{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:IKPDQ6BOAU23XVGWM3KPZGH4VV","short_pith_number":"pith:IKPDQ6BO","schema_version":"1.0","canonical_sha256":"429e38782e0535bbd4d666d4fc98fcad52a2d1db0cdf4a9d795cc7124935ed2c","source":{"kind":"arxiv","id":"2106.05546","version":2},"attestation_state":"computed","paper":{"title":"Progressive Multi-Granularity Training for Non-Autoregressive Translation","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dacheng Tao, Derek F. Wong, Liang Ding, Longyue Wang, Xuebo Liu, Zhaopeng Tu","submitted_at":"2021-06-10T07:16:07Z","abstract_excerpt":"Non-autoregressive translation (NAT) significantly accelerates the inference process via predicting the entire target sequence. However, recent studies show that NAT is weak at learning high-mode of knowledge such as one-to-many translations. We argue that modes can be divided into various granularities which can be learned from easy to hard. In this study, we empirically show that NAT models are prone to learn fine-grained lower-mode knowledge, such as words and phrases, compared with sentences. Based on this observation, we propose progressive multi-granularity training for NAT. More specifi"},"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":"2106.05546","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.CL","submitted_at":"2021-06-10T07:16:07Z","cross_cats_sorted":[],"title_canon_sha256":"188aa253fe261767d945e3bf41375157a39598197050a76bc38c3f3957807a43","abstract_canon_sha256":"8e817fe6491c70a52609feda612fd2703e3352fb7855d149deee542e4cf33bfe"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:48:24.112363Z","signature_b64":"UA6Epit3+W2TkTdvvil3xbdfPlZEKjfVtswx2wpPV7QvswTg8QCzZFQdScA+VKnwxbI0C2an/s/CHi9s5DDgCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"429e38782e0535bbd4d666d4fc98fcad52a2d1db0cdf4a9d795cc7124935ed2c","last_reissued_at":"2026-07-05T02:48:24.111903Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:48:24.111903Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Progressive Multi-Granularity Training for Non-Autoregressive Translation","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dacheng Tao, Derek F. Wong, Liang Ding, Longyue Wang, Xuebo Liu, Zhaopeng Tu","submitted_at":"2021-06-10T07:16:07Z","abstract_excerpt":"Non-autoregressive translation (NAT) significantly accelerates the inference process via predicting the entire target sequence. However, recent studies show that NAT is weak at learning high-mode of knowledge such as one-to-many translations. We argue that modes can be divided into various granularities which can be learned from easy to hard. In this study, we empirically show that NAT models are prone to learn fine-grained lower-mode knowledge, such as words and phrases, compared with sentences. Based on this observation, we propose progressive multi-granularity training for NAT. More specifi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.05546","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/2106.05546/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":"2106.05546","created_at":"2026-07-05T02:48:24.111961+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.05546v2","created_at":"2026-07-05T02:48:24.111961+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.05546","created_at":"2026-07-05T02:48:24.111961+00:00"},{"alias_kind":"pith_short_12","alias_value":"IKPDQ6BOAU23","created_at":"2026-07-05T02:48:24.111961+00:00"},{"alias_kind":"pith_short_16","alias_value":"IKPDQ6BOAU23XVGW","created_at":"2026-07-05T02:48:24.111961+00:00"},{"alias_kind":"pith_short_8","alias_value":"IKPDQ6BO","created_at":"2026-07-05T02:48:24.111961+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/IKPDQ6BOAU23XVGWM3KPZGH4VV","json":"https://pith.science/pith/IKPDQ6BOAU23XVGWM3KPZGH4VV.json","graph_json":"https://pith.science/api/pith-number/IKPDQ6BOAU23XVGWM3KPZGH4VV/graph.json","events_json":"https://pith.science/api/pith-number/IKPDQ6BOAU23XVGWM3KPZGH4VV/events.json","paper":"https://pith.science/paper/IKPDQ6BO"},"agent_actions":{"view_html":"https://pith.science/pith/IKPDQ6BOAU23XVGWM3KPZGH4VV","download_json":"https://pith.science/pith/IKPDQ6BOAU23XVGWM3KPZGH4VV.json","view_paper":"https://pith.science/paper/IKPDQ6BO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.05546&json=true","fetch_graph":"https://pith.science/api/pith-number/IKPDQ6BOAU23XVGWM3KPZGH4VV/graph.json","fetch_events":"https://pith.science/api/pith-number/IKPDQ6BOAU23XVGWM3KPZGH4VV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IKPDQ6BOAU23XVGWM3KPZGH4VV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IKPDQ6BOAU23XVGWM3KPZGH4VV/action/storage_attestation","attest_author":"https://pith.science/pith/IKPDQ6BOAU23XVGWM3KPZGH4VV/action/author_attestation","sign_citation":"https://pith.science/pith/IKPDQ6BOAU23XVGWM3KPZGH4VV/action/citation_signature","submit_replication":"https://pith.science/pith/IKPDQ6BOAU23XVGWM3KPZGH4VV/action/replication_record"}},"created_at":"2026-07-05T02:48:24.111961+00:00","updated_at":"2026-07-05T02:48:24.111961+00:00"}