{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:FLJWGOEGHVBQNQF7ZR5MJMHEMZ","short_pith_number":"pith:FLJWGOEG","schema_version":"1.0","canonical_sha256":"2ad36338863d4306c0bfcc7ac4b0e4666a904130fca748e47afea1c11433d5af","source":{"kind":"arxiv","id":"2107.08212","version":1},"attestation_state":"computed","paper":{"title":"On the Copying Behaviors of Pre-Training for Neural Machine Translation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Derek F. Wong, Liang Ding, Lidia S. Chao, Longyue Wang, Shuming Shi, Xuebo Liu, Zhaopeng Tu","submitted_at":"2021-07-17T10:02:30Z","abstract_excerpt":"Previous studies have shown that initializing neural machine translation (NMT) models with the pre-trained language models (LM) can speed up the model training and boost the model performance. In this work, we identify a critical side-effect of pre-training for NMT, which is due to the discrepancy between the training objectives of LM-based pre-training and NMT. Since the LM objective learns to reconstruct a few source tokens and copy most of them, the pre-training initialization would affect the copying behaviors of NMT models. We provide a quantitative analysis of copying behaviors by introd"},"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":"2107.08212","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2021-07-17T10:02:30Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"4bc450a74b9381c140856ce0cecfd7256cc6975711ce5dcc5836f9f639ea8523","abstract_canon_sha256":"1212cf6537847745eac4e7535afcc950081af40220a22aa774b2ec5f968640e4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:58:46.779925Z","signature_b64":"Hx7nZlJFzjExl3a6MZ7OoizxKo2SxKmm3F1hfl+jqS4jHKTQdThO4rV50EIJPmhmIvyML1t5ElA0BlV4BYgxBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2ad36338863d4306c0bfcc7ac4b0e4666a904130fca748e47afea1c11433d5af","last_reissued_at":"2026-07-05T02:58:46.779501Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:58:46.779501Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On the Copying Behaviors of Pre-Training for Neural Machine Translation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Derek F. Wong, Liang Ding, Lidia S. Chao, Longyue Wang, Shuming Shi, Xuebo Liu, Zhaopeng Tu","submitted_at":"2021-07-17T10:02:30Z","abstract_excerpt":"Previous studies have shown that initializing neural machine translation (NMT) models with the pre-trained language models (LM) can speed up the model training and boost the model performance. In this work, we identify a critical side-effect of pre-training for NMT, which is due to the discrepancy between the training objectives of LM-based pre-training and NMT. Since the LM objective learns to reconstruct a few source tokens and copy most of them, the pre-training initialization would affect the copying behaviors of NMT models. We provide a quantitative analysis of copying behaviors by introd"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2107.08212","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/2107.08212/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":"2107.08212","created_at":"2026-07-05T02:58:46.779564+00:00"},{"alias_kind":"arxiv_version","alias_value":"2107.08212v1","created_at":"2026-07-05T02:58:46.779564+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2107.08212","created_at":"2026-07-05T02:58:46.779564+00:00"},{"alias_kind":"pith_short_12","alias_value":"FLJWGOEGHVBQ","created_at":"2026-07-05T02:58:46.779564+00:00"},{"alias_kind":"pith_short_16","alias_value":"FLJWGOEGHVBQNQF7","created_at":"2026-07-05T02:58:46.779564+00:00"},{"alias_kind":"pith_short_8","alias_value":"FLJWGOEG","created_at":"2026-07-05T02:58:46.779564+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/FLJWGOEGHVBQNQF7ZR5MJMHEMZ","json":"https://pith.science/pith/FLJWGOEGHVBQNQF7ZR5MJMHEMZ.json","graph_json":"https://pith.science/api/pith-number/FLJWGOEGHVBQNQF7ZR5MJMHEMZ/graph.json","events_json":"https://pith.science/api/pith-number/FLJWGOEGHVBQNQF7ZR5MJMHEMZ/events.json","paper":"https://pith.science/paper/FLJWGOEG"},"agent_actions":{"view_html":"https://pith.science/pith/FLJWGOEGHVBQNQF7ZR5MJMHEMZ","download_json":"https://pith.science/pith/FLJWGOEGHVBQNQF7ZR5MJMHEMZ.json","view_paper":"https://pith.science/paper/FLJWGOEG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2107.08212&json=true","fetch_graph":"https://pith.science/api/pith-number/FLJWGOEGHVBQNQF7ZR5MJMHEMZ/graph.json","fetch_events":"https://pith.science/api/pith-number/FLJWGOEGHVBQNQF7ZR5MJMHEMZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FLJWGOEGHVBQNQF7ZR5MJMHEMZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FLJWGOEGHVBQNQF7ZR5MJMHEMZ/action/storage_attestation","attest_author":"https://pith.science/pith/FLJWGOEGHVBQNQF7ZR5MJMHEMZ/action/author_attestation","sign_citation":"https://pith.science/pith/FLJWGOEGHVBQNQF7ZR5MJMHEMZ/action/citation_signature","submit_replication":"https://pith.science/pith/FLJWGOEGHVBQNQF7ZR5MJMHEMZ/action/replication_record"}},"created_at":"2026-07-05T02:58:46.779564+00:00","updated_at":"2026-07-05T02:58:46.779564+00:00"}