{
  "bm25_reference": {
    "ndcg@10": 0.226,
    "recall@100": 0.572,
    "source": "MIRACL paper (TACL 2023), Table 5, full corpus"
  },
  "checks": {
    "determinism": {
      "bge-v2-m3": {
        "batch_size": 32,
        "depth": 100,
        "difference_points": 0.0,
        "identical_rankings": 305,
        "n_queries": 305,
        "ndcg@10_first": 0.36008335712152306,
        "ndcg@10_repeat": 0.36008335712152306,
        "passed": true,
        "run_id": "quality-determinism-bge-v2-m3-20261010T140311Z-331ebf",
        "stage": "hybrid-71k",
        "tolerance_points": 0.1
      },
      "jina-v3.5": {
        "batch_size": null,
        "depth": 100,
        "difference_points": 0.0,
        "identical_rankings": 305,
        "n_queries": 305,
        "ndcg@10_first": 0.37660312839974847,
        "ndcg@10_repeat": 0.37660312839974847,
        "passed": true,
        "run_id": "quality-determinism-jina-v3.5-20261010T142422Z-07a25e",
        "stage": "hybrid-71k",
        "tolerance_points": 0.1
      },
      "qwen3-0.6b": {
        "batch_size": 8,
        "depth": 100,
        "difference_points": 0.0,
        "identical_rankings": 305,
        "n_queries": 305,
        "ndcg@10_first": 0.3646704503914691,
        "ndcg@10_repeat": 0.3646704503914691,
        "passed": true,
        "run_id": "quality-determinism-qwen3-0.6b-20261010T140745Z-5a7df0",
        "stage": "hybrid-71k",
        "tolerance_points": 0.1
      }
    },
    "dtype": {
      "bge-v2-m3": {
        "batch_size_gpu": 32,
        "difference_points": 0.0,
        "gpu_value_used_for_cpu": true,
        "identical_rankings": 1,
        "n_queries": 20,
        "ndcg@10_cpu_fp32": 0.7610539990914337,
        "ndcg@10_gpu": 0.7610539990914337,
        "run_id": "quality-dtype-bge-v2-m3-20261011T065250Z-875f4f",
        "stage": "hybrid-71k",
        "tolerance_points": 0.5
      },
      "gte-base": {
        "batch_size_gpu": 32,
        "difference_points": 0.0,
        "gpu_value_used_for_cpu": true,
        "identical_rankings": 2,
        "n_queries": 20,
        "ndcg@10_cpu_fp32": 0.6865317842440822,
        "ndcg@10_gpu": 0.6865317842440822,
        "run_id": "quality-dtype-gte-base-20261011T064302Z-16d7c9",
        "stage": "hybrid-71k",
        "tolerance_points": 0.5
      },
      "jina-v3.5": {
        "batch_size_gpu": null,
        "difference_points": 0.12749845543877925,
        "gpu_value_used_for_cpu": true,
        "identical_rankings": 0,
        "n_queries": 20,
        "ndcg@10_cpu_fp32": 0.7711561073178305,
        "ndcg@10_gpu": 0.7724310918722183,
        "run_id": "quality-dtype-jina-v3.5-20261011T073653Z-f68b38",
        "stage": "hybrid-71k",
        "tolerance_points": 0.5
      },
      "qwen3-0.6b": {
        "batch_size_gpu": 8,
        "difference_points": 0.0,
        "gpu_value_used_for_cpu": true,
        "identical_rankings": 0,
        "n_queries": 20,
        "ndcg@10_cpu_fp32": 0.7305128439846746,
        "ndcg@10_gpu": 0.7305128439846746,
        "run_id": "quality-dtype-qwen3-0.6b-20261011T081457Z-09da93",
        "stage": "hybrid-71k",
        "tolerance_points": 0.5
      }
    },
    "evaluation": {
      "difference": 0.0,
      "n_queries": 305,
      "ours_ndcg@10": 0.17498289422243518,
      "passed": true,
      "pytrec_eval_ndcg@10": 0.17498289422243518,
      "run_id": "validation-evaluation-hybrid-71k-20261010T035449Z-c16a8b",
      "run_sha256": "3032e2cc871b0b82c144f6e5358876c3647988d331d47f910ab9e1fa982b1b8a",
      "stage": "hybrid-71k",
      "tolerance": 0.0001
    },
    "pointwise": {
      "voyage-rerank-3": {
        "depths": [
          20,
          100
        ],
        "max_abs_difference": 0.0,
        "n_queries": 5,
        "pointwise": true,
        "run_id": "validation-voyage-rerank-3-pointwise-20261010T035432Z-e3c788",
        "tolerance": 0.001
      },
      "voyage-rerank-3-lite": {
        "depths": [
          20,
          100
        ],
        "max_abs_difference": 0.0,
        "n_queries": 5,
        "pointwise": true,
        "run_id": "validation-voyage-rerank-3-lite-pointwise-20261010T035437Z-91c411",
        "tolerance": 0.001
      }
    }
  },
  "costs": {
    "checks": {
      "inputs": {
        "api_run_info": {
          "voyage-rerank-3": {
            "keyword-71k": "078d3ba24a1ae0b1936c21bc1807f83de37dc1ddc284874dc0281452f91ed452",
            "keyword-full": "8a26b977508cb4368ea5a1ca640ba6fbb0eae3f6f382ec9853d33e2df0257b12",
            "vector-71k": "5cba92a73de0d83cb4d51a811210416a890044b013b17b14fd2ab9834d9f9923"
          },
          "voyage-rerank-3-lite": {
            "keyword-71k": "16c2a8c27d467339786a93216f4bab1d835398ba8f9d6d38e7899ffc1038ccf4",
            "keyword-full": "76439f492fb9ad212df7a6ee280df00d946b44095384c3c320ead3af354d5a5c",
            "vector-71k": "e96f67e9d05f78af30263899deaba5afa847df395d2ee4b33c0c846285d3ddae"
          }
        },
        "first_stage_runs": {
          "hybrid-71k": "3032e2cc871b0b82c144f6e5358876c3647988d331d47f910ab9e1fa982b1b8a",
          "keyword-71k": "a03aeed0d2bc63e1c79953090984accf978ae6b9c974d02fd8431c5d72e2642a",
          "keyword-full": "69458380199a02c657b5b5231188fb4ed3a6e5792340ba332213ff04db6cbaf8",
          "vector-71k": "ef12845ce3afbdc8d17d906e6859eba9e234a7355f41899830d55a6d472a850c"
        },
        "latency": {
          "bge-v2-m3-cuda-desktop-se": "latency-bge-v2-m3-cuda-desktop-se-20261011T043232Z-d00b64",
          "first-stage-hybrid-71k-desktop-se": "latency-first-stage-hybrid-71k-desktop-se-20261010T152234Z-80ffe8",
          "gte-base-cuda-desktop-se": "latency-gte-base-cuda-desktop-se-20261011T043900Z-c4aece",
          "jina-v3.5-cuda-desktop-se": "latency-jina-v3.5-cuda-desktop-se-20261011T041221Z-de1b7f",
          "mxbai-large-v2-cuda-desktop-se": "latency-mxbai-large-v2-cuda-desktop-se-20261011T030622Z-f837da",
          "querit-4b-cuda-desktop-se": "latency-querit-4b-cuda-desktop-se-20261011T001254Z-186e05",
          "qwen3-0.6b-cuda-desktop-se": "latency-qwen3-0.6b-cuda-desktop-se-20261011T085531Z-bfd2bd",
          "qwen3-4b-cuda-desktop-se": "latency-qwen3-4b-cuda-desktop-se-20261011T020021Z-aa5503",
          "voyage-rerank-3-api-desktop-se-afternoon": "latency-voyage-rerank-3-api-desktop-se-afternoon-20261011T100517Z-9099a1",
          "voyage-rerank-3-api-desktop-se-evening": "latency-voyage-rerank-3-api-desktop-se-evening-20261010T162902Z-f848c5",
          "voyage-rerank-3-api-desktop-se-morning": "latency-voyage-rerank-3-api-desktop-se-morning-20261011T063414Z-47493a",
          "voyage-rerank-3-lite-api-desktop-se-afternoon": "latency-voyage-rerank-3-lite-api-desktop-se-afternoon-20261011T100852Z-9066d4",
          "voyage-rerank-3-lite-api-desktop-se-evening": "latency-voyage-rerank-3-lite-api-desktop-se-evening-20261010T163244Z-facd01",
          "voyage-rerank-3-lite-api-desktop-se-morning": "latency-voyage-rerank-3-lite-api-desktop-se-morning-20261011T063750Z-d5f19f"
        },
        "reranker_runs": {
          "hybrid-71k--bge-v2-m3": "bc0959664514e50a32696682f63dd437e2b7a1e59cc43b3a1c55c67d25ce9b31",
          "hybrid-71k--gte-base": "1cf6cad186e086b7e7d00249e884845f562febf23e345adf63401c41b3776320",
          "hybrid-71k--jina-v3.5": "c7910a55cac133dec5345383d1a96760db613e6986b74fd17df77eef78d77344",
          "hybrid-71k--mxbai-large-v2": "3fbe579073f2568b3dfa0cb891d85ff05374476606ffda00ac9510479b8dcc31",
          "hybrid-71k--querit-4b": "1d95196e560d0f067af56dde53498fe1cbb9c00dd154837fa7d607e4654e359c",
          "hybrid-71k--qwen3-0.6b": "a8d8786a0e034ae5a164afc4e22a15c6fc488a6d436286b67fa8c6325d2bb5bf",
          "hybrid-71k--qwen3-4b": "4a9458d131e44d4eb0b5f523f19e0dda48dcc55970b40e15641deee340208bc7",
          "hybrid-71k--voyage-rerank-3": "3e949aeb397e8a053376542666e591d5993eea5b5e8847b14e62b2a01fb64cc2",
          "hybrid-71k--voyage-rerank-3-lite": "03fe1dc6f8ca69149121c99041f3755e0ec59ae75378c8579916e705dfb0d809",
          "keyword-71k--bge-v2-m3": "58946ee5d98c7dbec37b8c83f5fb9da6b777799bac053bc398a7abdd65ffcf0f",
          "keyword-71k--gte-base": "d5bcb9f174ffc10c382a8004dced7a2fd9d2d2418285d5ebf757f18b8e99017c",
          "keyword-71k--jina-v3.5": "4fb39f2c415a380e314f33da59d96e5049a041eca1289dbfee7d99c59f6e36bc",
          "keyword-71k--mxbai-large-v2": "022572dd1a703819362027e28224a2474cd0e68b19e3ff8862b3b082ff8dc301",
          "keyword-71k--querit-4b": "875c9efc5d166172e0af96213a0c5b85e0f58e595aec76c147a2c0f52098efef",
          "keyword-71k--qwen3-0.6b": "45d0a510488a6676fd880278eca4751ceb9e38c1d00518a916ca1d0c90e9d25e",
          "keyword-71k--qwen3-4b": "6b77b343195d77f0b483234f07b62ad6e4d345a64f491df1718314d7ffd6bb96",
          "keyword-71k--voyage-rerank-3": "0d152a740dc8496a3c74775169ee6dc9fb035486ac5998398b8ce112a54fdb88",
          "keyword-71k--voyage-rerank-3-lite": "21f25d03dfb31ff7cee28d8f1f750aaaf5449829918e3370e5498ffced329ea4",
          "keyword-full--bge-v2-m3": "a4de05a7d199359656a92686ccd866c32c0b6e176cb935b4b02064c2a502e329",
          "keyword-full--gte-base": "1e19e8f947bf47bb8295595d59a7fc2d63a65944fa17aa9dfcdccc278fbe3714",
          "keyword-full--mxbai-large-v2": "eb5656eac01ce129aa660125a560e589390eb723a08a2489c2d50f0a98ff886c",
          "keyword-full--querit-4b": "b00e2def7a609bd8de313be8c154cf15f6765c1dd0ed3b8a0588547788bd5c14",
          "keyword-full--qwen3-0.6b": "f531df892644d9cbf4040b0298082d5269a6216aa477c20ed85f31a71937ce8d",
          "keyword-full--qwen3-4b": "ca03597659b4ce2c9399e8148c982b25230617ec3706815baad8148756c9d46d",
          "keyword-full--voyage-rerank-3": "79dce8b26fbe40258bbf1456d3e22290bc58a118acb3398ebb53809eb7fcf30d",
          "keyword-full--voyage-rerank-3-lite": "fa9a109cf5f0819671a3c1d23f3ad6a561ef9f93a034e54c0c74832188521d3c",
          "vector-71k--bge-v2-m3": "9d1b0a2cf8aed81e2d64ebb73110d29313a52a3a80703e0f12e1cb4889558edf",
          "vector-71k--gte-base": "1485207854146931ecf4de29077da81364499762eeb9b60328c5a8d5c163a1d4",
          "vector-71k--jina-v3.5": "4fc5abd57717d7af953bfb166e0aa05148384dbd721708f5f32687b3c51479f9",
          "vector-71k--mxbai-large-v2": "a45f2ffef2584bbdc3e46a2c777dce55df8eecd2cfbfaeee63a833915e4b560b",
          "vector-71k--querit-4b": "fd8c37435fda57a221b18217bc872a051e5a01cd41e9e41a7b93a22d8068b44e",
          "vector-71k--qwen3-0.6b": "2b0ddf7465bd930a575774428dc5b3e7d5961248d2aa9a5a8a5ba629ff9ba467",
          "vector-71k--qwen3-4b": "9c1e387eadb56ae1ce76b6ba5891cacba211514e2e7195967665aefd10f695f2",
          "vector-71k--voyage-rerank-3": "8ded261a994412dbecaca17d94e5a49cceae4a9de6ed2deebd74bc21c7d6b87e",
          "vector-71k--voyage-rerank-3-lite": "60493d5095ad819f685557a3405237e188e10e58bd07708000d3cda080411c93"
        }
      }
    },
    "context": {
      "hybrid-71k--bge-v2-m3": {
        "hit@5_reranked": 0.5114754098360655,
        "k_needed_hit": 26,
        "k_needed_recall": 32,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 7.694756907160391,
          "deepseek/deepseek-flash": 0.5771067680370293
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 1.1542135360740586
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 9.89325888063479,
          "deepseek/deepseek-flash": 0.7419944160476091
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 1.4839888320952181
        },
        "recall@5_reranked": 0.36545667447306796,
        "saved_passages_hit": 21,
        "saved_passages_recall": 27
      },
      "hybrid-71k--gte-base": {
        "hit@5_reranked": 0.4852459016393443,
        "k_needed_hit": 20,
        "k_needed_recall": 27,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 5.496254933685994,
          "deepseek/deepseek-flash": 0.41221912002644945
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 0.8244382400528989
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 8.061173902739457,
          "deepseek/deepseek-flash": 0.6045880427054593
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 1.2091760854109186
        },
        "recall@5_reranked": 0.35117096018735366,
        "saved_passages_hit": 15,
        "saved_passages_recall": 22
      },
      "hybrid-71k--jina-v3.5": {
        "hit@5_reranked": 0.5180327868852459,
        "k_needed_hit": 30,
        "k_needed_recall": 38,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 9.160424889476658,
          "deepseek/deepseek-flash": 0.6870318667107492
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 1.3740637334214985
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 12.091760854109186,
          "deepseek/deepseek-flash": 0.9068820640581889
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 1.8137641281163779
        },
        "recall@5_reranked": 0.3731693989071038,
        "saved_passages_hit": 25,
        "saved_passages_recall": 33
      },
      "hybrid-71k--mxbai-large-v2": {
        "hit@5_reranked": 0.4819672131147541,
        "k_needed_hit": 20,
        "k_needed_recall": 23,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 5.496254933685994,
          "deepseek/deepseek-flash": 0.41221912002644945
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 0.8244382400528989
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 6.595505920423192,
          "deepseek/deepseek-flash": 0.4946629440317394
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 0.9893258880634788
        },
        "recall@5_reranked": 0.3360421545667447,
        "saved_passages_hit": 15,
        "saved_passages_recall": 18
      },
      "hybrid-71k--querit-4b": {
        "hit@5_reranked": 0.49836065573770494,
        "k_needed_hit": 24,
        "k_needed_recall": 29,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 6.961922916002259,
          "deepseek/deepseek-flash": 0.5221442187001694
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 1.0442884374003387
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 8.794007893897591,
          "deepseek/deepseek-flash": 0.6595505920423194
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 1.3191011840846387
        },
        "recall@5_reranked": 0.35725995316159254,
        "saved_passages_hit": 19,
        "saved_passages_recall": 24
      },
      "hybrid-71k--qwen3-0.6b": {
        "hit@5_reranked": 0.5081967213114754,
        "k_needed_hit": 24,
        "k_needed_recall": 33,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 6.961922916002259,
          "deepseek/deepseek-flash": 0.5221442187001694
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 1.0442884374003387
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 10.259675876213857,
          "deepseek/deepseek-flash": 0.769475690716039
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 1.538951381432078
        },
        "recall@5_reranked": 0.36669008587041374,
        "saved_passages_hit": 19,
        "saved_passages_recall": 28
      },
      "hybrid-71k--qwen3-4b": {
        "hit@5_reranked": 0.5180327868852459,
        "k_needed_hit": 30,
        "k_needed_recall": 38,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 9.160424889476658,
          "deepseek/deepseek-flash": 0.6870318667107492
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 1.3740637334214985
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 12.091760854109186,
          "deepseek/deepseek-flash": 0.9068820640581889
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 1.8137641281163779
        },
        "recall@5_reranked": 0.3769086651053864,
        "saved_passages_hit": 25,
        "saved_passages_recall": 33
      },
      "hybrid-71k--voyage-rerank-3": {
        "hit@5_reranked": 0.5081967213114754,
        "k_needed_hit": 24,
        "k_needed_recall": 37,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 6.961922916002259,
          "deepseek/deepseek-flash": 0.5221442187001694
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 1.0442884374003387
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 11.725343858530119,
          "deepseek/deepseek-flash": 0.8794007893897589
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 1.7588015787795177
        },
        "recall@5_reranked": 0.371928181108509,
        "saved_passages_hit": 19,
        "saved_passages_recall": 32
      },
      "hybrid-71k--voyage-rerank-3-lite": {
        "hit@5_reranked": 0.5016393442622951,
        "k_needed_hit": 24,
        "k_needed_recall": 32,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 6.961922916002259,
          "deepseek/deepseek-flash": 0.5221442187001694
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 1.0442884374003387
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 9.89325888063479,
          "deepseek/deepseek-flash": 0.7419944160476091
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 1.4839888320952181
        },
        "recall@5_reranked": 0.3614988290398127,
        "saved_passages_hit": 19,
        "saved_passages_recall": 27
      },
      "keyword-71k--bge-v2-m3": {
        "hit@5_reranked": 0.33114754098360655,
        "k_needed_hit": 60,
        "k_needed_recall": 57,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 20.534752280415777,
          "deepseek/deepseek-flash": 1.5401064210311834
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 3.080212842062367
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 19.41467488330219,
          "deepseek/deepseek-flash": 1.456100616247664
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 2.912201232495328
        },
        "recall@5_reranked": 0.20192818110850896,
        "saved_passages_hit": 55,
        "saved_passages_recall": 52
      },
      "keyword-71k--gte-base": {
        "hit@5_reranked": 0.31475409836065577,
        "k_needed_hit": 57,
        "k_needed_recall": 48,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 19.41467488330219,
          "deepseek/deepseek-flash": 1.456100616247664
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 2.912201232495328
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 16.054442691961427,
          "deepseek/deepseek-flash": 1.204083201897107
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 2.408166403794214
        },
        "recall@5_reranked": 0.18635441061670568,
        "saved_passages_hit": 52,
        "saved_passages_recall": 43
      },
      "keyword-71k--jina-v3.5": {
        "hit@5_reranked": 0.3442622950819672,
        "k_needed_hit": 75,
        "k_needed_recall": 62,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 26.13513926598372,
          "deepseek/deepseek-flash": 1.9601354449487787
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 3.9202708898975573
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 21.28147054515817,
          "deepseek/deepseek-flash": 1.5961102908868627
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 3.1922205817737255
        },
        "recall@5_reranked": 0.21266978922716628,
        "saved_passages_hit": 70,
        "saved_passages_recall": 57
      },
      "keyword-71k--mxbai-large-v2": {
        "hit@5_reranked": 0.30491803278688523,
        "k_needed_hit": 53,
        "k_needed_recall": 47,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 17.921238353817408,
          "deepseek/deepseek-flash": 1.3440928765363056
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 2.688185753072611
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 15.68108355959023,
          "deepseek/deepseek-flash": 1.176081266969267
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 2.352162533938534
        },
        "recall@5_reranked": 0.18309133489461357,
        "saved_passages_hit": 48,
        "saved_passages_recall": 42
      },
      "keyword-71k--querit-4b": {
        "hit@5_reranked": 0.3180327868852459,
        "k_needed_hit": 58,
        "k_needed_recall": 56,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 19.78803401567339,
          "deepseek/deepseek-flash": 1.484102551175504
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 2.968205102351008
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 19.041315750931,
          "deepseek/deepseek-flash": 1.4280986813198246
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 2.8561973626396493
        },
        "recall@5_reranked": 0.19563622170179545,
        "saved_passages_hit": 53,
        "saved_passages_recall": 51
      },
      "keyword-71k--qwen3-0.6b": {
        "hit@5_reranked": 0.33114754098360655,
        "k_needed_hit": 60,
        "k_needed_recall": 57,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 20.534752280415777,
          "deepseek/deepseek-flash": 1.5401064210311834
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 3.080212842062367
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 19.41467488330219,
          "deepseek/deepseek-flash": 1.456100616247664
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 2.912201232495328
        },
        "recall@5_reranked": 0.199032006245121,
        "saved_passages_hit": 55,
        "saved_passages_recall": 52
      },
      "keyword-71k--qwen3-4b": {
        "hit@5_reranked": 0.3344262295081967,
        "k_needed_hit": 62,
        "k_needed_recall": 60,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 21.28147054515817,
          "deepseek/deepseek-flash": 1.5961102908868627
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 3.1922205817737255
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 20.534752280415777,
          "deepseek/deepseek-flash": 1.5401064210311834
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 3.080212842062367
        },
        "recall@5_reranked": 0.21034348165495703,
        "saved_passages_hit": 57,
        "saved_passages_recall": 55
      },
      "keyword-71k--voyage-rerank-3": {
        "hit@5_reranked": 0.32786885245901637,
        "k_needed_hit": 60,
        "k_needed_recall": 57,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 20.534752280415777,
          "deepseek/deepseek-flash": 1.5401064210311834
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 3.080212842062367
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 19.41467488330219,
          "deepseek/deepseek-flash": 1.456100616247664
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 2.912201232495328
        },
        "recall@5_reranked": 0.20348165495706477,
        "saved_passages_hit": 55,
        "saved_passages_recall": 52
      },
      "keyword-71k--voyage-rerank-3-lite": {
        "hit@5_reranked": 0.3114754098360656,
        "k_needed_hit": 57,
        "k_needed_recall": 55,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 19.41467488330219,
          "deepseek/deepseek-flash": 1.456100616247664
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 2.912201232495328
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 18.6679566185598,
          "deepseek/deepseek-flash": 1.400096746391985
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 2.80019349278397
        },
        "recall@5_reranked": 0.1942232630757221,
        "saved_passages_hit": 52,
        "saved_passages_recall": 50
      },
      "keyword-full--bge-v2-m3": {
        "hit@5_reranked": 0.2360655737704918,
        "k_needed_hit": 72,
        "k_needed_recall": 71,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 31.868469252843873,
          "deepseek/deepseek-flash": 2.3901351939632907
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 4.780270387926581
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 31.39282045802531,
          "deepseek/deepseek-flash": 2.354461534351898
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 4.708923068703796
        },
        "recall@5_reranked": 0.14453551912568305,
        "saved_passages_hit": 67,
        "saved_passages_recall": 66
      },
      "keyword-full--gte-base": {
        "hit@5_reranked": 0.22295081967213115,
        "k_needed_hit": 61,
        "k_needed_recall": 54,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 26.636332509839654,
          "deepseek/deepseek-flash": 1.9977249382379736
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 3.995449876475947
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 23.306790946109697,
          "deepseek/deepseek-flash": 1.748009320958227
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 3.496018641916454
        },
        "recall@5_reranked": 0.13473067915690867,
        "saved_passages_hit": 56,
        "saved_passages_recall": 49
      },
      "keyword-full--jina-v3.5": {
        "status": "not measurable at depth 100"
      },
      "keyword-full--mxbai-large-v2": {
        "hit@5_reranked": 0.23278688524590163,
        "k_needed_hit": 71,
        "k_needed_recall": 65,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 31.39282045802531,
          "deepseek/deepseek-flash": 2.354461534351898
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 4.708923068703796
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 28.538927689113915,
          "deepseek/deepseek-flash": 2.1404195766835437
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 4.280839153367087
        },
        "recall@5_reranked": 0.14073380171740826,
        "saved_passages_hit": 66,
        "saved_passages_recall": 60
      },
      "keyword-full--querit-4b": {
        "hit@5_reranked": 0.23934426229508196,
        "k_needed_hit": 76,
        "k_needed_recall": 71,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 33.77106443211814,
          "deepseek/deepseek-flash": 2.53282983240886
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 5.06565966481772
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 31.39282045802531,
          "deepseek/deepseek-flash": 2.354461534351898
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 4.708923068703796
        },
        "recall@5_reranked": 0.14455893832943012,
        "saved_passages_hit": 71,
        "saved_passages_recall": 66
      },
      "keyword-full--qwen3-0.6b": {
        "hit@5_reranked": 0.23934426229508196,
        "k_needed_hit": 76,
        "k_needed_recall": 71,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 33.77106443211814,
          "deepseek/deepseek-flash": 2.53282983240886
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 5.06565966481772
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 31.39282045802531,
          "deepseek/deepseek-flash": 2.354461534351898
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 4.708923068703796
        },
        "recall@5_reranked": 0.14483996877439498,
        "saved_passages_hit": 71,
        "saved_passages_recall": 66
      },
      "keyword-full--qwen3-4b": {
        "hit@5_reranked": 0.2459016393442623,
        "k_needed_hit": 78,
        "k_needed_recall": 76,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 34.722362021755266,
          "deepseek/deepseek-flash": 2.604177151631645
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 5.20835430326329
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 33.77106443211814,
          "deepseek/deepseek-flash": 2.53282983240886
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 5.06565966481772
        },
        "recall@5_reranked": 0.15325526932084307,
        "saved_passages_hit": 73,
        "saved_passages_recall": 71
      },
      "keyword-full--voyage-rerank-3": {
        "hit@5_reranked": 0.2360655737704918,
        "k_needed_hit": 72,
        "k_needed_recall": 71,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 31.868469252843873,
          "deepseek/deepseek-flash": 2.3901351939632907
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 4.780270387926581
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 31.39282045802531,
          "deepseek/deepseek-flash": 2.354461534351898
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 4.708923068703796
        },
        "recall@5_reranked": 0.1456830601092896,
        "saved_passages_hit": 67,
        "saved_passages_recall": 66
      },
      "keyword-full--voyage-rerank-3-lite": {
        "hit@5_reranked": 0.22950819672131148,
        "k_needed_hit": 71,
        "k_needed_recall": 64,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 31.39282045802531,
          "deepseek/deepseek-flash": 2.354461534351898
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 4.708923068703796
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 28.063278894295355,
          "deepseek/deepseek-flash": 2.104745917072151
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 4.209491834144302
        },
        "recall@5_reranked": 0.13882903981264635,
        "saved_passages_hit": 66,
        "saved_passages_recall": 59
      },
      "vector-71k--bge-v2-m3": {
        "hit@5_reranked": 0.8524590163934426,
        "k_needed_hit": 8,
        "k_needed_recall": 7,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 0.8922471860800565,
          "deepseek/deepseek-flash": 0.06691853895600426
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 0.13383707791200852
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 0.5948314573867043,
          "deepseek/deepseek-flash": 0.04461235930400283
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 0.08922471860800565
        },
        "recall@5_reranked": 0.632305490502212,
        "saved_passages_hit": 3,
        "saved_passages_recall": 2
      },
      "vector-71k--gte-base": {
        "hit@5_reranked": 0.7836065573770492,
        "k_needed_hit": 4,
        "k_needed_recall": 5,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 0.0,
          "deepseek/deepseek-flash": 0.0
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 0.0
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 0.0,
          "deepseek/deepseek-flash": 0.0
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 0.0
        },
        "recall@5_reranked": 0.552311995836586,
        "saved_passages_hit": 0,
        "saved_passages_recall": 0
      },
      "vector-71k--jina-v3.5": {
        "hit@5_reranked": 0.8524590163934426,
        "k_needed_hit": 8,
        "k_needed_recall": 8,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 0.8922471860800565,
          "deepseek/deepseek-flash": 0.06691853895600426
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 0.13383707791200852
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 0.8922471860800565,
          "deepseek/deepseek-flash": 0.06691853895600426
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 0.13383707791200852
        },
        "recall@5_reranked": 0.6440658339838669,
        "saved_passages_hit": 3,
        "saved_passages_recall": 3
      },
      "vector-71k--mxbai-large-v2": {
        "hit@5_reranked": 0.7639344262295082,
        "k_needed_hit": 4,
        "k_needed_recall": 5,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 0.0,
          "deepseek/deepseek-flash": 0.0
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 0.0
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 0.0,
          "deepseek/deepseek-flash": 0.0
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 0.0
        },
        "recall@5_reranked": 0.532352328909706,
        "saved_passages_hit": 0,
        "saved_passages_recall": 0
      },
      "vector-71k--querit-4b": {
        "hit@5_reranked": 0.8295081967213115,
        "k_needed_hit": 6,
        "k_needed_recall": 7,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 0.29741572869335214,
          "deepseek/deepseek-flash": 0.022306179652001414
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 0.04461235930400283
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 0.5948314573867043,
          "deepseek/deepseek-flash": 0.04461235930400283
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 0.08922471860800565
        },
        "recall@5_reranked": 0.6261930783242259,
        "saved_passages_hit": 1,
        "saved_passages_recall": 2
      },
      "vector-71k--qwen3-0.6b": {
        "hit@5_reranked": 0.8721311475409836,
        "k_needed_hit": 9,
        "k_needed_recall": 8,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 1.1896629147734086,
          "deepseek/deepseek-flash": 0.08922471860800565
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 0.1784494372160113
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 0.8922471860800565,
          "deepseek/deepseek-flash": 0.06691853895600426
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 0.13383707791200852
        },
        "recall@5_reranked": 0.6599557637262555,
        "saved_passages_hit": 4,
        "saved_passages_recall": 3
      },
      "vector-71k--qwen3-4b": {
        "hit@5_reranked": 0.8819672131147541,
        "k_needed_hit": 10,
        "k_needed_recall": 9,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 1.4870786434667609,
          "deepseek/deepseek-flash": 0.11153089826000706
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 0.22306179652001412
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 1.1896629147734086,
          "deepseek/deepseek-flash": 0.08922471860800565
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 0.1784494372160113
        },
        "recall@5_reranked": 0.6840202966432474,
        "saved_passages_hit": 5,
        "saved_passages_recall": 4
      },
      "vector-71k--voyage-rerank-3": {
        "hit@5_reranked": 0.8524590163934426,
        "k_needed_hit": 8,
        "k_needed_recall": 8,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 0.8922471860800565,
          "deepseek/deepseek-flash": 0.06691853895600426
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 0.13383707791200852
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 0.8922471860800565,
          "deepseek/deepseek-flash": 0.06691853895600426
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 0.13383707791200852
        },
        "recall@5_reranked": 0.6480783242258652,
        "saved_passages_hit": 3,
        "saved_passages_recall": 3
      },
      "vector-71k--voyage-rerank-3-lite": {
        "hit@5_reranked": 0.839344262295082,
        "k_needed_hit": 7,
        "k_needed_recall": 7,
        "llm_eur_saved_per_1000": {
          "anthropic/claude-sonnet-5.5": 0.5948314573867043,
          "deepseek/deepseek-flash": 0.04461235930400283
        },
        "llm_eur_saved_per_1000_peak": {
          "deepseek/deepseek-flash": 0.08922471860800565
        },
        "llm_eur_saved_per_1000_recall": {
          "anthropic/claude-sonnet-5.5": 0.5948314573867043,
          "deepseek/deepseek-flash": 0.04461235930400283
        },
        "llm_eur_saved_per_1000_recall_peak": {
          "deepseek/deepseek-flash": 0.08922471860800565
        },
        "recall@5_reranked": 0.6118293000260213,
        "saved_passages_hit": 2,
        "saved_passages_recall": 2
      }
    },
    "costs": {
      "api": {
        "voyage-rerank-3": {
          "hybrid-71k": {
            "k@100": {
              "eur_per_1000": 0.9534382927178939,
              "n_results": 3,
              "source": "latency",
              "tokenizer_deviation_pct": 2.0234357978206052,
              "tokenizer_estimate": 21800.83606557377,
              "tokens_per_query": 21368.45901639344,
              "tokens_per_query_max": 21368.45901639344,
              "tokens_per_query_min": 21368.45901639344
            },
            "k@20": {
              "eur_per_1000": 0.18955097240061675,
              "n_results": 3,
              "source": "latency",
              "tokenizer_deviation_pct": 2.0316337193777123,
              "tokenizer_estimate": 4334.524590163935,
              "tokens_per_query": 4248.216393442623,
              "tokens_per_query_max": 4248.216393442623,
              "tokens_per_query_min": 4248.216393442623
            }
          },
          "keyword-71k": {
            "k@100": {
              "eur_per_1000": 0.9684090466103685,
              "n_queries": 304,
              "source": "quality run",
              "tokens_per_query": 21703.98355263158
            }
          },
          "keyword-full": {
            "k@100": {
              "eur_per_1000": 1.2264181488770114,
              "n_queries": 304,
              "source": "quality run",
              "tokens_per_query": 27486.48355263158
            }
          },
          "vector-71k": {
            "k@100": {
              "eur_per_1000": 0.7805761257874149,
              "n_queries": 305,
              "source": "quality run",
              "tokens_per_query": 17494.27213114754
            }
          }
        },
        "voyage-rerank-3-lite": {
          "hybrid-71k": {
            "k@100": {
              "eur_per_1000": 0.3813753170871576,
              "n_results": 3,
              "source": "latency",
              "tokenizer_deviation_pct": 2.0234357978206052,
              "tokenizer_estimate": 21800.83606557377,
              "tokens_per_query": 21368.45901639344,
              "tokens_per_query_max": 21368.45901639344,
              "tokens_per_query_min": 21368.45901639344
            },
            "k@20": {
              "eur_per_1000": 0.07582038896024669,
              "n_results": 3,
              "source": "latency",
              "tokenizer_deviation_pct": 2.0316337193777123,
              "tokenizer_estimate": 4334.524590163935,
              "tokens_per_query": 4248.216393442623,
              "tokens_per_query_max": 4248.216393442623,
              "tokens_per_query_min": 4248.216393442623
            }
          },
          "keyword-71k": {
            "k@100": {
              "eur_per_1000": 0.3873636186441474,
              "n_queries": 304,
              "source": "quality run",
              "tokens_per_query": 21703.98355263158
            }
          },
          "keyword-full": {
            "k@100": {
              "eur_per_1000": 0.4905672595508046,
              "n_queries": 304,
              "source": "quality run",
              "tokens_per_query": 27486.48355263158
            }
          },
          "vector-71k": {
            "k@100": {
              "eur_per_1000": 0.31223045031496593,
              "n_queries": 305,
              "source": "quality run",
              "tokens_per_query": 17494.27213114754
            }
          }
        }
      },
      "break_even": {
        "bge-v2-m3-cuda-desktop-se-k@100": {
          "approximation": true,
          "break_even_vs_rerank-3": 224450.81305678052,
          "break_even_vs_rerank-3-lite": 561127.0326419512,
          "capacity_50": 2665384.942904679,
          "capacity_full": 5330769.885809358,
          "eur_per_month": 214.0,
          "k": 100,
          "model": "bge-v2-m3",
          "server": "gex45",
          "status_chat": "reachable",
          "status_chat_vs_rerank-3-lite": "reachable",
          "status_live": "not reachable",
          "status_live_vs_rerank-3-lite": "not reachable"
        },
        "bge-v2-m3-cuda-desktop-se-k@20": {
          "approximation": true,
          "break_even_vs_rerank-3": 1128983.9207351052,
          "break_even_vs_rerank-3-lite": 2822459.8018377633,
          "capacity_50": 15595448.798455572,
          "capacity_full": 31190897.596911144,
          "eur_per_month": 214.0,
          "k": 20,
          "model": "bge-v2-m3",
          "server": "gex45",
          "status_chat": "reachable",
          "status_chat_vs_rerank-3-lite": "reachable",
          "status_live": "reachable",
          "status_live_vs_rerank-3-lite": "reachable"
        },
        "gte-base-cuda-desktop-se-k@100": {
          "approximation": true,
          "break_even_vs_rerank-3": 224450.81305678052,
          "break_even_vs_rerank-3-lite": 561127.0326419512,
          "capacity_50": 5488996.783029616,
          "capacity_full": 10977993.566059232,
          "eur_per_month": 214.0,
          "k": 100,
          "model": "gte-base",
          "server": "gex45",
          "status_chat": "reachable",
          "status_chat_vs_rerank-3-lite": "reachable",
          "status_live": "reachable",
          "status_live_vs_rerank-3-lite": "reachable"
        },
        "gte-base-cuda-desktop-se-k@20": {
          "approximation": true,
          "break_even_vs_rerank-3": 1128983.9207351052,
          "break_even_vs_rerank-3-lite": 2822459.8018377633,
          "capacity_50": 32669886.874571107,
          "capacity_full": 65339773.749142215,
          "eur_per_month": 214.0,
          "k": 20,
          "model": "gte-base",
          "server": "gex45",
          "status_chat": "reachable",
          "status_chat_vs_rerank-3-lite": "reachable",
          "status_live": "reachable",
          "status_live_vs_rerank-3-lite": "reachable"
        },
        "jina-v3.5-cuda-desktop-se-k@100": {
          "approximation": true,
          "break_even_vs_rerank-3": 224450.81305678052,
          "break_even_vs_rerank-3-lite": 561127.0326419512,
          "capacity_50": 418675.2916504048,
          "capacity_full": 837350.5833008096,
          "eur_per_month": 214.0,
          "k": 100,
          "model": "jina-v3.5",
          "server": "gex45",
          "status_chat": "not reachable",
          "status_chat_vs_rerank-3-lite": "not reachable",
          "status_live": "not reachable",
          "status_live_vs_rerank-3-lite": "not reachable"
        },
        "jina-v3.5-cuda-desktop-se-k@20": {
          "approximation": true,
          "break_even_vs_rerank-3": 1128983.9207351052,
          "break_even_vs_rerank-3-lite": 2822459.8018377633,
          "capacity_50": 6375494.62611351,
          "capacity_full": 12750989.25222702,
          "eur_per_month": 214.0,
          "k": 20,
          "model": "jina-v3.5",
          "server": "gex45",
          "status_chat": "reachable",
          "status_chat_vs_rerank-3-lite": "reachable",
          "status_live": "reachable",
          "status_live_vs_rerank-3-lite": "reachable"
        },
        "mxbai-large-v2-cuda-desktop-se-k@100": {
          "approximation": true,
          "break_even_vs_rerank-3": 224450.81305678052,
          "break_even_vs_rerank-3-lite": 561127.0326419512,
          "capacity_50": 568343.8878708056,
          "capacity_full": 1136687.7757416111,
          "eur_per_month": 214.0,
          "k": 100,
          "model": "mxbai-large-v2",
          "server": "gex45",
          "status_chat": "not reachable",
          "status_chat_vs_rerank-3-lite": "not reachable",
          "status_live": "not reachable",
          "status_live_vs_rerank-3-lite": "not reachable"
        },
        "mxbai-large-v2-cuda-desktop-se-k@20": {
          "approximation": true,
          "break_even_vs_rerank-3": 1128983.9207351052,
          "break_even_vs_rerank-3-lite": 2822459.8018377633,
          "capacity_50": 2271543.9541040836,
          "capacity_full": 4543087.908208167,
          "eur_per_month": 214.0,
          "k": 20,
          "model": "mxbai-large-v2",
          "server": "gex45",
          "status_chat": "reachable",
          "status_chat_vs_rerank-3-lite": "not reachable",
          "status_live": "not reachable",
          "status_live_vs_rerank-3-lite": "not reachable"
        },
        "querit-4b-cuda-desktop-se-k@100": {
          "approximation": true,
          "break_even_vs_rerank-3": 224450.81305678052,
          "break_even_vs_rerank-3-lite": 561127.0326419512,
          "capacity_50": 98718.78887982092,
          "capacity_full": 197437.57775964183,
          "eur_per_month": 214.0,
          "k": 100,
          "model": "querit-4b",
          "server": "gex45",
          "status_chat": "not reachable",
          "status_chat_vs_rerank-3-lite": "not reachable",
          "status_live": "not reachable",
          "status_live_vs_rerank-3-lite": "not reachable"
        },
        "querit-4b-cuda-desktop-se-k@20": {
          "approximation": true,
          "break_even_vs_rerank-3": 1128983.9207351052,
          "break_even_vs_rerank-3-lite": 2822459.8018377633,
          "capacity_50": 489421.7829323102,
          "capacity_full": 978843.5658646204,
          "eur_per_month": 214.0,
          "k": 20,
          "model": "querit-4b",
          "server": "gex45",
          "status_chat": "not reachable",
          "status_chat_vs_rerank-3-lite": "not reachable",
          "status_live": "not reachable",
          "status_live_vs_rerank-3-lite": "not reachable"
        },
        "qwen3-0.6b-cuda-desktop-se-k@100": {
          "approximation": true,
          "break_even_vs_rerank-3": 224450.81305678052,
          "break_even_vs_rerank-3-lite": 561127.0326419512,
          "capacity_50": 1001956.3006692568,
          "capacity_full": 2003912.6013385137,
          "eur_per_month": 214.0,
          "k": 100,
          "model": "qwen3-0.6b",
          "server": "gex45",
          "status_chat": "not reachable",
          "status_chat_vs_rerank-3-lite": "not reachable",
          "status_live": "not reachable",
          "status_live_vs_rerank-3-lite": "not reachable"
        },
        "qwen3-0.6b-cuda-desktop-se-k@20": {
          "approximation": true,
          "break_even_vs_rerank-3": 1128983.9207351052,
          "break_even_vs_rerank-3-lite": 2822459.8018377633,
          "capacity_50": 3482179.5100111053,
          "capacity_full": 6964359.020022211,
          "eur_per_month": 214.0,
          "k": 20,
          "model": "qwen3-0.6b",
          "server": "gex45",
          "status_chat": "reachable",
          "status_chat_vs_rerank-3-lite": "reachable",
          "status_live": "not reachable",
          "status_live_vs_rerank-3-lite": "not reachable"
        },
        "qwen3-4b-cuda-desktop-se-k@100": {
          "approximation": true,
          "break_even_vs_rerank-3": 224450.81305678052,
          "break_even_vs_rerank-3-lite": 561127.0326419512,
          "capacity_50": 333222.3453825137,
          "capacity_full": 666444.6907650274,
          "eur_per_month": 214.0,
          "k": 100,
          "model": "qwen3-4b",
          "server": "gex45",
          "status_chat": "not reachable",
          "status_chat_vs_rerank-3-lite": "not reachable",
          "status_live": "not reachable",
          "status_live_vs_rerank-3-lite": "not reachable"
        },
        "qwen3-4b-cuda-desktop-se-k@20": {
          "approximation": true,
          "break_even_vs_rerank-3": 1128983.9207351052,
          "break_even_vs_rerank-3-lite": 2822459.8018377633,
          "capacity_50": 1402217.4492120992,
          "capacity_full": 2804434.8984241984,
          "eur_per_month": 214.0,
          "k": 20,
          "model": "qwen3-4b",
          "server": "gex45",
          "status_chat": "reachable",
          "status_chat_vs_rerank-3-lite": "not reachable",
          "status_live": "not reachable",
          "status_live_vs_rerank-3-lite": "not reachable"
        }
      },
      "first_stage_p95_ms": 67.2347182946396,
      "fx": {
        "date": "2026-10-09",
        "note": "ECB reference rate of Friday 2026-10-09, the last one before the measurement days (Saturday 2026-10-10, Sunday 2026-10-11); the ECB publishes no rate on weekends",
        "source": "ECB reference rate",
        "usd_per_eur": 1.1206,
        "weekday": "Friday"
      },
      "mean_passage_tokens": {
        "hybrid-71k": 205.30344262295083,
        "keyword-71k": 209.1931218675811,
        "keyword-full": 266.5060197368421,
        "vector-71k": 166.64203278688524
      },
      "prices": {
        "budgets": {
          "chat_rerank_p95_ms": 2000,
          "live_search_total_p95_ms": 500,
          "note": "assumptions: live search a few hundred milliseconds in total, chat 1 to 2 s for the reranking step"
        },
        "fx": {
          "date": "2026-10-09",
          "source": "ECB reference rate",
          "usd_per_eur": 1.1206
        },
        "hetzner": {
          "gex45": {
            "date": "2026-10-09",
            "eur_per_month": 212.3,
            "extra_eur_per_month": 1.7,
            "gpu": "RTX PRO 4000 Blackwell SFF 24 GB",
            "note": "price as checked on 2026-10-09; not re-checkable on 2026-10-11 (JavaScript-only page)",
            "setup_eur": 209.0,
            "source": "hetzner.com GPU server overview, website-price-api.hetzner.com"
          }
        },
        "llm": [
          {
            "date": "2026-10-11",
            "id": "anthropic/claude-sonnet-5.5",
            "source": "https://claude.com/pricing",
            "usd_per_million_input": 2.0
          },
          {
            "date": "2026-10-11",
            "id": "deepseek/deepseek-flash",
            "note": "DeepSeek-V4.1-Flash, cache miss, off-peak; legacy name deepseek-v4-flash is billed at this price",
            "source": "https://api-docs.deepseek.com/quick_start/pricing",
            "usd_per_million_input": 0.15,
            "usd_per_million_input_peak": 0.3
          }
        ],
        "voyage": {
          "rerank-3": {
            "date": "2026-10-11",
            "source": "https://docs.voyageai.com/docs/pricing",
            "usd_per_million_tokens": 0.05
          },
          "rerank-3-lite": {
            "date": "2026-10-11",
            "source": "https://docs.voyageai.com/docs/pricing",
            "usd_per_million_tokens": 0.02
          }
        }
      },
      "prices_sha256": "ed1be9a6716182ba4f567fc56b5d8a10d11a33c8ccf901872d97fb2ae2f4c93d",
      "usd_per_eur": 1.1206,
      "voyage_tokenizer_revision": "2d2c29517b5c3bbf130c879fda12b1678e598158"
    },
    "created_utc": "2026-10-11T16:42:12+00:00",
    "notes": {
      "gpu_approximation": "GPU latency was measured on an RTX 5070 Ti, not on the GEX45 (RTX PRO 4000 Blackwell SFF 24 GB): its break-even values are an approximation",
      "passage_tokens": "mean_passage_tokens: per stage, voyageai/rerank-3 at revision 2d2c29517b5c3bbf130c879fda12b1678e598158 on title and text of the top 100 candidates of all queries",
      "saving": "saved passages and the LLM saving are computed for Hit@5 (k_needed_hit) and Recall@5 (k_needed_recall); an LLM with a peak price is also priced at it (keys ending in _peak)",
      "setup_fee": "the GEX45 monthly price is 214.00 EUR; the one-off setup fee of 209 EUR is not part of it"
    },
    "run_id": "costs-round1-20261011T164212Z-95734e"
  },
  "cpu_indicative": {
    "conditions": {
      "cpu": "AMD Ryzen 9 5950X 16-Core Processor",
      "memory_cap": {
        "gb": 30,
        "source": "DEVIATIONS.md 2026-10-11"
      },
      "n_queries": 10,
      "passes": 1,
      "threads": 8,
      "warmup_calls": 2
    },
    "models": {
      "bge-v2-m3": {
        "created_utc": "2026-10-11T04:59:51+00:00",
        "k@100": {
          "mean_ms": 61409.83661879727,
          "n": 10,
          "p50_ms": 62146.97118299955,
          "p95_ms": 62642.68078599707
        },
        "k@20": {
          "mean_ms": 9977.319651597645,
          "n": 10,
          "p50_ms": 10241.914545491454,
          "p95_ms": 12916.846485299175
        },
        "loadavg_before": [
          7.18212890625,
          5.61181640625,
          3.3896484375
        ],
        "run_id": "latency-bge-v2-m3-cpu-desktop-se-cpu8-20261011T045951Z-76c895"
      },
      "gte-base": {
        "created_utc": "2026-10-11T04:45:37+00:00",
        "k@100": {
          "mean_ms": 28298.464546899777,
          "n": 10,
          "p50_ms": 28612.30463501124,
          "p95_ms": 28819.95504853985
        },
        "k@20": {
          "mean_ms": 4223.670570898685,
          "n": 10,
          "p50_ms": 4512.1684889891185,
          "p95_ms": 5357.161884418747
        },
        "loadavg_before": [
          1.12451171875,
          1.54296875,
          1.46630859375
        ],
        "run_id": "latency-gte-base-cpu-desktop-se-cpu8-20261011T044537Z-764ee9"
      },
      "jina-v3.5": {
        "created_utc": "2026-10-11T06:06:24+00:00",
        "k@100": {
          "mean_ms": 188551.24795570446,
          "n": 10,
          "p50_ms": 195979.08959601773,
          "p95_ms": 283557.6599151027
        },
        "k@20": {
          "mean_ms": 19275.705510302214,
          "n": 10,
          "p50_ms": 18282.998731010593,
          "p95_ms": 34405.58793949019
        },
        "loadavg_before": [
          6.3681640625,
          7.09912109375,
          7.091796875
        ],
        "run_id": "latency-jina-v3.5-cpu-desktop-se-cpu8-20261011T060624Z-016a59"
      },
      "qwen3-0.6b": {
        "created_utc": "2026-10-11T09:24:40+00:00",
        "k@100": {
          "status": "not measurable in 30 GB"
        },
        "k@20": {
          "mean_ms": 44060.33865789941,
          "n": 10,
          "p50_ms": 35533.5564614943,
          "p95_ms": 98836.73847516466
        },
        "loadavg_before": [
          0.7578125,
          0.5498046875,
          1.03759765625
        ],
        "run_id": "latency-qwen3-0.6b-cpu-desktop-se-cpu8-20261011T092440Z-101896"
      }
    }
  },
  "first_stage": {
    "hybrid-71k": {
      "contamination": [
        "MIRACLRetrieval",
        "MIRACLRetrievalHardNegatives",
        "MIRACLReranking"
      ],
      "counts": {
        "queries_with_fewer_than_100": 0,
        "queries_without_hits": 0
      },
      "created_utc": "2026-10-10T00:07:29+00:00",
      "exploratory": false,
      "latency": {
        "mean_ms": 38.601024595694874,
        "n": 915,
        "p50_ms": 35.02669600129593,
        "p95_ms": 55.96083429700229
      },
      "latency_created_utc": "2026-10-10T15:22:34+00:00",
      "latency_run_id": "latency-first-stage-hybrid-71k-desktop-se-20261010T152234Z-80ffe8",
      "metrics": {
        "ceiling_ndcg@10": {
          "mean": 0.46115157793366746
        },
        "hit@10": {
          "mean": 0.4360655737704918
        },
        "hit@5": {
          "ci_high": 0.3901639344262295,
          "ci_low": 0.28524590163934427,
          "mean": 0.3377049180327869
        },
        "judged@10": {
          "mean": 0.12688524590163935
        },
        "mrr@10": {
          "mean": 0.18533437418683318
        },
        "ndcg@10": {
          "ci_high": 0.20373374554993806,
          "ci_low": 0.1476800243657697,
          "mean": 0.17498289422243518
        },
        "ndcg_condensed@10": {
          "mean": 0.38735964250551674
        },
        "recall@10": {
          "mean": 0.26903200624512097
        },
        "recall@100": {
          "mean": 0.4350039032006245
        },
        "recall@20": {
          "mean": 0.32904761904761903
        },
        "recall@5": {
          "mean": 0.19079625292740046
        }
      },
      "query_embedding": {
        "batch_size": 1,
        "device": "cuda",
        "dtype": "float16",
        "mean_ms": 10.062300876788719,
        "model": "Snowflake/snowflake-arctic-embed-l-v2.0",
        "n": 915,
        "p50_ms": 9.828702997765504,
        "p95_ms": 11.273883997637313,
        "revision": "ac6544c8a46e00af67e330e85a9028c66b8cfd9a",
        "warmup": 20
      },
      "run_id": "first_stage-hybrid-71k-20261010T000729Z-108e4c"
    },
    "keyword-71k": {
      "contamination": [],
      "counts": {
        "queries_with_fewer_than_100": 3,
        "queries_without_hits": 1
      },
      "created_utc": "2026-10-10T00:07:21+00:00",
      "exploratory": false,
      "latency": {
        "mean_ms": 38.32853186802107,
        "n": 915,
        "p50_ms": 36.75935500359628,
        "p95_ms": 55.03031899861525
      },
      "latency_created_utc": "2026-10-10T15:21:26+00:00",
      "latency_run_id": "latency-first-stage-keyword-71k-desktop-se-20261010T152126Z-330905",
      "metrics": {
        "ceiling_ndcg@10": {
          "mean": 0.2638523417479578
        },
        "hit@10": {
          "mean": 0.18688524590163935
        },
        "hit@5": {
          "ci_high": 0.1737704918032787,
          "ci_low": 0.09836065573770492,
          "mean": 0.13442622950819672
        },
        "judged@10": {
          "mean": 0.07081967213114755
        },
        "mrr@10": {
          "mean": 0.09334894613583139
        },
        "ndcg@10": {
          "ci_high": 0.09605649644661068,
          "ci_low": 0.05430560617418975,
          "mean": 0.07428238494756044
        },
        "ndcg_condensed@10": {
          "mean": 0.20465572369341348
        },
        "recall@10": {
          "mean": 0.10108508977361437
        },
        "recall@100": {
          "mean": 0.23757220921155345
        },
        "recall@20": {
          "mean": 0.14451990632318498
        },
        "recall@5": {
          "mean": 0.06959406713505073
        }
      },
      "query_embedding": null,
      "run_id": "first_stage-keyword-71k-20261010T000721Z-52e3c2"
    },
    "keyword-71k-ltr": {
      "contamination": [],
      "counts": {
        "queries_with_fewer_than_100": 3,
        "queries_without_hits": 1
      },
      "created_utc": "2026-10-10T16:21:29+00:00",
      "exploratory": true,
      "latency": null,
      "latency_created_utc": null,
      "latency_run_id": null,
      "metrics": {
        "ceiling_ndcg@10": {
          "mean": 0.5014375832052528
        },
        "hit@10": {
          "mean": 0.3114754098360656
        },
        "hit@5": {
          "ci_high": 0.2754098360655738,
          "ci_low": 0.18360655737704917,
          "mean": 0.22950819672131148
        },
        "judged@10": {
          "mean": 0.1321311475409836
        },
        "mrr@10": {
          "mean": 0.14824225865209473
        },
        "ndcg@10": {
          "ci_high": 0.14879409253958079,
          "ci_low": 0.09836524572260068,
          "mean": 0.12245521303522566
        },
        "ndcg_condensed@10": {
          "mean": 0.37037365727189425
        },
        "recall@10": {
          "mean": 0.17128024980483997
        },
        "recall@100": {
          "mean": 0.4625696070778038
        },
        "recall@20": {
          "mean": 0.2454215456674473
        },
        "recall@5": {
          "mean": 0.11541764246682279
        }
      },
      "query_embedding": null,
      "run_id": "first_stage-keyword-71k-ltr-20261010T162129Z-6dea2f"
    },
    "keyword-full": {
      "contamination": [],
      "counts": {
        "queries_with_fewer_than_100": 1,
        "queries_without_hits": 1
      },
      "created_utc": "2026-10-10T01:48:27+00:00",
      "exploratory": false,
      "latency": {
        "mean_ms": 160.51128488536568,
        "n": 915,
        "p50_ms": 56.46387599699665,
        "p95_ms": 764.8886294016844
      },
      "latency_created_utc": "2026-10-10T15:25:12+00:00",
      "latency_run_id": "latency-first-stage-keyword-full-desktop-se-20261010T152512Z-e352d2",
      "metrics": {
        "ceiling_ndcg@10": {
          "mean": 0.18570729421947257
        },
        "hit@10": {
          "mean": 0.14098360655737704
        },
        "hit@5": {
          "ci_high": 0.14426229508196722,
          "ci_low": 0.07540983606557378,
          "mean": 0.10819672131147541
        },
        "judged@10": {
          "mean": 0.05114754098360656
        },
        "mrr@10": {
          "mean": 0.0804072339318241
        },
        "ndcg@10": {
          "ci_high": 0.08247661951521222,
          "ci_low": 0.042417668801672095,
          "mean": 0.0614607226564852
        },
        "ndcg_condensed@10": {
          "mean": 0.14786485122088336
        },
        "recall@10": {
          "mean": 0.07630757220921154
        },
        "recall@100": {
          "mean": 0.1662217017954723
        },
        "recall@20": {
          "mean": 0.10241217798594845
        },
        "recall@5": {
          "mean": 0.057572209211553475
        }
      },
      "query_embedding": null,
      "run_id": "first_stage-keyword-full-20261010T014827Z-cc6118"
    },
    "keyword-plain-71k": {
      "contamination": [],
      "counts": {
        "queries_with_fewer_than_100": 287,
        "queries_without_hits": 0
      },
      "created_utc": "2026-10-10T00:07:33+00:00",
      "exploratory": false,
      "latency": {
        "mean_ms": 27.576348629857723,
        "n": 915,
        "p50_ms": 21.44671000132803,
        "p95_ms": 76.7096120980568
      },
      "latency_created_utc": "2026-10-10T15:25:38+00:00",
      "latency_run_id": "latency-first-stage-keyword-plain-71k-desktop-se-20261010T152538Z-2237d2",
      "metrics": {
        "ceiling_ndcg@10": {
          "mean": 0.015728790487001005
        },
        "hit@10": {
          "mean": 0.03278688524590164
        },
        "hit@5": {
          "ci_high": 0.04918032786885246,
          "ci_low": 0.013114754098360656,
          "mean": 0.029508196721311476
        },
        "judged@10": {
          "mean": 0.007540983606557377
        },
        "mrr@10": {
          "mean": 0.025819672131147543
        },
        "ndcg@10": {
          "ci_high": 0.020784422170357714,
          "ci_low": 0.004857981133367159,
          "mean": 0.012118580232116774
        },
        "ndcg_condensed@10": {
          "mean": 0.013986005307201258
        },
        "recall@10": {
          "mean": 0.010140515222482435
        },
        "recall@100": {
          "mean": 0.011779859484777517
        },
        "recall@20": {
          "mean": 0.011779859484777517
        },
        "recall@5": {
          "mean": 0.008501170960187352
        }
      },
      "query_embedding": null,
      "run_id": "first_stage-keyword-plain-71k-20261010T000733Z-b317ea"
    },
    "vector-71k": {
      "contamination": [
        "MIRACLRetrieval",
        "MIRACLRetrievalHardNegatives",
        "MIRACLReranking"
      ],
      "counts": {
        "queries_with_fewer_than_100": 0,
        "queries_without_hits": 0
      },
      "created_utc": "2026-10-10T00:07:25+00:00",
      "exploratory": false,
      "latency": {
        "mean_ms": 33.95838939454063,
        "n": 915,
        "p50_ms": 31.73468599561602,
        "p95_ms": 41.02490899240365
      },
      "latency_created_utc": "2026-10-10T15:21:58+00:00",
      "latency_run_id": "latency-first-stage-vector-71k-desktop-se-20261010T152158Z-b8b773",
      "metrics": {
        "ceiling_ndcg@10": {
          "mean": 0.9716164506014489
        },
        "hit@10": {
          "mean": 0.8852459016393442
        },
        "hit@5": {
          "ci_high": 0.8622950819672132,
          "ci_low": 0.7770491803278688,
          "mean": 0.819672131147541
        },
        "judged@10": {
          "mean": 0.3737704918032787
        },
        "mrr@10": {
          "mean": 0.6627582617746551
        },
        "ndcg@10": {
          "ci_high": 0.6265068816908699,
          "ci_low": 0.5532981146569897,
          "mean": 0.5903147920817311
        },
        "ndcg_condensed@10": {
          "mean": 0.8171232816985772
        },
        "recall@10": {
          "mean": 0.7042583918813426
        },
        "recall@100": {
          "mean": 0.9652354931043456
        },
        "recall@20": {
          "mean": 0.8045771532656778
        },
        "recall@5": {
          "mean": 0.5768214936247723
        }
      },
      "query_embedding": {
        "batch_size": 1,
        "device": "cuda",
        "dtype": "float16",
        "mean_ms": 10.062300876788719,
        "model": "Snowflake/snowflake-arctic-embed-l-v2.0",
        "n": 915,
        "p50_ms": 9.828702997765504,
        "p95_ms": 11.273883997637313,
        "revision": "ac6544c8a46e00af67e330e85a9028c66b8cfd9a",
        "warmup": 20
      },
      "run_id": "first_stage-vector-71k-20261010T000725Z-1fc254"
    }
  },
  "hardware": {
    "gpu": "NVIDIA GeForce RTX 5070 Ti"
  },
  "judgments": {
    "checks": {
      "answered": 72,
      "control_agreed": 13,
      "control_agreement": 0.8666666666666667,
      "control_answered": 15,
      "control_total": 15,
      "control_warning": false,
      "judge": {
        "date": "2026-10-10",
        "kind": "llm",
        "labels_sha256": "01e4c73defadd585a3855136c2be7144863f5a9e66c2ed80d55e7285b328e124",
        "model": "claude-opus-5-5",
        "pairs_sha256": "9d23d6678b11205cbc58a320ff49f7475b3fcd0155e589853c4e5971cd86b67d",
        "prompt_sha256": "5fea598f8f99864a5ad740e2d20196e280266990ea4ba45fbb430d7bb482ee3f"
      },
      "sample_sha256": "70d6de4659f40174054f1ca7f2b423287acb2860de0cb4b0e4f384855fcfd1dc",
      "seed": 20261009,
      "sources": {
        "control-irrelevant": {
          "corpus": "71k",
          "corpus_sha256": "8e33b4da3abaf1bf7101d1d68f50d4c3c59fda65010153d95a7412a5c0d1196b",
          "qrels_sha256": "5d4a0d7ac42c7839a0513ba6deb6b27fc62803f318be8a7c411e10dbbde7578c"
        },
        "control-relevant": {
          "corpus": "71k",
          "corpus_sha256": "8e33b4da3abaf1bf7101d1d68f50d4c3c59fda65010153d95a7412a5c0d1196b",
          "qrels_sha256": "5d4a0d7ac42c7839a0513ba6deb6b27fc62803f318be8a7c411e10dbbde7578c"
        },
        "hybrid-71k:jina-v3.5": {
          "model": "jina-v3.5",
          "run_sha256": "c7910a55cac133dec5345383d1a96760db613e6986b74fd17df77eef78d77344",
          "stage": "hybrid-71k"
        },
        "hybrid-71k:qwen3-4b": {
          "model": "qwen3-4b",
          "run_sha256": "4a9458d131e44d4eb0b5f523f19e0dda48dcc55970b40e15641deee340208bc7",
          "stage": "hybrid-71k"
        },
        "hybrid-71k:voyage-rerank-3": {
          "model": "voyage-rerank-3",
          "run_sha256": "3e949aeb397e8a053376542666e591d5993eea5b5e8847b14e62b2a01fb64cc2",
          "stage": "hybrid-71k"
        },
        "keyword-full:qwen3-4b": {
          "model": "qwen3-4b",
          "run_sha256": "ca03597659b4ce2c9399e8148c982b25230617ec3706815baad8148756c9d46d",
          "stage": "keyword-full"
        }
      },
      "unclear": 0
    },
    "created_utc": "2026-10-10T18:21:00+00:00",
    "metrics": {
      "control-irrelevant": {
        "judged": 7,
        "rate": 0.2857142857142857,
        "relevant": 2,
        "unclear": 0,
        "wilson_high": 0.6410709098517873,
        "wilson_low": 0.0822171657090155
      },
      "control-relevant": {
        "judged": 8,
        "rate": 1.0,
        "relevant": 8,
        "unclear": 0,
        "wilson_high": 1.0,
        "wilson_low": 0.6755843804891231
      },
      "hybrid-71k:jina-v3.5": {
        "judged": 15,
        "rate": 0.4,
        "relevant": 6,
        "unclear": 0,
        "wilson_high": 0.642535572434923,
        "wilson_low": 0.19824228082594664
      },
      "hybrid-71k:qwen3-4b": {
        "judged": 15,
        "rate": 0.4,
        "relevant": 6,
        "unclear": 0,
        "wilson_high": 0.642535572434923,
        "wilson_low": 0.19824228082594664
      },
      "hybrid-71k:voyage-rerank-3": {
        "judged": 15,
        "rate": 0.4666666666666667,
        "relevant": 7,
        "unclear": 0,
        "wilson_high": 0.6988336984894921,
        "wilson_low": 0.24809225259746442
      },
      "keyword-full:qwen3-4b": {
        "judged": 15,
        "rate": 0.06666666666666667,
        "relevant": 1,
        "unclear": 0,
        "wilson_high": 0.2981707755242399,
        "wilson_low": 0.011866588606194856
      }
    },
    "notes": {
      "judge_bias": "an LLM judge may favour the same passages as LLM-based rerankers; the control pairs measure agreement with the MIRACL annotators, not this bias",
      "unclear_answers": "unclear answers are left out of the rates and counted separately, per group and in total"
    },
    "run_id": "judgments-judged-sample-20261010T182100Z-ca8dbb"
  },
  "latency": {
    "api:desktop-se:afternoon": {
      "voyage-rerank-3": {
        "created_utc": "2026-10-11T10:05:17+00:00",
        "finished_local": "2026-10-11T12:05:17+02:00",
        "k@100": {
          "mean_ms": 351.57125757164414,
          "n": 301,
          "p50_ms": 341.25486901029944,
          "p50_ms_per_pass": [
            341.25486901029944
          ],
          "p95_ms": 456.9225680024829
        },
        "k@20": {
          "mean_ms": 288.2902025973203,
          "n": 305,
          "p50_ms": 278.45161501318216,
          "p50_ms_per_pass": [
            278.45161501318216
          ],
          "p95_ms": 366.03759081335744
        },
        "run_id": "latency-voyage-rerank-3-api-desktop-se-afternoon-20261011T100517Z-9099a1",
        "tcp_connect_ms": 12.186575986561365,
        "training_opt_out": "confirmed",
        "training_opt_out_checked": "2026-10-11",
        "weekday": "Sunday"
      },
      "voyage-rerank-3-lite": {
        "created_utc": "2026-10-11T10:08:52+00:00",
        "finished_local": "2026-10-11T12:08:52+02:00",
        "k@100": {
          "mean_ms": 351.9095244924309,
          "n": 305,
          "p50_ms": 344.37996201450005,
          "p50_ms_per_pass": [
            344.37996201450005
          ],
          "p95_ms": 435.10518818511633
        },
        "k@20": {
          "mean_ms": 294.4341976490429,
          "n": 305,
          "p50_ms": 287.0092080265749,
          "p50_ms_per_pass": [
            287.0092080265749
          ],
          "p95_ms": 356.0759455896914
        },
        "run_id": "latency-voyage-rerank-3-lite-api-desktop-se-afternoon-20261011T100852Z-9066d4",
        "tcp_connect_ms": 12.089155497960746,
        "training_opt_out": "confirmed",
        "training_opt_out_checked": "2026-10-11",
        "weekday": "Sunday"
      }
    },
    "api:desktop-se:evening": {
      "voyage-rerank-3": {
        "created_utc": "2026-10-10T16:29:02+00:00",
        "finished_local": "2026-10-10T18:29:02+02:00",
        "k@100": {
          "mean_ms": 359.9083460498047,
          "n": 301,
          "p50_ms": 342.206155997701,
          "p50_ms_per_pass": [
            342.206155997701
          ],
          "p95_ms": 504.94668100145645
        },
        "k@20": {
          "mean_ms": 288.0009136493337,
          "n": 305,
          "p50_ms": 278.48049098975025,
          "p50_ms_per_pass": [
            278.48049098975025
          ],
          "p95_ms": 355.46070059936034
        },
        "run_id": "latency-voyage-rerank-3-api-desktop-se-evening-20261010T162902Z-f848c5",
        "tcp_connect_ms": 12.122179003199562,
        "training_opt_out": "unconfirmed",
        "training_opt_out_checked": "2026-10-09",
        "weekday": "Saturday"
      },
      "voyage-rerank-3-lite": {
        "created_utc": "2026-10-10T16:32:44+00:00",
        "finished_local": "2026-10-10T18:32:44+02:00",
        "k@100": {
          "mean_ms": 369.36725611129555,
          "n": 305,
          "p50_ms": 353.2199600012973,
          "p50_ms_per_pass": [
            353.2199600012973
          ],
          "p95_ms": 481.75052179140044
        },
        "k@20": {
          "mean_ms": 297.00609048528855,
          "n": 305,
          "p50_ms": 289.27212199778296,
          "p50_ms_per_pass": [
            289.27212199778296
          ],
          "p95_ms": 352.49391680117697
        },
        "run_id": "latency-voyage-rerank-3-lite-api-desktop-se-evening-20261010T163244Z-facd01",
        "tcp_connect_ms": 12.181218007754069,
        "training_opt_out": "unconfirmed",
        "training_opt_out_checked": "2026-10-09",
        "weekday": "Saturday"
      }
    },
    "api:desktop-se:morning": {
      "voyage-rerank-3": {
        "created_utc": "2026-10-11T06:34:14+00:00",
        "finished_local": "2026-10-11T08:34:14+02:00",
        "k@100": {
          "mean_ms": 350.45696019239114,
          "n": 301,
          "p50_ms": 340.6027519959025,
          "p50_ms_per_pass": [
            340.6027519959025
          ],
          "p95_ms": 433.673435007222
        },
        "k@20": {
          "mean_ms": 286.0286203151966,
          "n": 305,
          "p50_ms": 278.1783709942829,
          "p50_ms_per_pass": [
            278.1783709942829
          ],
          "p95_ms": 328.9324154029601
        },
        "run_id": "latency-voyage-rerank-3-api-desktop-se-morning-20261011T063414Z-47493a",
        "tcp_connect_ms": 12.221959492308088,
        "training_opt_out": "unconfirmed",
        "training_opt_out_checked": "2026-10-09",
        "weekday": "Sunday"
      },
      "voyage-rerank-3-lite": {
        "created_utc": "2026-10-11T06:37:50+00:00",
        "finished_local": "2026-10-11T08:37:50+02:00",
        "k@100": {
          "mean_ms": 356.2167529388881,
          "n": 305,
          "p50_ms": 350.03936299472116,
          "p50_ms_per_pass": [
            350.03936299472116
          ],
          "p95_ms": 414.25477258162573
        },
        "k@20": {
          "mean_ms": 293.16903815335655,
          "n": 305,
          "p50_ms": 286.7322310048621,
          "p50_ms_per_pass": [
            286.7322310048621
          ],
          "p95_ms": 361.3657616020646
        },
        "run_id": "latency-voyage-rerank-3-lite-api-desktop-se-morning-20261011T063750Z-d5f19f",
        "tcp_connect_ms": 11.974098495556973,
        "training_opt_out": "unconfirmed",
        "training_opt_out_checked": "2026-10-09",
        "weekday": "Sunday"
      }
    },
    "cuda:desktop-se": {
      "bge-v2-m3": {
        "created_utc": "2026-10-11T04:32:32+00:00",
        "k@100": {
          "mean_ms": 486.2337064858058,
          "n": 915,
          "p50_ms": 496.5486779983621,
          "p50_ms_per_pass": [
            512.6457090082113,
            496.0023900202941,
            494.9810719990637
          ],
          "p95_ms": 516.5834453160642
        },
        "k@20": {
          "mean_ms": 83.10116731801548,
          "n": 915,
          "p50_ms": 85.87239900953136,
          "p50_ms_per_pass": [
            84.49011397897266,
            87.4492930015549,
            87.9313990008086
          ],
          "p95_ms": 107.95813818695024
        },
        "k@50": {
          "mean_ms": 238.89692419894354,
          "n": 915,
          "p50_ms": 259.28083798498847,
          "p50_ms_per_pass": [
            259.2426020128187,
            259.23776798299514,
            259.2926849902142
          ],
          "p95_ms": 264.3223113118438
        },
        "run_id": "latency-bge-v2-m3-cuda-desktop-se-20261011T043232Z-d00b64"
      },
      "gte-base": {
        "created_utc": "2026-10-11T04:39:00+00:00",
        "k@100": {
          "mean_ms": 236.1087191755797,
          "n": 915,
          "p50_ms": 244.47898898506537,
          "p50_ms_per_pass": [
            244.489878998138,
            244.49402201571502,
            244.46565899415873
          ],
          "p95_ms": 247.25005190703087
        },
        "k@20": {
          "mean_ms": 39.66955885019464,
          "n": 915,
          "p50_ms": 41.41516599338502,
          "p50_ms_per_pass": [
            41.41516599338502,
            41.490216011879966,
            41.36783702415414
          ],
          "p95_ms": 50.12729699665215
        },
        "k@50": {
          "mean_ms": 112.82657481716625,
          "n": 915,
          "p50_ms": 123.08858399046585,
          "p50_ms_per_pass": [
            122.99052599701099,
            123.09000597451814,
            123.14969999715686
          ],
          "p95_ms": 125.39473618380725
        },
        "run_id": "latency-gte-base-cuda-desktop-se-20261011T043900Z-c4aece"
      },
      "jina-v3.5": {
        "created_utc": "2026-10-11T04:12:21+00:00",
        "k@100": {
          "mean_ms": 3095.477631104546,
          "n": 915,
          "p50_ms": 2957.4151929991785,
          "p50_ms_per_pass": [
            2956.885023013456,
            2957.7737099898513,
            2957.4151929991785
          ],
          "p95_ms": 5466.718614599085
        },
        "k@20": {
          "mean_ms": 203.27834560344368,
          "n": 915,
          "p50_ms": 186.30112300161272,
          "p50_ms_per_pass": [
            186.18972599506378,
            186.40104599762708,
            186.34896699222736
          ],
          "p95_ms": 337.6763823936926
        },
        "k@50": {
          "mean_ms": 905.8828080700242,
          "n": 915,
          "p50_ms": 835.7520879944786,
          "p50_ms_per_pass": [
            835.4137660062406,
            835.9546060091816,
            835.2819999854546
          ],
          "p95_ms": 1588.7781331024598
        },
        "run_id": "latency-jina-v3.5-cuda-desktop-se-20261011T041221Z-de1b7f"
      },
      "mxbai-large-v2": {
        "created_utc": "2026-10-11T03:06:22+00:00",
        "k@100": {
          "mean_ms": 2280.3095584527573,
          "n": 915,
          "p50_ms": 2095.3549850091804,
          "p50_ms_per_pass": [
            2094.8534989729524,
            2095.3549850091804,
            2096.0386439983267
          ],
          "p95_ms": 4342.567619209876
        },
        "k@20": {
          "mean_ms": 570.5370559343429,
          "n": 915,
          "p50_ms": 515.1620070100762,
          "p50_ms_per_pass": [
            515.0743809936102,
            515.1913350273389,
            514.9787080008537
          ],
          "p95_ms": 996.3384669885272
        },
        "k@50": {
          "mean_ms": 1355.6866292745992,
          "n": 915,
          "p50_ms": 1220.2538960264064,
          "p50_ms_per_pass": [
            1221.4215340209194,
            1220.2538960264064,
            1219.4071479898412
          ],
          "p95_ms": 2221.4463036943926
        },
        "run_id": "latency-mxbai-large-v2-cuda-desktop-se-20261011T030622Z-f837da"
      },
      "querit-4b": {
        "created_utc": "2026-10-11T00:12:54+00:00",
        "k@100": {
          "mean_ms": 13128.199957737883,
          "n": 915,
          "p50_ms": 13118.437324999832,
          "p50_ms_per_pass": [
            13125.792626990005,
            13120.229057996767,
            13108.918849990005
          ],
          "p95_ms": 13134.274961202755
        },
        "k@20": {
          "mean_ms": 2648.022718227162,
          "n": 915,
          "p50_ms": 2639.4875170080923,
          "p50_ms_per_pass": [
            2644.4881490024272,
            2641.3081909995526,
            2637.334397993982
          ],
          "p95_ms": 2687.7645314962137
        },
        "k@50": {
          "mean_ms": 6586.41369046368,
          "n": 915,
          "p50_ms": 6578.078530001221,
          "p50_ms_per_pass": [
            6576.649719005218,
            6576.906135014724,
            6581.500080006663
          ],
          "p95_ms": 6651.63715671515
        },
        "run_id": "latency-querit-4b-cuda-desktop-se-20261011T001254Z-186e05"
      },
      "qwen3-0.6b": {
        "created_utc": "2026-10-11T08:55:31+00:00",
        "k@100": {
          "mean_ms": 1293.469584586011,
          "n": 915,
          "p50_ms": 1166.8686180200893,
          "p50_ms_per_pass": [
            1175.879354996141,
            1163.237714994466,
            1146.6391159920022
          ],
          "p95_ms": 2647.571201398386
        },
        "k@20": {
          "mean_ms": 372.18069782848926,
          "n": 915,
          "p50_ms": 322.7822380140424,
          "p50_ms_per_pass": [
            326.3335970113985,
            322.7822380140424,
            319.6459099999629
          ],
          "p95_ms": 692.6114955916995
        },
        "k@50": {
          "mean_ms": 768.7205537607197,
          "n": 915,
          "p50_ms": 693.5476949729491,
          "p50_ms_per_pass": [
            688.9273619744927,
            685.8392379945144,
            694.2495509865694
          ],
          "p95_ms": 1268.9767349831525
        },
        "run_id": "latency-qwen3-0.6b-cuda-desktop-se-20261011T085531Z-bfd2bd"
      },
      "qwen3-4b": {
        "created_utc": "2026-10-11T02:00:21+00:00",
        "k@100": {
          "mean_ms": 3889.2949946447657,
          "n": 915,
          "p50_ms": 3799.245742993662,
          "p50_ms_per_pass": [
            3797.9197569948155,
            3799.245742993662,
            3800.098120002076
          ],
          "p95_ms": 5509.06642670161
        },
        "k@20": {
          "mean_ms": 924.250372670956,
          "n": 915,
          "p50_ms": 864.0762289869599,
          "p50_ms_per_pass": [
            864.396727993153,
            863.6506010079756,
            864.0762289869599
          ],
          "p95_ms": 1418.8344406022225
        },
        "k@50": {
          "mean_ms": 2047.7207791860167,
          "n": 915,
          "p50_ms": 1976.3904010178521,
          "p50_ms_per_pass": [
            1974.6302219864447,
            1976.3904010178521,
            1976.4167079993058
          ],
          "p95_ms": 2918.79690430942
        },
        "run_id": "latency-qwen3-4b-cuda-desktop-se-20261011T020021Z-aa5503"
      }
    }
  },
  "measured": "2026-10-11",
  "measured_from": "2026-10-09",
  "models": {
    "bge-v2-m3": {
      "architecture": "encoder",
      "commercial_use": "yes",
      "contamination": [
        "MIRACLReranking",
        "MIRACLRetrieval",
        "MIRACLRetrievalHardNegatives"
      ],
      "contamination_known": true,
      "hf_id": "BAAI/bge-reranker-v2-m3",
      "license": "Apache-2.0",
      "params_m": 568,
      "rated": true,
      "revision": "953dc6f6f85a1b2dbfca4c34a2796e7dde08d41e",
      "validation": {
        "deciding_reference": "model_card",
        "mean_spearman": 0.9996338560136244,
        "min_spearman": 0.9931991809667617,
        "n_queries": 20,
        "passed": true,
        "reason": null,
        "run_id": "validation-bge-v2-m3-20261010T033230Z-0cbba3",
        "status": "ok",
        "threshold": 0.99
      }
    },
    "gte-base": {
      "architecture": "encoder",
      "commercial_use": "yes",
      "contamination": [],
      "contamination_known": false,
      "hf_id": "Alibaba-NLP/gte-multilingual-reranker-base",
      "license": "Apache-2.0",
      "params_m": 306,
      "rated": true,
      "revision": "8215cf04918ba6f7b6a62bb44238ce2953d8831c",
      "validation": {
        "deciding_reference": "transformers457",
        "mean_spearman": 0.9999698466367711,
        "min_spearman": 0.9998829859538008,
        "n_queries": 20,
        "passed": true,
        "reason": null,
        "run_id": "validation-gte-base-20261010T033154Z-7ae9bd",
        "status": "ok",
        "threshold": 0.99
      }
    },
    "jina-v3.5": {
      "architecture": "listwise",
      "commercial_use": "license required",
      "contamination": [],
      "contamination_known": false,
      "hf_id": "jinaai/jina-reranker-v3.5",
      "license": "CC-BY-NC-4.0",
      "params_m": 600,
      "rated": true,
      "revision": "e8a93f33f0b22108f8c2364f8484ce3422552fbc",
      "validation": {
        "deciding_reference": "transformers457",
        "mean_spearman": 0.9999999999999998,
        "min_spearman": 0.9999999999999999,
        "n_queries": 20,
        "passed": true,
        "reason": null,
        "run_id": "validation-jina-v3.5-20261010T033454Z-79be40",
        "status": "ok",
        "threshold": 0.99
      }
    },
    "mxbai-large-v2": {
      "architecture": "llm",
      "commercial_use": "yes",
      "contamination": [],
      "contamination_known": true,
      "hf_id": "mixedbread-ai/mxbai-rerank-large-v2",
      "license": "Apache-2.0",
      "params_m": 1500,
      "rated": true,
      "revision": "ca7e1ee484c37c0ddd8d178a9a5c33cec575c5e6",
      "validation": {
        "deciding_reference": "vendor_package",
        "mean_spearman": 0.9991151682705217,
        "min_spearman": 0.9965865768232341,
        "n_queries": 20,
        "passed": true,
        "reason": null,
        "run_id": "validation-mxbai-large-v2-20261010T040457Z-eac0e2",
        "status": "ok",
        "threshold": 0.99
      }
    },
    "querit-4b": {
      "architecture": "llm",
      "commercial_use": "yes",
      "contamination": [
        "MIRACLReranking"
      ],
      "contamination_known": true,
      "hf_id": "Querit/Querit-4B",
      "license": "Apache-2.0 (model metadata only)",
      "params_m": 4000,
      "rated": true,
      "revision": "e557bb5ce6555f16b33dbb7d2ca62e8d27c14c13",
      "validation": {
        "deciding_reference": "transformers457",
        "mean_spearman": 1.0,
        "min_spearman": 0.9999999999999999,
        "n_queries": 20,
        "passed": true,
        "reason": null,
        "run_id": "validation-querit-4b-20261010T035119Z-0e6de8",
        "status": "ok",
        "threshold": 0.99
      }
    },
    "qwen3-0.6b": {
      "architecture": "llm",
      "commercial_use": "yes",
      "contamination": [
        "MIRACLReranking"
      ],
      "contamination_known": true,
      "hf_id": "Qwen/Qwen3-Reranker-0.6B",
      "license": "Apache-2.0",
      "params_m": 600,
      "rated": true,
      "revision": "e61197ed45024b0ed8a2d74b80b4d909f1255473",
      "validation": {
        "deciding_reference": "model_card",
        "mean_spearman": 0.9999303916435401,
        "min_spearman": 0.999741967270338,
        "n_queries": 20,
        "passed": true,
        "reason": null,
        "run_id": "validation-qwen3-0.6b-20261010T033614Z-8c7b96",
        "status": "ok",
        "threshold": 0.99
      }
    },
    "qwen3-4b": {
      "architecture": "llm",
      "commercial_use": "yes",
      "contamination": [
        "MIRACLReranking"
      ],
      "contamination_known": true,
      "hf_id": "Qwen/Qwen3-Reranker-4B",
      "license": "Apache-2.0",
      "params_m": 4000,
      "rated": true,
      "revision": "22e683669bc0f0bd69640a1354a6d0aebcfeede5",
      "validation": {
        "deciding_reference": "model_card",
        "mean_spearman": 0.9999260397248848,
        "min_spearman": 0.9995799304304103,
        "n_queries": 20,
        "passed": true,
        "reason": null,
        "run_id": "validation-qwen3-4b-20261010T034214Z-70f71a",
        "status": "ok",
        "threshold": 0.99
      }
    },
    "voyage-rerank-3": {
      "architecture": "api",
      "commercial_use": "yes (paid API)",
      "contamination": [],
      "contamination_known": false,
      "hf_id": "",
      "license": "Voyage AI terms of service",
      "params_m": 0,
      "rated": true,
      "revision": "2026-10-10",
      "validation": null
    },
    "voyage-rerank-3-lite": {
      "architecture": "api",
      "commercial_use": "yes (paid API)",
      "contamination": [],
      "contamination_known": false,
      "hf_id": "",
      "license": "Voyage AI terms of service",
      "params_m": 0,
      "rated": true,
      "revision": "2026-10-10",
      "validation": null
    }
  },
  "pitfall": {
    "checks": {
      "exploratory": {
        "vs_keyword-71k": [
          "keyword-71k-ltr"
        ]
      },
      "holm_families": {
        "vs_correct": [
          "no-instruction",
          "right-padding",
          "max256-correct",
          "max256-tail-cut"
        ],
        "vs_keyword-plain-71k": [
          "keyword-71k"
        ]
      },
      "manual_correct_vs_sentence_transformers_points": 0.05088816735001056,
      "manual_correct_vs_sentence_transformers_tolerance_points": 0.2,
      "sentence_transformers_result": "quality-hybrid-71k--qwen3-0.6b-20261010T140052Z-630b08"
    },
    "created_utc": "2026-10-10T16:21:29+00:00",
    "metrics": {
      "qwen3": {
        "correct": {
          "ndcg@10": {
            "ci_high": 0.40752976091595533,
            "ci_low": 0.3219647843769938,
            "mean": 0.3651793320649692
          }
        },
        "max256-correct": {
          "delta_vs_correct_points": -0.678963179482378,
          "ndcg@10": {
            "ci_high": 0.39976499294660767,
            "ci_low": 0.3162039922945238,
            "mean": 0.35838970027014544
          },
          "p_vs_correct_holm": 0.04279572042795721,
          "wording_vs_correct": "measurably worse"
        },
        "max256-tail-cut": {
          "delta_vs_correct_points": -16.029970456824746,
          "ndcg@10": {
            "ci_high": 0.2390141440629392,
            "ci_low": 0.1716763257799117,
            "mean": 0.20487962749672176
          },
          "p_vs_correct_holm": 0.00039996000399960006,
          "wording_vs_correct": "clearly worse"
        },
        "no-instruction": {
          "delta_vs_correct_points": 1.0376198573177065,
          "ndcg@10": {
            "ci_high": 0.4182490407665978,
            "ci_low": 0.33147474003716565,
            "mean": 0.3755555306381463
          },
          "p_vs_correct_holm": 0.04279572042795721,
          "wording_vs_correct": "measurably better"
        },
        "right-padding": {
          "delta_vs_correct_points": -22.946530706613004,
          "ndcg@10": {
            "ci_high": 0.16331167507635358,
            "ci_low": 0.10991854835400777,
            "mean": 0.13571402499883917
          },
          "p_vs_correct_holm": 0.00039996000399960006,
          "wording_vs_correct": "clearly worse"
        }
      },
      "typesense": {
        "delta_points": 6.216380471544366,
        "delta_vs_keyword-71k_points": 4.817282808766523,
        "exploratory": true,
        "keyword-71k": {
          "ci_high": 0.09605649644661068,
          "ci_low": 0.05430560617418975,
          "mean": 0.07428238494756044
        },
        "keyword-71k-ltr": {
          "ci_high": 0.14879409253958079,
          "ci_low": 0.09836524572260068,
          "mean": 0.12245521303522566
        },
        "keyword-plain-71k": {
          "ci_high": 0.020784422170357714,
          "ci_low": 0.004857981133367159,
          "mean": 0.012118580232116774
        },
        "p": 9.999000099990002e-05,
        "p_vs_keyword-71k": 9.999000099990002e-05
      }
    },
    "model": "qwen3-0.6b",
    "run_id": "pitfall-qwen3-0.6b-and-typesense-20261010T162129Z-e19e60",
    "stage": "hybrid-71k"
  },
  "plan": {
    "commit": "dec01291a2ce67e259385f3a69145b6b5f041f50",
    "timestamp_file": "PLAN.md.ots"
  },
  "quality": {
    "hybrid-71k": {
      "best": "qwen3-4b",
      "created_utc": "2026-10-12T08:05:07+00:00",
      "models": {
        "bge-v2-m3": {
          "comparison": {
            "contamination": [
              "MIRACLReranking",
              "MIRACLRetrieval",
              "MIRACLRetrievalHardNegatives"
            ],
            "contamination_known": true,
            "delta_best_minus_model_points": 1.8893224920613128,
            "delta_vs_first_stage_points": 18.51004628990879,
            "ndcg@10": 0.36008335712152306,
            "p_vs_best_holm": 0.0221977802219778,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "measurably better",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:51+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.5344262295081967
            },
            "hit@5": {
              "ci_high": 0.5672131147540984,
              "ci_low": 0.4557377049180328,
              "mean": 0.5114754098360655
            },
            "judged@10": {
              "mean": 0.2318032786885246
            },
            "mrr@10": {
              "mean": 0.4143377569607078
            },
            "ndcg@10": {
              "ci_high": 0.40337042214914215,
              "ci_low": 0.316580551630309,
              "mean": 0.36008335712152306
            },
            "ndcg_condensed@10": {
              "mean": 0.4158026723259443
            },
            "recall@10": {
              "mean": 0.4077985948477752
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.4246760343481655
            },
            "recall@5": {
              "mean": 0.36545667447306796
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.4852459016393443
            },
            "hit@5": {
              "ci_high": 0.5344262295081967,
              "ci_low": 0.42295081967213116,
              "mean": 0.4786885245901639
            },
            "judged@10": {
              "mean": 0.1714754098360656
            },
            "mrr@10": {
              "mean": 0.402080405932865
            },
            "ndcg@10": {
              "ci_high": 0.3572896929905259,
              "ci_low": 0.27448655465035726,
              "mean": 0.31570222031762757
            },
            "ndcg_condensed@10": {
              "mean": 0.4049005899749167
            },
            "recall@10": {
              "mean": 0.32546448087431695
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.32904761904761903
            },
            "recall@5": {
              "mean": 0.3121779859484778
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.521311475409836
            },
            "hit@5": {
              "ci_high": 0.5573770491803278,
              "ci_low": 0.4459016393442623,
              "mean": 0.5016393442622951
            },
            "judged@10": {
              "mean": 0.21311475409836064
            },
            "mrr@10": {
              "mean": 0.4124173822534478
            },
            "ndcg@10": {
              "ci_high": 0.3859164073897641,
              "ci_low": 0.3011918683440086,
              "mean": 0.34362146924940334
            },
            "ndcg_condensed@10": {
              "mean": 0.4133391543621477
            },
            "recall@10": {
              "mean": 0.3764090554254489
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.38811865729898515
            },
            "recall@5": {
              "mean": 0.34523809523809523
            }
          },
          "run_id": "quality-hybrid-71k--bge-v2-m3-20261010T140051Z-46dcf5"
        },
        "gte-base": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 4.072329027371851,
            "delta_vs_first_stage_points": 16.32703975459825,
            "ndcg@10": 0.3382532917684177,
            "p_vs_best_holm": 0.0007999200079992001,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "clearly better",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:51+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.5180327868852459
            },
            "hit@5": {
              "ci_high": 0.5409836065573771,
              "ci_low": 0.42950819672131146,
              "mean": 0.4852459016393443
            },
            "judged@10": {
              "mean": 0.2249180327868852
            },
            "mrr@10": {
              "mean": 0.38870543845953687
            },
            "ndcg@10": {
              "ci_high": 0.380730736331945,
              "ci_low": 0.2965280656652474,
              "mean": 0.3382532917684177
            },
            "ndcg_condensed@10": {
              "mean": 0.403892805556557
            },
            "recall@10": {
              "mean": 0.3869242779078845
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.40893052302888366
            },
            "recall@5": {
              "mean": 0.35117096018735366
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.4786885245901639
            },
            "hit@5": {
              "ci_high": 0.521311475409836,
              "ci_low": 0.4098360655737705,
              "mean": 0.46557377049180326
            },
            "judged@10": {
              "mean": 0.17049180327868851
            },
            "mrr@10": {
              "mean": 0.38132058287795995
            },
            "ndcg@10": {
              "ci_high": 0.3431720353876738,
              "ci_low": 0.2616289319406668,
              "mean": 0.30222054605140375
            },
            "ndcg_condensed@10": {
              "mean": 0.3986028532803199
            },
            "recall@10": {
              "mean": 0.3196955503512881
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.32904761904761903
            },
            "recall@5": {
              "mean": 0.3023965651834504
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.5147540983606558
            },
            "hit@5": {
              "ci_high": 0.5344262295081967,
              "ci_low": 0.42295081967213116,
              "mean": 0.4786885245901639
            },
            "judged@10": {
              "mean": 0.21114754098360658
            },
            "mrr@10": {
              "mean": 0.38910616705698675
            },
            "ndcg@10": {
              "ci_high": 0.3672679305029315,
              "ci_low": 0.28545287514836065,
              "mean": 0.32616612826363844
            },
            "ndcg_condensed@10": {
              "mean": 0.40368732179906697
            },
            "recall@10": {
              "mean": 0.3643871975019516
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.3842622950819672
            },
            "recall@5": {
              "mean": 0.3289305230288837
            }
          },
          "run_id": "quality-hybrid-71k--gte-base-20261010T140051Z-7180df"
        },
        "jina-v3.5": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 0.2373453642387724,
            "delta_vs_first_stage_points": 20.16202341773133,
            "ndcg@10": 0.37660312839974847,
            "p_vs_best_holm": 1.0,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:52+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.5409836065573771
            },
            "hit@5": {
              "ci_high": 0.5737704918032787,
              "ci_low": 0.46229508196721314,
              "mean": 0.5180327868852459
            },
            "judged@10": {
              "mean": 0.22426229508196724
            },
            "mrr@10": {
              "mean": 0.4402172781680978
            },
            "ndcg@10": {
              "ci_high": 0.42111354523236333,
              "ci_low": 0.33174636796680995,
              "mean": 0.37660312839974847
            },
            "ndcg_condensed@10": {
              "mean": 0.42716691736164847
            },
            "recall@10": {
              "mean": 0.41007025761124116
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.426448087431694
            },
            "recall@5": {
              "mean": 0.3731693989071038
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.4852459016393443
            },
            "hit@5": {
              "ci_high": 0.5344262295081967,
              "ci_low": 0.42295081967213116,
              "mean": 0.4786885245901639
            },
            "judged@10": {
              "mean": 0.17114754098360657
            },
            "mrr@10": {
              "mean": 0.4123263075722092
            },
            "ndcg@10": {
              "ci_high": 0.3653403073400263,
              "ci_low": 0.28081773832357393,
              "mean": 0.32304259505980254
            },
            "ndcg_condensed@10": {
              "mean": 0.4142163186886747
            },
            "recall@10": {
              "mean": 0.3262060889929742
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.32904761904761903
            },
            "recall@5": {
              "mean": 0.31616705698672914
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.5311475409836065
            },
            "hit@5": {
              "ci_high": 0.5672131147540984,
              "ci_low": 0.4557377049180328,
              "mean": 0.5114754098360655
            },
            "judged@10": {
              "mean": 0.21114754098360658
            },
            "mrr@10": {
              "mean": 0.42845563361956807
            },
            "ndcg@10": {
              "ci_high": 0.39856321936303185,
              "ci_low": 0.31336769276708204,
              "mean": 0.35632762512353494
            },
            "ndcg_condensed@10": {
              "mean": 0.42184864023118657
            },
            "recall@10": {
              "mean": 0.382263856362217
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.3947853239656518
            },
            "recall@5": {
              "mean": 0.35393442622950816
            }
          },
          "run_id": "quality-hybrid-71k--jina-v3.5-20261010T140052Z-7a3767"
        },
        "mxbai-large-v2": {
          "comparison": {
            "contamination": [],
            "contamination_known": true,
            "delta_best_minus_model_points": 4.810090219205715,
            "delta_vs_first_stage_points": 15.589278562764386,
            "ndcg@10": 0.33087567985007904,
            "p_vs_best_holm": 0.0007999200079992001,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "clearly better",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:52+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.5278688524590164
            },
            "hit@5": {
              "ci_high": 0.5377049180327869,
              "ci_low": 0.4262295081967213,
              "mean": 0.4819672131147541
            },
            "judged@10": {
              "mean": 0.20852459016393446
            },
            "mrr@10": {
              "mean": 0.37970465781941193
            },
            "ndcg@10": {
              "ci_high": 0.37178025675123,
              "ci_low": 0.2896679045818184,
              "mean": 0.33087567985007904
            },
            "ndcg_condensed@10": {
              "mean": 0.4180121281279578
            },
            "recall@10": {
              "mean": 0.3938095238095238
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.41797814207650275
            },
            "recall@5": {
              "mean": 0.3360421545667447
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.4852459016393443
            },
            "hit@5": {
              "ci_high": 0.521311475409836,
              "ci_low": 0.4098360655737705,
              "mean": 0.46557377049180326
            },
            "judged@10": {
              "mean": 0.16852459016393445
            },
            "mrr@10": {
              "mean": 0.3813505074160812
            },
            "ndcg@10": {
              "ci_high": 0.34029516991695286,
              "ci_low": 0.26025747582144254,
              "mean": 0.3003036590571089
            },
            "ndcg_condensed@10": {
              "mean": 0.40515290123240477
            },
            "recall@10": {
              "mean": 0.3251912568306011
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.32904761904761903
            },
            "recall@5": {
              "mean": 0.3003590944574551
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.521311475409836
            },
            "hit@5": {
              "ci_high": 0.5442622950819672,
              "ci_low": 0.43278688524590164,
              "mean": 0.4885245901639344
            },
            "judged@10": {
              "mean": 0.20032786885245904
            },
            "mrr@10": {
              "mean": 0.3821194379391101
            },
            "ndcg@10": {
              "ci_high": 0.3601186851851023,
              "ci_low": 0.2798782735736859,
              "mean": 0.3204013317761722
            },
            "ndcg_condensed@10": {
              "mean": 0.41301599836340525
            },
            "recall@10": {
              "mean": 0.37220140515222483
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.3867213114754098
            },
            "recall@5": {
              "mean": 0.33079625292740045
            }
          },
          "run_id": "quality-hybrid-71k--mxbai-large-v2-20261010T140052Z-b98d5c"
        },
        "querit-4b": {
          "comparison": {
            "contamination": [
              "MIRACLReranking"
            ],
            "contamination_known": true,
            "delta_best_minus_model_points": 1.6730573537623195,
            "delta_vs_first_stage_points": 18.726311428207783,
            "ndcg@10": 0.362246008504513,
            "p_vs_best_holm": 0.098990100989901,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:53+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.5245901639344263
            },
            "hit@5": {
              "ci_high": 0.5540983606557377,
              "ci_low": 0.4426229508196721,
              "mean": 0.49836065573770494
            },
            "judged@10": {
              "mean": 0.22131147540983606
            },
            "mrr@10": {
              "mean": 0.42163283892792086
            },
            "ndcg@10": {
              "ci_high": 0.40600109820040436,
              "ci_low": 0.3185421184349729,
              "mean": 0.362246008504513
            },
            "ndcg_condensed@10": {
              "mean": 0.4289074217349377
            },
            "recall@10": {
              "mean": 0.39870413739266203
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.4190944574551132
            },
            "recall@5": {
              "mean": 0.35725995316159254
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.4786885245901639
            },
            "hit@5": {
              "ci_high": 0.521311475409836,
              "ci_low": 0.4098360655737705,
              "mean": 0.46557377049180326
            },
            "judged@10": {
              "mean": 0.16983606557377048
            },
            "mrr@10": {
              "mean": 0.40784933645589383
            },
            "ndcg@10": {
              "ci_high": 0.35808383056892196,
              "ci_low": 0.274296837022535,
              "mean": 0.3161083815907691
            },
            "ndcg_condensed@10": {
              "mean": 0.41400721211315655
            },
            "recall@10": {
              "mean": 0.320624512099922
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.32904761904761903
            },
            "recall@5": {
              "mean": 0.3048711943793911
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.5180327868852459
            },
            "hit@5": {
              "ci_high": 0.5475409836065573,
              "ci_low": 0.4360655737704918,
              "mean": 0.4918032786885246
            },
            "judged@10": {
              "mean": 0.2081967213114754
            },
            "mrr@10": {
              "mean": 0.4183827738745772
            },
            "ndcg@10": {
              "ci_high": 0.3901997491647364,
              "ci_low": 0.3041222086047349,
              "mean": 0.3470618053247245
            },
            "ndcg_condensed@10": {
              "mean": 0.42439398088919744
            },
            "recall@10": {
              "mean": 0.3737314597970336
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.3893989071038251
            },
            "recall@5": {
              "mean": 0.33794691647150665
            }
          },
          "run_id": "quality-hybrid-71k--querit-4b-20261010T140053Z-1a1656"
        },
        "qwen3-0.6b": {
          "comparison": {
            "contamination": [
              "MIRACLReranking"
            ],
            "contamination_known": true,
            "delta_best_minus_model_points": 1.4306131650667075,
            "delta_vs_first_stage_points": 18.968755616903394,
            "ndcg@10": 0.3646704503914691,
            "p_vs_best_holm": 0.18198180181981802,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:52+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.5409836065573771
            },
            "hit@5": {
              "ci_high": 0.5639344262295082,
              "ci_low": 0.4524590163934426,
              "mean": 0.5081967213114754
            },
            "judged@10": {
              "mean": 0.23278688524590163
            },
            "mrr@10": {
              "mean": 0.42804189435336976
            },
            "ndcg@10": {
              "ci_high": 0.4069485885629406,
              "ci_low": 0.3214945232820251,
              "mean": 0.3646704503914691
            },
            "ndcg_condensed@10": {
              "mean": 0.4140002262898512
            },
            "recall@10": {
              "mean": 0.4091569086651054
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.43067915690866504
            },
            "recall@5": {
              "mean": 0.36669008587041374
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.4819672131147541
            },
            "hit@5": {
              "ci_high": 0.5311475409836065,
              "ci_low": 0.419672131147541,
              "mean": 0.47540983606557374
            },
            "judged@10": {
              "mean": 0.1737704918032787
            },
            "mrr@10": {
              "mean": 0.41082357533177205
            },
            "ndcg@10": {
              "ci_high": 0.3591312949496711,
              "ci_low": 0.27497392121112196,
              "mean": 0.3168942360370384
            },
            "ndcg_condensed@10": {
              "mean": 0.4055902213220764
            },
            "recall@10": {
              "mean": 0.32152224824355974
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.32904761904761903
            },
            "recall@5": {
              "mean": 0.31192818110850895
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.5311475409836065
            },
            "hit@5": {
              "ci_high": 0.5639344262295082,
              "ci_low": 0.4524590163934426,
              "mean": 0.5081967213114754
            },
            "judged@10": {
              "mean": 0.21672131147540985
            },
            "mrr@10": {
              "mean": 0.4275956284153005
            },
            "ndcg@10": {
              "ci_high": 0.3924800131497411,
              "ci_low": 0.3087793397717643,
              "mean": 0.35126229657161756
            },
            "ndcg_condensed@10": {
              "mean": 0.41356988927259075
            },
            "recall@10": {
              "mean": 0.3814832162373146
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.39377049180327867
            },
            "recall@5": {
              "mean": 0.35076502732240433
            }
          },
          "run_id": "quality-hybrid-71k--qwen3-0.6b-20261010T140052Z-630b08"
        },
        "qwen3-4b": {
          "comparison": {
            "contamination": [
              "MIRACLReranking"
            ],
            "contamination_known": true,
            "delta_vs_first_stage_points": 20.3993687819701,
            "ndcg@10": 0.3789765820421362,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:52+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.5377049180327869
            },
            "hit@5": {
              "ci_high": 0.5737704918032787,
              "ci_low": 0.46229508196721314,
              "mean": 0.5180327868852459
            },
            "judged@10": {
              "mean": 0.2249180327868852
            },
            "mrr@10": {
              "mean": 0.43776346604215455
            },
            "ndcg@10": {
              "ci_high": 0.42327084848945346,
              "ci_low": 0.3335835133656596,
              "mean": 0.3789765820421362
            },
            "ndcg_condensed@10": {
              "mean": 0.43030236372933583
            },
            "recall@10": {
              "mean": 0.4120296643247463
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.4311943793911008
            },
            "recall@5": {
              "mean": 0.3769086651053864
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.4852459016393443
            },
            "hit@5": {
              "ci_high": 0.5344262295081967,
              "ci_low": 0.42295081967213116,
              "mean": 0.4786885245901639
            },
            "judged@10": {
              "mean": 0.1718032786885246
            },
            "mrr@10": {
              "mean": 0.41980874316939887
            },
            "ndcg@10": {
              "ci_high": 0.36818598598312297,
              "ci_low": 0.28345387256428095,
              "mean": 0.32587402485603323
            },
            "ndcg_condensed@10": {
              "mean": 0.41296471351181036
            },
            "recall@10": {
              "mean": 0.32701795472287276
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.32904761904761903
            },
            "recall@5": {
              "mean": 0.31575331772053084
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.5311475409836065
            },
            "hit@5": {
              "ci_high": 0.5704918032786885,
              "ci_low": 0.45901639344262296,
              "mean": 0.5147540983606558
            },
            "judged@10": {
              "mean": 0.20885245901639346
            },
            "mrr@10": {
              "mean": 0.4353135571168357
            },
            "ndcg@10": {
              "ci_high": 0.40605130097266307,
              "ci_low": 0.3181130532518667,
              "mean": 0.36252903677420156
            },
            "ndcg_condensed@10": {
              "mean": 0.4261417451192491
            },
            "recall@10": {
              "mean": 0.385480093676815
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.39439500390320065
            },
            "recall@5": {
              "mean": 0.35669008587041373
            }
          },
          "run_id": "quality-hybrid-71k--qwen3-4b-20261010T140052Z-ea48ec"
        },
        "voyage-rerank-3": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 0.02168988170075381,
            "delta_vs_first_stage_points": 20.37767890026935,
            "ndcg@10": 0.37875968322512865,
            "p_vs_best_holm": 1.0,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:53+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.5344262295081967
            },
            "hit@5": {
              "ci_high": 0.5639344262295082,
              "ci_low": 0.4524590163934426,
              "mean": 0.5081967213114754
            },
            "judged@10": {
              "mean": 0.22098360655737706
            },
            "mrr@10": {
              "mean": 0.4371935987509758
            },
            "ndcg@10": {
              "ci_high": 0.42435078803782683,
              "ci_low": 0.3338204410058941,
              "mean": 0.37875968322512865
            },
            "ndcg_condensed@10": {
              "mean": 0.43627723931454465
            },
            "recall@10": {
              "mean": 0.41048399687743947
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.4251912568306011
            },
            "recall@5": {
              "mean": 0.371928181108509
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.4852459016393443
            },
            "hit@5": {
              "ci_high": 0.5344262295081967,
              "ci_low": 0.42295081967213116,
              "mean": 0.4786885245901639
            },
            "judged@10": {
              "mean": 0.16950819672131148
            },
            "mrr@10": {
              "mean": 0.41112412177985946
            },
            "ndcg@10": {
              "ci_high": 0.36532199743824795,
              "ci_low": 0.2797767348359059,
              "mean": 0.32233633333764605
            },
            "ndcg_condensed@10": {
              "mean": 0.4167127961661786
            },
            "recall@10": {
              "mean": 0.3253005464480874
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.32904761904761903
            },
            "recall@5": {
              "mean": 0.3159953161592506
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.5278688524590164
            },
            "hit@5": {
              "ci_high": 0.5540983606557377,
              "ci_low": 0.4426229508196721,
              "mean": 0.49836065573770494
            },
            "judged@10": {
              "mean": 0.20622950819672134
            },
            "mrr@10": {
              "mean": 0.42998959146500126
            },
            "ndcg@10": {
              "ci_high": 0.40108130856867885,
              "ci_low": 0.3139666211240512,
              "mean": 0.357706963347204
            },
            "ndcg_condensed@10": {
              "mean": 0.429391218382141
            },
            "recall@10": {
              "mean": 0.37971896955503515
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.3910694769711163
            },
            "recall@5": {
              "mean": 0.34988290398126465
            }
          },
          "run_id": "quality-hybrid-71k--voyage-rerank-3-20261010T140053Z-7dfa4f"
        },
        "voyage-rerank-3-lite": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 1.259636873862946,
            "delta_vs_first_stage_points": 19.139731908107155,
            "ndcg@10": 0.36638021330350673,
            "p_vs_best_holm": 0.3020697930206979,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:53+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.5344262295081967
            },
            "hit@5": {
              "ci_high": 0.5573770491803278,
              "ci_low": 0.4459016393442623,
              "mean": 0.5016393442622951
            },
            "judged@10": {
              "mean": 0.219016393442623
            },
            "mrr@10": {
              "mean": 0.42319021597710127
            },
            "ndcg@10": {
              "ci_high": 0.410862503601567,
              "ci_low": 0.3224374035525684,
              "mean": 0.36638021330350673
            },
            "ndcg_condensed@10": {
              "mean": 0.4270414960062045
            },
            "recall@10": {
              "mean": 0.40626854020296643
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.42535519125683063
            },
            "recall@5": {
              "mean": 0.3614988290398127
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.4852459016393443
            },
            "hit@5": {
              "ci_high": 0.5278688524590164,
              "ci_low": 0.4163934426229508,
              "mean": 0.4721311475409836
            },
            "judged@10": {
              "mean": 0.16852459016393442
            },
            "mrr@10": {
              "mean": 0.40027712724434034
            },
            "ndcg@10": {
              "ci_high": 0.35668888981512986,
              "ci_low": 0.27314854799720745,
              "mean": 0.3145864667230756
            },
            "ndcg_condensed@10": {
              "mean": 0.41226837421356566
            },
            "recall@10": {
              "mean": 0.32409836065573766
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.32904761904761903
            },
            "recall@5": {
              "mean": 0.3097423887587822
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.5278688524590164
            },
            "hit@5": {
              "ci_high": 0.5475409836065573,
              "ci_low": 0.4360655737704918,
              "mean": 0.4918032786885246
            },
            "judged@10": {
              "mean": 0.20754098360655737
            },
            "mrr@10": {
              "mean": 0.4198842050481394
            },
            "ndcg@10": {
              "ci_high": 0.39251805545155116,
              "ci_low": 0.3065070697208379,
              "mean": 0.3493652399524116
            },
            "ndcg_condensed@10": {
              "mean": 0.4215704197265436
            },
            "recall@10": {
              "mean": 0.3781576893052303
            },
            "recall@100": {
              "mean": 0.4350039032006245
            },
            "recall@20": {
              "mean": 0.39030444964871197
            },
            "recall@5": {
              "mean": 0.34128024980483995
            }
          },
          "run_id": "quality-hybrid-71k--voyage-rerank-3-lite-20261010T140053Z-50bf1f"
        }
      },
      "not_measured": {},
      "run_id": "quality-summary-hybrid-71k-20261012T080507Z-95eada",
      "winner": null
    },
    "keyword-71k": {
      "best": "qwen3-4b",
      "created_utc": "2026-10-12T08:05:07+00:00",
      "models": {
        "bge-v2-m3": {
          "comparison": {
            "contamination": [
              "MIRACLReranking",
              "MIRACLRetrieval",
              "MIRACLRetrievalHardNegatives"
            ],
            "contamination_known": true,
            "delta_best_minus_model_points": 1.1284801096421548,
            "delta_vs_first_stage_points": 13.368436930619065,
            "ndcg@10": 0.20796675425375108,
            "p_vs_best_holm": 0.0917908209179082,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:47+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.34098360655737703
            },
            "hit@5": {
              "ci_high": 0.3836065573770492,
              "ci_low": 0.2786885245901639,
              "mean": 0.33114754098360655
            },
            "judged@10": {
              "mean": 0.15180327868852458
            },
            "mrr@10": {
              "mean": 0.27283112151964606
            },
            "ndcg@10": {
              "ci_high": 0.2458495512308586,
              "ci_low": 0.17092672191628222,
              "mean": 0.20796675425375108
            },
            "ndcg_condensed@10": {
              "mean": 0.23699463564986453
            },
            "recall@10": {
              "mean": 0.21669008587041372
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.2292661982825917
            },
            "recall@5": {
              "mean": 0.20192818110850896
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.25245901639344265
            },
            "hit@5": {
              "ci_high": 0.29508196721311475,
              "ci_low": 0.19672131147540983,
              "mean": 0.2459016393442623
            },
            "judged@10": {
              "mean": 0.0977049180327869
            },
            "mrr@10": {
              "mean": 0.2098360655737705
            },
            "ndcg@10": {
              "ci_high": 0.17978968256960412,
              "ci_low": 0.11630895432720136,
              "mean": 0.14698249246964376
            },
            "ndcg_condensed@10": {
              "mean": 0.2217611735343921
            },
            "recall@10": {
              "mean": 0.14241217798594846
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.14451990632318498
            },
            "recall@5": {
              "mean": 0.1389383294301327
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.28852459016393445
            },
            "hit@5": {
              "ci_high": 0.32786885245901637,
              "ci_low": 0.2262295081967213,
              "mean": 0.2754098360655738
            },
            "judged@10": {
              "mean": 0.12590163934426232
            },
            "mrr@10": {
              "mean": 0.22826437678896697
            },
            "ndcg@10": {
              "ci_high": 0.20525158862484802,
              "ci_low": 0.13605064565715067,
              "mean": 0.1700005959710838
            },
            "ndcg_condensed@10": {
              "mean": 0.22726357369241995
            },
            "recall@10": {
              "mean": 0.17491022638563622
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.186775956284153
            },
            "recall@5": {
              "mean": 0.1624199843871975
            }
          },
          "run_id": "quality-keyword-71k--bge-v2-m3-20261010T140047Z-7230a0"
        },
        "gte-base": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 1.6916138653092134,
            "delta_vs_first_stage_points": 12.805303174952003,
            "ndcg@10": 0.2023354166970805,
            "p_vs_best_holm": 0.0811918808119188,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:47+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.34098360655737703
            },
            "hit@5": {
              "ci_high": 0.36721311475409835,
              "ci_low": 0.26229508196721313,
              "mean": 0.31475409836065577
            },
            "judged@10": {
              "mean": 0.14688524590163934
            },
            "mrr@10": {
              "mean": 0.27446526151444184
            },
            "ndcg@10": {
              "ci_high": 0.2396010385199934,
              "ci_low": 0.1660895361060079,
              "mean": 0.2023354166970805
            },
            "ndcg_condensed@10": {
              "mean": 0.23124379564285252
            },
            "recall@10": {
              "mean": 0.20865729898516786
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.22311475409836065
            },
            "recall@5": {
              "mean": 0.18635441061670568
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.24918032786885247
            },
            "hit@5": {
              "ci_high": 0.28852459016393445,
              "ci_low": 0.19344262295081968,
              "mean": 0.23934426229508196
            },
            "judged@10": {
              "mean": 0.09868852459016395
            },
            "mrr@10": {
              "mean": 0.2094574551131928
            },
            "ndcg@10": {
              "ci_high": 0.17540194525595906,
              "ci_low": 0.1136425527212194,
              "mean": 0.14360815643714964
            },
            "ndcg_condensed@10": {
              "mean": 0.2186462434810603
            },
            "recall@10": {
              "mean": 0.13975800156128024
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.14451990632318498
            },
            "recall@5": {
              "mean": 0.1314519906323185
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.28524590163934427
            },
            "hit@5": {
              "ci_high": 0.31475409836065577,
              "ci_low": 0.21639344262295082,
              "mean": 0.26557377049180325
            },
            "judged@10": {
              "mean": 0.12852459016393444
            },
            "mrr@10": {
              "mean": 0.23554514702055687
            },
            "ndcg@10": {
              "ci_high": 0.20294926428747556,
              "ci_low": 0.1349914195441349,
              "mean": 0.1680840305728503
            },
            "ndcg_condensed@10": {
              "mean": 0.22392731101674432
            },
            "recall@10": {
              "mean": 0.16950039032006242
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.18065573770491805
            },
            "recall@5": {
              "mean": 0.15504293520686963
            }
          },
          "run_id": "quality-keyword-71k--gte-base-20261010T140047Z-9a9cec"
        },
        "jina-v3.5": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 0.31027737584549964,
            "delta_vs_first_stage_points": 14.186639664415718,
            "ndcg@10": 0.21614878159171763,
            "p_vs_best_holm": 0.7073292670732927,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:47+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.3475409836065574
            },
            "hit@5": {
              "ci_high": 0.39672131147540984,
              "ci_low": 0.29180327868852457,
              "mean": 0.3442622950819672
            },
            "judged@10": {
              "mean": 0.14688524590163934
            },
            "mrr@10": {
              "mean": 0.28298724954462656
            },
            "ndcg@10": {
              "ci_high": 0.254476750006573,
              "ci_low": 0.1785537516126157,
              "mean": 0.21614878159171763
            },
            "ndcg_condensed@10": {
              "mean": 0.23672540897280736
            },
            "recall@10": {
              "mean": 0.22171740827478534
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.2332006245120999
            },
            "recall@5": {
              "mean": 0.21266978922716628
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.24918032786885247
            },
            "hit@5": {
              "ci_high": 0.29508196721311475,
              "ci_low": 0.19672131147540983,
              "mean": 0.2459016393442623
            },
            "judged@10": {
              "mean": 0.09836065573770492
            },
            "mrr@10": {
              "mean": 0.21677595628415303
            },
            "ndcg@10": {
              "ci_high": 0.18257909496675145,
              "ci_low": 0.11804569640219416,
              "mean": 0.14981444359463866
            },
            "ndcg_condensed@10": {
              "mean": 0.222838268801342
            },
            "recall@10": {
              "mean": 0.1402263856362217
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.14451990632318498
            },
            "recall@5": {
              "mean": 0.1367525370804059
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.29508196721311475
            },
            "hit@5": {
              "ci_high": 0.33450819672131027,
              "ci_low": 0.2360655737704918,
              "mean": 0.28524590163934427
            },
            "judged@10": {
              "mean": 0.12819672131147541
            },
            "mrr@10": {
              "mean": 0.24120218579234973
            },
            "ndcg@10": {
              "ci_high": 0.21325194649198842,
              "ci_low": 0.1432600064312427,
              "mean": 0.17766862311306839
            },
            "ndcg_condensed@10": {
              "mean": 0.2279328904592979
            },
            "recall@10": {
              "mean": 0.17729117876658856
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.18568306010928962
            },
            "recall@5": {
              "mean": 0.1697423887587822
            }
          },
          "run_id": "quality-keyword-71k--jina-v3.5-20261010T140047Z-8b7d0a"
        },
        "mxbai-large-v2": {
          "comparison": {
            "contamination": [],
            "contamination_known": true,
            "delta_best_minus_model_points": 2.675093145080634,
            "delta_vs_first_stage_points": 11.821823895180584,
            "ndcg@10": 0.19250062389936629,
            "p_vs_best_holm": 0.0007999200079992001,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "clearly better",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:48+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.3377049180327869
            },
            "hit@5": {
              "ci_high": 0.35737704918032787,
              "ci_low": 0.25245901639344265,
              "mean": 0.30491803278688523
            },
            "judged@10": {
              "mean": 0.13540983606557377
            },
            "mrr@10": {
              "mean": 0.25129586260733805
            },
            "ndcg@10": {
              "ci_high": 0.22838494964584355,
              "ci_low": 0.15797767812830196,
              "mean": 0.19250062389936629
            },
            "ndcg_condensed@10": {
              "mean": 0.23748583993581746
            },
            "recall@10": {
              "mean": 0.21056986729117877
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.22557377049180327
            },
            "recall@5": {
              "mean": 0.18309133489461357
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.25245901639344265
            },
            "hit@5": {
              "ci_high": 0.28524590163934427,
              "ci_low": 0.1901639344262295,
              "mean": 0.2360655737704918
            },
            "judged@10": {
              "mean": 0.09770491803278689
            },
            "mrr@10": {
              "mean": 0.1976541764246682
            },
            "ndcg@10": {
              "ci_high": 0.17065091365999221,
              "ci_low": 0.10982308331723394,
              "mean": 0.13945017873057144
            },
            "ndcg_condensed@10": {
              "mean": 0.21936887982533065
            },
            "recall@10": {
              "mean": 0.14233411397345822
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.14451990632318498
            },
            "recall@5": {
              "mean": 0.13234192037470727
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.29180327868852457
            },
            "hit@5": {
              "ci_high": 0.3180327868852459,
              "ci_low": 0.21967213114754097,
              "mean": 0.26885245901639343
            },
            "judged@10": {
              "mean": 0.11967213114754098
            },
            "mrr@10": {
              "mean": 0.22091465001301067
            },
            "ndcg@10": {
              "ci_high": 0.19661370910885906,
              "ci_low": 0.13021071513619178,
              "mean": 0.1627127208847503
            },
            "ndcg_condensed@10": {
              "mean": 0.22615493711535406
            },
            "recall@10": {
              "mean": 0.1737626854020297
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.18393442622950817
            },
            "recall@5": {
              "mean": 0.1549726775956284
            }
          },
          "run_id": "quality-keyword-71k--mxbai-large-v2-20261010T140048Z-7e33c3"
        },
        "querit-4b": {
          "comparison": {
            "contamination": [
              "MIRACLReranking"
            ],
            "contamination_known": true,
            "delta_best_minus_model_points": 0.9983119466368956,
            "delta_vs_first_stage_points": 13.498605093624322,
            "ndcg@10": 0.20926843588380367,
            "p_vs_best_holm": 0.27917208279172084,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:48+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.3344262295081967
            },
            "hit@5": {
              "ci_high": 0.3704918032786885,
              "ci_low": 0.26557377049180325,
              "mean": 0.3180327868852459
            },
            "judged@10": {
              "mean": 0.14557377049180328
            },
            "mrr@10": {
              "mean": 0.2747905282331512
            },
            "ndcg@10": {
              "ci_high": 0.24805679216996104,
              "ci_low": 0.1715720843659151,
              "mean": 0.20926843588380367
            },
            "ndcg_condensed@10": {
              "mean": 0.24143574352371056
            },
            "recall@10": {
              "mean": 0.21141295862607334
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.23156128024980482
            },
            "recall@5": {
              "mean": 0.19563622170179545
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.24918032786885247
            },
            "hit@5": {
              "ci_high": 0.28524590163934427,
              "ci_low": 0.1901639344262295,
              "mean": 0.2360655737704918
            },
            "judged@10": {
              "mean": 0.09770491803278689
            },
            "mrr@10": {
              "mean": 0.20995706479313037
            },
            "ndcg@10": {
              "ci_high": 0.17939743494661564,
              "ci_low": 0.1150408254144826,
              "mean": 0.14611059614274993
            },
            "ndcg_condensed@10": {
              "mean": 0.2239232893286737
            },
            "recall@10": {
              "mean": 0.13866510538641685
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.14451990632318498
            },
            "recall@5": {
              "mean": 0.13362997658079626
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.29180327868852457
            },
            "hit@5": {
              "ci_high": 0.3180327868852459,
              "ci_low": 0.21967213114754097,
              "mean": 0.26885245901639343
            },
            "judged@10": {
              "mean": 0.12754098360655738
            },
            "mrr@10": {
              "mean": 0.23511319281811083
            },
            "ndcg@10": {
              "ci_high": 0.20971222961028466,
              "ci_low": 0.13859832805919556,
              "mean": 0.17327323069484168
            },
            "ndcg_condensed@10": {
              "mean": 0.230950579014053
            },
            "recall@10": {
              "mean": 0.17389539422326308
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.18568306010928962
            },
            "recall@5": {
              "mean": 0.15877439500390317
            }
          },
          "run_id": "quality-keyword-71k--querit-4b-20261010T140048Z-13edb8"
        },
        "qwen3-0.6b": {
          "comparison": {
            "contamination": [
              "MIRACLReranking"
            ],
            "contamination_known": true,
            "delta_best_minus_model_points": 0.7677551709576985,
            "delta_vs_first_stage_points": 13.72916186930352,
            "ndcg@10": 0.21157400364059564,
            "p_vs_best_holm": 0.3533646635336466,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:48+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.3540983606557377
            },
            "hit@5": {
              "ci_high": 0.3836065573770492,
              "ci_low": 0.2786885245901639,
              "mean": 0.33114754098360655
            },
            "judged@10": {
              "mean": 0.15245901639344261
            },
            "mrr@10": {
              "mean": 0.28004033307311993
            },
            "ndcg@10": {
              "ci_high": 0.24927066607275247,
              "ci_low": 0.17481064938739987,
              "mean": 0.21157400364059564
            },
            "ndcg_condensed@10": {
              "mean": 0.2367359706431128
            },
            "recall@10": {
              "mean": 0.22225604996096804
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.23426229508196716
            },
            "recall@5": {
              "mean": 0.199032006245121
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.25245901639344265
            },
            "hit@5": {
              "ci_high": 0.29180327868852457,
              "ci_low": 0.19672131147540983,
              "mean": 0.24262295081967214
            },
            "judged@10": {
              "mean": 0.1
            },
            "mrr@10": {
              "mean": 0.21468904501691388
            },
            "ndcg@10": {
              "ci_high": 0.17812701867460315,
              "ci_low": 0.11596023597707562,
              "mean": 0.14653768811764603
            },
            "ndcg_condensed@10": {
              "mean": 0.22188490540484399
            },
            "recall@10": {
              "mean": 0.14085089773614362
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.14451990632318498
            },
            "recall@5": {
              "mean": 0.133551912568306
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.2983606557377049
            },
            "hit@5": {
              "ci_high": 0.3344262295081967,
              "ci_low": 0.2360655737704918,
              "mean": 0.28524590163934427
            },
            "judged@10": {
              "mean": 0.13245901639344265
            },
            "mrr@10": {
              "mean": 0.24639734582357528
            },
            "ndcg@10": {
              "ci_high": 0.2141539096579519,
              "ci_low": 0.1443127086966573,
              "mean": 0.17863489825943782
            },
            "ndcg_condensed@10": {
              "mean": 0.2302806839728451
            },
            "recall@10": {
              "mean": 0.17935987509758
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.18412177985948477
            },
            "recall@5": {
              "mean": 0.16943793911007027
            }
          },
          "run_id": "quality-keyword-71k--qwen3-0.6b-20261010T140048Z-2f5584"
        },
        "qwen3-4b": {
          "comparison": {
            "contamination": [
              "MIRACLReranking"
            ],
            "contamination_known": true,
            "delta_vs_first_stage_points": 14.49691704026122,
            "ndcg@10": 0.21925155535017263,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:48+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.35081967213114756
            },
            "hit@5": {
              "ci_high": 0.3901639344262295,
              "ci_low": 0.2819672131147541,
              "mean": 0.3344262295081967
            },
            "judged@10": {
              "mean": 0.14885245901639346
            },
            "mrr@10": {
              "mean": 0.2826541764246682
            },
            "ndcg@10": {
              "ci_high": 0.2586610459958759,
              "ci_low": 0.18086466923470787,
              "mean": 0.21925155535017263
            },
            "ndcg_condensed@10": {
              "mean": 0.24553639463294044
            },
            "recall@10": {
              "mean": 0.22597189695550354
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.23554254488680715
            },
            "recall@5": {
              "mean": 0.21034348165495703
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.25245901639344265
            },
            "hit@5": {
              "ci_high": 0.29180327868852457,
              "ci_low": 0.19672131147540983,
              "mean": 0.24262295081967214
            },
            "judged@10": {
              "mean": 0.09770491803278689
            },
            "mrr@10": {
              "mean": 0.21347775175644026
            },
            "ndcg@10": {
              "ci_high": 0.18159452510199617,
              "ci_low": 0.11746152851103583,
              "mean": 0.1487511289661505
            },
            "ndcg_condensed@10": {
              "mean": 0.22451231096206362
            },
            "recall@10": {
              "mean": 0.14249024199843868
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.14451990632318498
            },
            "recall@5": {
              "mean": 0.1370257611241218
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.2983606557377049
            },
            "hit@5": {
              "ci_high": 0.3442622950819672,
              "ci_low": 0.23934426229508196,
              "mean": 0.29180327868852457
            },
            "judged@10": {
              "mean": 0.12721311475409836
            },
            "mrr@10": {
              "mean": 0.24188914910226386
            },
            "ndcg@10": {
              "ci_high": 0.21466708149819036,
              "ci_low": 0.14385984579358166,
              "mean": 0.17848399571639093
            },
            "ndcg_condensed@10": {
              "mean": 0.23387204206131706
            },
            "recall@10": {
              "mean": 0.17881342701014832
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.18521467603434819
            },
            "recall@5": {
              "mean": 0.16943793911007027
            }
          },
          "run_id": "quality-keyword-71k--qwen3-4b-20261010T140048Z-5af098"
        },
        "voyage-rerank-3": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 0.4533453184885544,
            "delta_vs_first_stage_points": 14.043571721772663,
            "ndcg@10": 0.21471810216528708,
            "p_vs_best_holm": 0.7073292670732927,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:49+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.3442622950819672
            },
            "hit@5": {
              "ci_high": 0.380327868852459,
              "ci_low": 0.2754098360655738,
              "mean": 0.32786885245901637
            },
            "judged@10": {
              "mean": 0.1422950819672131
            },
            "mrr@10": {
              "mean": 0.2782825917252147
            },
            "ndcg@10": {
              "ci_high": 0.25368979341347114,
              "ci_low": 0.1766081026701181,
              "mean": 0.21471810216528708
            },
            "ndcg_condensed@10": {
              "mean": 0.24672083508171044
            },
            "recall@10": {
              "mean": 0.21933645589383294
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.22912568306010928
            },
            "recall@5": {
              "mean": 0.20348165495706477
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.25245901639344265
            },
            "hit@5": {
              "ci_high": 0.29180327868852457,
              "ci_low": 0.19672131147540983,
              "mean": 0.24262295081967214
            },
            "judged@10": {
              "mean": 0.09606557377049181
            },
            "mrr@10": {
              "mean": 0.21082357533177204
            },
            "ndcg@10": {
              "ci_high": 0.17983818480808522,
              "ci_low": 0.11658282971892497,
              "mean": 0.1475207995266199
            },
            "ndcg_condensed@10": {
              "mean": 0.2262228168274266
            },
            "recall@10": {
              "mean": 0.14350507416081187
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.14451990632318498
            },
            "recall@5": {
              "mean": 0.13554254488680717
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.29180327868852457
            },
            "hit@5": {
              "ci_high": 0.3180327868852459,
              "ci_low": 0.21967213114754097,
              "mean": 0.26885245901639343
            },
            "judged@10": {
              "mean": 0.12262295081967213
            },
            "mrr@10": {
              "mean": 0.23753447827218316
            },
            "ndcg@10": {
              "ci_high": 0.20938414845246395,
              "ci_low": 0.14000525444451697,
              "mean": 0.17383937069398447
            },
            "ndcg_condensed@10": {
              "mean": 0.23409791165454785
            },
            "recall@10": {
              "mean": 0.1756752537080406
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.18568306010928962
            },
            "recall@5": {
              "mean": 0.15858704137392662
            }
          },
          "run_id": "quality-keyword-71k--voyage-rerank-3-20261010T140049Z-e38cf5"
        },
        "voyage-rerank-3-lite": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 1.30792333106573,
            "delta_vs_first_stage_points": 13.18899370919549,
            "ndcg@10": 0.20617232203951533,
            "p_vs_best_holm": 0.1334866513348665,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:49+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.35081967213114756
            },
            "hit@5": {
              "ci_high": 0.3639344262295082,
              "ci_low": 0.25901639344262295,
              "mean": 0.3114754098360656
            },
            "judged@10": {
              "mean": 0.14327868852459016
            },
            "mrr@10": {
              "mean": 0.2681537861046058
            },
            "ndcg@10": {
              "ci_high": 0.2439439677722593,
              "ci_low": 0.170084227303951,
              "mean": 0.20617232203951533
            },
            "ndcg_condensed@10": {
              "mean": 0.2422751001303
            },
            "recall@10": {
              "mean": 0.22188134270101484
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.2297814207650273
            },
            "recall@5": {
              "mean": 0.1942232630757221
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.25245901639344265
            },
            "hit@5": {
              "ci_high": 0.29508196721311475,
              "ci_low": 0.19672131147540983,
              "mean": 0.2459016393442623
            },
            "judged@10": {
              "mean": 0.09606557377049181
            },
            "mrr@10": {
              "mean": 0.20508977361436379
            },
            "ndcg@10": {
              "ci_high": 0.17515653015317442,
              "ci_low": 0.11315110803421424,
              "mean": 0.14331223912120375
            },
            "ndcg_condensed@10": {
              "mean": 0.2234052014834114
            },
            "recall@10": {
              "mean": 0.14233411397345822
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.14451990632318498
            },
            "recall@5": {
              "mean": 0.13702576112412176
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.29508196721311475
            },
            "hit@5": {
              "ci_high": 0.3180327868852459,
              "ci_low": 0.21967213114754097,
              "mean": 0.26885245901639343
            },
            "judged@10": {
              "mean": 0.12360655737704919
            },
            "mrr@10": {
              "mean": 0.23302628155087174
            },
            "ndcg@10": {
              "ci_high": 0.2030867682857751,
              "ci_low": 0.1354718691767485,
              "mean": 0.16883034364124921
            },
            "ndcg_condensed@10": {
              "mean": 0.23095379502871838
            },
            "recall@10": {
              "mean": 0.17458235753317722
            },
            "recall@100": {
              "mean": 0.23757220921155345
            },
            "recall@20": {
              "mean": 0.18459016393442623
            },
            "recall@5": {
              "mean": 0.15730679156908664
            }
          },
          "run_id": "quality-keyword-71k--voyage-rerank-3-lite-20261010T140049Z-b3ebcd"
        }
      },
      "not_measured": {},
      "run_id": "quality-summary-keyword-71k-20261012T080507Z-ab7e07",
      "winner": null
    },
    "keyword-full": {
      "best": "qwen3-4b",
      "created_utc": "2026-10-12T08:05:08+00:00",
      "models": {
        "bge-v2-m3": {
          "comparison": {
            "contamination": [
              "MIRACLReranking",
              "MIRACLRetrieval",
              "MIRACLRetrievalHardNegatives"
            ],
            "contamination_known": true,
            "delta_best_minus_model_points": 0.7959718270794308,
            "delta_vs_first_stage_points": 9.178887710433736,
            "ndcg@10": 0.15324959976082256,
            "p_vs_best_holm": 0.132986701329867,
            "p_vs_first_stage_holm": 0.0007999200079992001,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:54+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.2459016393442623
            },
            "hit@5": {
              "ci_high": 0.28524590163934427,
              "ci_low": 0.1901639344262295,
              "mean": 0.2360655737704918
            },
            "judged@10": {
              "mean": 0.10524590163934426
            },
            "mrr@10": {
              "mean": 0.20323315118397087
            },
            "ndcg@10": {
              "ci_high": 0.1876121372889568,
              "ci_low": 0.12067956733669885,
              "mean": 0.15324959976082256
            },
            "ndcg_condensed@10": {
              "mean": 0.16983045745168313
            },
            "recall@10": {
              "mean": 0.15714285714285714
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.16294301327088212
            },
            "recall@5": {
              "mean": 0.14453551912568305
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.18032786885245902
            },
            "hit@5": {
              "ci_high": 0.22295081967213115,
              "ci_low": 0.13442622950819672,
              "mean": 0.17704918032786884
            },
            "judged@10": {
              "mean": 0.06557377049180328
            },
            "mrr@10": {
              "mean": 0.14986338797814205
            },
            "ndcg@10": {
              "ci_high": 0.13114649268286377,
              "ci_low": 0.07635708004719584,
              "mean": 0.10340429113425216
            },
            "ndcg_condensed@10": {
              "mean": 0.15365514259587465
            },
            "recall@10": {
              "mean": 0.10030444964871194
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.10241217798594845
            },
            "recall@5": {
              "mean": 0.09765027322404371
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.21639344262295082
            },
            "hit@5": {
              "ci_high": 0.25245901639344265,
              "ci_low": 0.16393442622950818,
              "mean": 0.20655737704918034
            },
            "judged@10": {
              "mean": 0.08655737704918032
            },
            "mrr@10": {
              "mean": 0.1762607338017174
            },
            "ndcg@10": {
              "ci_high": 0.1590677230404351,
              "ci_low": 0.09798472460511233,
              "mean": 0.12769846888253736
            },
            "ndcg_condensed@10": {
              "mean": 0.16009584888012687
            },
            "recall@10": {
              "mean": 0.12743950039032007
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.1322014051522248
            },
            "recall@5": {
              "mean": 0.11882903981264636
            }
          },
          "run_id": "quality-keyword-full--bge-v2-m3-20261010T140054Z-eae4a7"
        },
        "gte-base": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 1.4127863554521614,
            "delta_vs_first_stage_points": 8.562073182061006,
            "ndcg@10": 0.14708145447709525,
            "p_vs_best_holm": 0.047995200479952004,
            "p_vs_first_stage_holm": 0.0007999200079992001,
            "wording_best_vs_model": "measurably better",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:53+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.24262295081967214
            },
            "hit@5": {
              "ci_high": 0.2721311475409836,
              "ci_low": 0.17704918032786884,
              "mean": 0.22295081967213115
            },
            "judged@10": {
              "mean": 0.1
            },
            "mrr@10": {
              "mean": 0.20153395784543326
            },
            "ndcg@10": {
              "ci_high": 0.1805240911286306,
              "ci_low": 0.11544359393597362,
              "mean": 0.14708145447709525
            },
            "ndcg_condensed@10": {
              "mean": 0.16577939928802604
            },
            "recall@10": {
              "mean": 0.14763466042154566
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.15624512099921933
            },
            "recall@5": {
              "mean": 0.13473067915690867
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.18032786885245902
            },
            "hit@5": {
              "ci_high": 0.2098360655737705,
              "ci_low": 0.12786885245901639,
              "mean": 0.16721311475409836
            },
            "judged@10": {
              "mean": 0.06688524590163934
            },
            "mrr@10": {
              "mean": 0.14694379391100704
            },
            "ndcg@10": {
              "ci_high": 0.1284729899929193,
              "ci_low": 0.07489654927780182,
              "mean": 0.10112476926716563
            },
            "ndcg_condensed@10": {
              "mean": 0.15197193221705046
            },
            "recall@10": {
              "mean": 0.10092896174863387
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.10241217798594845
            },
            "recall@5": {
              "mean": 0.08989071038251366
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.20655737704918034
            },
            "hit@5": {
              "ci_high": 0.24262295081967214,
              "ci_low": 0.1540983606557377,
              "mean": 0.19672131147540983
            },
            "judged@10": {
              "mean": 0.08557377049180329
            },
            "mrr@10": {
              "mean": 0.17743950039032005
            },
            "ndcg@10": {
              "ci_high": 0.154536258238579,
              "ci_low": 0.09520342568812486,
              "mean": 0.12395809742353789
            },
            "ndcg_condensed@10": {
              "mean": 0.1577460976402245
            },
            "recall@10": {
              "mean": 0.12104605776736924
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.13110850897736145
            },
            "recall@5": {
              "mean": 0.11309133489461358
            }
          },
          "run_id": "quality-keyword-full--gte-base-20261010T140053Z-580b0f"
        },
        "jina-v3.5": {
          "comparison": null,
          "created_utc": "2026-10-10T14:00:54+00:00",
          "depth@20": {
            "hit@10": {
              "mean": 0.18032786885245902
            },
            "hit@5": {
              "ci_high": 0.22295081967213115,
              "ci_low": 0.13442622950819672,
              "mean": 0.17704918032786884
            },
            "judged@10": {
              "mean": 0.06655737704918033
            },
            "mrr@10": {
              "mean": 0.15699453551912568
            },
            "ndcg@10": {
              "ci_high": 0.13585047440745623,
              "ci_low": 0.07928442451759551,
              "mean": 0.10695834023282147
            },
            "ndcg_condensed@10": {
              "mean": 0.15462917727046047
            },
            "recall@10": {
              "mean": 0.09921155347384854
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.10241217798594845
            },
            "recall@5": {
              "mean": 0.09546448087431694
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.21639344262295082
            },
            "hit@5": {
              "ci_high": 0.25573770491803277,
              "ci_low": 0.16393442622950818,
              "mean": 0.2098360655737705
            },
            "judged@10": {
              "mean": 0.08262295081967214
            },
            "mrr@10": {
              "mean": 0.18858834244080147
            },
            "ndcg@10": {
              "ci_high": 0.16615712755776582,
              "ci_low": 0.10301505578552911,
              "mean": 0.1336880605119036
            },
            "ndcg_condensed@10": {
              "mean": 0.1624169357838719
            },
            "recall@10": {
              "mean": 0.12544886807181888
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.1322014051522248
            },
            "recall@5": {
              "mean": 0.1205776736924278
            }
          },
          "run_id": "quality-keyword-full--jina-v3.5-20261010T140054Z-59ad2e"
        },
        "mxbai-large-v2": {
          "comparison": {
            "contamination": [],
            "contamination_known": true,
            "delta_best_minus_model_points": 1.3201322306043957,
            "delta_vs_first_stage_points": 8.654727306908772,
            "ndcg@10": 0.1480079957255729,
            "p_vs_best_holm": 0.0027997200279972004,
            "p_vs_first_stage_holm": 0.0007999200079992001,
            "wording_best_vs_model": "measurably better",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:54+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.24918032786885247
            },
            "hit@5": {
              "ci_high": 0.2819672131147541,
              "ci_low": 0.18688524590163935,
              "mean": 0.23278688524590163
            },
            "judged@10": {
              "mean": 0.09573770491803278
            },
            "mrr@10": {
              "mean": 0.19702966432474628
            },
            "ndcg@10": {
              "ci_high": 0.18140476168809122,
              "ci_low": 0.11627323536666193,
              "mean": 0.1480079957255729
            },
            "ndcg_condensed@10": {
              "mean": 0.16925034556148405
            },
            "recall@10": {
              "mean": 0.15583138173302105
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.162256049960968
            },
            "recall@5": {
              "mean": 0.14073380171740826
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.18032786885245902
            },
            "hit@5": {
              "ci_high": 0.21311475409836064,
              "ci_low": 0.12786885245901639,
              "mean": 0.17049180327868851
            },
            "judged@10": {
              "mean": 0.06622950819672133
            },
            "mrr@10": {
              "mean": 0.14122040072859746
            },
            "ndcg@10": {
              "ci_high": 0.12724534712529592,
              "ci_low": 0.07372439637977753,
              "mean": 0.0999149408943261
            },
            "ndcg_condensed@10": {
              "mean": 0.1522846087782386
            },
            "recall@10": {
              "mean": 0.10022638563622169
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.10241217798594845
            },
            "recall@5": {
              "mean": 0.09433255269320842
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.21639344262295082
            },
            "hit@5": {
              "ci_high": 0.25245901639344265,
              "ci_low": 0.16065573770491803,
              "mean": 0.20655737704918034
            },
            "judged@10": {
              "mean": 0.08295081967213114
            },
            "mrr@10": {
              "mean": 0.16469945355191257
            },
            "ndcg@10": {
              "ci_high": 0.15163608388116134,
              "ci_low": 0.09277004298998619,
              "mean": 0.12145672704910448
            },
            "ndcg_condensed@10": {
              "mean": 0.1586243496643319
            },
            "recall@10": {
              "mean": 0.12678376268540203
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.13110850897736145
            },
            "recall@5": {
              "mean": 0.11797033567525368
            }
          },
          "run_id": "quality-keyword-full--mxbai-large-v2-20261010T140054Z-5ce6de"
        },
        "querit-4b": {
          "comparison": {
            "contamination": [
              "MIRACLReranking"
            ],
            "contamination_known": true,
            "delta_best_minus_model_points": 0.56837768068905,
            "delta_vs_first_stage_points": 9.406481856824117,
            "ndcg@10": 0.15552554122472637,
            "p_vs_best_holm": 0.3735626437356264,
            "p_vs_first_stage_holm": 0.0007999200079992001,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:55+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.24918032786885247
            },
            "hit@5": {
              "ci_high": 0.28852459016393445,
              "ci_low": 0.19344262295081968,
              "mean": 0.23934426229508196
            },
            "judged@10": {
              "mean": 0.10327868852459017
            },
            "mrr@10": {
              "mean": 0.20549180327868852
            },
            "ndcg@10": {
              "ci_high": 0.18988863862009214,
              "ci_low": 0.12259765387149046,
              "mean": 0.15552554122472637
            },
            "ndcg_condensed@10": {
              "mean": 0.1727162086603017
            },
            "recall@10": {
              "mean": 0.1591647150663544
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.1651288056206089
            },
            "recall@5": {
              "mean": 0.14455893832943012
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.18032786885245902
            },
            "hit@5": {
              "ci_high": 0.21639344262295082,
              "ci_low": 0.13114754098360656,
              "mean": 0.1737704918032787
            },
            "judged@10": {
              "mean": 0.06622950819672131
            },
            "mrr@10": {
              "mean": 0.15106557377049182
            },
            "ndcg@10": {
              "ci_high": 0.1334002991893428,
              "ci_low": 0.07750803025464365,
              "mean": 0.10495360038423385
            },
            "ndcg_condensed@10": {
              "mean": 0.15524471914086818
            },
            "recall@10": {
              "mean": 0.09983606557377049
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.10241217798594845
            },
            "recall@5": {
              "mean": 0.09671350507416081
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.21967213114754097
            },
            "hit@5": {
              "ci_high": 0.25245901639344265,
              "ci_low": 0.16065573770491803,
              "mean": 0.20655737704918034
            },
            "judged@10": {
              "mean": 0.0862295081967213
            },
            "mrr@10": {
              "mean": 0.17828259172521466
            },
            "ndcg@10": {
              "ci_high": 0.1629730529351945,
              "ci_low": 0.10053450406058406,
              "mean": 0.13096653968638636
            },
            "ndcg_condensed@10": {
              "mean": 0.1626340262208967
            },
            "recall@10": {
              "mean": 0.12962529274004683
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.1322014051522248
            },
            "recall@5": {
              "mean": 0.12217798594847774
            }
          },
          "run_id": "quality-keyword-full--querit-4b-20261010T140055Z-7b302f"
        },
        "qwen3-0.6b": {
          "comparison": {
            "contamination": [
              "MIRACLReranking"
            ],
            "contamination_known": true,
            "delta_best_minus_model_points": 0.6611628838818912,
            "delta_vs_first_stage_points": 9.313696653631276,
            "ndcg@10": 0.15459768919279795,
            "p_vs_best_holm": 0.33956604339566043,
            "p_vs_first_stage_holm": 0.0007999200079992001,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:54+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.24918032786885247
            },
            "hit@5": {
              "ci_high": 0.28852459016393445,
              "ci_low": 0.19344262295081968,
              "mean": 0.23934426229508196
            },
            "judged@10": {
              "mean": 0.10754098360655737
            },
            "mrr@10": {
              "mean": 0.20978922716627635
            },
            "ndcg@10": {
              "ci_high": 0.18876323179423424,
              "ci_low": 0.12173553181898808,
              "mean": 0.15459768919279795
            },
            "ndcg_condensed@10": {
              "mean": 0.16949611689005795
            },
            "recall@10": {
              "mean": 0.15530835284933645
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.16356752537080405
            },
            "recall@5": {
              "mean": 0.14483996877439498
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.18032786885245902
            },
            "hit@5": {
              "ci_high": 0.21639344262295082,
              "ci_low": 0.13114754098360656,
              "mean": 0.1737704918032787
            },
            "judged@10": {
              "mean": 0.06655737704918033
            },
            "mrr@10": {
              "mean": 0.14951209992193598
            },
            "ndcg@10": {
              "ci_high": 0.13122288275678337,
              "ci_low": 0.07610595421532484,
              "mean": 0.10291801101493535
            },
            "ndcg_condensed@10": {
              "mean": 0.15309488666667448
            },
            "recall@10": {
              "mean": 0.09928961748633879
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.10241217798594845
            },
            "recall@5": {
              "mean": 0.09472287275565963
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.21639344262295082
            },
            "hit@5": {
              "ci_high": 0.25245901639344265,
              "ci_low": 0.16065573770491803,
              "mean": 0.20655737704918034
            },
            "judged@10": {
              "mean": 0.08852459016393442
            },
            "mrr@10": {
              "mean": 0.17987639864689045
            },
            "ndcg@10": {
              "ci_high": 0.15846821078093493,
              "ci_low": 0.09759381019878155,
              "mean": 0.12723659924663488
            },
            "ndcg_condensed@10": {
              "mean": 0.15789384973529244
            },
            "recall@10": {
              "mean": 0.1258782201405152
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.12954722872755658
            },
            "recall@5": {
              "mean": 0.11846994535519126
            }
          },
          "run_id": "quality-keyword-full--qwen3-0.6b-20261010T140054Z-85e548"
        },
        "qwen3-4b": {
          "comparison": {
            "contamination": [
              "MIRACLReranking"
            ],
            "contamination_known": true,
            "delta_vs_first_stage_points": 9.974859537513167,
            "ndcg@10": 0.16120931803161687,
            "p_vs_first_stage_holm": 0.0007999200079992001,
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:55+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.24918032786885247
            },
            "hit@5": {
              "ci_high": 0.29508196721311475,
              "ci_low": 0.19991803278688533,
              "mean": 0.2459016393442623
            },
            "judged@10": {
              "mean": 0.10065573770491804
            },
            "mrr@10": {
              "mean": 0.21385636221701793
            },
            "ndcg@10": {
              "ci_high": 0.1971036885204342,
              "ci_low": 0.1272055640349873,
              "mean": 0.16120931803161687
            },
            "ndcg_condensed@10": {
              "mean": 0.1753596866728576
            },
            "recall@10": {
              "mean": 0.15760343481654954
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.16466042154566743
            },
            "recall@5": {
              "mean": 0.15325526932084307
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.18032786885245902
            },
            "hit@5": {
              "ci_high": 0.21311475409836064,
              "ci_low": 0.12786885245901639,
              "mean": 0.17049180327868851
            },
            "judged@10": {
              "mean": 0.06622950819672133
            },
            "mrr@10": {
              "mean": 0.15571818891491024
            },
            "ndcg@10": {
              "ci_high": 0.13640789412236545,
              "ci_low": 0.07952166832867467,
              "mean": 0.10747000530098098
            },
            "ndcg_condensed@10": {
              "mean": 0.15569753423392294
            },
            "recall@10": {
              "mean": 0.10038251366120218
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.10241217798594845
            },
            "recall@5": {
              "mean": 0.09491803278688524
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.21639344262295082
            },
            "hit@5": {
              "ci_high": 0.25573770491803277,
              "ci_low": 0.16393442622950818,
              "mean": 0.2098360655737705
            },
            "judged@10": {
              "mean": 0.08295081967213115
            },
            "mrr@10": {
              "mean": 0.18749024199843872
            },
            "ndcg@10": {
              "ci_high": 0.16592334017183824,
              "ci_low": 0.10259831114166858,
              "mean": 0.13324822925148228
            },
            "ndcg_condensed@10": {
              "mean": 0.16390896727904114
            },
            "recall@10": {
              "mean": 0.12642466822794693
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.13173302107728338
            },
            "recall@5": {
              "mean": 0.12295081967213115
            }
          },
          "run_id": "quality-keyword-full--qwen3-4b-20261010T140055Z-c2ecbd"
        },
        "voyage-rerank-3": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 0.26278534493688055,
            "delta_vs_first_stage_points": 9.712074192576287,
            "ndcg@10": 0.15858146458224806,
            "p_vs_best_holm": 0.39416058394160586,
            "p_vs_first_stage_holm": 0.0007999200079992001,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:55+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.25573770491803277
            },
            "hit@5": {
              "ci_high": 0.28524590163934427,
              "ci_low": 0.1901639344262295,
              "mean": 0.2360655737704918
            },
            "judged@10": {
              "mean": 0.09868852459016395
            },
            "mrr@10": {
              "mean": 0.21355971896955503
            },
            "ndcg@10": {
              "ci_high": 0.19403500460552764,
              "ci_low": 0.12506885805394333,
              "mean": 0.15858146458224806
            },
            "ndcg_condensed@10": {
              "mean": 0.17629901622057914
            },
            "recall@10": {
              "mean": 0.1582903981264637
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.16338017174082747
            },
            "recall@5": {
              "mean": 0.1456830601092896
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.18032786885245902
            },
            "hit@5": {
              "ci_high": 0.21311475409836064,
              "ci_low": 0.12786885245901639,
              "mean": 0.17049180327868851
            },
            "judged@10": {
              "mean": 0.06590163934426228
            },
            "mrr@10": {
              "mean": 0.15257611241217797
            },
            "ndcg@10": {
              "ci_high": 0.13502222084249912,
              "ci_low": 0.07845336714664575,
              "mean": 0.10611515883574905
            },
            "ndcg_condensed@10": {
              "mean": 0.15689886018388052
            },
            "recall@10": {
              "mean": 0.10139734582357532
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.10241217798594845
            },
            "recall@5": {
              "mean": 0.09433255269320842
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.21639344262295082
            },
            "hit@5": {
              "ci_high": 0.25573770491803277,
              "ci_low": 0.16393442622950818,
              "mean": 0.2098360655737705
            },
            "judged@10": {
              "mean": 0.08262295081967214
            },
            "mrr@10": {
              "mean": 0.1823809523809524
            },
            "ndcg@10": {
              "ci_high": 0.1622064454625992,
              "ci_low": 0.10037347499708084,
              "mean": 0.13049073189962007
            },
            "ndcg_condensed@10": {
              "mean": 0.16681024070941672
            },
            "recall@10": {
              "mean": 0.12733021077283374
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.1322014051522248
            },
            "recall@5": {
              "mean": 0.12199063231850117
            }
          },
          "run_id": "quality-keyword-full--voyage-rerank-3-20261010T140055Z-3a6828"
        },
        "voyage-rerank-3-lite": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 0.8356265832463738,
            "delta_vs_first_stage_points": 9.139232954266793,
            "ndcg@10": 0.15285305219915313,
            "p_vs_best_holm": 0.16798320167983202,
            "p_vs_first_stage_holm": 0.0007999200079992001,
            "wording_best_vs_model": "on par",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:55+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.25573770491803277
            },
            "hit@5": {
              "ci_high": 0.2786885245901639,
              "ci_low": 0.18360655737704917,
              "mean": 0.22950819672131148
            },
            "judged@10": {
              "mean": 0.0980327868852459
            },
            "mrr@10": {
              "mean": 0.20547619047619045
            },
            "ndcg@10": {
              "ci_high": 0.1870632009574082,
              "ci_low": 0.12015883591189834,
              "mean": 0.15285305219915313
            },
            "ndcg_condensed@10": {
              "mean": 0.17311217929281097
            },
            "recall@10": {
              "mean": 0.15730679156908664
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.16272443403590944
            },
            "recall@5": {
              "mean": 0.13882903981264635
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.18032786885245902
            },
            "hit@5": {
              "ci_high": 0.21639344262295082,
              "ci_low": 0.13114754098360656,
              "mean": 0.1737704918032787
            },
            "judged@10": {
              "mean": 0.06622950819672133
            },
            "mrr@10": {
              "mean": 0.146615925058548
            },
            "ndcg@10": {
              "ci_high": 0.12984925015962645,
              "ci_low": 0.0753069854807992,
              "mean": 0.10206586700519445
            },
            "ndcg_condensed@10": {
              "mean": 0.15411511429629096
            },
            "recall@10": {
              "mean": 0.10022638563622169
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.10241217798594845
            },
            "recall@5": {
              "mean": 0.09581576893052302
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.21639344262295082
            },
            "hit@5": {
              "ci_high": 0.24918032786885247,
              "ci_low": 0.15737704918032788,
              "mean": 0.20327868852459016
            },
            "judged@10": {
              "mean": 0.08163934426229508
            },
            "mrr@10": {
              "mean": 0.17981264637002342
            },
            "ndcg@10": {
              "ci_high": 0.1582900550030593,
              "ci_low": 0.09749057816107252,
              "mean": 0.12711566871317556
            },
            "ndcg_condensed@10": {
              "mean": 0.16350046890238082
            },
            "recall@10": {
              "mean": 0.12569086651053865
            },
            "recall@100": {
              "mean": 0.1662217017954723
            },
            "recall@20": {
              "mean": 0.1322014051522248
            },
            "recall@5": {
              "mean": 0.11562841530054645
            }
          },
          "run_id": "quality-keyword-full--voyage-rerank-3-lite-20261010T140055Z-ad9a59"
        }
      },
      "not_measured": {
        "jina-v3.5": {
          "depth@100": "vendor listwise block exceeds 16 GB VRAM for at least 13 of 304 queries (prompts >= 45.2k tokens OOM; 35.5k fit)"
        }
      },
      "run_id": "quality-summary-keyword-full-20261012T080508Z-368030",
      "winner": null
    },
    "vector-71k": {
      "best": "qwen3-4b",
      "created_utc": "2026-10-12T08:05:07+00:00",
      "models": {
        "bge-v2-m3": {
          "comparison": {
            "contamination": [
              "MIRACLReranking",
              "MIRACLRetrieval",
              "MIRACLRetrievalHardNegatives"
            ],
            "contamination_known": true,
            "delta_best_minus_model_points": 4.4539439515354555,
            "delta_vs_first_stage_points": 4.749198979332736,
            "ndcg@10": 0.6378067818750585,
            "p_vs_best_holm": 0.0011998800119988001,
            "p_vs_first_stage_holm": 0.0035996400359964006,
            "wording_best_vs_model": "clearly better",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:49+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.9049180327868852
            },
            "hit@5": {
              "ci_high": 0.8918032786885246,
              "ci_low": 0.8131147540983606,
              "mean": 0.8524590163934426
            },
            "judged@10": {
              "mean": 0.37901639344262295
            },
            "mrr@10": {
              "mean": 0.6919971376528753
            },
            "ndcg@10": {
              "ci_high": 0.67411429739069,
              "ci_low": 0.6001374171462157,
              "mean": 0.6378067818750585
            },
            "ndcg_condensed@10": {
              "mean": 0.8351385818380394
            },
            "recall@10": {
              "mean": 0.7463882383554514
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8469372885766329
            },
            "recall@5": {
              "mean": 0.632305490502212
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.8819672131147541
            },
            "hit@5": {
              "ci_high": 0.8918032786885246,
              "ci_low": 0.8131147540983606,
              "mean": 0.8524590163934426
            },
            "judged@10": {
              "mean": 0.3885245901639344
            },
            "mrr@10": {
              "mean": 0.6885688264376788
            },
            "ndcg@10": {
              "ci_high": 0.6682886627716289,
              "ci_low": 0.5924248833338733,
              "mean": 0.6308555537779529
            },
            "ndcg_condensed@10": {
              "mean": 0.8327037231634649
            },
            "recall@10": {
              "mean": 0.7325956284153006
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8045771532656778
            },
            "recall@5": {
              "mean": 0.6315716887848035
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.9016393442622951
            },
            "hit@5": {
              "ci_high": 0.8918032786885246,
              "ci_low": 0.8131147540983606,
              "mean": 0.8524590163934426
            },
            "judged@10": {
              "mean": 0.38688524590163936
            },
            "mrr@10": {
              "mean": 0.6933320322664585
            },
            "ndcg@10": {
              "ci_high": 0.6733505576169614,
              "ci_low": 0.5991343373599449,
              "mean": 0.6368482763401818
            },
            "ndcg_condensed@10": {
              "mean": 0.8337023821043765
            },
            "recall@10": {
              "mean": 0.7434608378870674
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8474837366640646
            },
            "recall@5": {
              "mean": 0.6305022118136873
            }
          },
          "run_id": "quality-vector-71k--bge-v2-m3-20261010T140049Z-9d7aab"
        },
        "gte-base": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 11.487950584923755,
            "delta_vs_first_stage_points": -2.284807654055565,
            "ndcg@10": 0.5674667155411754,
            "p_vs_best_holm": 0.0007999200079992001,
            "p_vs_first_stage_holm": 0.12488751124887511,
            "wording_best_vs_model": "clearly better",
            "wording_vs_first_stage": "on par"
          },
          "created_utc": "2026-10-10T14:00:49+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.8721311475409836
            },
            "hit@5": {
              "ci_high": 0.8295081967213115,
              "ci_low": 0.7377049180327869,
              "mean": 0.7836065573770492
            },
            "judged@10": {
              "mean": 0.36426229508196717
            },
            "mrr@10": {
              "mean": 0.6166185271922978
            },
            "ndcg@10": {
              "ci_high": 0.6047538282770902,
              "ci_low": 0.5297907144336069,
              "mean": 0.5674667155411754
            },
            "ndcg_condensed@10": {
              "mean": 0.80009864631187
            },
            "recall@10": {
              "mean": 0.6919346864428831
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.801975019516003
            },
            "recall@5": {
              "mean": 0.552311995836586
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.8721311475409836
            },
            "hit@5": {
              "ci_high": 0.839344262295082,
              "ci_low": 0.7475409836065574,
              "mean": 0.7934426229508197
            },
            "judged@10": {
              "mean": 0.3842622950819672
            },
            "mrr@10": {
              "mean": 0.6253642987249544
            },
            "ndcg@10": {
              "ci_high": 0.618961395416209,
              "ci_low": 0.5439891239799626,
              "mean": 0.5814561727701212
            },
            "ndcg_condensed@10": {
              "mean": 0.8056012226383644
            },
            "recall@10": {
              "mean": 0.7105412438199323
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8045771532656778
            },
            "recall@5": {
              "mean": 0.5735415040333073
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.8754098360655738
            },
            "hit@5": {
              "ci_high": 0.8360655737704918,
              "ci_low": 0.7442622950819672,
              "mean": 0.7901639344262295
            },
            "judged@10": {
              "mean": 0.3744262295081967
            },
            "mrr@10": {
              "mean": 0.6205503512880562
            },
            "ndcg@10": {
              "ci_high": 0.6110628913281227,
              "ci_low": 0.5366775762458359,
              "mean": 0.5740975352535735
            },
            "ndcg_condensed@10": {
              "mean": 0.8026997614652696
            },
            "recall@10": {
              "mean": 0.7024811345303148
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8181342701014832
            },
            "recall@5": {
              "mean": 0.5666289357272964
            }
          },
          "run_id": "quality-vector-71k--gte-base-20261010T140049Z-d0c904"
        },
        "jina-v3.5": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 4.087316917778594,
            "delta_vs_first_stage_points": 5.115826013089597,
            "ndcg@10": 0.6414730522126271,
            "p_vs_best_holm": 0.0029997000299970002,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "clearly better",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:50+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.9081967213114754
            },
            "hit@5": {
              "ci_high": 0.8918032786885246,
              "ci_low": 0.8131147540983606,
              "mean": 0.8524590163934426
            },
            "judged@10": {
              "mean": 0.3678688524590164
            },
            "mrr@10": {
              "mean": 0.69747983346344
            },
            "ndcg@10": {
              "ci_high": 0.6771751394545925,
              "ci_low": 0.6055945512387628,
              "mean": 0.6414730522126271
            },
            "ndcg_condensed@10": {
              "mean": 0.8448905406981201
            },
            "recall@10": {
              "mean": 0.755647931303669
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8448373666406453
            },
            "recall@5": {
              "mean": 0.6440658339838669
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.8885245901639345
            },
            "hit@5": {
              "ci_high": 0.8819672131147541,
              "ci_low": 0.8,
              "mean": 0.8426229508196721
            },
            "judged@10": {
              "mean": 0.3888524590163934
            },
            "mrr@10": {
              "mean": 0.6946383034087952
            },
            "ndcg@10": {
              "ci_high": 0.676165336944166,
              "ci_low": 0.601984524425781,
              "mean": 0.6394433771761813
            },
            "ndcg_condensed@10": {
              "mean": 0.8443026209146947
            },
            "recall@10": {
              "mean": 0.7439227166276345
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8045771532656778
            },
            "recall@5": {
              "mean": 0.6462516263335935
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.9180327868852459
            },
            "hit@5": {
              "ci_high": 0.8885245901639345,
              "ci_low": 0.8098360655737705,
              "mean": 0.8491803278688524
            },
            "judged@10": {
              "mean": 0.38
            },
            "mrr@10": {
              "mean": 0.6970686963309914
            },
            "ndcg@10": {
              "ci_high": 0.6821538136501765,
              "ci_low": 0.6115662192593385,
              "mean": 0.6471846699929054
            },
            "ndcg_condensed@10": {
              "mean": 0.8431071238028902
            },
            "recall@10": {
              "mean": 0.7668813427010148
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8541582097319803
            },
            "recall@5": {
              "mean": 0.6453773093937029
            }
          },
          "run_id": "quality-vector-71k--jina-v3.5-20261010T140050Z-1856e8"
        },
        "mxbai-large-v2": {
          "comparison": {
            "contamination": [],
            "contamination_known": true,
            "delta_best_minus_model_points": 14.259358268771594,
            "delta_vs_first_stage_points": -5.056215337903403,
            "ndcg@10": 0.5397526387026971,
            "p_vs_best_holm": 0.0007999200079992001,
            "p_vs_first_stage_holm": 0.0035996400359964006,
            "wording_best_vs_model": "clearly better",
            "wording_vs_first_stage": "clearly worse"
          },
          "created_utc": "2026-10-10T14:00:50+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.839344262295082
            },
            "hit@5": {
              "ci_high": 0.8098360655737705,
              "ci_low": 0.7147540983606557,
              "mean": 0.7639344262295082
            },
            "judged@10": {
              "mean": 0.3075409836065574
            },
            "mrr@10": {
              "mean": 0.59655997918293
            },
            "ndcg@10": {
              "ci_high": 0.577871969447336,
              "ci_low": 0.5021764504847352,
              "mean": 0.5397526387026971
            },
            "ndcg_condensed@10": {
              "mean": 0.8240041341229625
            },
            "recall@10": {
              "mean": 0.6619463960447567
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.7999154306531356
            },
            "recall@5": {
              "mean": 0.532352328909706
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.8754098360655738
            },
            "hit@5": {
              "ci_high": 0.819672131147541,
              "ci_low": 0.7245901639344262,
              "mean": 0.7737704918032787
            },
            "judged@10": {
              "mean": 0.36
            },
            "mrr@10": {
              "mean": 0.6164754098360655
            },
            "ndcg@10": {
              "ci_high": 0.609059393604139,
              "ci_low": 0.53569023457762,
              "mean": 0.5721945829034601
            },
            "ndcg_condensed@10": {
              "mean": 0.819982416599322
            },
            "recall@10": {
              "mean": 0.7120673952641166
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8045771532656778
            },
            "recall@5": {
              "mean": 0.5552016653655999
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.8491803278688524
            },
            "hit@5": {
              "ci_high": 0.8131147540983606,
              "ci_low": 0.7180327868852459,
              "mean": 0.7672131147540984
            },
            "judged@10": {
              "mean": 0.3265573770491803
            },
            "mrr@10": {
              "mean": 0.6028181108508978
            },
            "ndcg@10": {
              "ci_high": 0.5881968658460042,
              "ci_low": 0.5131816814520416,
              "mean": 0.5503631621834044
            },
            "ndcg_condensed@10": {
              "mean": 0.8232645824547447
            },
            "recall@10": {
              "mean": 0.6782110330470986
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8145420244600573
            },
            "recall@5": {
              "mean": 0.5406583398386677
            }
          },
          "run_id": "quality-vector-71k--mxbai-large-v2-20261010T140050Z-c649bf"
        },
        "querit-4b": {
          "comparison": {
            "contamination": [
              "MIRACLReranking"
            ],
            "contamination_known": true,
            "delta_best_minus_model_points": 4.907023706335901,
            "delta_vs_first_stage_points": 4.296119224532291,
            "ndcg@10": 0.633275984327054,
            "p_vs_best_holm": 0.0011998800119988001,
            "p_vs_first_stage_holm": 0.0035996400359964006,
            "wording_best_vs_model": "clearly better",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:50+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.8786885245901639
            },
            "hit@5": {
              "ci_high": 0.8688524590163934,
              "ci_low": 0.7868852459016393,
              "mean": 0.8295081967213115
            },
            "judged@10": {
              "mean": 0.3518032786885246
            },
            "mrr@10": {
              "mean": 0.6878584439240177
            },
            "ndcg@10": {
              "ci_high": 0.6707712454339817,
              "ci_low": 0.5958101551696311,
              "mean": 0.633275984327054
            },
            "ndcg_condensed@10": {
              "mean": 0.8625024593639037
            },
            "recall@10": {
              "mean": 0.7356791569086651
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8284985688264377
            },
            "recall@5": {
              "mean": 0.6261930783242259
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.8819672131147541
            },
            "hit@5": {
              "ci_high": 0.8786885245901639,
              "ci_low": 0.7967213114754098,
              "mean": 0.839344262295082
            },
            "judged@10": {
              "mean": 0.3701639344262295
            },
            "mrr@10": {
              "mean": 0.6934361176164456
            },
            "ndcg@10": {
              "ci_high": 0.674537619118934,
              "ci_low": 0.5988182624859166,
              "mean": 0.6370932209333862
            },
            "ndcg_condensed@10": {
              "mean": 0.8578057704077697
            },
            "recall@10": {
              "mean": 0.7348568826437679
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8045771532656778
            },
            "recall@5": {
              "mean": 0.6340931563882384
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.8786885245901639
            },
            "hit@5": {
              "ci_high": 0.8721311475409836,
              "ci_low": 0.7868852459016393,
              "mean": 0.8295081967213115
            },
            "judged@10": {
              "mean": 0.3554098360655738
            },
            "mrr@10": {
              "mean": 0.6881680978402289
            },
            "ndcg@10": {
              "ci_high": 0.6693086465439604,
              "ci_low": 0.5938707492863832,
              "mean": 0.6318191865678772
            },
            "ndcg_condensed@10": {
              "mean": 0.860479560982111
            },
            "recall@10": {
              "mean": 0.7315027322404372
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8259302628155087
            },
            "recall@5": {
              "mean": 0.6271220400728597
            }
          },
          "run_id": "quality-vector-71k--querit-4b-20261010T140050Z-3b27ed"
        },
        "qwen3-0.6b": {
          "comparison": {
            "contamination": [
              "MIRACLReranking"
            ],
            "contamination_known": true,
            "delta_best_minus_model_points": 4.449820961778695,
            "delta_vs_first_stage_points": 4.753321969089496,
            "ndcg@10": 0.6378480117726261,
            "p_vs_best_holm": 0.0029997000299970002,
            "p_vs_first_stage_holm": 0.005099490050994901,
            "wording_best_vs_model": "clearly better",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:50+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.9344262295081968
            },
            "hit@5": {
              "ci_high": 0.9081967213114754,
              "ci_low": 0.8327868852459016,
              "mean": 0.8721311475409836
            },
            "judged@10": {
              "mean": 0.38491803278688524
            },
            "mrr@10": {
              "mean": 0.6783008066614624
            },
            "ndcg@10": {
              "ci_high": 0.6716597586936816,
              "ci_low": 0.6037861846135041,
              "mean": 0.6378480117726261
            },
            "ndcg_condensed@10": {
              "mean": 0.8380814480733674
            },
            "recall@10": {
              "mean": 0.7771922976841009
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8863674212854542
            },
            "recall@5": {
              "mean": 0.6599557637262555
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.9016393442622951
            },
            "hit@5": {
              "ci_high": 0.8819672131147541,
              "ci_low": 0.8,
              "mean": 0.8426229508196721
            },
            "judged@10": {
              "mean": 0.39901639344262296
            },
            "mrr@10": {
              "mean": 0.6772404371584699
            },
            "ndcg@10": {
              "ci_high": 0.6681747634107194,
              "ci_low": 0.5965125199383119,
              "mean": 0.6328005557054308
            },
            "ndcg_condensed@10": {
              "mean": 0.8306955058581681
            },
            "recall@10": {
              "mean": 0.755647931303669
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8045771532656778
            },
            "recall@5": {
              "mean": 0.649471766848816
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.9278688524590164
            },
            "hit@5": {
              "ci_high": 0.898360655737705,
              "ci_low": 0.8229508196721311,
              "mean": 0.8622950819672132
            },
            "judged@10": {
              "mean": 0.39311475409836066
            },
            "mrr@10": {
              "mean": 0.6749986989331251
            },
            "ndcg@10": {
              "ci_high": 0.6729771872254016,
              "ci_low": 0.6044359231660911,
              "mean": 0.638788489556897
            },
            "ndcg_condensed@10": {
              "mean": 0.833543808126558
            },
            "recall@10": {
              "mean": 0.7805074160811866
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8755477491543064
            },
            "recall@5": {
              "mean": 0.6563492063492063
            }
          },
          "run_id": "quality-vector-71k--qwen3-0.6b-20261010T140050Z-3b2735"
        },
        "qwen3-4b": {
          "comparison": {
            "contamination": [
              "MIRACLReranking"
            ],
            "contamination_known": true,
            "delta_vs_first_stage_points": 9.203142930868191,
            "ndcg@10": 0.682346221390413,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:50+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.9377049180327869
            },
            "hit@5": {
              "ci_high": 0.9180327868852459,
              "ci_low": 0.8459016393442623,
              "mean": 0.8819672131147541
            },
            "judged@10": {
              "mean": 0.36950819672131147
            },
            "mrr@10": {
              "mean": 0.7326424668227947
            },
            "ndcg@10": {
              "ci_high": 0.7164572473158747,
              "ci_low": 0.6477023930108017,
              "mean": 0.682346221390413
            },
            "ndcg_condensed@10": {
              "mean": 0.8749943502501049
            },
            "recall@10": {
              "mean": 0.804367681498829
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.891376528753578
            },
            "recall@5": {
              "mean": 0.6840202966432474
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.9081967213114754
            },
            "hit@5": {
              "ci_high": 0.8885245901639345,
              "ci_low": 0.8098360655737705,
              "mean": 0.8491803278688524
            },
            "judged@10": {
              "mean": 0.3816393442622951
            },
            "mrr@10": {
              "mean": 0.720536039552433
            },
            "ndcg@10": {
              "ci_high": 0.7028292442990848,
              "ci_low": 0.6289827346183486,
              "mean": 0.6661395536260493
            },
            "ndcg_condensed@10": {
              "mean": 0.8577181499165303
            },
            "recall@10": {
              "mean": 0.7672482435597189
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8045771532656778
            },
            "recall@5": {
              "mean": 0.6673302107728337
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.9278688524590164
            },
            "hit@5": {
              "ci_high": 0.9049180327868852,
              "ci_low": 0.8295081967213115,
              "mean": 0.8688524590163934
            },
            "judged@10": {
              "mean": 0.3770491803278688
            },
            "mrr@10": {
              "mean": 0.73220010408535
            },
            "ndcg@10": {
              "ci_high": 0.7173139290042012,
              "ci_low": 0.6467005887740274,
              "mean": 0.6826100234134929
            },
            "ndcg_condensed@10": {
              "mean": 0.8706027878392347
            },
            "recall@10": {
              "mean": 0.801174863387978
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8776333593546709
            },
            "recall@5": {
              "mean": 0.679047619047619
            }
          },
          "run_id": "quality-vector-71k--qwen3-4b-20261010T140050Z-1dfdc4"
        },
        "voyage-rerank-3": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 2.3914284963181376,
            "delta_vs_first_stage_points": 6.811714434550053,
            "ndcg@10": 0.6584319364272316,
            "p_vs_best_holm": 0.0347965203479652,
            "p_vs_first_stage_holm": 0.0008999100089991002,
            "wording_best_vs_model": "clearly better",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:51+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.9081967213114754
            },
            "hit@5": {
              "ci_high": 0.8918032786885246,
              "ci_low": 0.8131147540983606,
              "mean": 0.8524590163934426
            },
            "judged@10": {
              "mean": 0.34950819672131145
            },
            "mrr@10": {
              "mean": 0.7232435597189696
            },
            "ndcg@10": {
              "ci_high": 0.6938436837188919,
              "ci_low": 0.6221351876224293,
              "mean": 0.6584319364272316
            },
            "ndcg_condensed@10": {
              "mean": 0.8781302295814465
            },
            "recall@10": {
              "mean": 0.7648946135831381
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8640072859744992
            },
            "recall@5": {
              "mean": 0.6480783242258652
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.9016393442622951
            },
            "hit@5": {
              "ci_high": 0.8885245901639345,
              "ci_low": 0.8065573770491803,
              "mean": 0.8491803278688524
            },
            "judged@10": {
              "mean": 0.37836065573770494
            },
            "mrr@10": {
              "mean": 0.722832422586521
            },
            "ndcg@10": {
              "ci_high": 0.6977824468639741,
              "ci_low": 0.6255043342364389,
              "mean": 0.6620686673058775
            },
            "ndcg_condensed@10": {
              "mean": 0.8669394025853439
            },
            "recall@10": {
              "mean": 0.7624290918553213
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8045771532656778
            },
            "recall@5": {
              "mean": 0.655104085349987
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.9114754098360656
            },
            "hit@5": {
              "ci_high": 0.8918032786885246,
              "ci_low": 0.8131147540983606,
              "mean": 0.8524590163934426
            },
            "judged@10": {
              "mean": 0.3586885245901639
            },
            "mrr@10": {
              "mean": 0.7256661462399168
            },
            "ndcg@10": {
              "ci_high": 0.6989355823855826,
              "ci_low": 0.6276843905150683,
              "mean": 0.6637719602525459
            },
            "ndcg_condensed@10": {
              "mean": 0.8764807067594783
            },
            "recall@10": {
              "mean": 0.7736377829820453
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8523367161072078
            },
            "recall@5": {
              "mean": 0.6491712204007285
            }
          },
          "run_id": "quality-vector-71k--voyage-rerank-3-20261010T140051Z-e1024d"
        },
        "voyage-rerank-3-lite": {
          "comparison": {
            "contamination": [],
            "contamination_known": false,
            "delta_best_minus_model_points": 5.513357208126002,
            "delta_vs_first_stage_points": 3.6897857227421893,
            "ndcg@10": 0.627212649309153,
            "p_vs_best_holm": 0.0011998800119988001,
            "p_vs_first_stage_holm": 0.015998400159984,
            "wording_best_vs_model": "clearly better",
            "wording_vs_first_stage": "clearly better"
          },
          "created_utc": "2026-10-10T14:00:51+00:00",
          "depth@100": {
            "hit@10": {
              "mean": 0.9016393442622951
            },
            "hit@5": {
              "ci_high": 0.8786885245901639,
              "ci_low": 0.7967213114754098,
              "mean": 0.839344262295082
            },
            "judged@10": {
              "mean": 0.3416393442622951
            },
            "mrr@10": {
              "mean": 0.6968774395003903
            },
            "ndcg@10": {
              "ci_high": 0.6630589452789711,
              "ci_low": 0.5912355110004343,
              "mean": 0.627212649309153
            },
            "ndcg_condensed@10": {
              "mean": 0.8591661655502605
            },
            "recall@10": {
              "mean": 0.7345901639344262
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8507832422586522
            },
            "recall@5": {
              "mean": 0.6118293000260213
            }
          },
          "depth@20": {
            "hit@10": {
              "mean": 0.8950819672131147
            },
            "hit@5": {
              "ci_high": 0.8754098360655738,
              "ci_low": 0.7934426229508197,
              "mean": 0.8360655737704918
            },
            "judged@10": {
              "mean": 0.37868852459016394
            },
            "mrr@10": {
              "mean": 0.7052498048399688
            },
            "ndcg@10": {
              "ci_high": 0.6786889970778162,
              "ci_low": 0.6067107324730893,
              "mean": 0.6427508753769122
            },
            "ndcg_condensed@10": {
              "mean": 0.8517304357650044
            },
            "recall@10": {
              "mean": 0.7512373145979703
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8045771532656778
            },
            "recall@5": {
              "mean": 0.624479573250065
            }
          },
          "depth@50": {
            "hit@10": {
              "mean": 0.9147540983606557
            },
            "hit@5": {
              "ci_high": 0.8885245901639345,
              "ci_low": 0.8098360655737705,
              "mean": 0.8491803278688524
            },
            "judged@10": {
              "mean": 0.3557377049180328
            },
            "mrr@10": {
              "mean": 0.7016770752016653
            },
            "ndcg@10": {
              "ci_high": 0.6747291376547622,
              "ci_low": 0.6039642948647348,
              "mean": 0.6393966532230647
            },
            "ndcg_condensed@10": {
              "mean": 0.8571626712776917
            },
            "recall@10": {
              "mean": 0.756120218579235
            },
            "recall@100": {
              "mean": 0.9652354931043456
            },
            "recall@20": {
              "mean": 0.8485662243039293
            },
            "recall@5": {
              "mean": 0.6168566224303929
            }
          },
          "run_id": "quality-vector-71k--voyage-rerank-3-lite-20261010T140051Z-23b020"
        }
      },
      "not_measured": {},
      "run_id": "quality-summary-vector-71k-20261012T080507Z-e8fbcc",
      "winner": "qwen3-4b"
    }
  },
  "repo_commit": "e715d27a1bf73df51f6cbddc99e936b293488590",
  "repo_dirty": false,
  "schema_version": 1
}
