{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:B4Z3QCSCW7NRQBMAY5XCXHYQPF","short_pith_number":"pith:B4Z3QCSC","schema_version":"1.0","canonical_sha256":"0f33b80a42b7db180580c76e2b9f10795a5361d970136e00037ddc9715c38eae","source":{"kind":"arxiv","id":"2012.05628","version":3},"attestation_state":"computed","paper":{"title":"As Good as New. How to Successfully Recycle English GPT-2 to Make Models for Other Languages","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Malvina Nissim, Wietse de Vries","submitted_at":"2020-12-10T12:27:16Z","abstract_excerpt":"Large generative language models have been very successful for English, but other languages lag behind, in part due to data and computational limitations. We propose a method that may overcome these problems by adapting existing pre-trained models to new languages. Specifically, we describe the adaptation of English GPT-2 to Italian and Dutch by retraining lexical embeddings without tuning the Transformer layers. As a result, we obtain lexical embeddings for Italian and Dutch that are aligned with the original English lexical embeddings. Additionally, we scale up complexity by transforming rel"},"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":"2012.05628","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2020-12-10T12:27:16Z","cross_cats_sorted":[],"title_canon_sha256":"9435f662eca25ef49c6233dc4433d83d335d0261528c9f7b96826edb00a9b5a9","abstract_canon_sha256":"db119f41823b1199c31213a345d1cf018b276f490a41e03e52fbb8ccdb49dd27"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:02:10.558845Z","signature_b64":"P9sjsxTaqdI8PgJ9BU9NA7lvNReWVdrB5QMsqidKkhZ4/Hnus6DNV4LIyz3k20OUZ29+9Uu/zh3bRcFEArqdDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0f33b80a42b7db180580c76e2b9f10795a5361d970136e00037ddc9715c38eae","last_reissued_at":"2026-07-05T03:02:10.558400Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:02:10.558400Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"As Good as New. How to Successfully Recycle English GPT-2 to Make Models for Other Languages","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Malvina Nissim, Wietse de Vries","submitted_at":"2020-12-10T12:27:16Z","abstract_excerpt":"Large generative language models have been very successful for English, but other languages lag behind, in part due to data and computational limitations. We propose a method that may overcome these problems by adapting existing pre-trained models to new languages. Specifically, we describe the adaptation of English GPT-2 to Italian and Dutch by retraining lexical embeddings without tuning the Transformer layers. As a result, we obtain lexical embeddings for Italian and Dutch that are aligned with the original English lexical embeddings. Additionally, we scale up complexity by transforming rel"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2012.05628","kind":"arxiv","version":3},"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/2012.05628/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":"2012.05628","created_at":"2026-07-05T03:02:10.558456+00:00"},{"alias_kind":"arxiv_version","alias_value":"2012.05628v3","created_at":"2026-07-05T03:02:10.558456+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2012.05628","created_at":"2026-07-05T03:02:10.558456+00:00"},{"alias_kind":"pith_short_12","alias_value":"B4Z3QCSCW7NR","created_at":"2026-07-05T03:02:10.558456+00:00"},{"alias_kind":"pith_short_16","alias_value":"B4Z3QCSCW7NRQBMA","created_at":"2026-07-05T03:02:10.558456+00:00"},{"alias_kind":"pith_short_8","alias_value":"B4Z3QCSC","created_at":"2026-07-05T03:02:10.558456+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.20369","citing_title":"CATCH-ME if you RAG: a dataset of Contextually Annotated multi-Turn Counterspeech against Hate and Misinformation Exchanges","ref_index":261,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22435","citing_title":"Assisted Counterspeech Writing at the Crossroads of Hate Speech and Misinformation","ref_index":251,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/B4Z3QCSCW7NRQBMAY5XCXHYQPF","json":"https://pith.science/pith/B4Z3QCSCW7NRQBMAY5XCXHYQPF.json","graph_json":"https://pith.science/api/pith-number/B4Z3QCSCW7NRQBMAY5XCXHYQPF/graph.json","events_json":"https://pith.science/api/pith-number/B4Z3QCSCW7NRQBMAY5XCXHYQPF/events.json","paper":"https://pith.science/paper/B4Z3QCSC"},"agent_actions":{"view_html":"https://pith.science/pith/B4Z3QCSCW7NRQBMAY5XCXHYQPF","download_json":"https://pith.science/pith/B4Z3QCSCW7NRQBMAY5XCXHYQPF.json","view_paper":"https://pith.science/paper/B4Z3QCSC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2012.05628&json=true","fetch_graph":"https://pith.science/api/pith-number/B4Z3QCSCW7NRQBMAY5XCXHYQPF/graph.json","fetch_events":"https://pith.science/api/pith-number/B4Z3QCSCW7NRQBMAY5XCXHYQPF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/B4Z3QCSCW7NRQBMAY5XCXHYQPF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/B4Z3QCSCW7NRQBMAY5XCXHYQPF/action/storage_attestation","attest_author":"https://pith.science/pith/B4Z3QCSCW7NRQBMAY5XCXHYQPF/action/author_attestation","sign_citation":"https://pith.science/pith/B4Z3QCSCW7NRQBMAY5XCXHYQPF/action/citation_signature","submit_replication":"https://pith.science/pith/B4Z3QCSCW7NRQBMAY5XCXHYQPF/action/replication_record"}},"created_at":"2026-07-05T03:02:10.558456+00:00","updated_at":"2026-07-05T03:02:10.558456+00:00"}