{"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/exploring-scaling-trends-in-llm-robustness","title":"Scaling Trends in Language Model Robustness","arxiv_id":"2407.18213","date":"2024-07-25","proceeding":null,"authors":["Nikolaus Howe","Ian McKenzie","Oskar Hollinsworth","Michał Zajac","Tom Tseng","Aaron Tucker","Pierre-Luc Bacon","Adam Gleave"],"abstract":"Increasing model size has unlocked a dazzling array of capabilities in modern language models. At the same time, even frontier models remain vulnerable to jailbreaks and prompt injections, despite concerted efforts to make them robust. As both attack and defense gain access to more compute, and as models become larger, what happens to robustness? We argue that to answer this question requires a \\emph{scaling} approach, which we employ in an extensive study of language model robustness across several classification tasks, model families, and adversarial attacks. We find that in the absence of explicit safety training, larger models are not consistently more robust; however, scale improves sample efficiency in adversarial training, though it worsens compute efficiency. Further, we find that increasing attack compute smoothly improves attack success rate against both undefended and adversarially trained models. Finally, after exploring robustness transfer across attacks and threat models, we combine attack and defense scaling rates to study the offense-defense balance. We find that while attack scaling outpaces adversarial training across all models studied, larger adversarially trained models might give defense the advantage in the long run. These results underscore the utility of the scaling lens, and provide a paradigm for evaluating future attacks and defenses on frontier models.","url_abs":"https://arxiv.org/abs/2407.18213v5","url_pdf":"https://arxiv.org/pdf/2407.18213v5.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":"exploring-scaling-trends-in-llm-robustness","repo_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"adversarial-robustness","task_name":"Adversarial Robustness"},{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"model","task_name":"model"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2407.18213","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2407.18213"}},"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/AlignmentResearch/scaling-llm-robustness-paper","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran":12,"unverified":1},"by_repo_kind":{"official":{"samples":13,"ran":12,"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":"da2a3e19a7673714","entry":"broadcast_list_of_bools","repo":"AlignmentResearch/scaling-llm-robustness-paper","repo_kind":"official","path":"robust_llm/dist_utils.py","file_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper/blob/HEAD/robust_llm/dist_utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"da2a3e19a7673714"}},{"code_sha256_prefix":"0da004cea081d237","entry":"broadcast_list_of_floats","repo":"AlignmentResearch/scaling-llm-robustness-paper","repo_kind":"official","path":"robust_llm/dist_utils.py","file_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper/blob/HEAD/robust_llm/dist_utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"0da004cea081d237"}},{"code_sha256_prefix":"258b3739406f6c1a","entry":"broadcast_list_of_ints","repo":"AlignmentResearch/scaling-llm-robustness-paper","repo_kind":"official","path":"robust_llm/dist_utils.py","file_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper/blob/HEAD/robust_llm/dist_utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"258b3739406f6c1a"}},{"code_sha256_prefix":"63c28d2b87badc61","entry":"check_mount_health","repo":"AlignmentResearch/scaling-llm-robustness-paper","repo_kind":"official","path":"robust_llm/file_utils.py","file_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper/blob/HEAD/robust_llm/file_utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"63c28d2b87badc61"}},{"code_sha256_prefix":"68bcf1dfa30aec29","entry":"equal_ignore_padding","repo":"AlignmentResearch/scaling-llm-robustness-paper","repo_kind":"official","path":"robust_llm/mocks.py","file_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper/blob/HEAD/robust_llm/mocks.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"68bcf1dfa30aec29"}},{"code_sha256_prefix":"71575d7c8c147ebc","entry":"format_attack_status","repo":"AlignmentResearch/scaling-llm-robustness-paper","repo_kind":"official","path":"robust_llm/logging_utils.py","file_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper/blob/HEAD/robust_llm/logging_utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"71575d7c8c147ebc"}},{"code_sha256_prefix":"d1b5c3c3f3ee2124","entry":"format_time","repo":"AlignmentResearch/scaling-llm-robustness-paper","repo_kind":"official","path":"robust_llm/logging_utils.py","file_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper/blob/HEAD/robust_llm/logging_utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"d1b5c3c3f3ee2124"}},{"code_sha256_prefix":"1fbe44dd312fb51a","entry":"format_training_status","repo":"AlignmentResearch/scaling-llm-robustness-paper","repo_kind":"official","path":"robust_llm/logging_utils.py","file_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper/blob/HEAD/robust_llm/logging_utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"1fbe44dd312fb51a"}},{"code_sha256_prefix":"e0e72f918568631c","entry":"get_common_prefix","repo":"AlignmentResearch/scaling-llm-robustness-paper","repo_kind":"official","path":"robust_llm/models/caching_wrapped_model.py","file_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper/blob/HEAD/robust_llm/models/caching_wrapped_model.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"e0e72f918568631c"}},{"code_sha256_prefix":"c246fe4a192305e9","entry":"get_common_prefix_for_batch","repo":"AlignmentResearch/scaling-llm-robustness-paper","repo_kind":"official","path":"robust_llm/models/caching_wrapped_model.py","file_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper/blob/HEAD/robust_llm/models/caching_wrapped_model.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"c246fe4a192305e9"}},{"code_sha256_prefix":"69abe1b5b8d47711","entry":"get_n_adv_tr_rounds","repo":"AlignmentResearch/scaling-llm-robustness-paper","repo_kind":"official","path":"robust_llm/experiment_utils.py","file_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper/blob/HEAD/robust_llm/experiment_utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"69abe1b5b8d47711"}},{"code_sha256_prefix":"cbdb4b91ddba8d13","entry":"get_rounds_to_evaluate","repo":"AlignmentResearch/scaling-llm-robustness-paper","repo_kind":"official","path":"robust_llm/experiment_utils.py","file_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper/blob/HEAD/robust_llm/experiment_utils.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"cbdb4b91ddba8d13"}},{"code_sha256_prefix":"34acb4359ddfd45a","entry":"assert_dicts_equal","repo":"AlignmentResearch/scaling-llm-robustness-paper","repo_kind":"official","path":"robust_llm/debug_utils.py","file_url":"https://github.com/AlignmentResearch/scaling-llm-robustness-paper/blob/HEAD/robust_llm/debug_utils.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"34acb4359ddfd45a"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}