{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/multilingual-pretraining-and-instruction","title":"Multilingual Pretraining and Instruction Tuning Improve Cross-Lingual Knowledge Alignment, But Only Shallowly","arxiv_id":"2404.04659","date":"2024-04-06","proceeding":null,"authors":["Changjiang Gao","Hongda Hu","Peng Hu","Jiajun Chen","Jixing Li","ShuJian Huang"],"abstract":"Despite their strong ability to retrieve knowledge in English, current large language models show imbalance abilities in different languages. Two approaches are proposed to address this, i.e., multilingual pretraining and multilingual instruction tuning. However, whether and how do such methods contribute to the cross-lingual knowledge alignment inside the models is unknown. In this paper, we propose CLiKA, a systematic framework to assess the cross-lingual knowledge alignment of LLMs in the Performance, Consistency and Conductivity levels, and explored the effect of multilingual pretraining and instruction tuning on the degree of alignment. Results show that: while both multilingual pretraining and instruction tuning are beneficial for cross-lingual knowledge alignment, the training strategy needs to be carefully designed. Namely, continued pretraining improves the alignment of the target language at the cost of other languages, while mixed pretraining affect other languages less. Also, the overall cross-lingual knowledge alignment, especially in the conductivity level, is unsatisfactory for all tested LLMs, and neither multilingual pretraining nor instruction tuning can substantially improve the cross-lingual knowledge conductivity.","url_abs":"https://arxiv.org/abs/2404.04659v1","url_pdf":"https://arxiv.org/pdf/2404.04659v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"multilingual-pretraining-and-instruction","repo_url":"https://github.com/rivergao/clika","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2404.04659","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2404.04659"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/rivergao/clika","reach":{"status":"ok"}}],"summary":{"ran":8,"unverified":2},"by_repo_kind":{"official":{"samples":10,"ran":8,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":10,"samples":[{"code_sha256_prefix":"3baabd9ba0f3d725","entry":"add_single_prompt","repo":"rivergao/clika","repo_kind":"official","path":"inference.py","file_url":"https://github.com/rivergao/clika/blob/HEAD/inference.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"3baabd9ba0f3d725"}},{"code_sha256_prefix":"587728d8906a8a0f","entry":"clean_output","repo":"rivergao/clika","repo_kind":"official","path":"x_copa/eval_xcopa_performance_fixed_en_prompt.py","file_url":"https://github.com/rivergao/clika/blob/HEAD/x_copa/eval_xcopa_performance_fixed_en_prompt.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"587728d8906a8a0f"}},{"code_sha256_prefix":"7f90c0248efc97be","entry":"clean_output","repo":"rivergao/clika","repo_kind":"official","path":"x_copa/generate_input_data.py","file_url":"https://github.com/rivergao/clika/blob/HEAD/x_copa/generate_input_data.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"7f90c0248efc97be"}},{"code_sha256_prefix":"21f6187362cebc43","entry":"generate_inputs","repo":"rivergao/clika","repo_kind":"official","path":"inference.py","file_url":"https://github.com/rivergao/clika/blob/HEAD/inference.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"21f6187362cebc43"}},{"code_sha256_prefix":"2418263f82ff96e3","entry":"generate_ouptut_data","repo":"rivergao/clika","repo_kind":"official","path":"x_geo/eval_x_geo_generate_input.py","file_url":"https://github.com/rivergao/clika/blob/HEAD/x_geo/eval_x_geo_generate_input.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"2418263f82ff96e3"}},{"code_sha256_prefix":"9aa04940752faf54","entry":"get_responese","repo":"rivergao/clika","repo_kind":"official","path":"x_copa/eval_xcopa_performance_fixed_en_prompt.py","file_url":"https://github.com/rivergao/clika/blob/HEAD/x_copa/eval_xcopa_performance_fixed_en_prompt.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"9aa04940752faf54"}},{"code_sha256_prefix":"6190a523dae843c0","entry":"get_start_idx","repo":"rivergao/clika","repo_kind":"official","path":"inference.py","file_url":"https://github.com/rivergao/clika/blob/HEAD/inference.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"6190a523dae843c0"}},{"code_sha256_prefix":"0b1fcc66ae076d2c","entry":"translate_google","repo":"rivergao/clika","repo_kind":"official","path":"x_csqa/csqa_translate.py","file_url":"https://github.com/rivergao/clika/blob/HEAD/x_csqa/csqa_translate.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"0b1fcc66ae076d2c"}},{"code_sha256_prefix":"029cd7b13ff5d84a","entry":"eval_res","repo":"rivergao/clika","repo_kind":"official","path":"res_statistic.py","file_url":"https://github.com/rivergao/clika/blob/HEAD/res_statistic.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"029cd7b13ff5d84a"}},{"code_sha256_prefix":"f509a2d7f2d6010a","entry":"translate_google","repo":"rivergao/clika","repo_kind":"official","path":"x_copa/x_copa_translate.py","file_url":"https://github.com/rivergao/clika/blob/HEAD/x_copa/x_copa_translate.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"f509a2d7f2d6010a"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}