{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:FC5OAH3FG3HFI5SEWHNH34QWE6","short_pith_number":"pith:FC5OAH3F","schema_version":"1.0","canonical_sha256":"28bae01f6536ce547644b1da7df216278e947277135bad2018e8fec6d9edd245","source":{"kind":"arxiv","id":"2410.10812","version":1},"attestation_state":"computed","paper":{"title":"HART: Efficient Visual Generation with Hybrid Autoregressive Transformer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Enze Xie, Han Cai, Haotian Tang, Junsong Chen, Junyu Chen, Shang Yang, Song Han, Yao Lu, Yecheng Wu, Zhuoyang Zhang","submitted_at":"2024-10-14T17:59:42Z","abstract_excerpt":"We introduce Hybrid Autoregressive Transformer (HART), an autoregressive (AR) visual generation model capable of directly generating 1024x1024 images, rivaling diffusion models in image generation quality. Existing AR models face limitations due to the poor image reconstruction quality of their discrete tokenizers and the prohibitive training costs associated with generating 1024px images. To address these challenges, we present the hybrid tokenizer, which decomposes the continuous latents from the autoencoder into two components: discrete tokens representing the big picture and continuous tok"},"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":"2410.10812","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-10-14T17:59:42Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"d7e0e4611e4af0f84477db85ceab326cf5b894b00d619c5c1950fd726980f6d3","abstract_canon_sha256":"4b2c85cad72055c69ca11d5d388787a26a4adaf9f8bef91091bf4f1f275f4cc8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:20:21.103789Z","signature_b64":"7Borvr9nm7+/4+D9XhZaZda/kh/JdEEwJOVJ+ZkkvNg0xf+mLDC1BBvG5gucZ6PXOUCwKDLqiwngEouavePhBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"28bae01f6536ce547644b1da7df216278e947277135bad2018e8fec6d9edd245","last_reissued_at":"2026-07-05T09:20:21.103330Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:20:21.103330Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"HART: Efficient Visual Generation with Hybrid Autoregressive Transformer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Enze Xie, Han Cai, Haotian Tang, Junsong Chen, Junyu Chen, Shang Yang, Song Han, Yao Lu, Yecheng Wu, Zhuoyang Zhang","submitted_at":"2024-10-14T17:59:42Z","abstract_excerpt":"We introduce Hybrid Autoregressive Transformer (HART), an autoregressive (AR) visual generation model capable of directly generating 1024x1024 images, rivaling diffusion models in image generation quality. Existing AR models face limitations due to the poor image reconstruction quality of their discrete tokenizers and the prohibitive training costs associated with generating 1024px images. To address these challenges, we present the hybrid tokenizer, which decomposes the continuous latents from the autoencoder into two components: discrete tokens representing the big picture and continuous tok"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.10812","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/2410.10812/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":"2410.10812","created_at":"2026-07-05T09:20:21.103390+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.10812v1","created_at":"2026-07-05T09:20:21.103390+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.10812","created_at":"2026-07-05T09:20:21.103390+00:00"},{"alias_kind":"pith_short_12","alias_value":"FC5OAH3FG3HF","created_at":"2026-07-05T09:20:21.103390+00:00"},{"alias_kind":"pith_short_16","alias_value":"FC5OAH3FG3HFI5SE","created_at":"2026-07-05T09:20:21.103390+00:00"},{"alias_kind":"pith_short_8","alias_value":"FC5OAH3F","created_at":"2026-07-05T09:20:21.103390+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":12,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25548","citing_title":"Concept Removal for Frontier Image Generative Models","ref_index":123,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08302","citing_title":"HACK++: Towards More Effective Head-Aware Key-Value Compression for Efficient Visual Autoregressive Modeling","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06078","citing_title":"Knowledge Distillation for Visual Autoregressive Models","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28453","citing_title":"DeVAR: Low-Dose CT Denoising via Visual Autoregressive Modeling","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06137","citing_title":"Autoregressive Visual Generation Needs a Prologue","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00310","citing_title":"Where to Refine, When to Stop: Rethinking Redundancy via Latent Discrepancy for Efficient Visual Autoregressive Generation","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2602.16169","citing_title":"Discrete Stochastic Localization for Non-autoregressive Generation","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2505.22394","citing_title":"PacTure: Efficient PBR Texture Generation on Packed Views with Visual Autoregressive Models","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09430","citing_title":"FlashAR: Efficient Post-Training Acceleration for Autoregressive Image Generation","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10045","citing_title":"ExtraVAR: Stage-Aware RoPE Remapping for Resolution Extrapolation in Visual Autoregressive Models","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09430","citing_title":"FlashAR: Efficient Post-Training Acceleration for Autoregressive Image Generation","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06137","citing_title":"Autoregressive Visual Generation Needs a Prologue","ref_index":42,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FC5OAH3FG3HFI5SEWHNH34QWE6","json":"https://pith.science/pith/FC5OAH3FG3HFI5SEWHNH34QWE6.json","graph_json":"https://pith.science/api/pith-number/FC5OAH3FG3HFI5SEWHNH34QWE6/graph.json","events_json":"https://pith.science/api/pith-number/FC5OAH3FG3HFI5SEWHNH34QWE6/events.json","paper":"https://pith.science/paper/FC5OAH3F"},"agent_actions":{"view_html":"https://pith.science/pith/FC5OAH3FG3HFI5SEWHNH34QWE6","download_json":"https://pith.science/pith/FC5OAH3FG3HFI5SEWHNH34QWE6.json","view_paper":"https://pith.science/paper/FC5OAH3F","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.10812&json=true","fetch_graph":"https://pith.science/api/pith-number/FC5OAH3FG3HFI5SEWHNH34QWE6/graph.json","fetch_events":"https://pith.science/api/pith-number/FC5OAH3FG3HFI5SEWHNH34QWE6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FC5OAH3FG3HFI5SEWHNH34QWE6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FC5OAH3FG3HFI5SEWHNH34QWE6/action/storage_attestation","attest_author":"https://pith.science/pith/FC5OAH3FG3HFI5SEWHNH34QWE6/action/author_attestation","sign_citation":"https://pith.science/pith/FC5OAH3FG3HFI5SEWHNH34QWE6/action/citation_signature","submit_replication":"https://pith.science/pith/FC5OAH3FG3HFI5SEWHNH34QWE6/action/replication_record"}},"created_at":"2026-07-05T09:20:21.103390+00:00","updated_at":"2026-07-05T09:20:21.103390+00:00"}