{"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/should-we-really-edit-language-models-on-the","title":"Should We Really Edit Language Models? On the Evaluation of Edited Language Models","arxiv_id":"2410.18785","date":"2024-10-24","proceeding":null,"authors":["Qi Li","Xiang Liu","Zhenheng Tang","Peijie Dong","Zeyu Li","Xinglin Pan","Xiaowen Chu"],"abstract":"Model editing has become an increasingly popular alternative for efficiently updating knowledge within language models. Current methods mainly focus on reliability, generalization, and locality, with many methods excelling across these criteria. Some recent works disclose the pitfalls of these editing methods such as knowledge distortion or conflict. However, the general abilities of post-edited language models remain unexplored. In this paper, we perform a comprehensive evaluation on various editing methods and different language models, and have following findings. (1) Existing editing methods lead to inevitable performance deterioration on general benchmarks, indicating that existing editing methods maintain the general abilities of the model within only a few dozen edits. When the number of edits is slightly large, the intrinsic knowledge structure of the model is disrupted or even completely damaged. (2) Instruction-tuned models are more robust to editing, showing less performance drop on general knowledge after editing. (3) Language model with large scale is more resistant to editing compared to small model. (4) The safety of the edited model, is significantly weakened, even for those safety-aligned models. Our findings indicate that current editing methods are only suitable for small-scale knowledge updates within language models, which motivates further research on more practical and reliable editing methods. The details of code and reproduction can be found in https://github.com/lqinfdim/EditingEvaluation.","url_abs":"https://arxiv.org/abs/2410.18785v1","url_pdf":"https://arxiv.org/pdf/2410.18785v1.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":"should-we-really-edit-language-models-on-the","repo_url":"https://github.com/lqinfdim/editingevaluation","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"general-knowledge","task_name":"General Knowledge"},{"task_slug":"model-editing","task_name":"Model Editing"}],"methods":[{"method_slug":"focus","method_name":"Focus"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2410.18785","atlas_url":"https://app.syntology.ai/?focus=2410.18785","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2410.18785"}},"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":"deterministic:regex_extraction","url":"https://github.com/zjunlp/EasyEdit","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/lqinfdim/editingevaluation","reach":null}],"summary":{"ran_draft_wrong":3,"ran_fixture":1,"unverified":6},"by_repo_kind":{"official":{"samples":1,"ran":1,"repositories":1},"found_in_text":{"samples":9,"ran":3,"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":1,"samples":[{"code_sha256_prefix":"27659a9c234ffb11","entry":"binary_log_probs","repo":"zjunlp/EasyEdit","repo_kind":"found_in_text","path":"easyeditor/trainer/losses.py","file_url":"https://github.com/zjunlp/EasyEdit/blob/HEAD/easyeditor/trainer/losses.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":"27659a9c234ffb11"}},{"code_sha256_prefix":"3d4ef690427737b3","entry":"get_mend_dataset_path","repo":"lqinfdim/editingevaluation","repo_kind":"official","path":"edit.py","file_url":"https://github.com/lqinfdim/editingevaluation/blob/HEAD/edit.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"3d4ef690427737b3"}},{"code_sha256_prefix":"e078eeed20f03838","entry":"masked_mean","repo":"zjunlp/EasyEdit","repo_kind":"found_in_text","path":"easyeditor/trainer/losses.py","file_url":"https://github.com/zjunlp/EasyEdit/blob/HEAD/easyeditor/trainer/losses.py","link_basis":"harvester_set","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"e078eeed20f03838"}},{"code_sha256_prefix":"70f6ab8bde55420e","entry":"recursive_copy","repo":"zjunlp/EasyEdit","repo_kind":"found_in_text","path":"easyeditor/util/nethook.py","file_url":"https://github.com/zjunlp/EasyEdit/blob/HEAD/easyeditor/util/nethook.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"70f6ab8bde55420e"}},{"code_sha256_prefix":"73316c06cd2dccd6","entry":"get_hidden_state","repo":"zjunlp/EasyEdit","repo_kind":"found_in_text","path":"easyeditor/util/nethook.py","file_url":"https://github.com/zjunlp/EasyEdit/blob/HEAD/easyeditor/util/nethook.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":"73316c06cd2dccd6"}},{"code_sha256_prefix":"f581ae26fe0ddd9b","entry":"get_model","repo":"zjunlp/EasyEdit","repo_kind":"found_in_text","path":"easyeditor/trainer/models.py","file_url":"https://github.com/zjunlp/EasyEdit/blob/HEAD/easyeditor/trainer/models.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":"f581ae26fe0ddd9b"}},{"code_sha256_prefix":"2defe42cf1a7f301","entry":"get_tokenizer","repo":"zjunlp/EasyEdit","repo_kind":"found_in_text","path":"easyeditor/trainer/models.py","file_url":"https://github.com/zjunlp/EasyEdit/blob/HEAD/easyeditor/trainer/models.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":"2defe42cf1a7f301"}},{"code_sha256_prefix":"bcc592aa5c1e2bb3","entry":"kl_loc_loss","repo":"zjunlp/EasyEdit","repo_kind":"found_in_text","path":"easyeditor/trainer/losses.py","file_url":"https://github.com/zjunlp/EasyEdit/blob/HEAD/easyeditor/trainer/losses.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":"bcc592aa5c1e2bb3"}},{"code_sha256_prefix":"6b9defc3514e3b3a","entry":"load_chat_template","repo":"zjunlp/EasyEdit","repo_kind":"found_in_text","path":"steer/models/get_model.py","file_url":"https://github.com/zjunlp/EasyEdit/blob/HEAD/steer/models/get_model.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":"6b9defc3514e3b3a"}},{"code_sha256_prefix":"a9cc0b3784406d31","entry":"replace_hidden_state","repo":"zjunlp/EasyEdit","repo_kind":"found_in_text","path":"easyeditor/util/nethook.py","file_url":"https://github.com/zjunlp/EasyEdit/blob/HEAD/easyeditor/util/nethook.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":"a9cc0b3784406d31"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}