{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:NCRB727PNALPUMOPNV37OE5X75","short_pith_number":"pith:NCRB727P","schema_version":"1.0","canonical_sha256":"68a21febef6816fa31cf6d77f713b7ff712276b51bd755e69facbcbd6ab5b038","source":{"kind":"arxiv","id":"2104.07429","version":2},"attestation_state":"computed","paper":{"title":"First the worst: Finding better gender translations during beam search","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bill Byrne, Danielle Saunders, Rosie Sallis","submitted_at":"2021-04-15T12:53:30Z","abstract_excerpt":"Neural machine translation inference procedures like beam search generate the most likely output under the model. This can exacerbate any demographic biases exhibited by the model. We focus on gender bias resulting from systematic errors in grammatical gender translation, which can lead to human referents being misrepresented or misgendered.\n  Most approaches to this problem adjust the training data or the model. By contrast, we experiment with simply adjusting the inference procedure. We experiment with reranking nbest lists using gender features obtained automatically from the source sentenc"},"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":"2104.07429","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-04-15T12:53:30Z","cross_cats_sorted":[],"title_canon_sha256":"f86ffdce4256a0ae2d79eb8310d335506b01a0e24f58c269905e5bdb2c649f2d","abstract_canon_sha256":"d1f6a2ef9a49cdc5a025a8ac1c2f981d4193c29debbe60480c3d50e690f967d3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:05:58.154881Z","signature_b64":"z5PBdyDA+6Xy9ifG+c3jMQkOxue8q68srO8+yyFgE4IYbSrejWhYo+u2YR00Qit1IUxwe4Iz3s+FxFXgAgy9Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"68a21febef6816fa31cf6d77f713b7ff712276b51bd755e69facbcbd6ab5b038","last_reissued_at":"2026-07-05T04:05:58.154380Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:05:58.154380Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"First the worst: Finding better gender translations during beam search","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bill Byrne, Danielle Saunders, Rosie Sallis","submitted_at":"2021-04-15T12:53:30Z","abstract_excerpt":"Neural machine translation inference procedures like beam search generate the most likely output under the model. This can exacerbate any demographic biases exhibited by the model. We focus on gender bias resulting from systematic errors in grammatical gender translation, which can lead to human referents being misrepresented or misgendered.\n  Most approaches to this problem adjust the training data or the model. By contrast, we experiment with simply adjusting the inference procedure. We experiment with reranking nbest lists using gender features obtained automatically from the source sentenc"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2104.07429","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/2104.07429/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":"2104.07429","created_at":"2026-07-05T04:05:58.154443+00:00"},{"alias_kind":"arxiv_version","alias_value":"2104.07429v2","created_at":"2026-07-05T04:05:58.154443+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2104.07429","created_at":"2026-07-05T04:05:58.154443+00:00"},{"alias_kind":"pith_short_12","alias_value":"NCRB727PNALP","created_at":"2026-07-05T04:05:58.154443+00:00"},{"alias_kind":"pith_short_16","alias_value":"NCRB727PNALPUMOP","created_at":"2026-07-05T04:05:58.154443+00:00"},{"alias_kind":"pith_short_8","alias_value":"NCRB727P","created_at":"2026-07-05T04:05:58.154443+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.23829","citing_title":"BiasFilter: An Inference-Time Debiasing Framework for Large Language Models","ref_index":57,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NCRB727PNALPUMOPNV37OE5X75","json":"https://pith.science/pith/NCRB727PNALPUMOPNV37OE5X75.json","graph_json":"https://pith.science/api/pith-number/NCRB727PNALPUMOPNV37OE5X75/graph.json","events_json":"https://pith.science/api/pith-number/NCRB727PNALPUMOPNV37OE5X75/events.json","paper":"https://pith.science/paper/NCRB727P"},"agent_actions":{"view_html":"https://pith.science/pith/NCRB727PNALPUMOPNV37OE5X75","download_json":"https://pith.science/pith/NCRB727PNALPUMOPNV37OE5X75.json","view_paper":"https://pith.science/paper/NCRB727P","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2104.07429&json=true","fetch_graph":"https://pith.science/api/pith-number/NCRB727PNALPUMOPNV37OE5X75/graph.json","fetch_events":"https://pith.science/api/pith-number/NCRB727PNALPUMOPNV37OE5X75/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NCRB727PNALPUMOPNV37OE5X75/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NCRB727PNALPUMOPNV37OE5X75/action/storage_attestation","attest_author":"https://pith.science/pith/NCRB727PNALPUMOPNV37OE5X75/action/author_attestation","sign_citation":"https://pith.science/pith/NCRB727PNALPUMOPNV37OE5X75/action/citation_signature","submit_replication":"https://pith.science/pith/NCRB727PNALPUMOPNV37OE5X75/action/replication_record"}},"created_at":"2026-07-05T04:05:58.154443+00:00","updated_at":"2026-07-05T04:05:58.154443+00:00"}