{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:YEHDVXKHMCLSFGMKD7RESEG2FW","short_pith_number":"pith:YEHDVXKH","schema_version":"1.0","canonical_sha256":"c10e3add47609722998a1fe24910da2d99f827665dad6e8d9bedf628f20fa60f","source":{"kind":"arxiv","id":"2203.08442","version":1},"attestation_state":"computed","paper":{"title":"Understanding and Improving Sequence-to-Sequence Pretraining for Neural Machine Translation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Michael Lyu, Shuming Shi, Wenxiang Jiao, Wenxuan Wang, Xing Wang, Yongchang Hao, Zhaopeng Tu","submitted_at":"2022-03-16T07:36:28Z","abstract_excerpt":"In this paper, we present a substantial step in better understanding the SOTA sequence-to-sequence (Seq2Seq) pretraining for neural machine translation~(NMT). We focus on studying the impact of the jointly pretrained decoder, which is the main difference between Seq2Seq pretraining and previous encoder-based pretraining approaches for NMT. By carefully designing experiments on three language pairs, we find that Seq2Seq pretraining is a double-edged sword: On one hand, it helps NMT models to produce more diverse translations and reduce adequacy-related translation errors. On the other hand, the"},"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":"2203.08442","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-03-16T07:36:28Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"2682bfe4c07e4c856513236095621b00ef3a9db6679d1a79593fbc6ccef17f7a","abstract_canon_sha256":"28507a0d0370aa188d3b7f159916460b3f83489f27e83acc639cdd26e3e3b533"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:05:44.695246Z","signature_b64":"a2RHqtWo/lGqtsERJqTrWqERcxjdF8DV8TXMibhPt9/fHfyUli/tAgMWihWtk5DrErUy0VnPhxhs+32cPaQGDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c10e3add47609722998a1fe24910da2d99f827665dad6e8d9bedf628f20fa60f","last_reissued_at":"2026-07-05T04:05:44.694706Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:05:44.694706Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Understanding and Improving Sequence-to-Sequence Pretraining for Neural Machine Translation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Michael Lyu, Shuming Shi, Wenxiang Jiao, Wenxuan Wang, Xing Wang, Yongchang Hao, Zhaopeng Tu","submitted_at":"2022-03-16T07:36:28Z","abstract_excerpt":"In this paper, we present a substantial step in better understanding the SOTA sequence-to-sequence (Seq2Seq) pretraining for neural machine translation~(NMT). We focus on studying the impact of the jointly pretrained decoder, which is the main difference between Seq2Seq pretraining and previous encoder-based pretraining approaches for NMT. By carefully designing experiments on three language pairs, we find that Seq2Seq pretraining is a double-edged sword: On one hand, it helps NMT models to produce more diverse translations and reduce adequacy-related translation errors. On the other hand, the"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.08442","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2203.08442/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":"2203.08442","created_at":"2026-07-05T04:05:44.694766+00:00"},{"alias_kind":"arxiv_version","alias_value":"2203.08442v1","created_at":"2026-07-05T04:05:44.694766+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.08442","created_at":"2026-07-05T04:05:44.694766+00:00"},{"alias_kind":"pith_short_12","alias_value":"YEHDVXKHMCLS","created_at":"2026-07-05T04:05:44.694766+00:00"},{"alias_kind":"pith_short_16","alias_value":"YEHDVXKHMCLSFGMK","created_at":"2026-07-05T04:05:44.694766+00:00"},{"alias_kind":"pith_short_8","alias_value":"YEHDVXKH","created_at":"2026-07-05T04:05:44.694766+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YEHDVXKHMCLSFGMKD7RESEG2FW","json":"https://pith.science/pith/YEHDVXKHMCLSFGMKD7RESEG2FW.json","graph_json":"https://pith.science/api/pith-number/YEHDVXKHMCLSFGMKD7RESEG2FW/graph.json","events_json":"https://pith.science/api/pith-number/YEHDVXKHMCLSFGMKD7RESEG2FW/events.json","paper":"https://pith.science/paper/YEHDVXKH"},"agent_actions":{"view_html":"https://pith.science/pith/YEHDVXKHMCLSFGMKD7RESEG2FW","download_json":"https://pith.science/pith/YEHDVXKHMCLSFGMKD7RESEG2FW.json","view_paper":"https://pith.science/paper/YEHDVXKH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2203.08442&json=true","fetch_graph":"https://pith.science/api/pith-number/YEHDVXKHMCLSFGMKD7RESEG2FW/graph.json","fetch_events":"https://pith.science/api/pith-number/YEHDVXKHMCLSFGMKD7RESEG2FW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YEHDVXKHMCLSFGMKD7RESEG2FW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YEHDVXKHMCLSFGMKD7RESEG2FW/action/storage_attestation","attest_author":"https://pith.science/pith/YEHDVXKHMCLSFGMKD7RESEG2FW/action/author_attestation","sign_citation":"https://pith.science/pith/YEHDVXKHMCLSFGMKD7RESEG2FW/action/citation_signature","submit_replication":"https://pith.science/pith/YEHDVXKHMCLSFGMKD7RESEG2FW/action/replication_record"}},"created_at":"2026-07-05T04:05:44.694766+00:00","updated_at":"2026-07-05T04:05:44.694766+00:00"}