{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:KNKL2ZL2OAWA434SQ4TUWHMWXE","short_pith_number":"pith:KNKL2ZL2","schema_version":"1.0","canonical_sha256":"5354bd657a702c0e6f9287274b1d96b92d66306dbaa653d850cbc27b44d53b3a","source":{"kind":"arxiv","id":"2508.15773","version":1},"attestation_state":"computed","paper":{"title":"Scaling Group Inference for Diverse and High-Quality Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.GR","cs.LG"],"primary_cat":"cs.CV","authors_text":"Daniil Ostashev, Gaurav Parmar, Jun-Yan Zhu, Kfir Aberman, Kuan-Chieh Wang, Or Patashnik, Srinivasa Narasimhan","submitted_at":"2025-08-21T17:59:57Z","abstract_excerpt":"Generative models typically sample outputs independently, and recent inference-time guidance and scaling algorithms focus on improving the quality of individual samples. However, in real-world applications, users are often presented with a set of multiple images (e.g., 4-8) for each prompt, where independent sampling tends to lead to redundant results, limiting user choices and hindering idea exploration. In this work, we introduce a scalable group inference method that improves both the diversity and quality of a group of samples. We formulate group inference as a quadratic integer assignment"},"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":"2508.15773","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-08-21T17:59:57Z","cross_cats_sorted":["cs.GR","cs.LG"],"title_canon_sha256":"7aed1608ecfe544157fef94eed442273aefccd17fd8834cb7f47d3a1cae1066c","abstract_canon_sha256":"cabff91ac90d4c8077d5135f1913f3f001f0b5815f6c382a31baf6c91aeb6490"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:57:18.624537Z","signature_b64":"vFYQp1dEcQp5xc+P3trG/wQKW+1WbEVdHEckBlBuX7xfelv0Xuh0IeziLHPsXtj00tMCA3a+RkqhIN9zHLMvAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5354bd657a702c0e6f9287274b1d96b92d66306dbaa653d850cbc27b44d53b3a","last_reissued_at":"2026-07-05T11:57:18.624083Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:57:18.624083Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scaling Group Inference for Diverse and High-Quality Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.GR","cs.LG"],"primary_cat":"cs.CV","authors_text":"Daniil Ostashev, Gaurav Parmar, Jun-Yan Zhu, Kfir Aberman, Kuan-Chieh Wang, Or Patashnik, Srinivasa Narasimhan","submitted_at":"2025-08-21T17:59:57Z","abstract_excerpt":"Generative models typically sample outputs independently, and recent inference-time guidance and scaling algorithms focus on improving the quality of individual samples. However, in real-world applications, users are often presented with a set of multiple images (e.g., 4-8) for each prompt, where independent sampling tends to lead to redundant results, limiting user choices and hindering idea exploration. In this work, we introduce a scalable group inference method that improves both the diversity and quality of a group of samples. We formulate group inference as a quadratic integer assignment"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.15773","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/2508.15773/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":"2508.15773","created_at":"2026-07-05T11:57:18.624139+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.15773v1","created_at":"2026-07-05T11:57:18.624139+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.15773","created_at":"2026-07-05T11:57:18.624139+00:00"},{"alias_kind":"pith_short_12","alias_value":"KNKL2ZL2OAWA","created_at":"2026-07-05T11:57:18.624139+00:00"},{"alias_kind":"pith_short_16","alias_value":"KNKL2ZL2OAWA434S","created_at":"2026-07-05T11:57:18.624139+00:00"},{"alias_kind":"pith_short_8","alias_value":"KNKL2ZL2","created_at":"2026-07-05T11:57:18.624139+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.27371","citing_title":"Don't Settle at the Mode! Mitigating Diversity Collapse in Pretrained Flow Models via Feature Self-Guidance","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2606.23679","citing_title":"Semantic Browsing: Controllable Diversity for Image Generation","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23178","citing_title":"Composing People Together: Iterative Pose-Image Generation for Multi-Person Interaction Scenes","ref_index":54,"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":47,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11494","citing_title":"STRIDE: Training-Free Diversity Guidance via PCA-Directed Feature Perturbation in Single-Step Diffusion Models","ref_index":30,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KNKL2ZL2OAWA434SQ4TUWHMWXE","json":"https://pith.science/pith/KNKL2ZL2OAWA434SQ4TUWHMWXE.json","graph_json":"https://pith.science/api/pith-number/KNKL2ZL2OAWA434SQ4TUWHMWXE/graph.json","events_json":"https://pith.science/api/pith-number/KNKL2ZL2OAWA434SQ4TUWHMWXE/events.json","paper":"https://pith.science/paper/KNKL2ZL2"},"agent_actions":{"view_html":"https://pith.science/pith/KNKL2ZL2OAWA434SQ4TUWHMWXE","download_json":"https://pith.science/pith/KNKL2ZL2OAWA434SQ4TUWHMWXE.json","view_paper":"https://pith.science/paper/KNKL2ZL2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.15773&json=true","fetch_graph":"https://pith.science/api/pith-number/KNKL2ZL2OAWA434SQ4TUWHMWXE/graph.json","fetch_events":"https://pith.science/api/pith-number/KNKL2ZL2OAWA434SQ4TUWHMWXE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KNKL2ZL2OAWA434SQ4TUWHMWXE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KNKL2ZL2OAWA434SQ4TUWHMWXE/action/storage_attestation","attest_author":"https://pith.science/pith/KNKL2ZL2OAWA434SQ4TUWHMWXE/action/author_attestation","sign_citation":"https://pith.science/pith/KNKL2ZL2OAWA434SQ4TUWHMWXE/action/citation_signature","submit_replication":"https://pith.science/pith/KNKL2ZL2OAWA434SQ4TUWHMWXE/action/replication_record"}},"created_at":"2026-07-05T11:57:18.624139+00:00","updated_at":"2026-07-05T11:57:18.624139+00:00"}