{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:NMTF73XTCMFJAWPCBTOVJJHIHZ","short_pith_number":"pith:NMTF73XT","schema_version":"1.0","canonical_sha256":"6b265feef3130a9059e20cdd54a4e83e696d7a30fe4d6ee9e0a3badbf246e0ba","source":{"kind":"arxiv","id":"2106.02171","version":1},"attestation_state":"computed","paper":{"title":"nmT5 -- Is parallel data still relevant for pre-training massively multilingual language models?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Aditya Siddhant, Linting Xue, Melvin Johnson, Mihir Kale, Noah Constant, Rami Al-Rfou","submitted_at":"2021-06-03T23:12:27Z","abstract_excerpt":"Recently, mT5 - a massively multilingual version of T5 - leveraged a unified text-to-text format to attain state-of-the-art results on a wide variety of multilingual NLP tasks. In this paper, we investigate the impact of incorporating parallel data into mT5 pre-training. We find that multi-tasking language modeling with objectives such as machine translation during pre-training is a straightforward way to improve performance on downstream multilingual and cross-lingual tasks. However, the gains start to diminish as the model capacity increases, suggesting that parallel data might not be as ess"},"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":"2106.02171","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-06-03T23:12:27Z","cross_cats_sorted":[],"title_canon_sha256":"0fb41a3b9d0e62f70f00929042c975e3355e5b2bbd20494ae05ace5b0e7951a8","abstract_canon_sha256":"038912992f7da1c9050128aa05727d49964e5793e193518209f26e8d9e967914"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:46:16.138222Z","signature_b64":"SyCBEe8lWzqFfwyZWBV9zu4GYzQ2XlvvpQ1RLgLxS8FyVw4RZsdSqW/LJYCVw1fCQtNFGnucbXPqlO+ohpuyCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6b265feef3130a9059e20cdd54a4e83e696d7a30fe4d6ee9e0a3badbf246e0ba","last_reissued_at":"2026-07-05T02:46:16.137814Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:46:16.137814Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"nmT5 -- Is parallel data still relevant for pre-training massively multilingual language models?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Aditya Siddhant, Linting Xue, Melvin Johnson, Mihir Kale, Noah Constant, Rami Al-Rfou","submitted_at":"2021-06-03T23:12:27Z","abstract_excerpt":"Recently, mT5 - a massively multilingual version of T5 - leveraged a unified text-to-text format to attain state-of-the-art results on a wide variety of multilingual NLP tasks. In this paper, we investigate the impact of incorporating parallel data into mT5 pre-training. We find that multi-tasking language modeling with objectives such as machine translation during pre-training is a straightforward way to improve performance on downstream multilingual and cross-lingual tasks. However, the gains start to diminish as the model capacity increases, suggesting that parallel data might not be as ess"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.02171","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/2106.02171/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":"2106.02171","created_at":"2026-07-05T02:46:16.137873+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.02171v1","created_at":"2026-07-05T02:46:16.137873+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.02171","created_at":"2026-07-05T02:46:16.137873+00:00"},{"alias_kind":"pith_short_12","alias_value":"NMTF73XTCMFJ","created_at":"2026-07-05T02:46:16.137873+00:00"},{"alias_kind":"pith_short_16","alias_value":"NMTF73XTCMFJAWPC","created_at":"2026-07-05T02:46:16.137873+00:00"},{"alias_kind":"pith_short_8","alias_value":"NMTF73XT","created_at":"2026-07-05T02:46:16.137873+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/NMTF73XTCMFJAWPCBTOVJJHIHZ","json":"https://pith.science/pith/NMTF73XTCMFJAWPCBTOVJJHIHZ.json","graph_json":"https://pith.science/api/pith-number/NMTF73XTCMFJAWPCBTOVJJHIHZ/graph.json","events_json":"https://pith.science/api/pith-number/NMTF73XTCMFJAWPCBTOVJJHIHZ/events.json","paper":"https://pith.science/paper/NMTF73XT"},"agent_actions":{"view_html":"https://pith.science/pith/NMTF73XTCMFJAWPCBTOVJJHIHZ","download_json":"https://pith.science/pith/NMTF73XTCMFJAWPCBTOVJJHIHZ.json","view_paper":"https://pith.science/paper/NMTF73XT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.02171&json=true","fetch_graph":"https://pith.science/api/pith-number/NMTF73XTCMFJAWPCBTOVJJHIHZ/graph.json","fetch_events":"https://pith.science/api/pith-number/NMTF73XTCMFJAWPCBTOVJJHIHZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NMTF73XTCMFJAWPCBTOVJJHIHZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NMTF73XTCMFJAWPCBTOVJJHIHZ/action/storage_attestation","attest_author":"https://pith.science/pith/NMTF73XTCMFJAWPCBTOVJJHIHZ/action/author_attestation","sign_citation":"https://pith.science/pith/NMTF73XTCMFJAWPCBTOVJJHIHZ/action/citation_signature","submit_replication":"https://pith.science/pith/NMTF73XTCMFJAWPCBTOVJJHIHZ/action/replication_record"}},"created_at":"2026-07-05T02:46:16.137873+00:00","updated_at":"2026-07-05T02:46:16.137873+00:00"}