{ "eval_file": "data/zero_overlap/eval.jsonl", "episodes": 48, "candidate_count": 24, "router": "checkpoints\\router_replay_v7_v2_128\\memory_router_v2.pt", "results": { "prior=1.00 blend=0.00": { "router": "prior=1.00 blend=0.00", "cases": 48, "answerable_cases": 40, "unknown_cases": 8, "accuracy_pct": 65.0, "wrong_attribute_pct": 35.0, "unknown_leak_pct": 75.0, "mean_records_selected": 6.25, "records_active_min": 23, "records_active_max": 24, "stop_reasons": { "evidence_found": 48 }, "top_score_answerable": { "n": 40, "min": 0.7, "p10": 0.77, "p50": 2.85, "p90": 8.52, "max": 8.52 }, "top_score_unknown": { "n": 8, "min": 0.78, "p10": 0.78, "p50": 2.85, "p90": 8.51, "max": 8.51 }, "seconds": 349.2 } }, "rows": { "prior=1.00 blend=0.00": [ { "id": "zero-overlap-000259", "attribute": "出生城市", "answerable": true, "expected": "VAL-M9VSJAEX", "reply": "您的出生城市是 VAL-M9VSJAEX。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.564332926884773, "need_memory": true }, { "id": "zero-overlap-000219", "attribute": "咖啡口味", "answerable": true, "expected": "VAL-M3W7MRJR", "reply": "关于您早上那杯咖啡的豆子偏好，当前记忆中没有相关记录。已知信息仅包含：user；午餐偏好；VAL-Y4Z5H2FG。", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 1, "top_score": 8.515033109300843, "need_memory": true }, { "id": "zero-overlap-000193", "attribute": "午餐偏好", "answerable": true, "expected": "VAL-HG84H5TD", "reply": "您的午餐偏好是 user；午餐偏好；VAL-HG84H5TD。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 0.6998433045297862, "need_memory": true }, { "id": "zero-overlap-000080", "attribute": "入职年份", "answerable": true, "expected": "VAL-GFT4HRX2", "reply": "关于您何时开始在这家公司上班，当前记忆中没有相关记录。已知信息：user；办公城市；VAL-TCNMN9HW。", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 1, "top_score": 8.51429933919514, "need_memory": true }, { "id": "zero-overlap-000285", "attribute": "起床时间", "answerable": true, "expected": "VAL-4K3RQQKF", "reply": "我的闹钟设定时间是 VAL-4K3RQQKF。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 3.359343914853775, "need_memory": true }, { "id": "zero-overlap-000246", "attribute": "常住城市", "answerable": false, "expected": "VAL-6BWJYDE4", "reply": "关于您小时候待的地方，当前长期记忆中未包含此信息。已知信息如下：user；常住城市；VAL-6BWJYDE4；办公城市；VAL-88V99MPM；宿舍楼号", "correct": false, "wrong_attribute": false, "leaked": true, "records_written": 24, "records_active": 23, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.6527658945475845, "need_memory": true }, { "id": "zero-overlap-000204", "attribute": "阅读工具", "answerable": true, "expected": "VAL-ZRLSPU5V", "reply": "您的阅读工具是 VAL-ZRLSPU5V。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.9886870104974497, "need_memory": true }, { "id": "zero-overlap-000078", "attribute": "常住城市", "answerable": true, "expected": "VAL-TQBY65U8", "reply": "我的常住城市是 VAL-TQBY65U8。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 1.2650369691848753, "need_memory": true }, { "id": "zero-overlap-000104", "attribute": "午餐偏好", "answerable": true, "expected": "VAL-SRFC842D", "reply": "您的午餐偏好是 VAL-SRFC842D。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 0.7423390823602677, "need_memory": true }, { "id": "zero-overlap-000281", "attribute": "咖啡口味", "answerable": true, "expected": "VAL-3GP4STZX", "reply": "关于您早上饮用的咖啡偏好哪种豆子，当前记忆中没有相关记录。已知信息仅包含：user；午餐偏好；VAL-XNN7T8EK。", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 1, "top_score": 8.515033109300843, "need_memory": true }, { "id": "zero-overlap-000106", "attribute": "咖啡口味", "answerable": false, "expected": "VAL-837DYYWV", "reply": "关于您的闹钟设置时间，当前长期记忆中未包含相关证据，无法回答。已知信息包括：user；闹钟设置时间；未知。", "correct": false, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 23, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 3.273330240117752, "need_memory": true }, { "id": "zero-overlap-000091", "attribute": "运动习惯", "answerable": true, "expected": "VAL-92MQHEDL", "reply": "您的运动习惯是 VAL-92MQHEDL。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 1.0058205354213714, "need_memory": true }, { "id": "zero-overlap-000088", "attribute": "设备型号", "answerable": true, "expected": "VAL-JY6S7EYF", "reply": "关于您使用的机器档次，当前长期记忆中未包含此信息，无法回答。已知信息仅涉及：user；常用编辑器；VAL-3KPKRM4J。", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 1, "top_score": 8.516791512356487, "need_memory": true }, { "id": "zero-overlap-000244", "attribute": "午餐偏好", "answerable": false, "expected": "VAL-2VLHC26V", "reply": "关于您服务的对象，当前长期记忆中未包含相关信息，无法回答。", "correct": false, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 23, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 0.7779876451194286, "need_memory": true }, { "id": "zero-overlap-000282", "attribute": "咖啡口味", "answerable": true, "expected": "VAL-Q2AQQMUP", "reply": "关于您早上那杯咖啡的豆子偏好，当前记忆中没有相关记录。已知信息仅包含：user；午餐偏好；VAL-5ATEHVSM。", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 1, "top_score": 8.515033109300843, "need_memory": true }, { "id": "zero-overlap-000056", "attribute": "主管姓名", "answerable": true, "expected": "VAL-VDL8H7Y9", "reply": "您的主管姓名是 user；主管姓名；VAL-VDL8H7Y9。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 1.9513123025616852, "need_memory": true }, { "id": "zero-overlap-000184", "attribute": "项目代号", "answerable": true, "expected": "VAL-3H5TQEC4", "reply": "您的团队名称是 VAL-Z98JC7BG。", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 3.191310502301591, "need_memory": true }, { "id": "zero-overlap-000198", "attribute": "手机尾号", "answerable": true, "expected": "VAL-7T3XVMVH", "reply": "您随身那台设备的末端数字是 4N，该设备的完整型号为 VAL-9MWPDS4N。", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 1, "top_score": 8.513663491707547, "need_memory": true }, { "id": "zero-overlap-000258", "attribute": "阅读工具", "answerable": true, "expected": "VAL-SHRY3GNG", "reply": "我翻书时依赖的阅读工具是 VAL-SHRY3GNG。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.9932317454522837, "need_memory": true }, { "id": "zero-overlap-000030", "attribute": "课程名称", "answerable": true, "expected": "VAL-BQY3K5EV", "reply": "您的课程名称是 VAL-BQY3K5EV。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 0.7088197509944439, "need_memory": true }, { "id": "zero-overlap-000196", "attribute": "手机尾号", "answerable": true, "expected": "VAL-ZHN2NMZV", "reply": "您随身那台设备的末端数字是 W9，该设备的完整型号为 VAL-5VBJ22W9。", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 1, "top_score": 8.513663491707547, "need_memory": true }, { "id": "zero-overlap-000268", "attribute": "宿舍楼号", "answerable": true, "expected": "VAL-QVYMWTLK", "reply": "关于您睡觉的地儿是第几排，当前长期记忆中未包含此信息。已知信息如下：user；宿舍楼号；VAL-QVYMWTLK。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.7659836388747943, "need_memory": true }, { "id": "zero-overlap-000006", "attribute": "档案标识", "answerable": true, "expected": "VAL-8EFN3KGR", "reply": "您的系统分配字符包括：紧急联系人姓氏 VAL-KLPQW2L2；课程名称 VAL-PVMQLCL9；团队名称 VAL-V4BVFB2N；客户名称 VAL-92LH7CHG", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 4.169416844133658, "need_memory": true }, { "id": "zero-overlap-000047", "attribute": "咖啡口味", "answerable": true, "expected": "VAL-PW46K9K9", "reply": "关于您早上饮用的咖啡偏好哪种豆子，当前记忆中没有相关记录。已知信息仅包含您的午餐偏好是 VAL-BFW4LVTV。", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 1, "top_score": 8.515033109300843, "need_memory": true }, { "id": "zero-overlap-000173", "attribute": "客户名称", "answerable": false, "expected": "VAL-P84HTKZR", "reply": "您随身那台设备的末端数字是 Y。\n设备型号：VAL-NARXVSSY", "correct": false, "wrong_attribute": false, "leaked": true, "records_written": 24, "records_active": 23, "stop_reason": "evidence_found", "records_selected": 1, "top_score": 8.513663491707547, "need_memory": true }, { "id": "zero-overlap-000000", "attribute": "出生城市", "answerable": true, "expected": "VAL-PHB8FNF9", "reply": "您的出生城市是 VAL-PHB8FNF9。user；出生城市；VAL-PHB8FNF9", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.8505235558902053, "need_memory": true }, { "id": "zero-overlap-000266", "attribute": "出生城市", "answerable": true, "expected": "VAL-K7FT5XFB", "reply": "您的出生城市是 VAL-K7FT5XFB。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.3051346188937454, "need_memory": true }, { "id": "zero-overlap-000051", "attribute": "项目代号", "answerable": false, "expected": "VAL-KG4RZYH5", "reply": "您的办公城市是 VAL-W9KGRD9U。", "correct": false, "wrong_attribute": false, "leaked": true, "records_written": 24, "records_active": 23, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 1.2606107878684996, "need_memory": true }, { "id": "zero-overlap-000061", "attribute": "常住城市", "answerable": true, "expected": "VAL-3SUNPNRD", "reply": "您的常住城市是 VAL-3SUNPNRD，通勤方式是 VAL-2CXTE3TB，出生城市是 VAL-GE828D22，办公城市是 VAL-XYRY8BUQ，入职年份", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 1.6213454890251158, "need_memory": true }, { "id": "zero-overlap-000214", "attribute": "客户名称", "answerable": true, "expected": "VAL-MV382EEF", "reply": "我服务的对象是 VAL-MV382EEF。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 0.7693164377659559, "need_memory": true }, { "id": "zero-overlap-000092", "attribute": "宿舍楼号", "answerable": true, "expected": "VAL-PLZZ659B", "reply": "关于您睡觉的地儿是第几排，当前长期记忆中未包含此信息。已知信息如下：user；宿舍楼号；VAL-PLZZ659B。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.8819549524530643, "need_memory": true }, { "id": "zero-overlap-000140", "attribute": "宿舍楼号", "answerable": true, "expected": "VAL-RMDSB297", "reply": "关于您睡觉的地儿是第几排，当前长期记忆中未包含此信息。已知信息如下：user；宿舍楼号；VAL-RMDSB297。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.942847951625275, "need_memory": true }, { "id": "zero-overlap-000164", "attribute": "客户名称", "answerable": true, "expected": "VAL-BB55YHW5", "reply": "您的客户名称是 VAL-BB55YHW5。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 0.77552460052073, "need_memory": true }, { "id": "zero-overlap-000213", "attribute": "通勤方式", "answerable": true, "expected": "VAL-M58DDQX2", "reply": "您的通勤方式是 VAL-M58DDQX2。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.1366783114608343, "need_memory": true }, { "id": "zero-overlap-000122", "attribute": "通勤方式", "answerable": true, "expected": "VAL-VULSNGJ3", "reply": "您的通勤方式是 VAL-VULSNGJ3。user；通勤方式；VAL-VULSNGJ3", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.130069571965557, "need_memory": true }, { "id": "zero-overlap-000168", "attribute": "常住城市", "answerable": true, "expected": "VAL-8TQ8737C", "reply": "您的常住城市是 VAL-8TQ8737C。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 1.4585979866981504, "need_memory": true }, { "id": "zero-overlap-000175", "attribute": "客户名称", "answerable": true, "expected": "VAL-QJNRRE8X", "reply": "您的客户名称是 VAL-QJNRRE8X。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 0.7712558816373348, "need_memory": true }, { "id": "zero-overlap-000144", "attribute": "入职年份", "answerable": false, "expected": "VAL-GLHQMRN6", "reply": "关于您睡觉的地儿是第几排，当前长期记忆中未包含此信息。已知信息如下：user；档案标识；VAL-AVX95V7L；手机尾号；VAL-LBRPQJRT", "correct": false, "wrong_attribute": false, "leaked": true, "records_written": 24, "records_active": 23, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.8524921381220096, "need_memory": true }, { "id": "zero-overlap-000002", "attribute": "午餐偏好", "answerable": true, "expected": "VAL-HEE2QMH2", "reply": "您的午餐偏好是 user；午餐偏好；VAL-HEE2QMH2。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 0.7196446585655213, "need_memory": true }, { "id": "zero-overlap-000171", "attribute": "常住城市", "answerable": true, "expected": "VAL-KZA246K9", "reply": "您的常住城市是 VAL-KZA246K9。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 1.286561994552612, "need_memory": true }, { "id": "zero-overlap-000154", "attribute": "入职年份", "answerable": true, "expected": "VAL-7Z9CBUX8", "reply": "关于您何时开始在这家公司上班，当前记忆中没有相关记录。已知信息：user；办公城市；VAL-BP7253WL。", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 1, "top_score": 8.51429933919514, "need_memory": true }, { "id": "zero-overlap-000234", "attribute": "工位楼层", "answerable": true, "expected": "VAL-87M7HVC7", "reply": "关于您坐在哪一间房的信息，当前长期记忆中未包含该数据。已知信息如下：user；宿舍楼号；VAL-JTSHA7PM。", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 3.1937497879965546, "need_memory": true }, { "id": "zero-overlap-000083", "attribute": "办公城市", "answerable": true, "expected": "VAL-YBVYV97X", "reply": "您的办公城市是 VAL-YBVYV97X。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.8081361769552853, "need_memory": true }, { "id": "zero-overlap-000079", "attribute": "出生城市", "answerable": true, "expected": "VAL-7R8WVAFG", "reply": "您的出生城市是 VAL-7R8WVAFG。", "correct": true, "wrong_attribute": false, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 2.7214163786326675, "need_memory": true }, { "id": "zero-overlap-000180", "attribute": "客户名称", "answerable": false, "expected": "VAL-3F87H6UT", "reply": "您随身那台设备的末端数字是 VAL-X5VZ48QF。", "correct": false, "wrong_attribute": false, "leaked": true, "records_written": 24, "records_active": 23, "stop_reason": "evidence_found", "records_selected": 1, "top_score": 8.513663491707547, "need_memory": true }, { "id": "zero-overlap-000025", "attribute": "紧急联系人姓氏", "answerable": true, "expected": "VAL-STECQ3PX", "reply": "您的默认语言是 VAL-PRVTUCP7，课程名称是 VAL-93ME7BJ5，出生城市是 VAL-KYBRBBKV，起床时间是 VAL-PJC3D3W9，午餐偏好是", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 8.516015553315961, "need_memory": true }, { "id": "zero-overlap-000187", "attribute": "手机尾号", "answerable": true, "expected": "VAL-PXRHJ49F", "reply": "您随身那台设备的末端数字是 H9，该设备的完整型号为 VAL-ZPUPV9H9。", "correct": false, "wrong_attribute": true, "leaked": false, "records_written": 24, "records_active": 24, "stop_reason": "evidence_found", "records_selected": 1, "top_score": 8.513663491707547, "need_memory": true }, { "id": "zero-overlap-000045", "attribute": "办公城市", "answerable": false, "expected": "VAL-CG4X6ZWD", "reply": "您的办公城市是 VAL-SZ4R5GT3。", "correct": false, "wrong_attribute": false, "leaked": true, "records_written": 24, "records_active": 23, "stop_reason": "evidence_found", "records_selected": 8, "top_score": 1.1886139500141142, "need_memory": true } ] } }