evalstone/P800模型能力评测结果 - DS4-Flash-INT8-NO-Thinking-2.0-FULL.csv

1	分类	Benchmark	得分	实测时间(h)	总样本数	延迟_mean(s)	输出TPS	请求QPS	输入tokens_mean	输出tokens_mean	累计总tokens	TTFT_mean(s)	TTFT P90	TTFT P99	TPOT_mean(s)	TPOT P90	TPOT P99
2	代码与工程	bigcodebench	0.9956	0.9915	1140	10.15508	32.99	0.0985	151.77	335.03	554946	0.26441	0.27379	0.42952	0.02957	0.03059	0.03143
3		humaneval	0.9044	0.1183	164	2.776502	42.58667	0.3602	163.0793	118.2398	45994	0.25521	0.304186	0.451449	0.021783	0.024498	0.027453
4		live_code_bench	0.6756	7.9486	100	10.41723	43.578	0.09644	518.1118	453.5784	1040894	0.253597	0.272382	0.464771	0.02093	0.023769	0.026398
5	推理与数学	aime24	0.6167	1.2217	30	42.44225	44.85	0.024108	119.3333	1901.842	55169	0.258426	0.274508	0.444511	0.021698	0.025218	hle
6		aime25	0.4639	2.0414	30	59.98314	46.03583	0.017092	188.4333	2761.283	80482	0.257489	0.273448	0.4417	0.021655	0.024778	0.026626
7		aime26	0.5611	1.4847	30	51.64098	44.66417	0.019925	139.7667	2305.069	60102	0.25465	0.268527	0.435013	0.021667	0.024596	0.026089
8		hmmt26	0.4015	1.5481	30	50.8562	44.73	0.020008	108.4242	2274.237	63775	0.255456	0.266018	0.431919	0.022514	0.025679	0.027263
9		imo_answerbench	0.36	1.2508	400	39.6005	44.94	0.0253	122.1	1779.8	753159	0.2466	0.2641	0.3167	0.0226	0.0257	0.0285
10		hle	0.0508	14.6863	3000	61.787217	34.42	0.0162	292.3252	2126.7916	6047792	0.275126	0.282138	0.474372	0.028997	0.029831	0.031026
11		gsm8k	0.9704	0.3178	1319	3.031098	43.8	0.3299	582.9507	132.7726	944039	0.253185	0.267307	0.45469	0.021127	0.02357	0.025568
12		competition_math	0.941	3.1817	500	8.681756	51	0.1152	296.5698	442.732	3696509	0.245456	0.263328	0.435642	0.019541	0.021677	0.024484
13		bbh	0.9032	1.8036	6513	3.561695	42.95	0.2808	916.98	152.9719	6966457	0.269139	0.276913	0.467974	0.023935	0.030921	0.065977
14		drop	0.8183	1.5336	9536	1.824962	36.31	0.548	1273.607	66.25849	12776955	0.297489	0.4593	0.505599	0.023644	0.028072	0.034489
15	知识与语言理解	gpqa_diamond	0.6944	0.3164	198	10.54948	41.335	0.0948	252.4343	436.0202	135770	0.245972	0.26274	0.456931	0.023819	0.027653	0.030237
16		mmlu_pro	0.8285	5.6167	12032	6.264368	31.05	0.1596	1312.528	194.5092	18132672	0.293428	0.437219	0.477191	0.031274	0.034267	0.036824
17		simple_qa	0.3814	1.9014	4326	1.88825	30.69	0.529590891	30.351826	57.955155	382016	0.26334	0.269454	0.452375	0.027793	0.031603	0.035322
18		mmlu	0.9045	3.0297	14042	2.6957	35.98	0.371	727.6	97	11532337	0.2643	0.2751	0.4663	0.026	0.0305	0.0348
19		cmmlu	0.9001	3.4935	11515	3.9225	36.65	0.2549	115.3	143.8	2983005	0.2553	0.2956	0.443	0.0266	0.0307	0.0353
20		arc	0.9476	0.2103	1172	0.662452	13.65	1.5095	102.0333	9.042841	394098	0.321567	0.437672	0.449777	0.066642	0.129754	0.132577
21		hellaswag	0.8666	0.4453	10042	0.597052	7.32	1.6749	219.8686	4.370444	2251808	0.341421	0.448449	0.461157	0.081246	0.133338	0.136106
22		trivia_qa	0.7768	5.3011	11313	9.163784	3.78	0.1091	11964.972	34.6152	95912700	4.722344	8.951851	16.512039	0.131478	0.292489	0.622806
23		winogrande	0.7845	0.0775	1267	0.695127	13.88	1.4386	76.11997	9.646409	108666	0.318495	0.433513	0.438349	0.069826	0.129328	0.130936
24	长上下文	longbench_v2	0.5686	3.0638	503	84.120374	2.75	0.0119	85638.8191	231.6441	43192843	25.148194	53.202063	64.267424	0.248375	0.534405	0.946107
25		openai_mrcr	0.7768	5.2828	2399	29.1532	12.95	0.0343	24774.8	377.6	60340773	6.6088	21.0867	37.983	0.0608	0.1125	0.2004
26	智能体与工具	tau2_bench	0.7684	1.7625
27		general_fc	0.6988	1.3803	2000	9.4082	39.21	0.1063	1619.2	368.9	3976219	0.4226	0.7249	1.2941	0.026	0.0307	0.0398
28		bfcl_v3	0.6643	5.23	4441	4.296605	25.32	0.2327	7268.4069	108.7864	97061732	0.418217	0.521432	1.106471	0.035785	0.039796	0.053177
29		总计	0.711992593	75.2394