{ "candidates": [ "XL-512 v6 best", "XL-512 v6 final", "V2-512 v6 best", "V2-512 v6 final" ], "baselines": [ "V2-128 deployed(v3)", "V2-512 v3 best", "V2-512 v3 final" ], "passed_axes": 66, "failed_axes": 22, "skipped_axes": 0, "dominates": false, "rows": [ { "axis": "Top-1 正确率", "candidate": "XL-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.807510946446615, "candidate_value": 0.957842146626249, "delta": 0.15033120017963397, "direction": "max", "pass": true }, { "axis": "Top-1 正确率", "candidate": "XL-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.807510946446615, "candidate_value": 0.9502077018075671, "delta": 0.14269675536095205, "direction": "max", "pass": true }, { "axis": "Top-1 正确率", "candidate": "V2-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.807510946446615, "candidate_value": 0.950488379925901, "delta": 0.14297743347928593, "direction": "max", "pass": true }, { "axis": "Top-1 正确率", "candidate": "V2-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.807510946446615, "candidate_value": 0.941843493881217, "delta": 0.134332547434602, "direction": "max", "pass": true }, { "axis": "Recall@1", "candidate": "XL-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.7698439429662064, "candidate_value": 0.9024082182553048, "delta": 0.13256427528909842, "direction": "max", "pass": true }, { "axis": "Recall@1", "candidate": "XL-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.7698439429662064, "candidate_value": 0.8948299090602897, "delta": 0.1249859660940833, "direction": "max", "pass": true }, { "axis": "Recall@1", "candidate": "V2-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.7698439429662064, "candidate_value": 0.8971034018187942, "delta": 0.12725945885258783, "direction": "max", "pass": true }, { "axis": "Recall@1", "candidate": "V2-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.7698439429662064, "candidate_value": 0.8864657011339396, "delta": 0.11662175816773324, "direction": "max", "pass": true }, { "axis": "Recall@3", "candidate": "XL-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.9501515661839003, "candidate_value": 0.9758897496351184, "delta": 0.025738183451218144, "direction": "max", "pass": true }, { "axis": "Recall@3", "candidate": "XL-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.9501515661839003, "candidate_value": 0.965336252385764, "delta": 0.01518468620186375, "direction": "max", "pass": true }, { "axis": "Recall@3", "candidate": "V2-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.9501515661839003, "candidate_value": 0.9884079937128102, "delta": 0.0382564275289099, "direction": "max", "pass": true }, { "axis": "Recall@3", "candidate": "V2-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.9501515661839003, "candidate_value": 0.9666273717300999, "delta": 0.01647580554619965, "direction": "max", "pass": true }, { "axis": "Recall@5", "candidate": "XL-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.9730829684517796, "candidate_value": 0.9834399910183003, "delta": 0.010357022566520713, "direction": "max", "pass": true }, { "axis": "Recall@5", "candidate": "XL-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.9730829684517796, "candidate_value": 0.9760862243179522, "delta": 0.0030032558661726716, "direction": "max", "pass": true }, { "axis": "Recall@5", "candidate": "V2-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.9730829684517796, "candidate_value": 0.9934882676546536, "delta": 0.02040529920287404, "direction": "max", "pass": true }, { "axis": "Recall@5", "candidate": "V2-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.9730829684517796, "candidate_value": 0.9764230380599529, "delta": 0.003340069608173346, "direction": "max", "pass": true }, { "axis": "MRR", "candidate": "XL-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.8845870538857731, "candidate_value": 0.9724377931581535, "delta": 0.08785073927238041, "direction": "max", "pass": true }, { "axis": "MRR", "candidate": "XL-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.8845870538857731, "candidate_value": 0.9664213223997663, "delta": 0.08183426851399322, "direction": "max", "pass": true }, { "axis": "MRR", "candidate": "V2-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.8845870538857731, "candidate_value": 0.9711571604054701, "delta": 0.08657010651969699, "direction": "max", "pass": true }, { "axis": "MRR", "candidate": "V2-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.8845870538857731, "candidate_value": 0.958724954839147, "delta": 0.07413790095337391, "direction": "max", "pass": true }, { "axis": "nDCG@3", "candidate": "XL-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.8953026178367461, "candidate_value": 0.9664212737428025, "delta": 0.07111865590605637, "direction": "max", "pass": true }, { "axis": "nDCG@3", "candidate": "XL-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.8953026178367461, "candidate_value": 0.9560053718886551, "delta": 0.06070275405190895, "direction": "max", "pass": true }, { "axis": "nDCG@3", "candidate": "V2-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.8953026178367461, "candidate_value": 0.9723689530227545, "delta": 0.07706633518600836, "direction": "max", "pass": true }, { "axis": "nDCG@3", "candidate": "V2-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.8953026178367461, "candidate_value": 0.9548399733910118, "delta": 0.05953735555426565, "direction": "max", "pass": true }, { "axis": "多跳证据全中(Top-3)", "candidate": "XL-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.9418996295048838, "candidate_value": 0.9686201863702706, "delta": 0.02672055686538677, "direction": "max", "pass": true }, { "axis": "多跳证据全中(Top-3)", "candidate": "XL-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.9418996295048838, "candidate_value": 0.9534635679802402, "delta": 0.011563938475356417, "direction": "max", "pass": true }, { "axis": "多跳证据全中(Top-3)", "candidate": "V2-512 v6 best", "baseline": "V2-512 v3 best", "baseline_value": 0.9418996295048838, "candidate_value": 0.9863590434489727, "delta": 0.04445941394408892, "direction": "max", "pass": true }, { "axis": "多跳证据全中(Top-3)", "candidate": "V2-512 v6 final", "baseline": "V2-512 v3 best", "baseline_value": 0.9418996295048838, "candidate_value": 0.9644100145952621, "delta": 0.022510385090378282, "direction": "max", "pass": true }, { "axis": "多跳证据全中(仅多正例)", "candidate": "XL-512 v6 best", "baseline": "V2-512 v3 final", "baseline_value": 0.849, "candidate_value": 0.8705, "delta": 0.021500000000000075, "direction": "max", "pass": true }, { "axis": "多跳证据全中(仅多正例)", "candidate": "XL-512 v6 final", "baseline": "V2-512 v3 final", "baseline_value": 0.849, "candidate_value": 0.7885, "delta": -0.0605, "direction": "max", "pass": false }, { "axis": "多跳证据全中(仅多正例)", "candidate": "V2-512 v6 best", "baseline": "V2-512 v3 final", "baseline_value": 0.849, "candidate_value": 0.9635, "delta": 0.11450000000000005, "direction": "max", "pass": true }, { "axis": "多跳证据全中(仅多正例)", "candidate": "V2-512 v6 final", "baseline": "V2-512 v3 final", "baseline_value": 0.849, "candidate_value": 0.9605, "delta": 0.11150000000000004, "direction": "max", "pass": true }, { "axis": "hop 正确率", "candidate": "XL-512 v6 best", "baseline": "V2-512 v3 final", "baseline_value": 0.8175182481751825, "candidate_value": 1.0, "delta": 0.18248175182481752, "direction": "max", "pass": true }, { "axis": "hop 正确率", "candidate": "XL-512 v6 final", "baseline": "V2-512 v3 final", "baseline_value": 0.8175182481751825, "candidate_value": 1.0, "delta": 0.18248175182481752, "direction": "max", "pass": true }, { "axis": "hop 正确率", "candidate": "V2-512 v6 best", "baseline": "V2-512 v3 final", "baseline_value": 0.8175182481751825, "candidate_value": 1.0, "delta": 0.18248175182481752, "direction": "max", "pass": true }, { "axis": "hop 正确率", "candidate": "V2-512 v6 final", "baseline": "V2-512 v3 final", "baseline_value": 0.8175182481751825, "candidate_value": 1.0, "delta": 0.18248175182481752, "direction": "max", "pass": true }, { "axis": "hop 欠预测率", "candidate": "XL-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.046817110138093636, "candidate_value": 0.0, "delta": -0.046817110138093636, "direction": "min", "pass": true }, { "axis": "hop 欠预测率", "candidate": "XL-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.046817110138093636, "candidate_value": 0.0, "delta": -0.046817110138093636, "direction": "min", "pass": true }, { "axis": "hop 欠预测率", "candidate": "V2-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.046817110138093636, "candidate_value": 0.0, "delta": -0.046817110138093636, "direction": "min", "pass": true }, { "axis": "hop 欠预测率", "candidate": "V2-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.046817110138093636, "candidate_value": 0.0, "delta": -0.046817110138093636, "direction": "min", "pass": true }, { "axis": "need F1 (门槛 0.50)", "candidate": "XL-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.895773512669387, "candidate_value": 1.0, "delta": 0.10422648733061302, "direction": "max", "pass": true }, { "axis": "need F1 (门槛 0.50)", "candidate": "XL-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.895773512669387, "candidate_value": 1.0, "delta": 0.10422648733061302, "direction": "max", "pass": true }, { "axis": "need F1 (门槛 0.50)", "candidate": "V2-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.895773512669387, "candidate_value": 0.999775507913346, "delta": 0.10400199524395903, "direction": "max", "pass": true }, { "axis": "need F1 (门槛 0.50)", "candidate": "V2-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.895773512669387, "candidate_value": 1.0, "delta": 0.10422648733061302, "direction": "max", "pass": true }, { "axis": "need 召回 (门槛 0.50)", "candidate": "XL-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.998203660042663, "candidate_value": 1.0, "delta": 0.0017963399573369676, "direction": "max", "pass": true }, { "axis": "need 召回 (门槛 0.50)", "candidate": "XL-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.998203660042663, "candidate_value": 1.0, "delta": 0.0017963399573369676, "direction": "max", "pass": true }, { "axis": "need 召回 (门槛 0.50)", "candidate": "V2-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.998203660042663, "candidate_value": 1.0, "delta": 0.0017963399573369676, "direction": "max", "pass": true }, { "axis": "need 召回 (门槛 0.50)", "candidate": "V2-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.998203660042663, "candidate_value": 1.0, "delta": 0.0017963399573369676, "direction": "max", "pass": true }, { "axis": "未知拒答率 (门槛 0.50)", "candidate": "XL-512 v6 best", "baseline": "V2-512 v3 final", "baseline_value": 0.5129079396005846, "candidate_value": 1.0, "delta": 0.48709206039941544, "direction": "max", "pass": true }, { "axis": "未知拒答率 (门槛 0.50)", "candidate": "XL-512 v6 final", "baseline": "V2-512 v3 final", "baseline_value": 0.5129079396005846, "candidate_value": 1.0, "delta": 0.48709206039941544, "direction": "max", "pass": true }, { "axis": "未知拒答率 (门槛 0.50)", "candidate": "V2-512 v6 best", "baseline": "V2-512 v3 final", "baseline_value": 0.5129079396005846, "candidate_value": 0.9980516317584024, "delta": 0.48514369215781783, "direction": "max", "pass": true }, { "axis": "未知拒答率 (门槛 0.50)", "candidate": "V2-512 v6 final", "baseline": "V2-512 v3 final", "baseline_value": 0.5129079396005846, "candidate_value": 1.0, "delta": 0.48709206039941544, "direction": "max", "pass": true }, { "axis": "已知问题被误拒率", "candidate": "XL-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.001796339957336926, "candidate_value": 0.0, "delta": -0.001796339957336926, "direction": "min", "pass": true }, { "axis": "已知问题被误拒率", "candidate": "XL-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.001796339957336926, "candidate_value": 0.0, "delta": -0.001796339957336926, "direction": "min", "pass": true }, { "axis": "已知问题被误拒率", "candidate": "V2-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.001796339957336926, "candidate_value": 0.0, "delta": -0.001796339957336926, "direction": "min", "pass": true }, { "axis": "已知问题被误拒率", "candidate": "V2-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.001796339957336926, "candidate_value": 0.0, "delta": -0.001796339957336926, "direction": "min", "pass": true }, { "axis": "未知问题被误读率", "candidate": "XL-512 v6 best", "baseline": "V2-512 v3 final", "baseline_value": 0.4870920603994155, "candidate_value": 0.0, "delta": -0.4870920603994155, "direction": "min", "pass": true }, { "axis": "未知问题被误读率", "candidate": "XL-512 v6 final", "baseline": "V2-512 v3 final", "baseline_value": 0.4870920603994155, "candidate_value": 0.0, "delta": -0.4870920603994155, "direction": "min", "pass": true }, { "axis": "未知问题被误读率", "candidate": "V2-512 v6 best", "baseline": "V2-512 v3 final", "baseline_value": 0.4870920603994155, "candidate_value": 0.001948368241597662, "delta": -0.48514369215781783, "direction": "min", "pass": true }, { "axis": "未知问题被误读率", "candidate": "V2-512 v6 final", "baseline": "V2-512 v3 final", "baseline_value": 0.4870920603994155, "candidate_value": 0.0, "delta": -0.4870920603994155, "direction": "min", "pass": true }, { "axis": "仲裁准确率", "candidate": "XL-512 v6 best", "baseline": "V2-512 v3 final", "baseline_value": 0.8175182481751825, "candidate_value": 1.0, "delta": 0.18248175182481752, "direction": "max", "pass": true }, { "axis": "仲裁准确率", "candidate": "XL-512 v6 final", "baseline": "V2-512 v3 final", "baseline_value": 0.8175182481751825, "candidate_value": 1.0, "delta": 0.18248175182481752, "direction": "max", "pass": true }, { "axis": "仲裁准确率", "candidate": "V2-512 v6 best", "baseline": "V2-512 v3 final", "baseline_value": 0.8175182481751825, "candidate_value": 0.9996350364963503, "delta": 0.18211678832116784, "direction": "max", "pass": true }, { "axis": "仲裁准确率", "candidate": "V2-512 v6 final", "baseline": "V2-512 v3 final", "baseline_value": 0.8175182481751825, "candidate_value": 1.0, "delta": 0.18248175182481752, "direction": "max", "pass": true }, { "axis": "平均分数余量", "candidate": "XL-512 v6 best", "baseline": "V2-512 v3 final", "baseline_value": 53.160489893678886, "candidate_value": 45.56406442258217, "delta": -7.596425471096715, "direction": "max", "pass": false }, { "axis": "平均分数余量", "candidate": "XL-512 v6 final", "baseline": "V2-512 v3 final", "baseline_value": 53.160489893678886, "candidate_value": 215.75331382084113, "delta": 162.59282392716224, "direction": "max", "pass": true }, { "axis": "平均分数余量", "candidate": "V2-512 v6 best", "baseline": "V2-512 v3 final", "baseline_value": 53.160489893678886, "candidate_value": 12.36926158705623, "delta": -40.791228306622656, "direction": "max", "pass": false }, { "axis": "平均分数余量", "candidate": "V2-512 v6 final", "baseline": "V2-512 v3 final", "baseline_value": 53.160489893678886, "candidate_value": 73.24820388849786, "delta": 20.08771399481897, "direction": "max", "pass": true }, { "axis": "单查询延迟 ms (GPU)", "candidate": "XL-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.8439246668907193, "candidate_value": 1.2161159999474573, "delta": 0.372191333056738, "direction": "min", "pass": false }, { "axis": "单查询延迟 ms (GPU)", "candidate": "XL-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.8439246668907193, "candidate_value": 1.2505186665900208, "delta": 0.40659399969930143, "direction": "min", "pass": false }, { "axis": "单查询延迟 ms (GPU)", "candidate": "V2-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.8439246668907193, "candidate_value": 0.8750353333986519, "delta": 0.031110666507932594, "direction": "min", "pass": false }, { "axis": "单查询延迟 ms (GPU)", "candidate": "V2-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 0.8439246668907193, "candidate_value": 0.8725079999809774, "delta": 0.028583333090258067, "direction": "min", "pass": false }, { "axis": "路由 QPS (GPU, 单查询)", "candidate": "XL-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 1184.9398876848934, "candidate_value": 822.2899789520122, "delta": -362.64990873288116, "direction": "max", "pass": false }, { "axis": "路由 QPS (GPU, 单查询)", "candidate": "XL-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 1184.9398876848934, "candidate_value": 799.6681910609555, "delta": -385.27169662393794, "direction": "max", "pass": false }, { "axis": "路由 QPS (GPU, 单查询)", "candidate": "V2-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 1184.9398876848934, "candidate_value": 1142.8109949754637, "delta": -42.12889270942969, "direction": "max", "pass": false }, { "axis": "路由 QPS (GPU, 单查询)", "candidate": "V2-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 1184.9398876848934, "candidate_value": 1146.121296334019, "delta": -38.818591350874385, "direction": "max", "pass": false }, { "axis": "批量 QPS (GPU, batch=64)", "candidate": "XL-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 72916.19191648123, "candidate_value": 57386.75086015149, "delta": -15529.441056329735, "direction": "max", "pass": false }, { "axis": "批量 QPS (GPU, batch=64)", "candidate": "XL-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 72916.19191648123, "candidate_value": 53291.58828255585, "delta": -19624.603633925377, "direction": "max", "pass": false }, { "axis": "批量 QPS (GPU, batch=64)", "candidate": "V2-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 72916.19191648123, "candidate_value": 73617.37369428946, "delta": 701.1817778082332, "direction": "max", "pass": true }, { "axis": "批量 QPS (GPU, batch=64)", "candidate": "V2-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 72916.19191648123, "candidate_value": 61636.8434007473, "delta": -11279.348515733931, "direction": "max", "pass": false }, { "axis": "批量 QPS (GPU, batch=256)", "candidate": "XL-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 264812.97583671653, "candidate_value": 96746.88596045674, "delta": -168066.0898762598, "direction": "max", "pass": false }, { "axis": "批量 QPS (GPU, batch=256)", "candidate": "XL-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 264812.97583671653, "candidate_value": 94345.88082639035, "delta": -170467.0950103262, "direction": "max", "pass": false }, { "axis": "批量 QPS (GPU, batch=256)", "candidate": "V2-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 264812.97583671653, "candidate_value": 161941.27100652116, "delta": -102871.70483019538, "direction": "max", "pass": false }, { "axis": "批量 QPS (GPU, batch=256)", "candidate": "V2-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 264812.97583671653, "candidate_value": 164167.81025689258, "delta": -100645.16557982395, "direction": "max", "pass": false }, { "axis": "地址字节/记录", "candidate": "XL-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 512, "candidate_value": 2048, "delta": 1536, "direction": "min", "pass": false }, { "axis": "地址字节/记录", "candidate": "XL-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 512, "candidate_value": 2048, "delta": 1536, "direction": "min", "pass": false }, { "axis": "地址字节/记录", "candidate": "V2-512 v6 best", "baseline": "V2-128 deployed(v3)", "baseline_value": 512, "candidate_value": 2048, "delta": 1536, "direction": "min", "pass": false }, { "axis": "地址字节/记录", "candidate": "V2-512 v6 final", "baseline": "V2-128 deployed(v3)", "baseline_value": 512, "candidate_value": 2048, "delta": 1536, "direction": "min", "pass": false } ] }