{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:Z7VUJ5FMEJCD3LDHHE6PYK72R5","short_pith_number":"pith:Z7VUJ5FM","schema_version":"1.0","canonical_sha256":"cfeb44f4ac22443dac67393cfc2bfa8f42f403bc8e8a2a9940df4edaf8c33f1e","source":{"kind":"arxiv","id":"2005.09093","version":2},"attestation_state":"computed","paper":{"title":"Are All Languages Created Equal in Multilingual BERT?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Mark Dredze, Shijie Wu","submitted_at":"2020-05-18T21:15:39Z","abstract_excerpt":"Multilingual BERT (mBERT) trained on 104 languages has shown surprisingly good cross-lingual performance on several NLP tasks, even without explicit cross-lingual signals. However, these evaluations have focused on cross-lingual transfer with high-resource languages, covering only a third of the languages covered by mBERT. We explore how mBERT performs on a much wider set of languages, focusing on the quality of representation for low-resource languages, measured by within-language performance. We consider three tasks: Named Entity Recognition (99 languages), Part-of-speech Tagging, and Depend"},"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":"2005.09093","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-05-18T21:15:39Z","cross_cats_sorted":[],"title_canon_sha256":"195ac804656bd89c16b4c4de15ea98a4ac58e86b7b5b1a02b91a4a79f32917f9","abstract_canon_sha256":"7f002addcf35d32b6143f911a6f02f14be78a2828763473500b511250adb1126"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:39:26.784215Z","signature_b64":"0RiG6ZxS6HUc7fLbwRgLOb05B73jaucNZvekr0Xnz69C91Eog847/WDSXX9PQqfzF9J4uNonhSswjerDoCuyAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cfeb44f4ac22443dac67393cfc2bfa8f42f403bc8e8a2a9940df4edaf8c33f1e","last_reissued_at":"2026-07-05T01:39:26.783767Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:39:26.783767Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Are All Languages Created Equal in Multilingual BERT?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Mark Dredze, Shijie Wu","submitted_at":"2020-05-18T21:15:39Z","abstract_excerpt":"Multilingual BERT (mBERT) trained on 104 languages has shown surprisingly good cross-lingual performance on several NLP tasks, even without explicit cross-lingual signals. However, these evaluations have focused on cross-lingual transfer with high-resource languages, covering only a third of the languages covered by mBERT. We explore how mBERT performs on a much wider set of languages, focusing on the quality of representation for low-resource languages, measured by within-language performance. We consider three tasks: Named Entity Recognition (99 languages), Part-of-speech Tagging, and Depend"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2005.09093","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/2005.09093/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":"2005.09093","created_at":"2026-07-05T01:39:26.783827+00:00"},{"alias_kind":"arxiv_version","alias_value":"2005.09093v2","created_at":"2026-07-05T01:39:26.783827+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2005.09093","created_at":"2026-07-05T01:39:26.783827+00:00"},{"alias_kind":"pith_short_12","alias_value":"Z7VUJ5FMEJCD","created_at":"2026-07-05T01:39:26.783827+00:00"},{"alias_kind":"pith_short_16","alias_value":"Z7VUJ5FMEJCD3LDH","created_at":"2026-07-05T01:39:26.783827+00:00"},{"alias_kind":"pith_short_8","alias_value":"Z7VUJ5FM","created_at":"2026-07-05T01:39:26.783827+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2506.06816","citing_title":"How do datasets, developers, and models affect biases in a low-resourced language?: The Case of the Bengali Language","ref_index":145,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02608","citing_title":"Dependency Parsing Across the Resource Spectrum: Evaluating Architectures on High and Low-Resource Languages","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Z7VUJ5FMEJCD3LDHHE6PYK72R5","json":"https://pith.science/pith/Z7VUJ5FMEJCD3LDHHE6PYK72R5.json","graph_json":"https://pith.science/api/pith-number/Z7VUJ5FMEJCD3LDHHE6PYK72R5/graph.json","events_json":"https://pith.science/api/pith-number/Z7VUJ5FMEJCD3LDHHE6PYK72R5/events.json","paper":"https://pith.science/paper/Z7VUJ5FM"},"agent_actions":{"view_html":"https://pith.science/pith/Z7VUJ5FMEJCD3LDHHE6PYK72R5","download_json":"https://pith.science/pith/Z7VUJ5FMEJCD3LDHHE6PYK72R5.json","view_paper":"https://pith.science/paper/Z7VUJ5FM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2005.09093&json=true","fetch_graph":"https://pith.science/api/pith-number/Z7VUJ5FMEJCD3LDHHE6PYK72R5/graph.json","fetch_events":"https://pith.science/api/pith-number/Z7VUJ5FMEJCD3LDHHE6PYK72R5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Z7VUJ5FMEJCD3LDHHE6PYK72R5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Z7VUJ5FMEJCD3LDHHE6PYK72R5/action/storage_attestation","attest_author":"https://pith.science/pith/Z7VUJ5FMEJCD3LDHHE6PYK72R5/action/author_attestation","sign_citation":"https://pith.science/pith/Z7VUJ5FMEJCD3LDHHE6PYK72R5/action/citation_signature","submit_replication":"https://pith.science/pith/Z7VUJ5FMEJCD3LDHHE6PYK72R5/action/replication_record"}},"created_at":"2026-07-05T01:39:26.783827+00:00","updated_at":"2026-07-05T01:39:26.783827+00:00"}