{"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/from-form-s-to-meaning-probing-the-semantic","title":"From Form(s) to Meaning: Probing the Semantic Depths of Language Models Using Multisense Consistency","arxiv_id":"2404.12145","date":"2024-04-18","proceeding":null,"authors":["Xenia Ohmer","Elia Bruni","Dieuwke Hupkes"],"abstract":"The staggering pace with which the capabilities of large language models (LLMs) are increasing, as measured by a range of commonly used natural language understanding (NLU) benchmarks, raises many questions regarding what \"understanding\" means for a language model and how it compares to human understanding. This is especially true since many LLMs are exclusively trained on text, casting doubt on whether their stellar benchmark performances are reflective of a true understanding of the problems represented by these benchmarks, or whether LLMs simply excel at uttering textual forms that correlate with what someone who understands the problem would say. In this philosophically inspired work, we aim to create some separation between form and meaning, with a series of tests that leverage the idea that world understanding should be consistent across presentational modes - inspired by Fregean senses - of the same meaning. Specifically, we focus on consistency across languages as well as paraphrases. Taking GPT-3.5 as our object of study, we evaluate multisense consistency across five different languages and various tasks. We start the evaluation in a controlled setting, asking the model for simple facts, and then proceed with an evaluation on four popular NLU benchmarks. We find that the model's multisense consistency is lacking and run several follow-up analyses to verify that this lack of consistency is due to a sense-dependent task understanding. We conclude that, in this aspect, the understanding of LLMs is still quite far from being consistent and human-like, and deliberate on how this impacts their utility in the context of learning about human language and understanding.","url_abs":"https://arxiv.org/abs/2404.12145v1","url_pdf":"https://arxiv.org/pdf/2404.12145v1.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":"from-form-s-to-meaning-probing-the-semantic","repo_url":"https://github.com/facebookresearch/multisense_consistency","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"form","task_name":"Form"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"natural-language-understanding","task_name":"Natural Language Understanding"}],"methods":[{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"attention-dropout","method_name":"Attention Dropout"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"cosine-annealing","method_name":"Cosine Annealing"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"focus","method_name":"Focus"},{"method_slug":"gpt-3","method_name":"GPT-3"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"linear-warmup-with-cosine-annealing","method_name":"Linear Warmup With Cosine Annealing"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"weight-decay","method_name":"Weight Decay"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2404.12145","atlas_url":"https://app.syntology.ai/?focus=2404.12145","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2404.12145"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-25T09:33:49+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/facebookresearch/multisense_consistency","reach":{"status":"ok","spdx":"NOASSERTION"}}],"summary":{"ran":9,"unverified":2},"by_repo_kind":{"official":{"samples":11,"ran":9,"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":11,"samples":[{"code_sha256_prefix":"08ff23d54b057d4a","entry":"accuracy_benchmarks","repo":"facebookresearch/multisense_consistency","repo_kind":"official","path":"analysis_utils/load_scores.py","file_url":"https://github.com/facebookresearch/multisense_consistency/blob/HEAD/analysis_utils/load_scores.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"08ff23d54b057d4a"}},{"code_sha256_prefix":"26e46b2668ac3bda","entry":"accuracy_facts","repo":"facebookresearch/multisense_consistency","repo_kind":"official","path":"analysis_utils/load_scores.py","file_url":"https://github.com/facebookresearch/multisense_consistency/blob/HEAD/analysis_utils/load_scores.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"26e46b2668ac3bda"}},{"code_sha256_prefix":"cedf458936035929","entry":"alternative_formulation","repo":"facebookresearch/multisense_consistency","repo_kind":"official","path":"template_data/addition/arithmetics_utils.py","file_url":"https://github.com/facebookresearch/multisense_consistency/blob/HEAD/template_data/addition/arithmetics_utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"cedf458936035929"}},{"code_sha256_prefix":"36e1ff0361e9c46b","entry":"cut_double_quotes","repo":"facebookresearch/multisense_consistency","repo_kind":"official","path":"utils/response_standardization.py","file_url":"https://github.com/facebookresearch/multisense_consistency/blob/HEAD/utils/response_standardization.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"36e1ff0361e9c46b"}},{"code_sha256_prefix":"f2a2dd2791ad4316","entry":"dutch_number_generator","repo":"facebookresearch/multisense_consistency","repo_kind":"official","path":"template_data/addition/arithmetics_utils.py","file_url":"https://github.com/facebookresearch/multisense_consistency/blob/HEAD/template_data/addition/arithmetics_utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"f2a2dd2791ad4316"}},{"code_sha256_prefix":"8325d0eddc238131","entry":"exact_match_score","repo":"facebookresearch/multisense_consistency","repo_kind":"official","path":"utils/eval_metrics.py","file_url":"https://github.com/facebookresearch/multisense_consistency/blob/HEAD/utils/eval_metrics.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"8325d0eddc238131"}},{"code_sha256_prefix":"a51c9ee8226e3742","entry":"f1_score","repo":"facebookresearch/multisense_consistency","repo_kind":"official","path":"utils/eval_metrics.py","file_url":"https://github.com/facebookresearch/multisense_consistency/blob/HEAD/utils/eval_metrics.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"a51c9ee8226e3742"}},{"code_sha256_prefix":"9f890f8bdba1b1cc","entry":"normalize_answer","repo":"facebookresearch/multisense_consistency","repo_kind":"official","path":"utils/eval_metrics.py","file_url":"https://github.com/facebookresearch/multisense_consistency/blob/HEAD/utils/eval_metrics.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"9f890f8bdba1b1cc"}},{"code_sha256_prefix":"7036e51ad025f0a9","entry":"standardize_qa_response","repo":"facebookresearch/multisense_consistency","repo_kind":"official","path":"utils/response_standardization.py","file_url":"https://github.com/facebookresearch/multisense_consistency/blob/HEAD/utils/response_standardization.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"7036e51ad025f0a9"}},{"code_sha256_prefix":"ffbb0b1462b308dd","entry":"number_to_words","repo":"facebookresearch/multisense_consistency","repo_kind":"official","path":"template_data/addition/arithmetics_utils.py","file_url":"https://github.com/facebookresearch/multisense_consistency/blob/HEAD/template_data/addition/arithmetics_utils.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"ffbb0b1462b308dd"}},{"code_sha256_prefix":"9cf2b2891163f7cc","entry":"standardize_classification_response","repo":"facebookresearch/multisense_consistency","repo_kind":"official","path":"utils/response_standardization.py","file_url":"https://github.com/facebookresearch/multisense_consistency/blob/HEAD/utils/response_standardization.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"9cf2b2891163f7cc"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}