{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:EU4SD4MMLBGF4IUUYELQDLV7WJ","short_pith_number":"pith:EU4SD4MM","schema_version":"1.0","canonical_sha256":"253921f18c584c5e2294c11701aebfb27679af55a5e3f1b6a5f578add643f5b1","source":{"kind":"arxiv","id":"2406.08070","version":2},"attestation_state":"computed","paper":{"title":"CFG++: Manifold-constrained Classifier Free Guidance for Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Geon Yeong Park, Hyelin Nam, Hyungjin Chung, Jeongsol Kim, Jong Chul Ye","submitted_at":"2024-06-12T10:40:10Z","abstract_excerpt":"Classifier-free guidance (CFG) is a fundamental tool in modern diffusion models for text-guided generation. Although effective, CFG has notable drawbacks. For instance, DDIM with CFG lacks invertibility, complicating image editing; furthermore, high guidance scales, essential for high-quality outputs, frequently result in issues like mode collapse. Contrary to the widespread belief that these are inherent limitations of diffusion models, this paper reveals that the problems actually stem from the off-manifold phenomenon associated with CFG, rather than the diffusion models themselves. More spe"},"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":"2406.08070","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-06-12T10:40:10Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"01b220c9bfe313db8c895f84535cb3ea9ba23c4c7f3de008624788e9333212d2","abstract_canon_sha256":"b6d8d766b7e185fabff6ff330236a90ec54057a741a8b22e1b743e914183be76"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:06:06.235200Z","signature_b64":"paj+nuLXb84r53JZ+w8x59OcNk1kgZfPEA7ir7i7sq7xk5ztfJJ4hbcASF0pModivPOps2UShgdd8R7/mBcABg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"253921f18c584c5e2294c11701aebfb27679af55a5e3f1b6a5f578add643f5b1","last_reissued_at":"2026-07-05T09:06:06.234759Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:06:06.234759Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CFG++: Manifold-constrained Classifier Free Guidance for Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Geon Yeong Park, Hyelin Nam, Hyungjin Chung, Jeongsol Kim, Jong Chul Ye","submitted_at":"2024-06-12T10:40:10Z","abstract_excerpt":"Classifier-free guidance (CFG) is a fundamental tool in modern diffusion models for text-guided generation. Although effective, CFG has notable drawbacks. For instance, DDIM with CFG lacks invertibility, complicating image editing; furthermore, high guidance scales, essential for high-quality outputs, frequently result in issues like mode collapse. Contrary to the widespread belief that these are inherent limitations of diffusion models, this paper reveals that the problems actually stem from the off-manifold phenomenon associated with CFG, rather than the diffusion models themselves. More spe"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.08070","kind":"arxiv","version":2},"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/2406.08070/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":"2406.08070","created_at":"2026-07-05T09:06:06.234821+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.08070v2","created_at":"2026-07-05T09:06:06.234821+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.08070","created_at":"2026-07-05T09:06:06.234821+00:00"},{"alias_kind":"pith_short_12","alias_value":"EU4SD4MMLBGF","created_at":"2026-07-05T09:06:06.234821+00:00"},{"alias_kind":"pith_short_16","alias_value":"EU4SD4MMLBGF4IUU","created_at":"2026-07-05T09:06:06.234821+00:00"},{"alias_kind":"pith_short_8","alias_value":"EU4SD4MM","created_at":"2026-07-05T09:06:06.234821+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":17,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.07665","citing_title":"Guidance Breaks the Fitted Operator: A Terminal-Fitted Repair for Classifier-Free Guidance","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2605.27919","citing_title":"Frequency-Guided Action Diffusion via Sub-Frequency Manifold Traversal","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22717","citing_title":"Live Music Diffusion Models: Efficient Fine-Tuning and Post-Training of Interactive Diffusion Music Generators","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2603.07561","citing_title":"PureCC: Pure Learning for Text-to-Image Concept Customization","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15803","citing_title":"Embedding-perturbed Exploration Preference Optimization for Flow Models","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20079","citing_title":"Probability-Conserving Flow Guidance","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2506.01666","citing_title":"Synthesis of discrete-continuous quantum circuits with multimodal diffusion models","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2506.14399","citing_title":"Factored Classifier-Free Guidance","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2601.00090","citing_title":"It's Never Too Late: Noise Optimization for Collapse Recovery in Trained Diffusion Models","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2603.06165","citing_title":"Reflective Flow Sampling Enhancement","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2604.02973","citing_title":"Exploring Motion-Language Alignment for Text-driven Motion Generation","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2604.26503","citing_title":"Delta Score Matters! Spatial Adaptive Multi Guidance in Diffusion Models","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2604.25289","citing_title":"Exploring Time Conditioning in Diffusion Generative Models from Disjoint Noisy Data Manifolds","ref_index":65,"is_internal_anchor":false},{"citing_arxiv_id":"2604.24238","citing_title":"GeoEdit: Local Frames for Fast, Training-Free On-Manifold Editing in Diffusion Models","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2604.23536","citing_title":"$Z^2$-Sampling: Zero-Cost Zigzag Trajectories for Semantic Alignment in Diffusion Models","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06124","citing_title":"P-Guide: Parameter-Efficient Prior Steering for Single-Pass CFG Inference","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06512","citing_title":"DCR: Counterfactual Attractor Guidance for Rare Compositional Generation","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/EU4SD4MMLBGF4IUUYELQDLV7WJ","json":"https://pith.science/pith/EU4SD4MMLBGF4IUUYELQDLV7WJ.json","graph_json":"https://pith.science/api/pith-number/EU4SD4MMLBGF4IUUYELQDLV7WJ/graph.json","events_json":"https://pith.science/api/pith-number/EU4SD4MMLBGF4IUUYELQDLV7WJ/events.json","paper":"https://pith.science/paper/EU4SD4MM"},"agent_actions":{"view_html":"https://pith.science/pith/EU4SD4MMLBGF4IUUYELQDLV7WJ","download_json":"https://pith.science/pith/EU4SD4MMLBGF4IUUYELQDLV7WJ.json","view_paper":"https://pith.science/paper/EU4SD4MM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.08070&json=true","fetch_graph":"https://pith.science/api/pith-number/EU4SD4MMLBGF4IUUYELQDLV7WJ/graph.json","fetch_events":"https://pith.science/api/pith-number/EU4SD4MMLBGF4IUUYELQDLV7WJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EU4SD4MMLBGF4IUUYELQDLV7WJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EU4SD4MMLBGF4IUUYELQDLV7WJ/action/storage_attestation","attest_author":"https://pith.science/pith/EU4SD4MMLBGF4IUUYELQDLV7WJ/action/author_attestation","sign_citation":"https://pith.science/pith/EU4SD4MMLBGF4IUUYELQDLV7WJ/action/citation_signature","submit_replication":"https://pith.science/pith/EU4SD4MMLBGF4IUUYELQDLV7WJ/action/replication_record"}},"created_at":"2026-07-05T09:06:06.234821+00:00","updated_at":"2026-07-05T09:06:06.234821+00:00"}