{"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":"/code/load-qrels","entry":"load_qrels","source":"Syntology graph, per-sample; not an archive number","read_at":"2026-09-24T18:15:14+00:00","claim":"Names are grouped by exact entry-name string. Same-named routines are NOT asserted to be equivalent; 'ran' means executed on a synthesized fixture, not correctness. n_samples_ran = sum of by_status over every status except 'unverified' (ran_draft_wrong and ran_fixture are failures of Syntology's instrument, not of the code); n_papers_ran = papers with at least one such sample.","status_vocabulary":{"ran_honours":"ran, honoured the contract we drafted","ran_violates":"ran, violated the contract we drafted","ran_draft_wrong":"ran; our contract draft was wrong, not the code","ran_fixture":"ran; our fixture could not drive it","ran":"ran on a synthesized input","unverified":"unverified (harvested, no recorded run)"},"n_papers":12,"n_papers_ran":7,"units":"n_samples, n_samples_ran, n_samples_fingerprinted and by_status count distinct code bodies (code_sha256); n_places and n_places_pointer_only count places, one per (paper, code body) pair, which is also the unit of the samples list","n_samples":12,"n_samples_ran":7,"n_samples_fingerprinted":0,"n_places":12,"n_places_pointer_only":6,"by_status":{"ran_honours":0,"ran_violates":0,"ran_draft_wrong":4,"ran_fixture":0,"ran":3,"unverified":5},"syntology":{"atlas_url":null,"mcp":null,"mcp_per_sample":{"tool":"get_code","arguments_in":"samples[].mcp_get_code"},"developers":"https://syntology.ai/developers"},"samples":[{"arxiv_id":"2609.01556","paper":"/paper/arxiv-2609-01556","title":"Retrieved but not ranked: surface-form bias in structural retrieval, from mathematics to agent trajectories","date":null,"month_inferred_from_arxiv_id":"2026-09","title_source":"syntology","repo":"nabirarashid/structural-retrieval","path":"src/data.py","file_url":"https://github.com/nabirarashid/structural-retrieval/blob/HEAD/src/data.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"code_sha256_prefix":"79ffe3d7b0c300c3","mcp_get_code":{"code_sha256":"79ffe3d7b0c300c3"}},{"arxiv_id":"2608.21365","paper":"/paper/arxiv-2608-21365","title":"KSE-Web: An Analysis of Hybrid Retrieval and LLM-Assisted Query Expansion for Low-Resource Khmer Semantic Search","date":null,"month_inferred_from_arxiv_id":"2026-08","title_source":"syntology","repo":"back-kh/Khmer-Semantic-Search","path":"codes/evaluate_retrieval.py","file_url":"https://github.com/back-kh/Khmer-Semantic-Search/blob/HEAD/codes/evaluate_retrieval.py","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"code_sha256_prefix":"8772de2956a1a1a4","mcp_get_code":{"code_sha256":"8772de2956a1a1a4"}},{"arxiv_id":"2607.24165","paper":"/paper/arxiv-2607-24165","title":"Do Current Retrievers Cover All the Evidence? A Controlled Study of Conjunctive Cross-Page Retrieval","date":null,"month_inferred_from_arxiv_id":"2026-07","title_source":"syntology","repo":"sunggukcha/n_clue","path":"code/data_release/score_public_release.py","file_url":"https://github.com/sunggukcha/n_clue/blob/HEAD/code/data_release/score_public_release.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"code_sha256_prefix":"91dfe1de3ed50c38","mcp_get_code":{"code_sha256":"91dfe1de3ed50c38"}},{"arxiv_id":"2605.03824","paper":"/paper/arxiv-2605-03824","title":"Reproducing Complex Set-Compositional Information Retrieval","date":null,"month_inferred_from_arxiv_id":"2026-05","title_source":"syntology","repo":"informagi/Complex-Set-Compositional-IR","path":"code/evaluation/evaluate.py","file_url":"https://github.com/informagi/Complex-Set-Compositional-IR/blob/HEAD/code/evaluation/evaluate.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"3769c42b408e1d80","mcp_get_code":{"code_sha256":"3769c42b408e1d80"}},{"arxiv_id":"2602.02444","paper":"/paper/arxiv-2602-02444","title":"RANKVIDEO: Reasoning Reranking for Text-to-Video Retrieval","date":null,"month_inferred_from_arxiv_id":"2026-02","title_source":"syntology","repo":"tskow99/RANKVIDEO-Reasoning-Reranker","path":"rankvideo/evaluate.py","file_url":"https://github.com/tskow99/RANKVIDEO-Reasoning-Reranker/blob/HEAD/rankvideo/evaluate.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"0ba56ee29da26ed0","mcp_get_code":{"code_sha256":"0ba56ee29da26ed0"}},{"arxiv_id":"2411.06877","paper":"/paper/llm-assisted-relevance-assessments-when","title":"LLM-Assisted Relevance Assessments: When Should We Ask LLMs for Help?","date":"2024-11-11","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"RikiyaT/LARA","path":"src/example-covid-itachi.py","file_url":"https://github.com/RikiyaT/LARA/blob/HEAD/src/example-covid-itachi.py","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"2764249ced7bbd4f","mcp_get_code":{"code_sha256":"2764249ced7bbd4f"}},{"arxiv_id":"2312.11361","paper":"/paper/nomiracl-knowing-when-you-don-t-know-for","title":"\"Knowing When You Don't Know\": A Multilingual Relevance Assessment Dataset for Robust Retrieval-Augmented Generation","date":"2023-12-18","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"project-miracl/nomiracl","path":"nomiracl/dataset/util.py","file_url":"https://github.com/project-miracl/nomiracl/blob/HEAD/nomiracl/dataset/util.py","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"code_sha256_prefix":"d5328a85425a2266","mcp_get_code":{"code_sha256":"d5328a85425a2266"}},{"arxiv_id":"2205.12487","paper":"/paper/end-to-end-multimodal-fact-checking-and","title":"End-to-End Multimodal Fact-Checking and Explanation Generation: A Challenging Dataset and Models","date":"2022-05-25","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"VT-NLP/Mocheg","path":"retrieval/eval/eval_msmarco_mocheg.py","file_url":"https://github.com/VT-NLP/Mocheg/blob/HEAD/retrieval/eval/eval_msmarco_mocheg.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"code_sha256_prefix":"188ada4a117ab111","mcp_get_code":{"code_sha256":"188ada4a117ab111"}},{"arxiv_id":"2204.10362","paper":"/paper/human-preferences-as-dueling-bandits","title":"Human Preferences as Dueling Bandits","date":"2022-04-21","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"claclark/preferences","path":"prefj.py","file_url":"https://github.com/claclark/preferences/blob/HEAD/prefj.py","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"code_sha256_prefix":"46d4b9ba90167c86","mcp_get_code":{"code_sha256":"46d4b9ba90167c86"}},{"arxiv_id":"1904.08375","paper":"/paper/document-expansion-by-query-prediction","title":"Document Expansion by Query Prediction","date":"2019-04-17","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"nyu-dl/dl4ir-doc2query","path":"convert_msmarco_to_opennmt.py","file_url":"https://github.com/nyu-dl/dl4ir-doc2query/blob/HEAD/convert_msmarco_to_opennmt.py","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"BSD-3-Clause","inline_ok":true,"code_sha256_prefix":"43fa12d3c9abf09c","mcp_get_code":{"code_sha256":"43fa12d3c9abf09c"}},{"arxiv_id":"1901.04085","paper":"/paper/passage-re-ranking-with-bert","title":"Passage Re-ranking with BERT","date":"2019-01-13","month_inferred_from_arxiv_id":null,"title_source":"archive","repo":"nyu-dl/dl4marco-bert","path":"convert_treccar_to_tfrecord.py","file_url":"https://github.com/nyu-dl/dl4marco-bert/blob/HEAD/convert_treccar_to_tfrecord.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"BSD-3-Clause","inline_ok":true,"code_sha256_prefix":"9eb250ad7f25c79e","mcp_get_code":{"code_sha256":"9eb250ad7f25c79e"}},{"arxiv_id":"2025.naacl-long.14","paper":null,"title":"arXiv:2025.naacl-long.14","date":null,"month_inferred_from_arxiv_id":null,"title_source":null,"repo":"vectara/mirage-bench","path":"mirage_bench/util.py","file_url":"https://github.com/vectara/mirage-bench/blob/HEAD/mirage_bench/util.py","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"code_sha256_prefix":"5f9cdb305ab94fdb","mcp_get_code":{"code_sha256":"5f9cdb305ab94fdb"}}]}