{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:AUXRPQCYK7T7ZCEVDVVJQSPEUO","short_pith_number":"pith:AUXRPQCY","schema_version":"1.0","canonical_sha256":"052f17c05857e7fc88951d6a9849e4a3bcf386aaa01a22fa5392869d8ddd839c","source":{"kind":"arxiv","id":"1908.07181","version":5},"attestation_state":"computed","paper":{"title":"Latent-Variable Non-Autoregressive Neural Machine Translation with Deterministic Inference Using a Delta Posterior","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Hideki Nakayama, Jason Lee, Kyunghyun Cho, Raphael Shu","submitted_at":"2019-08-20T06:14:18Z","abstract_excerpt":"Although neural machine translation models reached high translation quality, the autoregressive nature makes inference difficult to parallelize and leads to high translation latency. Inspired by recent refinement-based approaches, we propose LaNMT, a latent-variable non-autoregressive model with continuous latent variables and deterministic inference procedure. In contrast to existing approaches, we use a deterministic inference algorithm to find the target sequence that maximizes the lowerbound to the log-probability. During inference, the length of translation automatically adapts itself. Ou"},"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":"1908.07181","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-08-20T06:14:18Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"644d20ef22191d552aec34023ad655f1578de86dbe31a95600917e2e1dc2caf6","abstract_canon_sha256":"0ef0ee7c4ea225a480da2666ef4fe66fe2f4bbb00cf222654fb7106e086353d5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:20:54.323617Z","signature_b64":"GiwVama6Vv603lRnp11I94QS8qZGVKCVXNjIljTxkrhdl/d+qGRxhJhuQ/7a/DfkS6oWLPXypaOGzBvqAZXXDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"052f17c05857e7fc88951d6a9849e4a3bcf386aaa01a22fa5392869d8ddd839c","last_reissued_at":"2026-07-05T00:20:54.323043Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:20:54.323043Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Latent-Variable Non-Autoregressive Neural Machine Translation with Deterministic Inference Using a Delta Posterior","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Hideki Nakayama, Jason Lee, Kyunghyun Cho, Raphael Shu","submitted_at":"2019-08-20T06:14:18Z","abstract_excerpt":"Although neural machine translation models reached high translation quality, the autoregressive nature makes inference difficult to parallelize and leads to high translation latency. Inspired by recent refinement-based approaches, we propose LaNMT, a latent-variable non-autoregressive model with continuous latent variables and deterministic inference procedure. In contrast to existing approaches, we use a deterministic inference algorithm to find the target sequence that maximizes the lowerbound to the log-probability. During inference, the length of translation automatically adapts itself. Ou"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1908.07181","kind":"arxiv","version":5},"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/1908.07181/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":"1908.07181","created_at":"2026-07-05T00:20:54.323101+00:00"},{"alias_kind":"arxiv_version","alias_value":"1908.07181v5","created_at":"2026-07-05T00:20:54.323101+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1908.07181","created_at":"2026-07-05T00:20:54.323101+00:00"},{"alias_kind":"pith_short_12","alias_value":"AUXRPQCYK7T7","created_at":"2026-07-05T00:20:54.323101+00:00"},{"alias_kind":"pith_short_16","alias_value":"AUXRPQCYK7T7ZCEV","created_at":"2026-07-05T00:20:54.323101+00:00"},{"alias_kind":"pith_short_8","alias_value":"AUXRPQCY","created_at":"2026-07-05T00:20:54.323101+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"1909.02480","citing_title":"FlowSeq: Non-Autoregressive Conditional Sequence Generation with Generative Flow","ref_index":31,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AUXRPQCYK7T7ZCEVDVVJQSPEUO","json":"https://pith.science/pith/AUXRPQCYK7T7ZCEVDVVJQSPEUO.json","graph_json":"https://pith.science/api/pith-number/AUXRPQCYK7T7ZCEVDVVJQSPEUO/graph.json","events_json":"https://pith.science/api/pith-number/AUXRPQCYK7T7ZCEVDVVJQSPEUO/events.json","paper":"https://pith.science/paper/AUXRPQCY"},"agent_actions":{"view_html":"https://pith.science/pith/AUXRPQCYK7T7ZCEVDVVJQSPEUO","download_json":"https://pith.science/pith/AUXRPQCYK7T7ZCEVDVVJQSPEUO.json","view_paper":"https://pith.science/paper/AUXRPQCY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1908.07181&json=true","fetch_graph":"https://pith.science/api/pith-number/AUXRPQCYK7T7ZCEVDVVJQSPEUO/graph.json","fetch_events":"https://pith.science/api/pith-number/AUXRPQCYK7T7ZCEVDVVJQSPEUO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AUXRPQCYK7T7ZCEVDVVJQSPEUO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AUXRPQCYK7T7ZCEVDVVJQSPEUO/action/storage_attestation","attest_author":"https://pith.science/pith/AUXRPQCYK7T7ZCEVDVVJQSPEUO/action/author_attestation","sign_citation":"https://pith.science/pith/AUXRPQCYK7T7ZCEVDVVJQSPEUO/action/citation_signature","submit_replication":"https://pith.science/pith/AUXRPQCYK7T7ZCEVDVVJQSPEUO/action/replication_record"}},"created_at":"2026-07-05T00:20:54.323101+00:00","updated_at":"2026-07-05T00:20:54.323101+00:00"}