{"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/query-focused-retrieval-heads-improve-long","title":"Query-Focused Retrieval Heads Improve Long-Context Reasoning and Re-ranking","arxiv_id":"2506.09944","date":"2025-06-11","proceeding":null,"authors":["Wuwei Zhang","Fangcong Yin","Howard Yen","Danqi Chen","Xi Ye"],"abstract":"Recent work has identified retrieval heads (Wu et al., 2025b), a subset of attention heads responsible for retrieving salient information in long-context language models (LMs), as measured by their copy-paste behavior in Needle-in-a-Haystack tasks. In this paper, we introduce QRHEAD (Query-Focused Retrieval Head), an improved set of attention heads that enhance retrieval from long context. We identify QRHEAD by aggregating attention scores with respect to the input query, using a handful of examples from real-world tasks (e.g., long-context QA). We further introduce QR- RETRIEVER, an efficient and effective retriever that uses the accumulated attention mass of QRHEAD as retrieval scores. We use QR- RETRIEVER for long-context reasoning by selecting the most relevant parts with the highest retrieval scores. On multi-hop reasoning tasks LongMemEval and CLIPPER, this yields over 10% performance gains over full context and outperforms strong dense retrievers. We also evaluate QRRETRIEVER as a re-ranker on the BEIR benchmark and find that it achieves strong zero-shot performance, outperforming other LLM-based re-rankers such as RankGPT. Further analysis shows that both the querycontext attention scoring and task selection are crucial for identifying QRHEAD with strong downstream utility. Overall, our work contributes a general-purpose retriever and offers interpretability insights into the long-context capabilities of LMs.","url_abs":"https://arxiv.org/abs/2506.09944v1","url_pdf":"https://arxiv.org/pdf/2506.09944v1.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":"query-focused-retrieval-heads-improve-long","repo_url":"https://github.com/princeton-pli/qrhead","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"re-ranking","task_name":"Re-Ranking"},{"task_slug":"retrieval","task_name":"Retrieval"}],"methods":[{"method_slug":"attention","method_name":"Attention"},{"method_slug":"copy-paste","method_name":"Copy-Paste"},{"method_slug":"set","method_name":"SET"},{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2506.09944","atlas_url":"https://app.syntology.ai/?focus=2506.09944","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2506.09944"}},"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/princeton-pli/QRHead","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"ran_fixture":2,"ran_draft_wrong":3,"unverified":13},"by_repo_kind":{"official":{"samples":18,"ran":5,"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":"30d7eec482ebf6b1","entry":"repeat_kv","repo":"princeton-pli/QRHead","repo_kind":"official","path":"src/qrretriever/custom_modeling_llama.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/src/qrretriever/custom_modeling_llama.py","link_basis":"harvester_set","language":"python","status":"ran_fixture","verification_level":2,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"30d7eec482ebf6b1"}},{"code_sha256_prefix":"d61c483a3c2b3156","entry":"apply_rotary_pos_emb","repo":"princeton-pli/QRHead","repo_kind":"official","path":"src/qrretriever/custom_modeling_qwen2.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/src/qrretriever/custom_modeling_qwen2.py","link_basis":"harvester_set","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"d61c483a3c2b3156"}},{"code_sha256_prefix":"bac65c3dafaec040","entry":"apply_rotary_pos_emb","repo":"princeton-pli/QRHead","repo_kind":"official","path":"src/qrretriever/custom_modeling_llama.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/src/qrretriever/custom_modeling_llama.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"bac65c3dafaec040"}},{"code_sha256_prefix":"8e4f7c8ecedd17fa","entry":"get_anscheck_prompt","repo":"princeton-pli/QRHead","repo_kind":"official","path":"exp_scripts/generation/eval_generation_lme.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/exp_scripts/generation/eval_generation_lme.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"8e4f7c8ecedd17fa"}},{"code_sha256_prefix":"b99eea6376d1e212","entry":"rotate_half","repo":"princeton-pli/QRHead","repo_kind":"official","path":"src/qrretriever/custom_modeling_llama.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/src/qrretriever/custom_modeling_llama.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"b99eea6376d1e212"}},{"code_sha256_prefix":"6485b6194332b150","entry":"compute_recall","repo":"princeton-pli/QRHead","repo_kind":"official","path":"exp_scripts/retrieval/eval_retrieval.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/exp_scripts/retrieval/eval_retrieval.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":"6485b6194332b150"}},{"code_sha256_prefix":"63bfa7b825473d3d","entry":"extract_relevant_rounds","repo":"princeton-pli/QRHead","repo_kind":"official","path":"exp_scripts/generation/run_generation_lme.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/exp_scripts/generation/run_generation_lme.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":"63bfa7b825473d3d"}},{"code_sha256_prefix":"f8bf3d9d30b922e6","entry":"generate_prompt","repo":"princeton-pli/QRHead","repo_kind":"official","path":"exp_scripts/generation/run_generation_lme.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/exp_scripts/generation/run_generation_lme.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":"f8bf3d9d30b922e6"}},{"code_sha256_prefix":"d81d0dbbee1e8828","entry":"get_doc_scores_per_head","repo":"princeton-pli/QRHead","repo_kind":"official","path":"exp_scripts/detection/detect_qrhead_beir.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/exp_scripts/detection/detect_qrhead_beir.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":"d81d0dbbee1e8828"}},{"code_sha256_prefix":"009654fea103956a","entry":"get_doc_scores_per_head","repo":"princeton-pli/QRHead","repo_kind":"official","path":"exp_scripts/detection/detect_qrhead_lme.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/exp_scripts/detection/detect_qrhead_lme.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":"009654fea103956a"}},{"code_sha256_prefix":"30515a995de52d2d","entry":"get_top_k_docs","repo":"princeton-pli/QRHead","repo_kind":"official","path":"exp_scripts/retrieval/eval_retrieval.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/exp_scripts/retrieval/eval_retrieval.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":"30515a995de52d2d"}},{"code_sha256_prefix":"84ed72a7f256783d","entry":"get_topk_doc_idx","repo":"princeton-pli/QRHead","repo_kind":"official","path":"exp_scripts/generation/run_generation_clipper.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/exp_scripts/generation/run_generation_clipper.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":"84ed72a7f256783d"}},{"code_sha256_prefix":"46d639c95b19b2e5","entry":"lme_eval","repo":"princeton-pli/QRHead","repo_kind":"official","path":"exp_scripts/detection/detect_qrhead_lme.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/exp_scripts/detection/detect_qrhead_lme.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":"46d639c95b19b2e5"}},{"code_sha256_prefix":"229e3b5a73665fb2","entry":"load_config","repo":"princeton-pli/QRHead","repo_kind":"official","path":"src/qrretriever/config.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/src/qrretriever/config.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":"229e3b5a73665fb2"}},{"code_sha256_prefix":"1b2be7c22837418d","entry":"merge_configs","repo":"princeton-pli/QRHead","repo_kind":"official","path":"src/qrretriever/config.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/src/qrretriever/config.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":"1b2be7c22837418d"}},{"code_sha256_prefix":"1340e53f675329ea","entry":"parse_file","repo":"princeton-pli/QRHead","repo_kind":"official","path":"exp_scripts/generation/eval_generation_clipper.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/exp_scripts/generation/eval_generation_clipper.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":"1340e53f675329ea"}},{"code_sha256_prefix":"ff68a23061cdf19a","entry":"prepare_prompt_for_generation","repo":"princeton-pli/QRHead","repo_kind":"official","path":"exp_scripts/generation/run_generation_clipper.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/exp_scripts/generation/run_generation_clipper.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":"ff68a23061cdf19a"}},{"code_sha256_prefix":"b3336deddac204eb","entry":"score_heads","repo":"princeton-pli/QRHead","repo_kind":"official","path":"exp_scripts/detection/detect_qrhead_lme.py","file_url":"https://github.com/princeton-pli/QRHead/blob/HEAD/exp_scripts/detection/detect_qrhead_lme.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":"b3336deddac204eb"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}