OpenLLMLeaderboard之DataLearner备份

大模型评测得分排行榜Open LLM Leaderboard中国站

为了方便大家更便捷查询，DataLearnerAI发布了DataLearnerAI-GPT：目前已经支持基于OpenLLMLeaderboard数据回答任意大模型评测结果数据地址如下：

https://chat.openai.com/g/g-8eu9KgtUm-datalearnerai-gpt

关于DataLearnerAI-GPT的详细介绍参考：https://www.datalearner.com/blog/1051699757266256

随着大量大型语言模型（LLMs）和聊天机器人每周都在发布，它们往往伴随着对性能的夸大宣称，要筛选出由开源社区所取得的真正进展以及哪个模型是当前的技术领先水平，可能会非常困难。

为此，HF推出了这个大模型开放评测追踪排行榜。📐 🤗 Open LLM Leaderboard 旨在追踪、排名和评估开源大型语言模型（LLMs）和聊天机器人在不同评测任务上的得分。

由于HuggingFace的访问稳定性和速度，我们提供了同步更新的结果。原网页请访问：https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard

Open LLM Leaderboard排行榜的各个评测任务介绍

AI2 Reasoning Challenge (25-shot)

一套小学科学问题。
HellaSwag (10-shot)

对于人类而言简单（大约95%）的常识推理测试，但对于最新技术模型而言具有挑战性。
MMLU (5-shot)

测试文本模型的多任务准确性，涵盖57项任务，包括小学数学、美国历史、计算机科学、法律等。
TruthfulQA (0-shot)

测试模型复制网络上常见虚假信息的倾向。注意：工具中的 TruthfulQA 实际上至少是6次尝试的任务。
Winogrande (5-shot)

大规模的、具有对抗性的、困难的 Winograd 基准测试，用于常识推理。
GSM8k (5-shot)

多样化的小学数学文字问题，用于测试模型解决多步骤数学推理问题的能力。
DROP (3-shot)

要求对段落内容进行离散推理的英语阅读理解基准测试。

下表中关于模型类型的图标解释如下：

🟢 ：预训练模型：新的、基础模型，针对给定的语料库进行训练。

🔶 ：微调模型：对预训练模型在更多数据上进行微调，例如更适合聊天的微调结果。

⭕ ：指令微调模型：指令微调，是特别针对任务指令数据集进行微调的模型。

🟦 ：强化学习微调模型：强化微调，通常会通过增加策略稍微改变模型的损失函数。

❓：表示未知

模型名称	模型类型	参数大小（亿）	平均分	ARC分数	Hellaswag分数	MMLU分数	TruthfulQA分数	Winogrande分数	GSM8K分数	DROP分数
dolphin-2.1-mistral-7b	⭕	71	53.47	64.42	84.92	63.32	55.56	77.74	20.77	7.56
speechless-code-mistral-7b-v1.0	🔶	71	53.47	60.58	83.75	62.98	47.9	78.69	19.18	21.19
based-30b	🔶	323	53.46	63.91	85.67	58.28	35.7	80.11	0.3	50.22
Wizard-Vicuna-30B-Uncensored-fp16	🔶	323	53.44	62.12	83.45	58.24	50.81	78.45	14.25	26.74
Wizard-Vicuna-30B-Uncensored	🔶	323	53.44	62.12	83.45	58.24	50.81	78.45	14.25	26.74
SlimOpenOrca-Mistral-7B	🔶	72	53.43	62.97	83.49	62.3	57.39	77.43	21.46	9.01
Chat-AYB-Platypus2-13B	⭕	130	53.39	60.49	84.03	57.83	54.52	75.77	2.96	38.12
dolphin-2.1-mistral-7b	🔶	71	53.37	63.99	85.0	63.44	55.57	77.9	20.09	7.61
Mistral-7B-SlimOrca	🔶	71	53.34	62.54	83.86	62.77	54.23	77.43	21.38	11.2
speechless-mistral-dolphin-orca-platypus-samantha-7b	❓	0	53.34	64.33	84.4	63.72	52.52	78.37	21.38	8.66
30B-Lazarus	🔶	323	53.33	64.93	84.27	56.47	58.65	78.37	7.73	22.9
MLewd-ReMM-L2-Chat-20B	🔶	200	53.33	62.46	85.62	59.13	55.63	77.19	10.92	22.33
speechless-hermes-coig-lite-13b	🔶	130	53.31	59.47	82.28	55.18	47.6	78.61	10.77	39.25
llama-65b	🟢	653	53.31	63.48	86.09	63.93	43.43	82.56	27.67	5.98
Mistral-7B-Instruct-v0.1	⭕	71	53.27	54.52	75.63	55.38	56.28	73.72	14.25	43.1
Pygmalion-2-13b-SuperCOT	🔶	130	53.27	63.23	83.68	54.9	53.14	77.51	6.29	34.13
OpenRP-13B	🔶	128	53.25	62.12	82.6	57.5	48.29	76.01	12.89	33.38
speechless-hermes-coig-lite-13b	🔶	130	53.22	59.56	82.26	55.3	47.56	78.53	9.86	39.5
oasst-rlhf-2-llama-30b-7k-steps-hf	🟦	323	53.18	61.35	83.8	57.89	51.18	78.77	31.46	7.78
chinese-alpaca-33b-merged	❓	324	53.09	59.3	78.43	57.69	52.45	76.09	8.04	39.67
Mistral-11B-TestBench9	🔶	107	53.06	64.08	84.24	64.0	56.19	78.45	16.15	8.35
Yi-6B	🟢	60	53.06	55.55	76.42	63.85	41.86	73.8	12.66	47.32
llama-30b-supercot	🔶	323	53.06	64.85	85.08	56.56	53.96	80.03	11.9	19.07
Llama-chat-AY-13B	🔶	130	53.04	62.8	83.23	60.01	55.95	75.93	12.13	21.25
LlongOrca-13B-16k	🔶	128	53.02	62.46	82.75	55.54	50.11	76.4	12.28	31.59
Mistral-11B-TestBench11	🔶	107	53.01	64.42	83.93	63.82	56.68	77.74	14.94	9.57
Llama2-chat-AYB-13B	🔶	130	53.01	63.4	84.79	59.34	55.62	76.24	11.3	20.41
Dolphin2.1-OpenOrca-7B	❓	0	53.0	63.91	84.26	62.66	53.84	78.22	19.94	8.17
chinese-alpaca-2-13b	🔶	130	53.0	58.7	79.74	55.1	50.22	75.69	10.46	41.06
Huginn-13b-FP16	🔶	128	52.97	60.58	82.53	53.71	54.46	73.72	4.32	41.44
Dolphin2.1-OpenOrca-7B	❓	0	52.92	64.16	84.25	62.7	53.83	77.66	19.71	8.15
MythoMix-Platypus2-13B-QLoRA-0.80-epoch	⭕	130	52.91	60.32	83.72	55.74	52.18	75.53	0.91	41.98
Nous-Hermes-Platypus2-13B-QLoRA-0.80-epoch	⭕	130	52.89	59.9	83.29	56.69	51.08	75.22	1.44	42.65
Samantha-Nebula-7B	⭕	72	52.87	57.0	82.25	54.21	49.58	73.09	11.37	42.57
UltraLM-13b-v2.0	🔶	128	52.85	62.63	81.49	56.17	49.48	76.48	10.99	32.69
TekniumAiroboros-Nebula-7B	⭕	72	52.82	57.17	81.72	55.25	51.64	73.24	9.4	41.33
Libra-19B	⭕	192	52.8	60.58	82.04	55.57	48.41	76.32	0.08	46.63
llama-2-13b-FINETUNE5_4w-r8-q_k_v_o	🔶	128	52.79	57.25	81.73	55.72	41.53	77.58	14.03	41.7
ReMM-Mistral-13B	🔶	128	52.76	62.2	83.82	55.43	53.32	74.51	12.05	27.96
llama-30b-2048-instruct-PL-lora_unload	🔶	323	52.75	63.82	84.7	61.49	52.49	79.79	17.89	9.08
vigogne-2-13b-instruct	🔶	130	52.74	61.18	83.25	55.92	51.08	77.35	2.05	38.37
MLewd-Chat-v2-13B	🔶	128	52.72	61.86	83.81	57.0	54.51	75.77	10.46	25.63
llama-2-13b-Beluga-QLoRA	⭕	128	52.7	59.22	81.92	56.67	48.23	77.19	1.29	44.41
trurl-2-13b-academic	⭕	128	52.7	57.94	79.55	55.2	43.46	76.56	10.92	45.28
Athena-v1	🔶	130	52.68	60.07	82.64	55.61	46.58	74.82	4.93	44.11
speechless-orca-platypus-coig-lite-4k-0.6e-13b	🔶	128	52.65	58.79	79.93	56.77	48.29	75.93	4.25	44.59
ReMM-L2-13B	🔶	130	52.58	59.73	83.1	54.11	49.94	74.51	2.96	43.7
ReMM-L2-13B-PIPPA	🔶	128	52.58	59.73	83.12	54.1	49.94	74.51	2.96	43.69
UndiMix-v1-13b	🔶	130	52.56	59.47	82.45	55.83	49.78	75.45	10.01	34.95
Llama-2-70B-chat-GPTQ	🔶	728	52.56	62.63	84.81	62.74	50.98	78.69	18.65	9.4
llama-30b-instruct-2048-PL-lora	🔶	323	52.55	63.31	84.66	61.66	53.35	79.08	16.83	8.94
CollectiveCognition-v1-Mistral-7B	⭕	71	52.55	62.37	85.5	62.76	54.48	77.58	17.89	7.22
airoboros-33b-gpt4	❓	323	52.54	63.74	84.87	58.54	47.06	77.03	12.66	23.9
speechless-llama2-dolphin-orca-platypus-13b	🔶	128	52.54	59.64	82.65	57.9	43.44	77.19	9.7	37.24
Slerpeno	🔶	130	52.5	61.69	84.1	56.77	48.05	76.4	12.51	28.0
zephyr-7b-alpha	🔶	71	52.4	61.01	84.04	61.39	57.9	78.61	14.03	9.82
magpie-13b	🔶	128	52.37	63.31	84.25	58.15	49.15	76.48	14.48	20.78
Nous-Hermes-Llama2-13b	🔶	128	52.35	61.52	83.29	55.11	50.38	75.45	10.08	30.61
WizardLM-30B-Uncensored	🔶	323	52.32	60.24	82.93	56.8	51.57	74.35	12.89	27.45
llama-2-13b-huangyt_Fintune_1_17w-q_k_v_o_proj	🔶	128	52.29	59.73	81.06	54.53	38.64	78.14	14.03	39.9
firefly-llama2-13b	🔶	128	52.25	59.13	81.99	55.49	51.57	74.66	11.22	31.69
orca_mini_v3_13b	🔶	128	52.23	63.14	82.35	56.52	51.81	76.48	13.12	22.23
orca_mini_v3_13b	⭕	128	52.23	63.14	82.35	56.52	51.81	76.48	13.12	22.23
MythoLogic-L2-13b	🔶	128	52.22	61.01	83.93	55.7	48.64	76.09	11.75	28.43
platypus-2-22b-relora	⭕	218	52.21	57.68	82.44	55.33	43.61	77.35	6.6	42.46
llama-2-13b-FINETUNE4_3.8w-r8-q_k_v_o	🔶	128	52.21	57.68	81.91	54.95	41.31	76.48	12.05	41.07
llama-2-13b-FINETUNE5_4w-r16-q_k_v_o	🔶	128	52.2	58.7	81.66	53.87	43.02	76.72	13.8	37.63
airoboros-33b-gpt4-1.2	🔶	323	52.2	64.42	84.93	60.35	49.18	77.51	9.78	19.21
13B-Chimera	❓	128	52.19	57.59	81.5	49.86	52.59	77.27	10.69	35.84
GodziLLa-30B	🔶	323	52.17	61.52	82.13	54.21	55.91	76.16	0.38	34.86
samantha-1.2-mistral-7b	⭕	71	52.16	64.08	85.08	63.91	50.4	78.53	16.98	6.13
Camel-Platypus2-13B	🔶	128	52.12	60.75	83.61	56.51	49.6	75.37	0.08	38.91
tigerbot-13b-base	🟢	130	52.11	53.84	77.05	53.57	44.06	74.98	17.06	44.21
platypus2-22b-relora	⭕	218	52.11	57.51	82.36	54.94	43.62	77.11	6.29	42.9
speechless-orca-platypus-coig-lite-4k-0.5e-13b	🔶	128	52.09	58.02	80.15	57.26	48.04	75.45	5.84	39.88
llama-2-13b-FINETUNE3_3.3w-r16-q_k_v_o	🔶	128	52.08	59.3	81.2	55.58	38.13	76.8	13.5	40.06
CreativityEngine	🔶	128	52.07	59.3	82.42	53.55	52.46	74.19	9.55	32.98
llama-2-13b-FINETUNE3_3.3w-r8-q_k_v_o	🔶	128	52.06	56.06	81.89	55.04	40.12	76.56	14.25	40.49
Nous-Hermes-Llama2-13b	🔶	128	52.03	61.26	83.26	55.04	50.41	75.37	9.17	29.71
firefly-llama2-13b-v1.2	🔶	128	52.03	60.67	80.46	56.51	51.03	74.82	11.75	28.94
Stheno-1.8-L2-13B	🔶	128	52.0	63.48	84.12	58.57	52.86	76.4	13.27	15.33
Llama-2-13b-orca-v1	🔶	128	51.99	62.2	82.32	57.67	49.6	76.8	12.89	22.47
Nous-Hermes-13B-Code	🔶	128	51.98	61.18	83.21	55.13	50.56	75.14	10.39	28.28
llama-2-13b-huangyt_Fintune_1_17w	🔶	128	51.96	59.47	81.0	54.31	38.17	77.27	13.27	40.25
CalliopeDS-v2-L2-13B	🔶	130	51.95	62.8	84.14	56.14	51.06	76.01	12.59	20.92
StableBeluga-13B	🔶	130	51.94	62.03	82.27	57.71	49.61	76.87	13.8	21.26
Llama-2-13b-orca-v1	🔶	128	51.94	62.03	82.27	57.71	49.61	76.87	13.8	21.26
gpt4-alpaca-lora-30b-HF	❓	323	51.93	64.85	85.72	58.51	52.24	80.19	15.54	6.44
llama-2-13b-huangyt_FINETUNE2_3w-q_k_v_o_proj	🔶	128	51.9	58.53	82.47	53.9	37.92	76.8	12.81	40.85
NewHope_HF_not_official	🔶	128	51.9	61.09	84.03	55.73	44.96	74.98	15.85	26.66
llama-2-13b-FINETUNE4_3.8w-r16-q_k_v_o_gate_up_down	🔶	128	51.89	57.25	81.49	55.9	39.79	75.77	12.05	40.95
llama-2-13b-FINETUNE4_3.8w-r16-q_k_v_o	🔶	128	51.89	56.23	81.98	55.87	39.76	76.72	11.52	41.18
Mistral-7B-OpenOrca-1k	⭕	71	51.88	62.97	84.66	62.2	52.96	78.61	11.98	9.74
gaodrew-llama-30b-instruct-2048-Open-Platypus-100steps	🔶	323	51.87	61.52	84.06	60.23	51.05	80.82	17.66	7.75
Emerhyst-20B	🔶	200	51.85	61.69	84.98	56.98	54.16	76.09	8.49	20.56
speechless-llama2-hermes-orca-platypus-wizardlm-13b	🔶	130	51.85	59.64	82.7	58.3	56.0	75.37	13.12	17.81
llama-2-13b-FINETUNE3_3.3w-r4-q_k_v_o	🔶	128	51.83	59.04	81.15	53.0	40.16	76.48	11.9	41.1
MegaMix-T1-13B	🔶	130	51.82	61.35	83.44	58.49	48.19	76.09	24.11	11.04
speechless-codellama-34b-v1.9	🔶	335	51.8	54.27	75.2	56.12	43.92	73.56	24.79	34.74
vigogne2-enno-13b-sft-lora-4bit	⭕	128	51.79	62.03	82.65	54.11	42.98	76.95	0.15	43.65