{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:A5ZWIUPYMMSPAQN26C5NLHXP5I","short_pith_number":"pith:A5ZWIUPY","schema_version":"1.0","canonical_sha256":"07736451f86324f041baf0bad59eefea3cd67c3d372836001d0dd21b2dda8841","source":{"kind":"arxiv","id":"1906.01604","version":1},"attestation_state":"computed","paper":{"title":"KERMIT: Generative Insertion-Based Modeling for Sequences","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","stat.ML"],"primary_cat":"cs.CL","authors_text":"Jakob Uszkoreit, Kelvin Guu, Mitchell Stern, Nikita Kitaev, William Chan","submitted_at":"2019-06-04T17:35:35Z","abstract_excerpt":"We present KERMIT, a simple insertion-based approach to generative modeling for sequences and sequence pairs. KERMIT models the joint distribution and its decompositions (i.e., marginals and conditionals) using a single neural network and, unlike much prior work, does not rely on a prespecified factorization of the data distribution. During training, one can feed KERMIT paired data $(x, y)$ to learn the joint distribution $p(x, y)$, and optionally mix in unpaired data $x$ or $y$ to refine the marginals $p(x)$ or $p(y)$. During inference, we have access to the conditionals $p(x \\mid y)$ and $p("},"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":"1906.01604","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-06-04T17:35:35Z","cross_cats_sorted":["cs.LG","stat.ML"],"title_canon_sha256":"cb49f33186661eeb0060e706c200add5e750a55d8be60d763ef2e054686f01f1","abstract_canon_sha256":"e3bb00b597506197bcf4d58a83e6ae96a8f3aa3ee43ae286297a0dec940716fc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:44:15.985410Z","signature_b64":"0hWNmDicQcNN0DM/ZxjBBex8Y4HLbGqNUstdabmIy6/k8Zdzuh/0pdj7dfDzP05JQe4dWRpSo5/CoeCgsKUzCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"07736451f86324f041baf0bad59eefea3cd67c3d372836001d0dd21b2dda8841","last_reissued_at":"2026-05-17T23:44:15.984825Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:44:15.984825Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"KERMIT: Generative Insertion-Based Modeling for Sequences","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","stat.ML"],"primary_cat":"cs.CL","authors_text":"Jakob Uszkoreit, Kelvin Guu, Mitchell Stern, Nikita Kitaev, William Chan","submitted_at":"2019-06-04T17:35:35Z","abstract_excerpt":"We present KERMIT, a simple insertion-based approach to generative modeling for sequences and sequence pairs. KERMIT models the joint distribution and its decompositions (i.e., marginals and conditionals) using a single neural network and, unlike much prior work, does not rely on a prespecified factorization of the data distribution. During training, one can feed KERMIT paired data $(x, y)$ to learn the joint distribution $p(x, y)$, and optionally mix in unpaired data $x$ or $y$ to refine the marginals $p(x)$ or $p(y)$. During inference, we have access to the conditionals $p(x \\mid y)$ and $p("},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1906.01604","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":""},"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":"1906.01604","created_at":"2026-05-17T23:44:15.984909+00:00"},{"alias_kind":"arxiv_version","alias_value":"1906.01604v1","created_at":"2026-05-17T23:44:15.984909+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1906.01604","created_at":"2026-05-17T23:44:15.984909+00:00"},{"alias_kind":"pith_short_12","alias_value":"A5ZWIUPYMMSP","created_at":"2026-05-18T12:33:12.712433+00:00"},{"alias_kind":"pith_short_16","alias_value":"A5ZWIUPYMMSPAQN2","created_at":"2026-05-18T12:33:12.712433+00:00"},{"alias_kind":"pith_short_8","alias_value":"A5ZWIUPY","created_at":"2026-05-18T12:33:12.712433+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2207.14255","citing_title":"Efficient Training of Language Models to Fill in the Middle","ref_index":99,"is_internal_anchor":true},{"citing_arxiv_id":"2204.05999","citing_title":"InCoder: A Generative Model for Code Infilling and Synthesis","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"1907.11692","citing_title":"RoBERTa: A Robustly Optimized BERT Pretraining Approach","ref_index":6,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/A5ZWIUPYMMSPAQN26C5NLHXP5I","json":"https://pith.science/pith/A5ZWIUPYMMSPAQN26C5NLHXP5I.json","graph_json":"https://pith.science/api/pith-number/A5ZWIUPYMMSPAQN26C5NLHXP5I/graph.json","events_json":"https://pith.science/api/pith-number/A5ZWIUPYMMSPAQN26C5NLHXP5I/events.json","paper":"https://pith.science/paper/A5ZWIUPY"},"agent_actions":{"view_html":"https://pith.science/pith/A5ZWIUPYMMSPAQN26C5NLHXP5I","download_json":"https://pith.science/pith/A5ZWIUPYMMSPAQN26C5NLHXP5I.json","view_paper":"https://pith.science/paper/A5ZWIUPY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1906.01604&json=true","fetch_graph":"https://pith.science/api/pith-number/A5ZWIUPYMMSPAQN26C5NLHXP5I/graph.json","fetch_events":"https://pith.science/api/pith-number/A5ZWIUPYMMSPAQN26C5NLHXP5I/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/A5ZWIUPYMMSPAQN26C5NLHXP5I/action/timestamp_anchor","attest_storage":"https://pith.science/pith/A5ZWIUPYMMSPAQN26C5NLHXP5I/action/storage_attestation","attest_author":"https://pith.science/pith/A5ZWIUPYMMSPAQN26C5NLHXP5I/action/author_attestation","sign_citation":"https://pith.science/pith/A5ZWIUPYMMSPAQN26C5NLHXP5I/action/citation_signature","submit_replication":"https://pith.science/pith/A5ZWIUPYMMSPAQN26C5NLHXP5I/action/replication_record"}},"created_at":"2026-05-17T23:44:15.984909+00:00","updated_at":"2026-05-17T23:44:15.984909+00:00"}