{"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/separating-tongue-from-thought-activation","title":"Separating Tongue from Thought: Activation Patching Reveals Language-Agnostic Concept Representations in Transformers","arxiv_id":"2411.08745","date":"2024-11-13","proceeding":null,"authors":["Clément Dumas","Chris Wendler","Veniamin Veselovsky","Giovanni Monea","Robert West"],"abstract":"A central question in multilingual language modeling is whether large language models (LLMs) develop a universal concept representation, disentangled from specific languages. In this paper, we address this question by analyzing latent representations (latents) during a word translation task in transformer-based LLMs. We strategically extract latents from a source translation prompt and insert them into the forward pass on a target translation prompt. By doing so, we find that the output language is encoded in the latent at an earlier layer than the concept to be translated. Building on this insight, we conduct two key experiments. First, we demonstrate that we can change the concept without changing the language and vice versa through activation patching alone. Second, we show that patching with the mean over latents across different languages does not impair and instead improves the models' performance in translating the concept. Our results provide evidence for the existence of language-agnostic concept representations within the investigated models.","url_abs":"https://arxiv.org/abs/2411.08745v3","url_pdf":"https://arxiv.org/pdf/2411.08745v3.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":"separating-tongue-from-thought-activation","repo_url":"https://github.com/butanium/llm-lang-agnostic","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"translation","task_name":"Translation"},{"task_slug":"word-translation","task_name":"Word Translation"}],"methods":[{"method_slug":"patching","method_name":"Patching"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2411.08745","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2411.08745"}},"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/butanium/llm-lang-agnostic","reach":{"status":"ok"}}],"summary":{"unverified":10},"by_repo_kind":{"official":{"samples":10,"ran":0,"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":"d9985dd3f6df938f","entry":"add_spaces","repo":"butanium/llm-lang-agnostic","repo_kind":"official","path":"src/prompt_tools.py","file_url":"https://github.com/butanium/llm-lang-agnostic/blob/HEAD/src/prompt_tools.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":"d9985dd3f6df938f"}},{"code_sha256_prefix":"f0cedcf96dce07ce","entry":"byte_string_to_list","repo":"butanium/llm-lang-agnostic","repo_kind":"official","path":"src/prompt_tools.py","file_url":"https://github.com/butanium/llm-lang-agnostic/blob/HEAD/src/prompt_tools.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":"f0cedcf96dce07ce"}},{"code_sha256_prefix":"d3d8b54b3ae77dd9","entry":"get_cloze_dataset","repo":"butanium/llm-lang-agnostic","repo_kind":"official","path":"src/load_dataset.py","file_url":"https://github.com/butanium/llm-lang-agnostic/blob/HEAD/src/load_dataset.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":"d3d8b54b3ae77dd9"}},{"code_sha256_prefix":"01d5c8d8d8623cf6","entry":"get_word_translation_dataset","repo":"butanium/llm-lang-agnostic","repo_kind":"official","path":"src/load_dataset.py","file_url":"https://github.com/butanium/llm-lang-agnostic/blob/HEAD/src/load_dataset.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":"01d5c8d8d8623cf6"}},{"code_sha256_prefix":"4a5f917729414302","entry":"k_subplots","repo":"butanium/llm-lang-agnostic","repo_kind":"official","path":"src/display_utils.py","file_url":"https://github.com/butanium/llm-lang-agnostic/blob/HEAD/src/display_utils.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":"4a5f917729414302"}},{"code_sha256_prefix":"407ca7792a91113c","entry":"load_cloze","repo":"butanium/llm-lang-agnostic","repo_kind":"official","path":"src/load_dataset.py","file_url":"https://github.com/butanium/llm-lang-agnostic/blob/HEAD/src/load_dataset.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":"407ca7792a91113c"}},{"code_sha256_prefix":"fa631efa3ef9c566","entry":"plot_ci_plus_heatmap","repo":"butanium/llm-lang-agnostic","repo_kind":"official","path":"src/display_utils.py","file_url":"https://github.com/butanium/llm-lang-agnostic/blob/HEAD/src/display_utils.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":"fa631efa3ef9c566"}},{"code_sha256_prefix":"db389b8baf11d2e3","entry":"repeat_prompt","repo":"butanium/llm-lang-agnostic","repo_kind":"official","path":"src/interventions.py","file_url":"https://github.com/butanium/llm-lang-agnostic/blob/HEAD/src/interventions.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":"db389b8baf11d2e3"}},{"code_sha256_prefix":"516d409723a7793f","entry":"token_prefixes","repo":"butanium/llm-lang-agnostic","repo_kind":"official","path":"src/prompt_tools.py","file_url":"https://github.com/butanium/llm-lang-agnostic/blob/HEAD/src/prompt_tools.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":"516d409723a7793f"}},{"code_sha256_prefix":"59c98f4117e2ce93","entry":"ulist","repo":"butanium/llm-lang-agnostic","repo_kind":"official","path":"src/utils.py","file_url":"https://github.com/butanium/llm-lang-agnostic/blob/HEAD/src/utils.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":"59c98f4117e2ce93"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}