{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:DZ7PMTNAIR7C6DKS3JMGN5JGMR","short_pith_number":"pith:DZ7PMTNA","schema_version":"1.0","canonical_sha256":"1e7ef64da0447e2f0d52da5866f52664647a610c743df7c0168d92e2cd51d5d3","source":{"kind":"arxiv","id":"2310.14189","version":1},"attestation_state":"computed","paper":{"title":"Improved Techniques for Training Consistency Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Prafulla Dhariwal, Yang Song","submitted_at":"2023-10-22T05:33:38Z","abstract_excerpt":"Consistency models are a nascent family of generative models that can sample high quality data in one step without the need for adversarial training. Current consistency models achieve optimal sample quality by distilling from pre-trained diffusion models and employing learned metrics such as LPIPS. However, distillation limits the quality of consistency models to that of the pre-trained diffusion model, and LPIPS causes undesirable bias in evaluation. To tackle these challenges, we present improved techniques for consistency training, where consistency models learn directly from data without "},"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":"2310.14189","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-22T05:33:38Z","cross_cats_sorted":[],"title_canon_sha256":"44e3a7d6224fa956d692ca60c104a827e35d4b966aa243f2cdbfb087c2522e63","abstract_canon_sha256":"c0c4eebbeba9f71bbbd2ddc91fe2f507bd630888844dc777a0b1187c3d2802a3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T04:58:47.941359Z","signature_b64":"d3Oo25wTQSbtWPq5GFZ9gezRDNNn/aVYZr3FuCoi6TsaFGca/kY2TLwJ+ayPq8VtFXblnevE9wtwEu43p1eoDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1e7ef64da0447e2f0d52da5866f52664647a610c743df7c0168d92e2cd51d5d3","last_reissued_at":"2026-05-21T04:58:47.939769Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T04:58:47.939769Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Improved Techniques for Training Consistency Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Prafulla Dhariwal, Yang Song","submitted_at":"2023-10-22T05:33:38Z","abstract_excerpt":"Consistency models are a nascent family of generative models that can sample high quality data in one step without the need for adversarial training. Current consistency models achieve optimal sample quality by distilling from pre-trained diffusion models and employing learned metrics such as LPIPS. However, distillation limits the quality of consistency models to that of the pre-trained diffusion model, and LPIPS causes undesirable bias in evaluation. To tackle these challenges, we present improved techniques for consistency training, where consistency models learn directly from data without "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.14189","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/2310.14189/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":"2310.14189","created_at":"2026-05-21T04:58:47.939849+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.14189v1","created_at":"2026-05-21T04:58:47.939849+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.14189","created_at":"2026-05-21T04:58:47.939849+00:00"},{"alias_kind":"pith_short_12","alias_value":"DZ7PMTNAIR7C","created_at":"2026-05-21T04:58:47.939849+00:00"},{"alias_kind":"pith_short_16","alias_value":"DZ7PMTNAIR7C6DKS","created_at":"2026-05-21T04:58:47.939849+00:00"},{"alias_kind":"pith_short_8","alias_value":"DZ7PMTNA","created_at":"2026-05-21T04:58:47.939849+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":26,"internal_anchor_count":26,"sample":[{"citing_arxiv_id":"2605.21489","citing_title":"Variance Reduction for Expectations with Diffusion Teachers","ref_index":72,"is_internal_anchor":true},{"citing_arxiv_id":"2605.23381","citing_title":"VDE: Training-Free Accelerating Rectified Flow Model via Velocity Decomposition and Estimation","ref_index":44,"is_internal_anchor":true},{"citing_arxiv_id":"2602.02214","citing_title":"Causal Forcing: Autoregressive Diffusion Distillation Done Right for High-Quality Real-Time Interactive Video Generation","ref_index":34,"is_internal_anchor":true},{"citing_arxiv_id":"2605.21981","citing_title":"RiT: Vanilla Diffusion Transformers Suffice in Representation Space","ref_index":29,"is_internal_anchor":true},{"citing_arxiv_id":"2602.02214","citing_title":"Causal Forcing: Autoregressive Diffusion Distillation Done Right for High-Quality Real-Time Interactive Video Generation","ref_index":34,"is_internal_anchor":true},{"citing_arxiv_id":"2605.21489","citing_title":"Variance Reduction for Expectations with Diffusion Teachers","ref_index":72,"is_internal_anchor":true},{"citing_arxiv_id":"2605.15592","citing_title":"Efficient Image Synthesis with Sphere Latent Encoder","ref_index":31,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17042","citing_title":"Thermal-Only Crowd Counting with Deployment-Time Privacy Protection","ref_index":53,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17899","citing_title":"DCFold: Efficient Protein Structure Generation with Single Forward Pass","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2605.16732","citing_title":"DiRotQ: Rotation-Aware Quantization for 4-bit Diffusion Transformers","ref_index":65,"is_internal_anchor":true},{"citing_arxiv_id":"2510.08431","citing_title":"Large Scale Diffusion Distillation via Score-Regularized Continuous-Time Consistency","ref_index":23,"is_internal_anchor":true},{"citing_arxiv_id":"2602.10764","citing_title":"Dual-End Consistency Model","ref_index":44,"is_internal_anchor":true},{"citing_arxiv_id":"2410.12557","citing_title":"One Step Diffusion via Shortcut Models","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2509.24527","citing_title":"Training Agents Inside of Scalable World Models","ref_index":78,"is_internal_anchor":true},{"citing_arxiv_id":"2503.00200","citing_title":"Unified Video Action Model","ref_index":39,"is_internal_anchor":true},{"citing_arxiv_id":"2605.11755","citing_title":"One-Step Generative Modeling via Wasserstein Gradient Flows","ref_index":55,"is_internal_anchor":true},{"citing_arxiv_id":"2604.25819","citing_title":"Mutual Forcing: Dual-Mode Self-Evolution for Fast Autoregressive Audio-Video Character Generation","ref_index":38,"is_internal_anchor":true},{"citing_arxiv_id":"2604.22379","citing_title":"Efficient Diffusion Distillation via Embedding Loss","ref_index":46,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05975","citing_title":"Physical Fidelity Reconstruction via Improved Consistency-Distilled Flow Matching for Dynamical Systems","ref_index":19,"is_internal_anchor":true},{"citing_arxiv_id":"2604.12273","citing_title":"SubFlow: Sub-mode Conditioned Flow Matching for Diverse One-Step Generation","ref_index":50,"is_internal_anchor":true},{"citing_arxiv_id":"2604.08837","citing_title":"Discrete Meanflow Training Curriculum","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2605.07020","citing_title":"FlashMol: High-Quality Molecule Generation in as Few as Four Steps","ref_index":32,"is_internal_anchor":true},{"citing_arxiv_id":"2605.07327","citing_title":"Teacher-Feature Drifting: One-Step Diffusion Distillation with Pretrained Diffusion Representations","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2604.08828","citing_title":"Post-Hoc Guidance for Consistency Models by Joint Flow Distribution Learning","ref_index":54,"is_internal_anchor":true},{"citing_arxiv_id":"2604.18168","citing_title":"Extending One-Step Image Generation from Class Labels to Text via Discriminative Text Representation","ref_index":23,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DZ7PMTNAIR7C6DKS3JMGN5JGMR","json":"https://pith.science/pith/DZ7PMTNAIR7C6DKS3JMGN5JGMR.json","graph_json":"https://pith.science/api/pith-number/DZ7PMTNAIR7C6DKS3JMGN5JGMR/graph.json","events_json":"https://pith.science/api/pith-number/DZ7PMTNAIR7C6DKS3JMGN5JGMR/events.json","paper":"https://pith.science/paper/DZ7PMTNA"},"agent_actions":{"view_html":"https://pith.science/pith/DZ7PMTNAIR7C6DKS3JMGN5JGMR","download_json":"https://pith.science/pith/DZ7PMTNAIR7C6DKS3JMGN5JGMR.json","view_paper":"https://pith.science/paper/DZ7PMTNA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.14189&json=true","fetch_graph":"https://pith.science/api/pith-number/DZ7PMTNAIR7C6DKS3JMGN5JGMR/graph.json","fetch_events":"https://pith.science/api/pith-number/DZ7PMTNAIR7C6DKS3JMGN5JGMR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DZ7PMTNAIR7C6DKS3JMGN5JGMR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DZ7PMTNAIR7C6DKS3JMGN5JGMR/action/storage_attestation","attest_author":"https://pith.science/pith/DZ7PMTNAIR7C6DKS3JMGN5JGMR/action/author_attestation","sign_citation":"https://pith.science/pith/DZ7PMTNAIR7C6DKS3JMGN5JGMR/action/citation_signature","submit_replication":"https://pith.science/pith/DZ7PMTNAIR7C6DKS3JMGN5JGMR/action/replication_record"}},"created_at":"2026-05-21T04:58:47.939849+00:00","updated_at":"2026-05-21T04:58:47.939849+00:00"}