{"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/copy-suppression-comprehensively","title":"Copy Suppression: Comprehensively Understanding an Attention Head","arxiv_id":"2310.04625","date":"2023-10-06","proceeding":null,"authors":["Callum McDougall","Arthur Conmy","Cody Rushing","Thomas McGrath","Neel Nanda"],"abstract":"We present a single attention head in GPT-2 Small that has one main role across the entire training distribution. If components in earlier layers predict a certain token, and this token appears earlier in the context, the head suppresses it: we call this copy suppression. Attention Head 10.7 (L10H7) suppresses naive copying behavior which improves overall model calibration. This explains why multiple prior works studying certain narrow tasks found negative heads that systematically favored the wrong answer. We uncover the mechanism that the Negative Heads use for copy suppression with weights-based evidence and are able to explain 76.9% of the impact of L10H7 in GPT-2 Small. To the best of our knowledge, this is the most comprehensive description of the complete role of a component in a language model to date. One major effect of copy suppression is its role in self-repair. Self-repair refers to how ablating crucial model components results in downstream neural network parts compensating for this ablation. Copy suppression leads to self-repair: if an initial overconfident copier is ablated, then there is nothing to suppress. We show that self-repair is implemented by several mechanisms, one of which is copy suppression, which explains 39% of the behavior in a narrow task. Interactive visualisations of the copy suppression phenomena may be seen at our web app https://copy-suppression.streamlit.app/","url_abs":"https://arxiv.org/abs/2310.04625v1","url_pdf":"https://arxiv.org/pdf/2310.04625v1.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":"copy-suppression-comprehensively","repo_url":"https://github.com/callummcdougall/seri-mats-2023-streamlit-pages","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"jax","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"language-modelling","task_name":"Language Modelling"}],"methods":[{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"attention-dropout","method_name":"Attention Dropout"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"cosine-annealing","method_name":"Cosine Annealing"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"discriminative-fine-tuning","method_name":"Discriminative Fine-Tuning"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"gpt-2","method_name":"GPT-2"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"linear-warmup-with-cosine-annealing","method_name":"Linear Warmup With Cosine Annealing"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"weight-decay","method_name":"Weight Decay"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2310.04625","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2310.04625"}},"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/callummcdougall/seri-mats-2023-streamlit-pages","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran":3,"unverified":2},"by_repo_kind":{"official":{"samples":5,"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":0,"samples":[{"code_sha256_prefix":"ba931d66d841af5d","entry":"get_duplicate_token_head_detection_pattern","repo":"callummcdougall/seri-mats-2023-streamlit-pages","repo_kind":"official","path":"transformer_lens/head_detector.py","file_url":"https://github.com/callummcdougall/seri-mats-2023-streamlit-pages/blob/HEAD/transformer_lens/head_detector.py","link_basis":"harvester_set","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":"ba931d66d841af5d"}},{"code_sha256_prefix":"79a2d1104ca128a0","entry":"get_previous_token_head_detection_pattern","repo":"callummcdougall/seri-mats-2023-streamlit-pages","repo_kind":"official","path":"transformer_lens/head_detector.py","file_url":"https://github.com/callummcdougall/seri-mats-2023-streamlit-pages/blob/HEAD/transformer_lens/head_detector.py","link_basis":"harvester_set","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":"79a2d1104ca128a0"}},{"code_sha256_prefix":"49d7660823635dab","entry":"sanity_check","repo":"callummcdougall/seri-mats-2023-streamlit-pages","repo_kind":"official","path":"transformer_lens/evals.py","file_url":"https://github.com/callummcdougall/seri-mats-2023-streamlit-pages/blob/HEAD/transformer_lens/evals.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":"49d7660823635dab"}},{"code_sha256_prefix":"a40a57f1e417e33c","entry":"convert_hf_model_config","repo":"callummcdougall/seri-mats-2023-streamlit-pages","repo_kind":"official","path":"transformer_lens/loading_from_pretrained.py","file_url":"https://github.com/callummcdougall/seri-mats-2023-streamlit-pages/blob/HEAD/transformer_lens/loading_from_pretrained.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":"a40a57f1e417e33c"}},{"code_sha256_prefix":"e5d367becb0bf75f","entry":"get_official_model_name","repo":"callummcdougall/seri-mats-2023-streamlit-pages","repo_kind":"official","path":"transformer_lens/loading_from_pretrained.py","file_url":"https://github.com/callummcdougall/seri-mats-2023-streamlit-pages/blob/HEAD/transformer_lens/loading_from_pretrained.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":"e5d367becb0bf75f"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}