{"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/simplicity-prevails-rethinking-negative","title":"Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearning","arxiv_id":"2410.07163","date":"2024-10-09","proceeding":null,"authors":["Chongyu Fan","Jiancheng Liu","Licong Lin","Jinghan Jia","Ruiqi Zhang","Song Mei","Sijia Liu"],"abstract":"In this work, we address the problem of large language model (LLM) unlearning, aiming to remove unwanted data influences and associated model capabilities (e.g., copyrighted data or harmful content generation) while preserving essential model utilities, without the need for retraining from scratch. Despite the growing need for LLM unlearning, a principled optimization framework remains lacking. To this end, we revisit the state-of-the-art approach, negative preference optimization (NPO), and identify the issue of reference model bias, which could undermine NPO's effectiveness, particularly when unlearning forget data of varying difficulty. Given that, we propose a simple yet effective unlearning optimization framework, called SimNPO, showing that 'simplicity' in removing the reliance on a reference model (through the lens of simple preference optimization) benefits unlearning. We also provide deeper insights into SimNPO's advantages, supported by analysis using mixtures of Markov chains. Furthermore, we present extensive experiments validating SimNPO's superiority over existing unlearning baselines in benchmarks like TOFU and MUSE, and robustness against relearning attacks. Codes are available at https://github.com/OPTML-Group/Unlearn-Simple.","url_abs":"https://arxiv.org/abs/2410.07163v2","url_pdf":"https://arxiv.org/pdf/2410.07163v2.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":"simplicity-prevails-rethinking-negative","repo_url":"https://github.com/OPTML-Group/Unlearn-Simple","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"simplicity-prevails-rethinking-negative","repo_url":"https://github.com/optml-group/unlearn-smooth","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"large-language-model","task_name":"Large Language Model"}],"methods":[{"method_slug":null,"method_name":"Identify the Issue"},{"method_slug":"tofu","method_name":"Tofu"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2410.07163","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2410.07163"}},"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":"deterministic:regex_extraction","url":"https://github.com/OPTML-Group/Unlearn-Simple","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/optml-group/unlearn-smooth","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran":7,"ran_draft_wrong":1,"unverified":2},"by_repo_kind":{"official":{"samples":10,"ran":8,"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":"cdd171f20cd6f53d","entry":"collate_fn","repo":"OPTML-Group/Unlearn-Simple","repo_kind":"official","path":"TOFU/data_module.py","file_url":"https://github.com/OPTML-Group/Unlearn-Simple/blob/HEAD/TOFU/data_module.py","link_basis":"harvester_set","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":"cdd171f20cd6f53d"}},{"code_sha256_prefix":"88a4f13abd54a00c","entry":"compute_metrics","repo":"OPTML-Group/Unlearn-Simple","repo_kind":"official","path":"WMDP/src/unlearn/base.py","file_url":"https://github.com/OPTML-Group/Unlearn-Simple/blob/HEAD/WMDP/src/unlearn/base.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":"88a4f13abd54a00c"}},{"code_sha256_prefix":"04d9dfc17842cf17","entry":"cross_entropy_loss","repo":"OPTML-Group/Unlearn-Simple","repo_kind":"official","path":"synthetic/unlearn_utils.py","file_url":"https://github.com/OPTML-Group/Unlearn-Simple/blob/HEAD/synthetic/unlearn_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":"04d9dfc17842cf17"}},{"code_sha256_prefix":"649fc48067a48529","entry":"find_all_linear_names","repo":"OPTML-Group/Unlearn-Simple","repo_kind":"official","path":"TOFU/finetune.py","file_url":"https://github.com/OPTML-Group/Unlearn-Simple/blob/HEAD/TOFU/finetune.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"649fc48067a48529"}},{"code_sha256_prefix":"75df7d35faaa9ff0","entry":"get_activation","repo":"OPTML-Group/Unlearn-Simple","repo_kind":"official","path":"synthetic/model.py","file_url":"https://github.com/OPTML-Group/Unlearn-Simple/blob/HEAD/synthetic/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":"75df7d35faaa9ff0"}},{"code_sha256_prefix":"650d5f6bfcd7d9bb","entry":"get_loss","repo":"OPTML-Group/Unlearn-Simple","repo_kind":"official","path":"WMDP/src/unlearn/base.py","file_url":"https://github.com/OPTML-Group/Unlearn-Simple/blob/HEAD/WMDP/src/unlearn/base.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":"650d5f6bfcd7d9bb"}},{"code_sha256_prefix":"f8e37979490cc066","entry":"grad_ascent_loss","repo":"OPTML-Group/Unlearn-Simple","repo_kind":"official","path":"synthetic/unlearn_utils.py","file_url":"https://github.com/OPTML-Group/Unlearn-Simple/blob/HEAD/synthetic/unlearn_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":"f8e37979490cc066"}},{"code_sha256_prefix":"3359083515d97775","entry":"grad_descent_loss","repo":"OPTML-Group/Unlearn-Simple","repo_kind":"official","path":"synthetic/unlearn_utils.py","file_url":"https://github.com/OPTML-Group/Unlearn-Simple/blob/HEAD/synthetic/unlearn_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":"3359083515d97775"}},{"code_sha256_prefix":"7dec325d088c0b76","entry":"convert_raw_data_to_model_format","repo":"OPTML-Group/Unlearn-Simple","repo_kind":"official","path":"TOFU/data_module.py","file_url":"https://github.com/OPTML-Group/Unlearn-Simple/blob/HEAD/TOFU/data_module.py","link_basis":"harvester_set","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":"7dec325d088c0b76"}},{"code_sha256_prefix":"52f769b837c6e32a","entry":"unlearncollector","repo":"OPTML-Group/Unlearn-Simple","repo_kind":"official","path":"TOFU/data_module.py","file_url":"https://github.com/OPTML-Group/Unlearn-Simple/blob/HEAD/TOFU/data_module.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":"52f769b837c6e32a"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}