{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:NJCUCNBDDQMHOBU6ZR2EULIRJD","short_pith_number":"pith:NJCUCNBD","schema_version":"1.0","canonical_sha256":"6a454134231c1877069ecc744a2d1148f5618831ea145d969e75227e09d9dbe1","source":{"kind":"arxiv","id":"2607.05381","version":1},"attestation_state":"computed","paper":{"title":"What Does a Discrete Diffusion Model Learn?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.IT","math.IT","stat.ML"],"primary_cat":"cs.LG","authors_text":"Aran Raoufi, Bernhard Sch\\\"olkopf, Rodrigo Casado Noguerales, Thomas Hofmann","submitted_at":"2026-07-06T17:56:11Z","abstract_excerpt":"What does a discrete diffusion model learn: a denoiser, a score ratio, or a bridge plug-in predictor? At the level of jump rates, these are one object in different coordinates, and reading a neural network in the wrong coordinate changes the process being trained and sampled. Starting with a rigorous derivation of the continuous-time Markov chain (CTMC) ELBO for any noising process, boundary terms included, we prove the \\emph{Oracle Distance} theorem: the negative ELBO is exactly equal to the data entropy plus the path KL from the oracle reverse process to the learned one, not merely a bound. "},"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":"2607.05381","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-06T17:56:11Z","cross_cats_sorted":["cs.AI","cs.CL","cs.IT","math.IT","stat.ML"],"title_canon_sha256":"d6fbda457cea46b097edd9556afcb18b89974c9f1d1782c4cb61a146a07c08aa","abstract_canon_sha256":"6ee6ca81a80f420948d153559f638917d8c2a2d4efe80777e81ca5be2502939b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T03:19:25.518415Z","signature_b64":"5NIdPK+VgTdZjDcRDx/Y/xjl4JxCnBUa83EBsSsL0sLKADmP2aIjGPq80YDm7GET6UMVM4PcxiksHE1sp0GeAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6a454134231c1877069ecc744a2d1148f5618831ea145d969e75227e09d9dbe1","last_reissued_at":"2026-07-07T03:19:25.517375Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T03:19:25.517375Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What Does a Discrete Diffusion Model Learn?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.IT","math.IT","stat.ML"],"primary_cat":"cs.LG","authors_text":"Aran Raoufi, Bernhard Sch\\\"olkopf, Rodrigo Casado Noguerales, Thomas Hofmann","submitted_at":"2026-07-06T17:56:11Z","abstract_excerpt":"What does a discrete diffusion model learn: a denoiser, a score ratio, or a bridge plug-in predictor? At the level of jump rates, these are one object in different coordinates, and reading a neural network in the wrong coordinate changes the process being trained and sampled. Starting with a rigorous derivation of the continuous-time Markov chain (CTMC) ELBO for any noising process, boundary terms included, we prove the \\emph{Oracle Distance} theorem: the negative ELBO is exactly equal to the data entropy plus the path KL from the oracle reverse process to the learned one, not merely a bound. "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.05381","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/2607.05381/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":"2607.05381","created_at":"2026-07-07T03:19:25.517506+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.05381v1","created_at":"2026-07-07T03:19:25.517506+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.05381","created_at":"2026-07-07T03:19:25.517506+00:00"},{"alias_kind":"pith_short_12","alias_value":"NJCUCNBDDQMH","created_at":"2026-07-07T03:19:25.517506+00:00"},{"alias_kind":"pith_short_16","alias_value":"NJCUCNBDDQMHOBU6","created_at":"2026-07-07T03:19:25.517506+00:00"},{"alias_kind":"pith_short_8","alias_value":"NJCUCNBD","created_at":"2026-07-07T03:19:25.517506+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/NJCUCNBDDQMHOBU6ZR2EULIRJD","json":"https://pith.science/pith/NJCUCNBDDQMHOBU6ZR2EULIRJD.json","graph_json":"https://pith.science/api/pith-number/NJCUCNBDDQMHOBU6ZR2EULIRJD/graph.json","events_json":"https://pith.science/api/pith-number/NJCUCNBDDQMHOBU6ZR2EULIRJD/events.json","paper":"https://pith.science/paper/NJCUCNBD"},"agent_actions":{"view_html":"https://pith.science/pith/NJCUCNBDDQMHOBU6ZR2EULIRJD","download_json":"https://pith.science/pith/NJCUCNBDDQMHOBU6ZR2EULIRJD.json","view_paper":"https://pith.science/paper/NJCUCNBD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.05381&json=true","fetch_graph":"https://pith.science/api/pith-number/NJCUCNBDDQMHOBU6ZR2EULIRJD/graph.json","fetch_events":"https://pith.science/api/pith-number/NJCUCNBDDQMHOBU6ZR2EULIRJD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NJCUCNBDDQMHOBU6ZR2EULIRJD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NJCUCNBDDQMHOBU6ZR2EULIRJD/action/storage_attestation","attest_author":"https://pith.science/pith/NJCUCNBDDQMHOBU6ZR2EULIRJD/action/author_attestation","sign_citation":"https://pith.science/pith/NJCUCNBDDQMHOBU6ZR2EULIRJD/action/citation_signature","submit_replication":"https://pith.science/pith/NJCUCNBDDQMHOBU6ZR2EULIRJD/action/replication_record"}},"created_at":"2026-07-07T03:19:25.517506+00:00","updated_at":"2026-07-07T03:19:25.517506+00:00"}