autoregressive Image Generation without Vector Quantization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"He Li, Kaiming He, Mingyang Deng, Tianhong Li, Yonglong Tian","submitted_at":"2024-06-17T17:59:58Z","abstract_excerpt":"Conventional wisdom holds that autoregressive models for image generation are typically accompanied by vector-quantized tokens. We observe that while a discrete-valued space can facilitate representing a categorical distribution, it is not a necessity for autoregressive modeling. In this work, we propose to model the per-token probability distribution using a diffusion procedure, which allows us to apply autoregressive models in a continuous-valued space. Rather than using categorical cross-entropy loss, we define a Diffusion Loss function to model the per-token probability. This approach elim"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.11838","kind":"arxiv","version":3},"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/2406.11838/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":"2406.11838","created_at":"2026-07-05T09:29:30.549958+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.11838v3","created_at":"2026-07-05T09:29:30.549958+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.11838","created_at":"2026-07-05T09:29:30.549958+00:00"},{"alias_kind":"pith_short_12","alias_value":"JUSI6Y6RDIT2","created_at":"2026-07-05T09:29:30.549958+00:00"},{"alias_kind":"pith_short_16","alias_value":"JUSI6Y6RDIT2FHAU","created_at":"2026-07-05T09:29:30.549958+00:00"},{"alias_kind":"pith_short_8","alias_value":"JUSI6Y6R","created_at":"2026-07-05T09:29:30.549958+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":18,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.01918","citing_title":"Zeus: Towards Tuning-Free Foundation Model for Time Series Analysis","ref_index":164,"is_internal_anchor":false},{"citing_arxiv_id":"2606.31734","citing_title":"MemLearner: Learning to Query Context memory for Video World Models","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18267","citing_title":"SRC-Flow: Compact Semantic Representations Enable Normalizing Flows for Image Generation","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29570","citing_title":"Hierarchical Policy Learning via Spectral Decomposition","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2502.00816","citing_title":"Sundial: A Family of Highly Capable Time Series Foundation Models","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2505.11334","citing_title":"MARRS: Masked Autoregressive Unit-based Reaction Synthesis","ref_index":16,"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":90,"is_internal_anchor":false},{"citing_arxiv_id":"2410.17891","citing_title":"Scaling Diffusion Language Models via Adaptation from Autoregressive Models","ref_index":150,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18267","citing_title":"SRC-Flow: Compact Semantic Representations Enable Normalizing Flows for Image Generation","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18390","citing_title":"Vision Foundation Models as Generalist Tokenizers for Image Generation","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2410.06885","citing_title":"F5-TTS: A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching","ref_index":117,"is_internal_anchor":false},{"citing_arxiv_id":"2410.13848","citing_title":"Janus: Decoupling Visual Encoding for Unified Multimodal Understanding and Generation","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2503.00200","citing_title":"Unified Video Action Model","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12011","citing_title":"CaloArt: Large-Patch x-Prediction Diffusion Transformers for High-Granularity Calorimeter Shower Generation","ref_index":61,"is_internal_anchor":false},{"citing_arxiv_id":"2506.15564","citing_title":"Show-o2: Improved Native Unified Multimodal Models","ref_index":61,"is_internal_anchor":false},{"citing_arxiv_id":"2604.09168","citing_title":"ELT: Elastic Looped Transformers for Visual Generation","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07230","citing_title":"CASCADE: Context-Aware Relaxation for Speculative Image Decoding","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18471","citing_title":"NI Sampling: Accelerating Discrete Diffusion Sampling by Token Order Optimization","ref_index":40,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JUSI6Y6RDIT2FHAUJ5QEUZYQ3V","json":"https://pith.science/pith/JUSI6Y6RDIT2FHAUJ5QEUZYQ3V.json","graph_json":"https://pith.science/api/pith-number/JUSI6Y6RDIT2FHAUJ5QEUZYQ3V/graph.json","events_json":"https://pith.science/api/pith-number/JUSI6Y6RDIT2FHAUJ5QEUZYQ3V/events.json","paper":"https://pith.science/paper/JUSI6Y6R"},"agent_actions":{"view_html":"https://pith.science/pith/JUSI6Y6RDIT2FHAUJ5QEUZYQ3V","download_json":"https://pith.science/pith/JUSI6Y6RDIT2FHAUJ5QEUZYQ3V.json","view_paper":"https://pith.science/paper/JUSI6Y6R","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.11838&json=true","fetch_graph":"https://pith.science/api/pith-number/JUSI6Y6RDIT2FHAUJ5QEUZYQ3V/graph.json","fetch_events":"https://pith.science/api/pith-number/JUSI6Y6RDIT2FHAUJ5QEUZYQ3V/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JUSI6Y6RDIT2FHAUJ5QEUZYQ3V/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JUSI6Y6RDIT2FHAUJ5QEUZYQ3V/action/storage_attestation","attest_author":"https://pith.science/pith/JUSI6Y6RDIT2FHAUJ5QEUZYQ3V/action/author_attestation","sign_citation":"https://pith.science/pith/JUSI6Y6RDIT2FHAUJ5QEUZYQ3V/action/citation_signature","submit_replication":"https://pith.science/pith/JUSI6Y6RDIT2FHAUJ5QEUZYQ3V/action/replication_record"}},"created_at":"2026-07-05T09:29:30.549958+00:00","updated_at":"2026-07-05T09:29:30.549958+00:00"}