{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:XUAUYIG62CRBK2MCFFJQVHUQZZ","short_pith_number":"pith:XUAUYIG6","schema_version":"1.0","canonical_sha256":"bd014c20ded0a215698229530a9e90ce5bc1b19305ca19b450789b0a09b9055a","source":{"kind":"arxiv","id":"2512.08505","version":2},"attestation_state":"computed","paper":{"title":"Early Estimation of Language to Latent Alignment in Diffusion Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Idan Szpektor, Joao Magalhaes, Regev Cohen, Vasco Ramos","submitted_at":"2025-12-09T11:45:01Z","abstract_excerpt":"Conditional diffusion models frequently suffer from language-image misalignments. Due to the ambiguity of intermediate noise corrupted latents, assessing prompt adherence currently requires completing the entire sampling trajectory. This late-stage evaluation incurs even higher computational costs during test-time scaling strategies, such as Best-of-N (BoN) sampling, as all misaligned trajectories must finish generation before being discarded. To tackle this, we propose NoisyCLIP, a noise-aware twin-tower model that enables early language-to-latent alignment estimation. By learning a vision en"},"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":"2512.08505","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-12-09T11:45:01Z","cross_cats_sorted":[],"title_canon_sha256":"7788f4672ef518c14abcb9567b42e65de56975fb51e2bfc8006f5d78071d59db","abstract_canon_sha256":"0c886d0d0035f6322f0353d2b9fdcc0c34e9e467bd92b77f6a9af2476da4103b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-30T01:17:30.574272Z","signature_b64":"aQm+FEFs8Znj4lLuybqVdsdw369p2xQ1+S47f+iCmAK8pVy7zNdUPxqamcDqz93muz5M3ZhNGS46VgGRVJDcAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bd014c20ded0a215698229530a9e90ce5bc1b19305ca19b450789b0a09b9055a","last_reissued_at":"2026-06-30T01:17:30.573604Z","signature_status":"signed_v1","first_computed_at":"2026-06-30T01:17:30.573604Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Early Estimation of Language to Latent Alignment in Diffusion Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Idan Szpektor, Joao Magalhaes, Regev Cohen, Vasco Ramos","submitted_at":"2025-12-09T11:45:01Z","abstract_excerpt":"Conditional diffusion models frequently suffer from language-image misalignments. Due to the ambiguity of intermediate noise corrupted latents, assessing prompt adherence currently requires completing the entire sampling trajectory. This late-stage evaluation incurs even higher computational costs during test-time scaling strategies, such as Best-of-N (BoN) sampling, as all misaligned trajectories must finish generation before being discarded. To tackle this, we propose NoisyCLIP, a noise-aware twin-tower model that enables early language-to-latent alignment estimation. By learning a vision en"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.08505","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/2512.08505/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":"2512.08505","created_at":"2026-06-30T01:17:30.573689+00:00"},{"alias_kind":"arxiv_version","alias_value":"2512.08505v2","created_at":"2026-06-30T01:17:30.573689+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.08505","created_at":"2026-06-30T01:17:30.573689+00:00"},{"alias_kind":"pith_short_12","alias_value":"XUAUYIG62CRB","created_at":"2026-06-30T01:17:30.573689+00:00"},{"alias_kind":"pith_short_16","alias_value":"XUAUYIG62CRBK2MC","created_at":"2026-06-30T01:17:30.573689+00:00"},{"alias_kind":"pith_short_8","alias_value":"XUAUYIG6","created_at":"2026-06-30T01:17:30.573689+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2606.09601","citing_title":"Assessing Sample Quality in Conditional Generation under Compositional Shift","ref_index":27,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XUAUYIG62CRBK2MCFFJQVHUQZZ","json":"https://pith.science/pith/XUAUYIG62CRBK2MCFFJQVHUQZZ.json","graph_json":"https://pith.science/api/pith-number/XUAUYIG62CRBK2MCFFJQVHUQZZ/graph.json","events_json":"https://pith.science/api/pith-number/XUAUYIG62CRBK2MCFFJQVHUQZZ/events.json","paper":"https://pith.science/paper/XUAUYIG6"},"agent_actions":{"view_html":"https://pith.science/pith/XUAUYIG62CRBK2MCFFJQVHUQZZ","download_json":"https://pith.science/pith/XUAUYIG62CRBK2MCFFJQVHUQZZ.json","view_paper":"https://pith.science/paper/XUAUYIG6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2512.08505&json=true","fetch_graph":"https://pith.science/api/pith-number/XUAUYIG62CRBK2MCFFJQVHUQZZ/graph.json","fetch_events":"https://pith.science/api/pith-number/XUAUYIG62CRBK2MCFFJQVHUQZZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XUAUYIG62CRBK2MCFFJQVHUQZZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XUAUYIG62CRBK2MCFFJQVHUQZZ/action/storage_attestation","attest_author":"https://pith.science/pith/XUAUYIG62CRBK2MCFFJQVHUQZZ/action/author_attestation","sign_citation":"https://pith.science/pith/XUAUYIG62CRBK2MCFFJQVHUQZZ/action/citation_signature","submit_replication":"https://pith.science/pith/XUAUYIG62CRBK2MCFFJQVHUQZZ/action/replication_record"}},"created_at":"2026-06-30T01:17:30.573689+00:00","updated_at":"2026-06-30T01:17:30.573689+00:00"}