{"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-2602-22175","title":"DySCO: Dynamic Attention-Scaling Decoding for Long-Context Language Models","arxiv_id":"2602.22175","date":"2026-02-25","proceeding":null,"authors":["Xi Ye","Wuwei Zhang","Fangcong Yin","Howard Yen","Danqi Chen"],"abstract":"Understanding and reasoning over long contexts is a crucial capability for language models (LMs). Although recent models support increasingly long context windows, their accuracy often deteriorates as input length grows. In practice, models often struggle to keep attention aligned with the most relevant context throughout decoding. In this work, we propose DYSCO, a novel decoding algorithm for improving long-context reasoning. DYSCO leverages retrieval heads--a subset of attention heads specialized for longcontext retrieval--to identify task-relevant tokens at each decoding step and explicitly up-weight them. By doing so, DYSCO dynamically adjusts attention during generation to better utilize relevant context. The method is training-free and can be applied directly to any off-the-shelf LMs. Across multiple instruction-tuned and reasoning models, DYSCO consistently improves performance on challenging long-context reasoning benchmarks, yielding relative gains of up to 25% on MRCR and LongBenchV2 at 128K context length with modest additional compute. Further analysis highlights the importance of both dynamic attention rescaling and retrievalhead guided selection for the effectiveness of the method, while providing interpretability insights into decoding-time attention behavior. Our code is available at https://github.com/princeton-pli/DySCO.","url_abs":"https://arxiv.org/abs/2602.22175","url_pdf":"https://arxiv.org/pdf/2602.22175","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":{"atlas_url":"https://app.syntology.ai/?focus=2602.22175","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2602.22175"}},"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":[{"mentioned_in_github":null,"is_official":null,"provenance":"deterministic:regex_extraction","mentioned_in_paper":null,"url":"https://github.com/princeton-pli/DySCO","reach":null}],"summary":{"ran":1,"ran_fixture":3,"ran_draft_wrong":2,"unverified":2},"by_repo_kind":{"found_in_text":{"samples":8,"ran":6,"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":"9a1efbfcb2edfa35","entry":"RescaleConfig","repo":"princeton-pli/DySCO","repo_kind":"found_in_text","path":"dysco/custom_mixin.py","file_url":"https://github.com/princeton-pli/DySCO/blob/HEAD/dysco/custom_mixin.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":"9a1efbfcb2edfa35"}},{"code_sha256_prefix":"778b5fb28b268923","entry":"_aggregate_head_attention","repo":"princeton-pli/DySCO","repo_kind":"found_in_text","path":"dysco/custom_mixin.py","file_url":"https://github.com/princeton-pli/DySCO/blob/HEAD/dysco/custom_mixin.py","link_basis":"first_harvest_node","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":"778b5fb28b268923"}},{"code_sha256_prefix":"55123587eebf5f9d","entry":"_build_intervention_vector","repo":"princeton-pli/DySCO","repo_kind":"found_in_text","path":"dysco/custom_mixin.py","file_url":"https://github.com/princeton-pli/DySCO/blob/HEAD/dysco/custom_mixin.py","link_basis":"first_harvest_node","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":"55123587eebf5f9d"}},{"code_sha256_prefix":"16ecf41fd208d319","entry":"_nucleus_mask","repo":"princeton-pli/DySCO","repo_kind":"found_in_text","path":"dysco/custom_mixin.py","file_url":"https://github.com/princeton-pli/DySCO/blob/HEAD/dysco/custom_mixin.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"16ecf41fd208d319"}},{"code_sha256_prefix":"383459431244f4c3","entry":"_select_important_tokens","repo":"princeton-pli/DySCO","repo_kind":"found_in_text","path":"dysco/custom_mixin.py","file_url":"https://github.com/princeton-pli/DySCO/blob/HEAD/dysco/custom_mixin.py","link_basis":"first_harvest_node","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":"383459431244f4c3"}},{"code_sha256_prefix":"8eae734f5f3c705c","entry":"obtain_template_sequence_mask","repo":"princeton-pli/DySCO","repo_kind":"found_in_text","path":"dysco/custom_mixin.py","file_url":"https://github.com/princeton-pli/DySCO/blob/HEAD/dysco/custom_mixin.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"8eae734f5f3c705c"}},{"code_sha256_prefix":"9f9368b64be89c74","entry":"CustomGenerationMixin","repo":"princeton-pli/DySCO","repo_kind":"found_in_text","path":"dysco/custom_mixin.py","file_url":"https://github.com/princeton-pli/DySCO/blob/HEAD/dysco/custom_mixin.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":"9f9368b64be89c74"}},{"code_sha256_prefix":"86e03cc35c95de82","entry":"_apply_importance_decay","repo":"princeton-pli/DySCO","repo_kind":"found_in_text","path":"dysco/custom_mixin.py","file_url":"https://github.com/princeton-pli/DySCO/blob/HEAD/dysco/custom_mixin.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":"86e03cc35c95de82"}}]},"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}