{"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/arxiv-2601-06787","title":"Garbage Attention in Large Language Models: BOS Sink Heads and Sink-aware Pruning","arxiv_id":"2601.06787","date":"2026-01-11","proceeding":null,"authors":["Jaewon Sok","Jewon Yeom","Seonghyeon Park","Jeongjae Park","Taesup Kim"],"abstract":"Large Language Models (LLMs) are known to contain significant redundancy, yet a systematic explanation for why certain components, particularly in higher layers, are more redundant has remained elusive. In this work, we identify the BOS sink phenomenon as a key mechanism driving this layer-wise sensitivity. We show that attention heads with high BOS sink scores are strongly associated with functional redundancy: such heads, especially in deeper layers, contribute little to predictive performance and effectively serve as \\emph{dumping grounds} for superfluous attention weights. This provides a concrete functional explanation for the structural redundancy reported in prior studies. Leveraging this insight, we introduce a simple pruning strategy that removes high-BOS sink heads. Experiments on Gemma-3, Llama-3.1, and Qwen3 demonstrate that this approach identifies redundant transformer components more reliably than weight- or activation-based criteria, while preserving performance close to dense baselines even under aggressive pruning. Moreover, we find that the behavior of sink heads remains stable across different sequence lengths. Overall, our results suggest that structural properties of attention offer a more intuitive and robust basis for model compression than magnitude-based methods.","url_abs":"https://arxiv.org/abs/2601.06787","url_pdf":"https://arxiv.org/pdf/2601.06787","source":{"archive":null,"snapshot":"2025-07-28","note":"not in the Papers with Code archive (frozen at the snapshot)","row_kind":"graph","title_abstract_authors_date":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)"},"code_links":[],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2601.06787","atlas_url":"https://app.syntology.ai/?focus=2601.06787","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2601.06787"}},"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":[{"mentioned_in_github":null,"is_official":null,"provenance":"deterministic:regex_extraction","mentioned_in_paper":null,"url":"https://github.com/EleutherAI/lm-evaluation-harness","reach":null},{"mentioned_in_github":null,"is_official":null,"provenance":"deterministic:regex_extraction","mentioned_in_paper":null,"url":"https://github.com/CASIA-LMC-Lab/FLAP","reach":null}],"summary":{"ran":2,"ran_honours":1,"unverified":7},"by_repo_kind":{"found_in_text":{"samples":10,"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":"faf763d25abb99d6","entry":"BiasGPT","repo":"CASIA-LMC-Lab/FLAP","repo_kind":"found_in_text","path":"lib/prune.py","file_url":"https://github.com/CASIA-LMC-Lab/FLAP/blob/HEAD/lib/prune.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"faf763d25abb99d6"}},{"code_sha256_prefix":"2d5c7a7358bcd465","entry":"TokenizerWrapper","repo":"CASIA-LMC-Lab/FLAP","repo_kind":"found_in_text","path":"lib/prune.py","file_url":"https://github.com/CASIA-LMC-Lab/FLAP/blob/HEAD/lib/prune.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"2d5c7a7358bcd465"}},{"code_sha256_prefix":"afa39472138ad65e","entry":"cal_remove_neuron","repo":"CASIA-LMC-Lab/FLAP","repo_kind":"found_in_text","path":"lib/prune.py","file_url":"https://github.com/CASIA-LMC-Lab/FLAP/blob/HEAD/lib/prune.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"afa39472138ad65e"}},{"code_sha256_prefix":"1e3629d8fc0ca6c7","entry":"compress","repo":"CASIA-LMC-Lab/FLAP","repo_kind":"found_in_text","path":"lib/prune.py","file_url":"https://github.com/CASIA-LMC-Lab/FLAP/blob/HEAD/lib/prune.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"1e3629d8fc0ca6c7"}},{"code_sha256_prefix":"6ccf0fe3fd4574d5","entry":"get_c4","repo":"CASIA-LMC-Lab/FLAP","repo_kind":"found_in_text","path":"lib/prune.py","file_url":"https://github.com/CASIA-LMC-Lab/FLAP/blob/HEAD/lib/prune.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"6ccf0fe3fd4574d5"}},{"code_sha256_prefix":"21ef6425a9e8b029","entry":"get_loaders","repo":"CASIA-LMC-Lab/FLAP","repo_kind":"found_in_text","path":"lib/prune.py","file_url":"https://github.com/CASIA-LMC-Lab/FLAP/blob/HEAD/lib/prune.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"21ef6425a9e8b029"}},{"code_sha256_prefix":"1d035152616b1d19","entry":"get_ptb","repo":"CASIA-LMC-Lab/FLAP","repo_kind":"found_in_text","path":"lib/prune.py","file_url":"https://github.com/CASIA-LMC-Lab/FLAP/blob/HEAD/lib/prune.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"1d035152616b1d19"}},{"code_sha256_prefix":"2d79e7ef9f571f3d","entry":"get_wikitext2","repo":"CASIA-LMC-Lab/FLAP","repo_kind":"found_in_text","path":"lib/prune.py","file_url":"https://github.com/CASIA-LMC-Lab/FLAP/blob/HEAD/lib/prune.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"2d79e7ef9f571f3d"}},{"code_sha256_prefix":"2519b3efd481ad47","entry":"prepare_calibration_input","repo":"CASIA-LMC-Lab/FLAP","repo_kind":"found_in_text","path":"lib/prune.py","file_url":"https://github.com/CASIA-LMC-Lab/FLAP/blob/HEAD/lib/prune.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"2519b3efd481ad47"}},{"code_sha256_prefix":"61d8056ebaa89693","entry":"prune_flap","repo":"CASIA-LMC-Lab/FLAP","repo_kind":"found_in_text","path":"lib/prune.py","file_url":"https://github.com/CASIA-LMC-Lab/FLAP/blob/HEAD/lib/prune.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"61d8056ebaa89693"}}]},"arxiv_metadata":{"licence":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)","fields":["title","abstract","authors","date"],"primary_category":"cs.CL","source":"arxiv_2026.jsonl"},"syntology_extracted_results":null}