{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:OFYD3BQBKHIJEVPBASA4I6MSI5","short_pith_number":"pith:OFYD3BQB","schema_version":"1.0","canonical_sha256":"71703d860151d09255e10481c4799247408b2f581a2ab7041c3411072b03fbe0","source":{"kind":"arxiv","id":"2103.03841","version":1},"attestation_state":"computed","paper":{"title":"Generating Images with Sparse Representations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.CV","authors_text":"Charlie Nash, Jacob Menick, Peter W. Battaglia, Sander Dieleman","submitted_at":"2021-03-05T17:56:03Z","abstract_excerpt":"The high dimensionality of images presents architecture and sampling-efficiency challenges for likelihood-based generative models. Previous approaches such as VQ-VAE use deep autoencoders to obtain compact representations, which are more practical as inputs for likelihood-based models. We present an alternative approach, inspired by common image compression methods like JPEG, and convert images to quantized discrete cosine transform (DCT) blocks, which are represented sparsely as a sequence of DCT channel, spatial location, and DCT coefficient triples. We propose a Transformer-based autoregres"},"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":"2103.03841","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2021-03-05T17:56:03Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"f6451f9109d333df08946c12c9ef17e3e2e5d8a210a01bbc178f367d63058032","abstract_canon_sha256":"227b4172302db37f827776511187d1dac804c004457a223bd689b7822d1b1ecf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:20:40.360156Z","signature_b64":"D0z9lzC+vpJxEUAVdQQYPlDPK+FiNvLl7hmErfceY9rkPUBN9DmbnYxem2wVVmmuPXwu+LU41x/H1u4LURxFCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"71703d860151d09255e10481c4799247408b2f581a2ab7041c3411072b03fbe0","last_reissued_at":"2026-07-05T02:20:40.359673Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:20:40.359673Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Generating Images with Sparse Representations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.CV","authors_text":"Charlie Nash, Jacob Menick, Peter W. Battaglia, Sander Dieleman","submitted_at":"2021-03-05T17:56:03Z","abstract_excerpt":"The high dimensionality of images presents architecture and sampling-efficiency challenges for likelihood-based generative models. Previous approaches such as VQ-VAE use deep autoencoders to obtain compact representations, which are more practical as inputs for likelihood-based models. We present an alternative approach, inspired by common image compression methods like JPEG, and convert images to quantized discrete cosine transform (DCT) blocks, which are represented sparsely as a sequence of DCT channel, spatial location, and DCT coefficient triples. We propose a Transformer-based autoregres"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2103.03841","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/2103.03841/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":"2103.03841","created_at":"2026-07-05T02:20:40.359730+00:00"},{"alias_kind":"arxiv_version","alias_value":"2103.03841v1","created_at":"2026-07-05T02:20:40.359730+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2103.03841","created_at":"2026-07-05T02:20:40.359730+00:00"},{"alias_kind":"pith_short_12","alias_value":"OFYD3BQBKHIJ","created_at":"2026-07-05T02:20:40.359730+00:00"},{"alias_kind":"pith_short_16","alias_value":"OFYD3BQBKHIJEVPB","created_at":"2026-07-05T02:20:40.359730+00:00"},{"alias_kind":"pith_short_8","alias_value":"OFYD3BQB","created_at":"2026-07-05T02:20:40.359730+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":27,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.31363","citing_title":"Language-Assisted Super-Resolution from Real-World Low-Resolution Patches","ref_index":86,"is_internal_anchor":false},{"citing_arxiv_id":"2607.02508","citing_title":"From SRA to Self-Flow: Data Augmentation or Self-Supervision?","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08788","citing_title":"MaskAlign: Token-Subset Representation Alignment for Efficient Diffusion Training","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2607.00927","citing_title":"Post-Training Pruning for Diffusion Transformers","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2606.31363","citing_title":"Language-Assisted Super-Resolution from Real-World Low-Resolution Patches","ref_index":86,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14819","citing_title":"The Velocity Deficit: Initial Energy Injection for Flow Matching","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15741","citing_title":"HyperDiT: Hyper-Connected Transformers for High-Fidelity Pixel-Space Diffusion","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2605.24631","citing_title":"Beyond Generative Priors: Minority Sampling with JEPA-Guided Diffusion","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27919","citing_title":"Frequency-Guided Action Diffusion via Sub-Frequency Manifold Traversal","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30332","citing_title":"Colored Noise Diffusion Sampling","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00746","citing_title":"Scaling Parallel Sequence Models to Foundation-Scale Vision Encoders","ref_index":86,"is_internal_anchor":false},{"citing_arxiv_id":"2509.23582","citing_title":"RobuQ: Pushing DiTs to W1.58A2 via Robust Activation Quantization","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":34,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15741","citing_title":"HyperDiT: Hyper-Connected Transformers for High-Fidelity Pixel-Space Diffusion","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16949","citing_title":"Beyond Point-Wise Matching: Structural Representation Alignment for Accelerating Diffusion Transformers","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2510.18457","citing_title":"VFM-VAE: Vision Foundation Models Can Be Good Tokenizers for Latent Diffusion Models","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2511.19365","citing_title":"DeCo: Frequency-Decoupled Pixel Diffusion for End-to-End Image Generation","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2110.04627","citing_title":"Vector-quantized Image Modeling with Improved VQGAN","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2601.14671","citing_title":"Mirai: Autoregressive Visual Generation Needs Foresight","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2105.05233","citing_title":"Diffusion Models Beat GANs on Image Synthesis","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10790","citing_title":"Elucidating Representation Degradation Problem in Diffusion Model Training","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2212.09748","citing_title":"Scalable Diffusion Models with Transformers","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2604.25299","citing_title":"The Thinking Pixel: Recursive Sparse Reasoning in Multimodal Diffusion Latents","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2406.06525","citing_title":"Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06339","citing_title":"Evolution of Video Generative Foundations","ref_index":127,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OFYD3BQBKHIJEVPBASA4I6MSI5","json":"https://pith.science/pith/OFYD3BQBKHIJEVPBASA4I6MSI5.json","graph_json":"https://pith.science/api/pith-number/OFYD3BQBKHIJEVPBASA4I6MSI5/graph.json","events_json":"https://pith.science/api/pith-number/OFYD3BQBKHIJEVPBASA4I6MSI5/events.json","paper":"https://pith.science/paper/OFYD3BQB"},"agent_actions":{"view_html":"https://pith.science/pith/OFYD3BQBKHIJEVPBASA4I6MSI5","download_json":"https://pith.science/pith/OFYD3BQBKHIJEVPBASA4I6MSI5.json","view_paper":"https://pith.science/paper/OFYD3BQB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2103.03841&json=true","fetch_graph":"https://pith.science/api/pith-number/OFYD3BQBKHIJEVPBASA4I6MSI5/graph.json","fetch_events":"https://pith.science/api/pith-number/OFYD3BQBKHIJEVPBASA4I6MSI5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OFYD3BQBKHIJEVPBASA4I6MSI5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OFYD3BQBKHIJEVPBASA4I6MSI5/action/storage_attestation","attest_author":"https://pith.science/pith/OFYD3BQBKHIJEVPBASA4I6MSI5/action/author_attestation","sign_citation":"https://pith.science/pith/OFYD3BQBKHIJEVPBASA4I6MSI5/action/citation_signature","submit_replication":"https://pith.science/pith/OFYD3BQBKHIJEVPBASA4I6MSI5/action/replication_record"}},"created_at":"2026-07-05T02:20:40.359730+00:00","updated_at":"2026-07-05T02:20:40.359730+00:00"}