Datasets:
scenario stringclasses 1
value | published_split stringclasses 1
value | source_split stringclasses 1
value | evaluation_type stringclasses 1
value | query_id stringlengths 15 102 | base_query_id stringclasses 0
values | is_validation bool 1
class | domain stringclasses 18
values | source_dataset stringclasses 18
values | source_sample_id stringlengths 15 102 | variant_type stringclasses 1
value | query_text large_stringlengths 7 39k | input_json large_stringlengths 18 39.7k | media_json large_stringclasses 1
value | provenance_json large_stringclasses 24
values | embedding_model stringclasses 1
value | embedding list | model_ids listlengths 10 10 | performance listlengths 10 10 | cost_cny listlengths 10 10 | latency_ms listlengths 10 10 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
chinese | train | train | train | AlignBench-generation_open-100 | null | false | AlignBench | AlignBench | AlignBench-generation_open-100 | original | 一条宽度为10m,深度为2m的矩形河道中,水以2m/s的速度流动。假设水的密度ρ为1000kg/m³,请给出解答：
a) 河道中水的流量,b) 河道中水的动能,c) 河道中水的总动能。 | {"text":"一条宽度为10m,深度为2m的矩形河道中,水以2m/s的速度流动。假设水的密度ρ为1000kg/m³,请给出解答：\n a) 河道中水的流量,b) 河道中水的动能,c) 河道中水的总动能。"} | [] | {"benchmark":"AlignBench","task_category":["Generation"],"kind":"original"} | BAAI/bge-m3 | [
0.0034384168684482574,
0.037582479417324066,
-0.05768416076898575,
0.0062284269370138645,
0.007879799231886864,
-0.018526552245020866,
-0.044909022748470306,
-0.020905958488583565,
0.007211013697087765,
-0.007465955335646868,
-0.01676052249968052,
-0.011413704603910446,
0.008750895038247108,... | [
"deepseek/deepseek-v4-flash",
"deepseek/deepseek-v4-pro",
"minimax/minimax-m2.7",
"minimax/minimax-m3",
"moonshot/kimi-k3",
"qwen/qwen3.7-max",
"qwen/qwen3.7-plus",
"qwen/qwen3.8-max",
"z-ai/glm-5.2",
"z-ai/glm-5.3"
] | [
0.699999988079071,
0.800000011920929,
0.800000011920929,
0.699999988079071,
0.800000011920929,
0.6000000238418579,
0.8999999761581421,
1,
1,
1
] | [
0.042561,
0.179253,
0.0306957,
0.1261512,
0.17584,
0.11856,
0.029896,
0.17112,
0.12412,
0.286576
] | [
1420413.646,
1038896.002,
4411102.973,
458766.092,
4403567.378,
3302731.031,
4892089.8889999995,
2981129.141,
5731107.698,
15529674.615
] |
chinese | train | train | train | AlignBench-generation_open-101 | null | false | AlignBench | AlignBench | AlignBench-generation_open-101 | original | 下面是一道多选题，它包含多个正确选项，请仔细查看选项并直接给出所有正确的答案。
哪些因素会影响物体在水中的浮力？
A. 物体的质量
B. 物体的体积
C. 水的密度 | {"text":"下面是一道多选题，它包含多个正确选项，请仔细查看选项并直接给出所有正确的答案。\n 哪些因素会影响物体在水中的浮力？\n A. 物体的质量\n B. 物体的体积\n C. 水的密度"} | [] | {"benchmark":"AlignBench","task_category":["Generation"],"kind":"original"} | BAAI/bge-m3 | [
0.01193390041589737,
0.04081503674387932,
-0.06803224980831146,
-0.00573762459680438,
-0.010950729250907898,
-0.03551159426569939,
-0.03350245580077171,
0.031539786607027054,
0.012936996296048164,
-0.00961801502853632,
-0.02431526966392994,
0.009463091380894184,
-0.008843454532325268,
-0.0... | [
"deepseek/deepseek-v4-flash",
"deepseek/deepseek-v4-pro",
"minimax/minimax-m2.7",
"minimax/minimax-m3",
"moonshot/kimi-k3",
"qwen/qwen3.7-max",
"qwen/qwen3.7-plus",
"qwen/qwen3.8-max",
"z-ai/glm-5.2",
"z-ai/glm-5.3"
] | [
1,
0.8999999761581421,
1,
0.8999999761581421,
0.699999988079071,
0.699999988079071,
1,
0.699999988079071,
0.699999988079071,
0.8999999761581421
] | [
0.003372,
0.008091,
0.0030093,
0.0067872,
0.02404,
0.042012,
0.015686,
0.009504,
0.033332,
0.045204
] | [
1313172.115,
310968.62700000004,
4144054.6969999997,
367965.716,
4131023.421,
3140027.25,
4722989.0770000005,
2710936.975,
5557699.068,
14839562.978
] |
chinese | train | train | train | AlignBench-generation_open-102 | null | false | AlignBench | AlignBench | AlignBench-generation_open-102 | original | 世界最大的海港是哪个海港？ | {"text":"世界最大的海港是哪个海港？"} | [] | {"benchmark":"AlignBench","task_category":["Generation"],"kind":"original"} | BAAI/bge-m3 | [
0.020801996812224388,
0.02039501816034317,
-0.026009758934378624,
-0.008262568153440952,
-0.016737347468733788,
0.010994532145559788,
-0.037320587784051895,
-0.04788618162274361,
0.02067515254020691,
-0.0007475424208678305,
0.014601526781916618,
0.026821594685316086,
-0.0033523410093039274,
... | [
"deepseek/deepseek-v4-flash",
"deepseek/deepseek-v4-pro",
"minimax/minimax-m2.7",
"minimax/minimax-m3",
"moonshot/kimi-k3",
"qwen/qwen3.7-max",
"qwen/qwen3.7-plus",
"qwen/qwen3.8-max",
"z-ai/glm-5.2",
"z-ai/glm-5.3"
] | [
1,
1,
0.5,
1,
1,
0.8999999761581421,
1,
0.8999999761581421,
1,
1
] | [
0.003207,
0.015021,
0.0130074,
0.0065058,
0.06512,
0.055848,
0.009016,
0.011064,
0.029624,
0.021196
] | [
1401770.131,
4014746.378,
4529121.095000001,
445400.78599999996,
4551499.664,
3388445.228,
4949277.829,
3008104.0429999996,
5780302.337,
1126739.1230000001
] |
chinese | train | train | train | AlignBench-generation_open-104 | null | false | AlignBench | AlignBench | AlignBench-generation_open-104 | original | 下面是一道多选题，它包含多个正确选项，请仔细查看选项并直接给出所有正确的答案。
下面哪些是海水淡化方法（）
A、蒸馏法
B、反渗透法
C、水合物法
D、冰冻法 | {"text":"下面是一道多选题，它包含多个正确选项，请仔细查看选项并直接给出所有正确的答案。\n 下面哪些是海水淡化方法（）\n A、蒸馏法\n B、反渗透法\n C、水合物法\n D、冰冻法"} | [] | {"benchmark":"AlignBench","task_category":["Generation"],"kind":"original"} | BAAI/bge-m3 | [
0.012650148943066597,
0.027920974418520927,
-0.03380533680319786,
-0.006213318090885878,
-0.013992627151310444,
-0.018807945773005486,
0.023878643289208412,
0.014611367136240005,
0.013833132572472095,
0.024382293224334717,
-0.017722269520163536,
-0.013114726170897484,
0.02784876711666584,
... | [
"deepseek/deepseek-v4-flash",
"deepseek/deepseek-v4-pro",
"minimax/minimax-m2.7",
"minimax/minimax-m3",
"moonshot/kimi-k3",
"qwen/qwen3.7-max",
"qwen/qwen3.7-plus",
"qwen/qwen3.8-max",
"z-ai/glm-5.2",
"z-ai/glm-5.3"
] | [
0.5,
0.4000000059604645,
0.30000001192092896,
0.5,
0.5,
0.5,
0.6000000238418579,
0.4000000059604645,
0.4000000059604645,
0.4000000059604645
] | [
0.001341,
0.003159,
0.006006,
0.0058338,
0.06698,
0.0162,
0.004756,
0.006336,
0.009932,
0.018024
] | [
1299540.303,
3731790.282,
4132454.758,
361997.387,
4121603.852,
3115158.132,
4680565.549,
2690109.809,
5525870.468,
14707116.908
] |
chinese | train | train | train | AlignBench-generation_open-105 | null | false | AlignBench | AlignBench | AlignBench-generation_open-105 | original | 下面是一道多选题，它包含多个正确选项，请仔细查看选项并直接给出所有正确的答案。
第一次和第二次世界大战爆发的导火线事件是？
A. 萨拉热窝事件
B. 波茨坦会议
C. 波兰偷袭
D. 孤立政策 | {"text":"下面是一道多选题，它包含多个正确选项，请仔细查看选项并直接给出所有正确的答案。\n 第一次和第二次世界大战爆发的导火线事件是？\n A. 萨拉热窝事件\n B. 波茨坦会议\n C. 波兰偷袭\n D. 孤立政策"} | [] | {"benchmark":"AlignBench","task_category":["Generation"],"kind":"original"} | BAAI/bge-m3 | [
0.06774086505174637,
0.001862619537860155,
-0.0658533126115799,
0.036766137927770615,
-0.018232813104987144,
-0.02153130993247032,
-0.009020081721246243,
-0.006530591286718845,
0.017466183751821518,
0.011050984263420105,
0.027185000479221344,
-0.002260406967252493,
0.005167292430996895,
-0... | [
"deepseek/deepseek-v4-flash",
"deepseek/deepseek-v4-pro",
"minimax/minimax-m2.7",
"minimax/minimax-m3",
"moonshot/kimi-k3",
"qwen/qwen3.7-max",
"qwen/qwen3.7-plus",
"qwen/qwen3.8-max",
"z-ai/glm-5.2",
"z-ai/glm-5.3"
] | [
1,
0.8999999761581421,
1,
0.8999999761581421,
1,
0.8999999761581421,
0.8999999761581421,
1,
0.8999999761581421,
1
] | [
0.007851,
0.012753,
0.0046704,
0.0088074,
0.04216,
0.027288,
0.006718,
0.01962,
0.05088,
0.052476
] | [
1362337.3180000002,
3885440.42,
4311534.735,
409789.778,
4317438.889,
3251480.974,
4821066.237,
2864046.638,
5682200.477,
15311875.312
] |
chinese | train | train | train | AlignBench-generation_open-107 | null | false | AlignBench | AlignBench | AlignBench-generation_open-107 | original | 板块构造学说是谁提出的？主要观点是什么？ | {"text":"板块构造学说是谁提出的？主要观点是什么？"} | [] | {"benchmark":"AlignBench","task_category":["Generation"],"kind":"original"} | BAAI/bge-m3 | [
-0.026792019605636597,
-0.0051191444508731365,
-0.013297738507390022,
0.03367234393954277,
0.016255715861916542,
-0.006690606940537691,
-0.03842083364725113,
-0.006047685164958239,
0.0037578302435576916,
-0.011371535249054432,
-0.006634525954723358,
0.04089149087667465,
-0.025997359305620193... | [
"deepseek/deepseek-v4-flash",
"deepseek/deepseek-v4-pro",
"minimax/minimax-m2.7",
"minimax/minimax-m3",
"moonshot/kimi-k3",
"qwen/qwen3.7-max",
"qwen/qwen3.7-plus",
"qwen/qwen3.8-max",
"z-ai/glm-5.2",
"z-ai/glm-5.3"
] | [
0.8999999761581421,
0.8999999761581421,
1,
1,
0.8999999761581421,
0.8999999761581421,
0.8999999761581421,
0.800000011920929,
1,
0.8999999761581421
] | [
0.00639,
0.031806,
0.0137466,
0.0147798,
0.26582,
0.100152,
0.017498,
0.042588,
0.054336,
0.083316
] | [
1335168.621,
3797431.637,
4251190.847,
390591.143,
4265622.058,
3202605.1659999997,
4769510.926,
2773821.5409999997,
5613221.361,
98581.959
] |
chinese | train | train | train | AlignBench-generation_open-108 | null | false | AlignBench | AlignBench | AlignBench-generation_open-108 | original | 惟知跃进是谁的口号？ | {"text":"惟知跃进是谁的口号？"} | [] | {"benchmark":"AlignBench","task_category":["Generation"],"kind":"original"} | BAAI/bge-m3 | [
-0.026676656678318977,
0.00369324185885489,
-0.006559711880981922,
0.0011632699752226472,
-0.005029201973229647,
-0.06069295108318329,
-0.005657801870256662,
0.013673890382051468,
-0.031810320913791656,
0.004280693829059601,
-0.026744535192847252,
-0.01820833794772625,
-0.027271291241049767,... | [
"deepseek/deepseek-v4-flash",
"deepseek/deepseek-v4-pro",
"minimax/minimax-m2.7",
"minimax/minimax-m3",
"moonshot/kimi-k3",
"qwen/qwen3.7-max",
"qwen/qwen3.7-plus",
"qwen/qwen3.8-max",
"z-ai/glm-5.2",
"z-ai/glm-5.3"
] | [
1,
0.8999999761581421,
0.30000001192092896,
0.30000001192092896,
0.8999999761581421,
0.30000001192092896,
0.20000000298023224,
0.30000001192092896,
1,
0.10000000149011612
] | [
0.001605,
0.003195,
0.0014385,
0.0083832,
0.07252,
0.09762,
0.032458,
0.0483,
0.020544,
0.008784
] | [
1323292.5210000002,
3746976.3680000002,
4163583.0340000005,
379213.196,
4180928.7830000003,
3196022.608,
91555.526,
2761332.265,
5587132.19,
14879066.290000001
] |
chinese | train | train | train | AlignBench-generation_open-109 | null | false | AlignBench | AlignBench | AlignBench-generation_open-109 | original | 什么是超新星？ | {"text":"什么是超新星？"} | [] | {"benchmark":"AlignBench","task_category":["Generation"],"kind":"original"} | BAAI/bge-m3 | [
-0.024349210783839226,
-0.03600218892097473,
-0.056103359907865524,
0.03659119829535484,
0.016003184020519257,
-0.036291297525167465,
-0.030122457072138786,
0.027094734832644463,
0.004578980151563883,
-0.0076813772320747375,
-0.022359661757946014,
0.03168077394366264,
-0.008781712502241135,
... | [
"deepseek/deepseek-v4-flash",
"deepseek/deepseek-v4-pro",
"minimax/minimax-m2.7",
"minimax/minimax-m3",
"moonshot/kimi-k3",
"qwen/qwen3.7-max",
"qwen/qwen3.7-plus",
"qwen/qwen3.8-max",
"z-ai/glm-5.2",
"z-ai/glm-5.3"
] | [
0.800000011920929,
1,
0.800000011920929,
1,
1,
0.800000011920929,
1,
0.8999999761581421,
0.800000011920929,
1
] | [
0.011835,
0.027864,
0.0190218,
0.0135618,
0.11558,
0.063564,
0.013428,
0.013164,
0.055288,
0.03258
] | [
1394748.411,
3968260.1169999996,
4470372.277000001,
435560.527,
4489060.470000001,
3346197.184,
4912531.79,
2947457.983,
497436.361,
15652300.233
] |
chinese | train | train | train | AlignBench-generation_open-11 | null | false | AlignBench | AlignBench | AlignBench-generation_open-11 | original | "法国、英国、西班牙、瑞士、德国、意大利、荷兰、比利时，其中任意2个(...TRUNCATED) | "{\"text\":\"法国、英国、西班牙、瑞士、德国、意大利、荷兰、比利时，其中(...TRUNCATED) | [] | {"benchmark":"AlignBench","task_category":["Generation"],"kind":"original"} | BAAI/bge-m3 | [-0.014275794848799706,0.029666796326637268,-0.011096927337348461,-0.027701865881681442,-0.010627190(...TRUNCATED) | ["deepseek/deepseek-v4-flash","deepseek/deepseek-v4-pro","minimax/minimax-m2.7","minimax/minimax-m3"(...TRUNCATED) | [
0.699999988079071,
0.699999988079071,
1,
1,
1,
0.8999999761581421,
1,
1,
1,
1
] | [
0.007893,
0.004752,
0.0160482,
0.0188958,
0.1075,
0.059496,
0.009102,
0.03246,
0.040064,
0.035556
] | [1351429.14,3832811.867,4321172.358,408035.02099999995,4292436.583,3227586.6289999997,4794786.917,28(...TRUNCATED) |
chinese | train | train | train | AlignBench-generation_open-111 | null | false | AlignBench | AlignBench | AlignBench-generation_open-111 | original | "有哪些在家庭经济学领域擅长理论模型推导、在核心期刊上发表过高被引文(...TRUNCATED) | "{\"text\":\"有哪些在家庭经济学领域擅长理论模型推导、在核心期刊上发表过(...TRUNCATED) | [] | {"benchmark":"AlignBench","task_category":["Generation"],"kind":"original"} | BAAI/bge-m3 | [-0.03772673010826111,0.0011747878743335605,-0.05109231919050217,-0.006364902015775442,0.02295780926(...TRUNCATED) | ["deepseek/deepseek-v4-flash","deepseek/deepseek-v4-pro","minimax/minimax-m2.7","minimax/minimax-m3"(...TRUNCATED) | [0.800000011920929,0.8999999761581421,0.6000000238418579,0.5,0.800000011920929,1.0,0.899999976158142(...TRUNCATED) | [
0.016542,
0.098631,
0.0118818,
0.0058926,
1.62962,
0.157272,
0.042778,
0.590784,
0.116828,
0.033152
] | [1430586.1930000002,414549.77999999997,3518504.119,460142.48,1889575.321,3481438.6909999996,5062656.(...TRUNCATED) |
End of preview. Expand in Data Studio
RobustRouteBench
RobustRouteBench is a unified public benchmark for general-text, multimodal, and Chinese-language model routing.
from datasets import load_dataset
text = load_dataset("SinapisAI/RobustRouteBench", "text")
multimodal = load_dataset("SinapisAI/RobustRouteBench", "multimodal")
chinese = load_dataset("SinapisAI/RobustRouteBench", "chinese")
Each scenario exposes separate train, validation, and test splits. The
500-row validation split is a convenience overlay containing 400 standard and
100 robust queries. Those rows intentionally remain in the complete test split,
so validation and test overlap by design.
The test split contains both standard and robustness evaluation rows. Use
evaluation_type (standard or robust) and variant_type to report slices.
Each row is query-centric. performance, cost_cny, and latency_ms are
aligned to model_ids. All costs use CNY; historical USD proxies use the frozen
benchmark conversion 1 USD = 7.2 CNY.
Source outcomes that were not successful have performance zero. No source diagnostic payloads or raw model responses are published. Official BGE-M3 embeddings are included as baseline features but alternative encoders are allowed.
The repository is self-contained. All MMR-4000 training images and all standard
and robustness evaluation media are stored under media/multimodal/; no other
dataset repository is required to train or evaluate a router. See LICENSES.md
for upstream attribution and terms. This compilation uses license: other; no
blanket license is asserted over all upstream questions or images.
- Downloads last month
- 40