{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:VA6WDHK66CSPJKURP3JQ2LTUGI","short_pith_number":"pith:VA6WDHK6","schema_version":"1.0","canonical_sha256":"a83d619d5ef0a4f4aa917ed30d2e7432113b09e8d87d536657fde2fcb87a8a35","source":{"kind":"arxiv","id":"1903.00802","version":1},"attestation_state":"computed","paper":{"title":"Calibration of Encoder Decoder Models for Neural Machine Translation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Aviral Kumar, Sunita Sarawagi","submitted_at":"2019-03-03T01:08:47Z","abstract_excerpt":"We study the calibration of several state of the art neural machine translation(NMT) systems built on attention-based encoder-decoder models. For structured outputs like in NMT, calibration is important not just for reliable confidence with predictions, but also for proper functioning of beam-search inference. We show that most modern NMT models are surprisingly miscalibrated even when conditioned on the true previous tokens. Our investigation leads to two main reasons -- severe miscalibration of EOS (end of sequence marker) and suppression of attention uncertainty. We design recalibration met"},"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":"1903.00802","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-03-03T01:08:47Z","cross_cats_sorted":["cs.CL","stat.ML"],"title_canon_sha256":"f4bb0e05e61604ccab120f4b635b9a84287b384f3e1f3312e758b87bb66a92bb","abstract_canon_sha256":"53f6a1a985dc8964f8792d9f0fff6e2ee1bde8f64e937e671b763995e80228ad"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:52:00.681982Z","signature_b64":"9ybuh08Z+Pk/ZbAGRWqM9DRSgn4EHc6Zw/sTWZnvSECEXBRNXBkKxdut/ij3cPXHsIk+8cFQfOMusGqpW4IDDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a83d619d5ef0a4f4aa917ed30d2e7432113b09e8d87d536657fde2fcb87a8a35","last_reissued_at":"2026-05-17T23:52:00.681514Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:52:00.681514Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Calibration of Encoder Decoder Models for Neural Machine Translation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Aviral Kumar, Sunita Sarawagi","submitted_at":"2019-03-03T01:08:47Z","abstract_excerpt":"We study the calibration of several state of the art neural machine translation(NMT) systems built on attention-based encoder-decoder models. For structured outputs like in NMT, calibration is important not just for reliable confidence with predictions, but also for proper functioning of beam-search inference. We show that most modern NMT models are surprisingly miscalibrated even when conditioned on the true previous tokens. Our investigation leads to two main reasons -- severe miscalibration of EOS (end of sequence marker) and suppression of attention uncertainty. We design recalibration met"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1903.00802","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":"1903.00802","created_at":"2026-05-17T23:52:00.681591+00:00"},{"alias_kind":"arxiv_version","alias_value":"1903.00802v1","created_at":"2026-05-17T23:52:00.681591+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1903.00802","created_at":"2026-05-17T23:52:00.681591+00:00"},{"alias_kind":"pith_short_12","alias_value":"VA6WDHK66CSP","created_at":"2026-05-18T12:33:30.264802+00:00"},{"alias_kind":"pith_short_16","alias_value":"VA6WDHK66CSPJKUR","created_at":"2026-05-18T12:33:30.264802+00:00"},{"alias_kind":"pith_short_8","alias_value":"VA6WDHK6","created_at":"2026-05-18T12:33:30.264802+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2605.20490","citing_title":"ECUAS$_n$: A family of metrics for principled evaluation of uncertainty-augmented systems","ref_index":52,"is_internal_anchor":true},{"citing_arxiv_id":"2605.20490","citing_title":"ECUAS$_n$: A family of metrics for principled evaluation of uncertainty-augmented systems","ref_index":52,"is_internal_anchor":true},{"citing_arxiv_id":"2604.08974","citing_title":"Confident in a Confidence Score: Investigating the Sensitivity of Confidence Scores to Supervised Fine-Tuning","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VA6WDHK66CSPJKURP3JQ2LTUGI","json":"https://pith.science/pith/VA6WDHK66CSPJKURP3JQ2LTUGI.json","graph_json":"https://pith.science/api/pith-number/VA6WDHK66CSPJKURP3JQ2LTUGI/graph.json","events_json":"https://pith.science/api/pith-number/VA6WDHK66CSPJKURP3JQ2LTUGI/events.json","paper":"https://pith.science/paper/VA6WDHK6"},"agent_actions":{"view_html":"https://pith.science/pith/VA6WDHK66CSPJKURP3JQ2LTUGI","download_json":"https://pith.science/pith/VA6WDHK66CSPJKURP3JQ2LTUGI.json","view_paper":"https://pith.science/paper/VA6WDHK6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1903.00802&json=true","fetch_graph":"https://pith.science/api/pith-number/VA6WDHK66CSPJKURP3JQ2LTUGI/graph.json","fetch_events":"https://pith.science/api/pith-number/VA6WDHK66CSPJKURP3JQ2LTUGI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VA6WDHK66CSPJKURP3JQ2LTUGI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VA6WDHK66CSPJKURP3JQ2LTUGI/action/storage_attestation","attest_author":"https://pith.science/pith/VA6WDHK66CSPJKURP3JQ2LTUGI/action/author_attestation","sign_citation":"https://pith.science/pith/VA6WDHK66CSPJKURP3JQ2LTUGI/action/citation_signature","submit_replication":"https://pith.science/pith/VA6WDHK66CSPJKURP3JQ2LTUGI/action/replication_record"}},"created_at":"2026-05-17T23:52:00.681591+00:00","updated_at":"2026-05-17T23:52:00.681591+00:00"}