{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:K44MCLWEZS754NDO46MNROQ55V","short_pith_number":"pith:K44MCLWE","schema_version":"1.0","canonical_sha256":"5738c12ec4ccbfde346ee798d8ba1ded4bba5171806420a7a17b6d51afcbe7a8","source":{"kind":"arxiv","id":"2401.01055","version":2},"attestation_state":"computed","paper":{"title":"LLaMA Beyond English: An Empirical Study on Language Capability Transfer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jun Zhao, Luhui Gao, Qi Zhang, Tao Gui, Xuanjing Huang, Zhihao Zhang","submitted_at":"2024-01-02T06:29:02Z","abstract_excerpt":"In recent times, substantial advancements have been witnessed in large language models (LLMs), exemplified by ChatGPT, showcasing remarkable proficiency across a range of complex tasks. However, many mainstream LLMs (e.g. LLaMA) are pretrained on English-dominant corpus, which limits their performance in other non-English languages. In this paper, we focus on how to effectively transfer the capabilities of language generation and following instructions to a non-English language. To answer this question, we conduct an extensive empirical investigation based on LLaMA, accumulating over 1440 GPU "},"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":"2401.01055","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-01-02T06:29:02Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"af06cc5072127fa375936e94a5443b31f22398f139f7faf439bed0c32cf7bb32","abstract_canon_sha256":"eca0fd054943ef4b3c41bff63d8b9a890e317a3d3620bbdad1996db530847b83"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:32:45.436470Z","signature_b64":"/c+/mFThj1cEu0cScu7dR5BKvUH0AFZS8Zo/7Na/omqqGNeoxjZnnDYeIVk9hc7B0rCiDHX2O8oUWboEOzoTAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5738c12ec4ccbfde346ee798d8ba1ded4bba5171806420a7a17b6d51afcbe7a8","last_reissued_at":"2026-07-05T07:32:45.435977Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:32:45.435977Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LLaMA Beyond English: An Empirical Study on Language Capability Transfer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Jun Zhao, Luhui Gao, Qi Zhang, Tao Gui, Xuanjing Huang, Zhihao Zhang","submitted_at":"2024-01-02T06:29:02Z","abstract_excerpt":"In recent times, substantial advancements have been witnessed in large language models (LLMs), exemplified by ChatGPT, showcasing remarkable proficiency across a range of complex tasks. However, many mainstream LLMs (e.g. LLaMA) are pretrained on English-dominant corpus, which limits their performance in other non-English languages. In this paper, we focus on how to effectively transfer the capabilities of language generation and following instructions to a non-English language. To answer this question, we conduct an extensive empirical investigation based on LLaMA, accumulating over 1440 GPU "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.01055","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/2401.01055/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":"2401.01055","created_at":"2026-07-05T07:32:45.436038+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.01055v2","created_at":"2026-07-05T07:32:45.436038+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.01055","created_at":"2026-07-05T07:32:45.436038+00:00"},{"alias_kind":"pith_short_12","alias_value":"K44MCLWEZS75","created_at":"2026-07-05T07:32:45.436038+00:00"},{"alias_kind":"pith_short_16","alias_value":"K44MCLWEZS754NDO","created_at":"2026-07-05T07:32:45.436038+00:00"},{"alias_kind":"pith_short_8","alias_value":"K44MCLWE","created_at":"2026-07-05T07:32:45.436038+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.06738","citing_title":"Modular Monolingual Adaptation using Pretrained Language Models","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15613","citing_title":"Toward LLMs Beyond English-Centric Development","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2602.14517","citing_title":"Large Language Models for Math Education in Low-Resource Languages: A Study in Sinhala and Tamil","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14366","citing_title":"Reinforcement Learning with Semantic Rewards Enables Low-Resource Language Expansion without Alignment Tax","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02971","citing_title":"Multilingual Safety Alignment via Self-Distillation","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02971","citing_title":"Multilingual Safety Alignment via Self-Distillation","ref_index":29,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/K44MCLWEZS754NDO46MNROQ55V","json":"https://pith.science/pith/K44MCLWEZS754NDO46MNROQ55V.json","graph_json":"https://pith.science/api/pith-number/K44MCLWEZS754NDO46MNROQ55V/graph.json","events_json":"https://pith.science/api/pith-number/K44MCLWEZS754NDO46MNROQ55V/events.json","paper":"https://pith.science/paper/K44MCLWE"},"agent_actions":{"view_html":"https://pith.science/pith/K44MCLWEZS754NDO46MNROQ55V","download_json":"https://pith.science/pith/K44MCLWEZS754NDO46MNROQ55V.json","view_paper":"https://pith.science/paper/K44MCLWE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.01055&json=true","fetch_graph":"https://pith.science/api/pith-number/K44MCLWEZS754NDO46MNROQ55V/graph.json","fetch_events":"https://pith.science/api/pith-number/K44MCLWEZS754NDO46MNROQ55V/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/K44MCLWEZS754NDO46MNROQ55V/action/timestamp_anchor","attest_storage":"https://pith.science/pith/K44MCLWEZS754NDO46MNROQ55V/action/storage_attestation","attest_author":"https://pith.science/pith/K44MCLWEZS754NDO46MNROQ55V/action/author_attestation","sign_citation":"https://pith.science/pith/K44MCLWEZS754NDO46MNROQ55V/action/citation_signature","submit_replication":"https://pith.science/pith/K44MCLWEZS754NDO46MNROQ55V/action/replication_record"}},"created_at":"2026-07-05T07:32:45.436038+00:00","updated_at":"2026-07-05T07:32:45.436038+00:00"}