{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:GJYC5TKFCACSI2GTDUCLGBV6DB","short_pith_number":"pith:GJYC5TKF","schema_version":"1.0","canonical_sha256":"32702ecd4510052468d31d04b306be185136803088bc2beb1ca8f87b01ac4a9c","source":{"kind":"arxiv","id":"2109.05611","version":2},"attestation_state":"computed","paper":{"title":"Levenshtein Training for Word-level Quality Estimation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Marcin Junczys-Dowmunt, Matt Post, Philipp Koehn, Shuoyang Ding","submitted_at":"2021-09-12T20:45:48Z","abstract_excerpt":"We propose a novel scheme to use the Levenshtein Transformer to perform the task of word-level quality estimation. A Levenshtein Transformer is a natural fit for this task: trained to perform decoding in an iterative manner, a Levenshtein Transformer can learn to post-edit without explicit supervision. To further minimize the mismatch between the translation task and the word-level QE task, we propose a two-stage transfer learning procedure on both augmented data and human post-editing data. We also propose heuristics to construct reference labels that are compatible with subword-level finetun"},"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":"2109.05611","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-09-12T20:45:48Z","cross_cats_sorted":[],"title_canon_sha256":"2d60233870b994e7b262126cf99fc1d9f83c99428a24bb952708ceee0f801ea3","abstract_canon_sha256":"f647c6cd2eb91fd6b2f9ac5b14eb614aca4579477cfb5fc35011590803745248"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:14:57.147834Z","signature_b64":"EPQG5FdPDgrRjF42rdkNNwzo+WbTiCVRQJCh5vOAMGnV6CAE9hz1Z3uLDFlW2Xk3sq7G0PlwCiS4BEWxgVR1Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"32702ecd4510052468d31d04b306be185136803088bc2beb1ca8f87b01ac4a9c","last_reissued_at":"2026-07-05T03:14:57.147405Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:14:57.147405Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Levenshtein Training for Word-level Quality Estimation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Marcin Junczys-Dowmunt, Matt Post, Philipp Koehn, Shuoyang Ding","submitted_at":"2021-09-12T20:45:48Z","abstract_excerpt":"We propose a novel scheme to use the Levenshtein Transformer to perform the task of word-level quality estimation. A Levenshtein Transformer is a natural fit for this task: trained to perform decoding in an iterative manner, a Levenshtein Transformer can learn to post-edit without explicit supervision. To further minimize the mismatch between the translation task and the word-level QE task, we propose a two-stage transfer learning procedure on both augmented data and human post-editing data. We also propose heuristics to construct reference labels that are compatible with subword-level finetun"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2109.05611","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/2109.05611/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":"2109.05611","created_at":"2026-07-05T03:14:57.147468+00:00"},{"alias_kind":"arxiv_version","alias_value":"2109.05611v2","created_at":"2026-07-05T03:14:57.147468+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2109.05611","created_at":"2026-07-05T03:14:57.147468+00:00"},{"alias_kind":"pith_short_12","alias_value":"GJYC5TKFCACS","created_at":"2026-07-05T03:14:57.147468+00:00"},{"alias_kind":"pith_short_16","alias_value":"GJYC5TKFCACSI2GT","created_at":"2026-07-05T03:14:57.147468+00:00"},{"alias_kind":"pith_short_8","alias_value":"GJYC5TKF","created_at":"2026-07-05T03:14:57.147468+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2508.18306","citing_title":"SALMAN: Stability Analysis of Language Models Through the Maps Between Graph-based Manifolds","ref_index":47,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GJYC5TKFCACSI2GTDUCLGBV6DB","json":"https://pith.science/pith/GJYC5TKFCACSI2GTDUCLGBV6DB.json","graph_json":"https://pith.science/api/pith-number/GJYC5TKFCACSI2GTDUCLGBV6DB/graph.json","events_json":"https://pith.science/api/pith-number/GJYC5TKFCACSI2GTDUCLGBV6DB/events.json","paper":"https://pith.science/paper/GJYC5TKF"},"agent_actions":{"view_html":"https://pith.science/pith/GJYC5TKFCACSI2GTDUCLGBV6DB","download_json":"https://pith.science/pith/GJYC5TKFCACSI2GTDUCLGBV6DB.json","view_paper":"https://pith.science/paper/GJYC5TKF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2109.05611&json=true","fetch_graph":"https://pith.science/api/pith-number/GJYC5TKFCACSI2GTDUCLGBV6DB/graph.json","fetch_events":"https://pith.science/api/pith-number/GJYC5TKFCACSI2GTDUCLGBV6DB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GJYC5TKFCACSI2GTDUCLGBV6DB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GJYC5TKFCACSI2GTDUCLGBV6DB/action/storage_attestation","attest_author":"https://pith.science/pith/GJYC5TKFCACSI2GTDUCLGBV6DB/action/author_attestation","sign_citation":"https://pith.science/pith/GJYC5TKFCACSI2GTDUCLGBV6DB/action/citation_signature","submit_replication":"https://pith.science/pith/GJYC5TKFCACSI2GTDUCLGBV6DB/action/replication_record"}},"created_at":"2026-07-05T03:14:57.147468+00:00","updated_at":"2026-07-05T03:14:57.147468+00:00"}