{"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/did-the-model-understand-the-question","title":"Did the Model Understand the Question?","arxiv_id":"1805.05492","date":"2018-05-14","proceeding":"ACL 2018 7","authors":["Pramod Kaushik Mudrakarta","Ankur Taly","Mukund Sundararajan","Kedar Dhamdhere"],"abstract":"We analyze state-of-the-art deep learning models for three tasks: question\nanswering on (1) images, (2) tables, and (3) passages of text. Using the notion\nof \\emph{attribution} (word importance), we find that these deep networks often\nignore important question terms. Leveraging such behavior, we perturb questions\nto craft a variety of adversarial examples. Our strongest attacks drop the\naccuracy of a visual question answering model from $61.1\\%$ to $19\\%$, and that\nof a tabular question answering model from $33.5\\%$ to $3.3\\%$. Additionally,\nwe show how attributions can strengthen attacks proposed by Jia and Liang\n(2017) on paragraph comprehension models. Our results demonstrate that\nattributions can augment standard measures of accuracy and empower\ninvestigation of model performance. When a model is accurate but for the wrong\nreasons, attributions can surface erroneous logic in the model that indicates\ninadequacies in the test data.","url_abs":"http://arxiv.org/abs/1805.05492v1","url_pdf":"http://arxiv.org/pdf/1805.05492v1.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":"did-the-model-understand-the-question","repo_url":"https://github.com/pramodkaushik/acl18_results","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"did-the-model-understand-the-question","repo_url":"https://github.com/ankurtaly/Attributions","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"did-the-model-understand-the-question","repo_url":"https://github.com/ankurtaly/Integrated-Gradients","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"did-the-model-understand-the-question","repo_url":"https://github.com/pramodkaushik/np_analysis","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"question-answering","task_name":"Question Answering"},{"task_slug":"visual-question-answering-1","task_name":"Visual Question Answering"},{"task_slug":"visual-question-answering","task_name":"Visual Question Answering (VQA)"},{"task_slug":"model","task_name":"model"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1805.05492","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1805.05492"}},"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/pramodkaushik/np_analysis","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/pramodkaushik/acl18_results","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ankurtaly/Attributions","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ankurtaly/Integrated-Gradients","reach":{"status":"ok"}}],"summary":{"unverified":10},"by_repo_kind":{"listed":{"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":0,"samples":[{"code_sha256_prefix":"77535dcfef8b0dbc","entry":"LSTMCell","repo":"pramodkaushik/np_analysis","repo_kind":"listed","path":"neural_programmer/nn_utils.py","file_url":"https://github.com/pramodkaushik/np_analysis/blob/HEAD/neural_programmer/nn_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"77535dcfef8b0dbc"}},{"code_sha256_prefix":"ad5bb040fe1f5334","entry":"apply_dropout","repo":"pramodkaushik/np_analysis","repo_kind":"listed","path":"neural_programmer/nn_utils.py","file_url":"https://github.com/pramodkaushik/np_analysis/blob/HEAD/neural_programmer/nn_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ad5bb040fe1f5334"}},{"code_sha256_prefix":"bc1a05fa2d44edab","entry":"convert_to_int_2d_and_pad","repo":"pramodkaushik/np_analysis","repo_kind":"listed","path":"neural_programmer/data_utils.py","file_url":"https://github.com/pramodkaushik/np_analysis/blob/HEAD/neural_programmer/data_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"bc1a05fa2d44edab"}},{"code_sha256_prefix":"58b53508826d68cb","entry":"correct_unicode","repo":"pramodkaushik/np_analysis","repo_kind":"listed","path":"neural_programmer/wiki_data.py","file_url":"https://github.com/pramodkaushik/np_analysis/blob/HEAD/neural_programmer/wiki_data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"58b53508826d68cb"}},{"code_sha256_prefix":"6955954c1e19d195","entry":"get_embedding","repo":"pramodkaushik/np_analysis","repo_kind":"listed","path":"neural_programmer/nn_utils.py","file_url":"https://github.com/pramodkaushik/np_analysis/blob/HEAD/neural_programmer/nn_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"6955954c1e19d195"}},{"code_sha256_prefix":"30c358b2ca0a8aea","entry":"is_nan_or_inf","repo":"pramodkaushik/np_analysis","repo_kind":"listed","path":"neural_programmer/wiki_data.py","file_url":"https://github.com/pramodkaushik/np_analysis/blob/HEAD/neural_programmer/wiki_data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"30c358b2ca0a8aea"}},{"code_sha256_prefix":"f74d08d012c8804a","entry":"restore_model","repo":"pramodkaushik/np_analysis","repo_kind":"listed","path":"neural_programmer/notebook_utils.py","file_url":"https://github.com/pramodkaushik/np_analysis/blob/HEAD/neural_programmer/notebook_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"f74d08d012c8804a"}},{"code_sha256_prefix":"947a25244430a341","entry":"return_index","repo":"pramodkaushik/np_analysis","repo_kind":"listed","path":"neural_programmer/data_utils.py","file_url":"https://github.com/pramodkaushik/np_analysis/blob/HEAD/neural_programmer/data_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"947a25244430a341"}},{"code_sha256_prefix":"8378d8e8e76a6e45","entry":"strip_accents","repo":"pramodkaushik/np_analysis","repo_kind":"listed","path":"neural_programmer/wiki_data.py","file_url":"https://github.com/pramodkaushik/np_analysis/blob/HEAD/neural_programmer/wiki_data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"8378d8e8e76a6e45"}},{"code_sha256_prefix":"a40d31c6f62a9b43","entry":"word_lookup","repo":"pramodkaushik/np_analysis","repo_kind":"listed","path":"neural_programmer/data_utils.py","file_url":"https://github.com/pramodkaushik/np_analysis/blob/HEAD/neural_programmer/data_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"a40d31c6f62a9b43"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}