{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:DQUCZZNNT6SUHBPVFFAFVU4MV2","short_pith_number":"pith:DQUCZZNN","schema_version":"1.0","canonical_sha256":"1c282ce5ad9fa54385f529405ad38cae96595d9d01b538040611d202a277ccb6","source":{"kind":"arxiv","id":"2607.21606","version":1},"attestation_state":"computed","paper":{"title":"TILT: Improving Compositional Generation in Diffusion Models with a Model-Intrinsic Reward","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Debottam Dutta, Jaehoon Hahm, Jianchong Chen, Romit Roy Choudhury","submitted_at":"2026-05-16T16:56:52Z","abstract_excerpt":"Recent advances in powerful text-to-image generation models have made it increasingly important to develop test-time methods that modify the sampling trajectory to produce images more faithful to complex compositional prompts. We present TILT, a training-free framework for compositional text-to-image generation via test-time reward alignment. We interpret compositional failures as overlap modes between joint and single-concept distributions, and define a reward that favors samples where all concepts are jointly present. This reward is intrinsic to the base model and does not require any extern"},"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":"2607.21606","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-05-16T16:56:52Z","cross_cats_sorted":[],"title_canon_sha256":"58f199f3fea5d36c41200043c4c989e83bc8e1ea2d04bedcb05f34ee4b0853ae","abstract_canon_sha256":"72e4e0e0b105cd47866cda57602e6d4e74e8beaf4354ff8fdd06da2cd3c57b65"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-27T00:20:10.658280Z","signature_b64":"KocoTKJ/5UoSFgsUI0Ta9uWMPCacwzgzik0uFUodLUAWcjy/A7ZoNnjFknwaGie87Cboj1KKHpFv5Wt2EXDeDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1c282ce5ad9fa54385f529405ad38cae96595d9d01b538040611d202a277ccb6","last_reissued_at":"2026-07-27T00:20:10.657350Z","signature_status":"signed_v1","first_computed_at":"2026-07-27T00:20:10.657350Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TILT: Improving Compositional Generation in Diffusion Models with a Model-Intrinsic Reward","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Debottam Dutta, Jaehoon Hahm, Jianchong Chen, Romit Roy Choudhury","submitted_at":"2026-05-16T16:56:52Z","abstract_excerpt":"Recent advances in powerful text-to-image generation models have made it increasingly important to develop test-time methods that modify the sampling trajectory to produce images more faithful to complex compositional prompts. We present TILT, a training-free framework for compositional text-to-image generation via test-time reward alignment. We interpret compositional failures as overlap modes between joint and single-concept distributions, and define a reward that favors samples where all concepts are jointly present. This reward is intrinsic to the base model and does not require any extern"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.21606","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/2607.21606/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":"2607.21606","created_at":"2026-07-27T00:20:10.657802+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.21606v1","created_at":"2026-07-27T00:20:10.657802+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.21606","created_at":"2026-07-27T00:20:10.657802+00:00"},{"alias_kind":"pith_short_12","alias_value":"DQUCZZNNT6SU","created_at":"2026-07-27T00:20:10.657802+00:00"},{"alias_kind":"pith_short_16","alias_value":"DQUCZZNNT6SUHBPV","created_at":"2026-07-27T00:20:10.657802+00:00"},{"alias_kind":"pith_short_8","alias_value":"DQUCZZNN","created_at":"2026-07-27T00:20:10.657802+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DQUCZZNNT6SUHBPVFFAFVU4MV2","json":"https://pith.science/pith/DQUCZZNNT6SUHBPVFFAFVU4MV2.json","graph_json":"https://pith.science/api/pith-number/DQUCZZNNT6SUHBPVFFAFVU4MV2/graph.json","events_json":"https://pith.science/api/pith-number/DQUCZZNNT6SUHBPVFFAFVU4MV2/events.json","paper":"https://pith.science/paper/DQUCZZNN"},"agent_actions":{"view_html":"https://pith.science/pith/DQUCZZNNT6SUHBPVFFAFVU4MV2","download_json":"https://pith.science/pith/DQUCZZNNT6SUHBPVFFAFVU4MV2.json","view_paper":"https://pith.science/paper/DQUCZZNN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.21606&json=true","fetch_graph":"https://pith.science/api/pith-number/DQUCZZNNT6SUHBPVFFAFVU4MV2/graph.json","fetch_events":"https://pith.science/api/pith-number/DQUCZZNNT6SUHBPVFFAFVU4MV2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DQUCZZNNT6SUHBPVFFAFVU4MV2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DQUCZZNNT6SUHBPVFFAFVU4MV2/action/storage_attestation","attest_author":"https://pith.science/pith/DQUCZZNNT6SUHBPVFFAFVU4MV2/action/author_attestation","sign_citation":"https://pith.science/pith/DQUCZZNNT6SUHBPVFFAFVU4MV2/action/citation_signature","submit_replication":"https://pith.science/pith/DQUCZZNNT6SUHBPVFFAFVU4MV2/action/replication_record"}},"created_at":"2026-07-27T00:20:10.657802+00:00","updated_at":"2026-07-27T00:20:10.657802+00:00"}