{"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/stop-reasoning-when-multimodal-llms-with","title":"Stop Reasoning! When Multimodal LLM with Chain-of-Thought Reasoning Meets Adversarial Image","arxiv_id":"2402.14899","date":"2024-02-22","proceeding":null,"authors":["Zefeng Wang","Zhen Han","Shuo Chen","Fan Xue","Zifeng Ding","Xun Xiao","Volker Tresp","Philip Torr","Jindong Gu"],"abstract":"Multimodal LLMs (MLLMs) with a great ability of text and image understanding have received great attention. To achieve better reasoning with MLLMs, Chain-of-Thought (CoT) reasoning has been widely explored, which further promotes MLLMs' explainability by giving intermediate reasoning steps. Despite the strong power demonstrated by MLLMs in multimodal reasoning, recent studies show that MLLMs still suffer from adversarial images. This raises the following open questions: Does CoT also enhance the adversarial robustness of MLLMs? What do the intermediate reasoning steps of CoT entail under adversarial attacks? To answer these questions, we first generalize existing attacks to CoT-based inferences by attacking the two main components, i.e., rationale and answer. We find that CoT indeed improves MLLMs' adversarial robustness against the existing attack methods by leveraging the multi-step reasoning process, but not substantially. Based on our findings, we further propose a novel attack method, termed as stop-reasoning attack, that attacks the model while bypassing the CoT reasoning process. Experiments on three MLLMs and two visual reasoning datasets verify the effectiveness of our proposed method. We show that stop-reasoning attack can result in misled predictions and outperform baseline attacks by a significant margin.","url_abs":"https://arxiv.org/abs/2402.14899v3","url_pdf":"https://arxiv.org/pdf/2402.14899v3.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":"stop-reasoning-when-multimodal-llms-with","repo_url":"https://github.com/aipenguin/stopreasoning","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"adversarial-robustness","task_name":"Adversarial Robustness"},{"task_slug":"multimodal-reasoning","task_name":"Multimodal Reasoning"},{"task_slug":"visual-reasoning","task_name":"Visual Reasoning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2402.14899","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2402.14899"}},"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/aipenguin/stopreasoning","reach":{"status":"ok"}}],"summary":{"ran_fixture":1,"ran_violates":1,"ran":3,"ran_draft_wrong":1,"unverified":1},"by_repo_kind":{"official":{"samples":7,"ran":6,"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":7,"samples":[{"code_sha256_prefix":"9b4dff79d5e6102c","entry":"apply_rotary_pos_emb","repo":"aipenguin/stopreasoning","repo_kind":"official","path":"minigpt4/models/modeling_llama.py","file_url":"https://github.com/aipenguin/stopreasoning/blob/HEAD/minigpt4/models/modeling_llama.py","link_basis":"harvester_set","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"9b4dff79d5e6102c"}},{"code_sha256_prefix":"4cb732f513d69dfd","entry":"disabled_train","repo":"aipenguin/stopreasoning","repo_kind":"official","path":"minigpt4/models/blip2.py","file_url":"https://github.com/aipenguin/stopreasoning/blob/HEAD/minigpt4/models/blip2.py","link_basis":"harvester_set","language":"python","status":"ran_violates","verification_level":1,"contract_check":"VIOLATES","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"4cb732f513d69dfd"}},{"code_sha256_prefix":"0cef706777d8cdff","entry":"load_data","repo":"aipenguin/stopreasoning","repo_kind":"official","path":"utils/get_accuracy.py","file_url":"https://github.com/aipenguin/stopreasoning/blob/HEAD/utils/get_accuracy.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":"0cef706777d8cdff"}},{"code_sha256_prefix":"d5adbb94c8913bad","entry":"load_loss","repo":"aipenguin/stopreasoning","repo_kind":"official","path":"utils/get_accuracy.py","file_url":"https://github.com/aipenguin/stopreasoning/blob/HEAD/utils/get_accuracy.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":"d5adbb94c8913bad"}},{"code_sha256_prefix":"b99eea6376d1e212","entry":"rotate_half","repo":"aipenguin/stopreasoning","repo_kind":"official","path":"minigpt4/models/modeling_llama.py","file_url":"https://github.com/aipenguin/stopreasoning/blob/HEAD/minigpt4/models/modeling_llama.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"b99eea6376d1e212"}},{"code_sha256_prefix":"cb33571427334815","entry":"tile","repo":"aipenguin/stopreasoning","repo_kind":"official","path":"minigpt4/models/base_model.py","file_url":"https://github.com/aipenguin/stopreasoning/blob/HEAD/minigpt4/models/base_model.py","link_basis":"harvester_set","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":"cb33571427334815"}},{"code_sha256_prefix":"0ec9fc2025c16f65","entry":"all_gather_with_grad","repo":"aipenguin/stopreasoning","repo_kind":"official","path":"minigpt4/models/base_model.py","file_url":"https://github.com/aipenguin/stopreasoning/blob/HEAD/minigpt4/models/base_model.py","link_basis":"harvester_set","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":"0ec9fc2025c16f65"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}