LLM Evaluation on HuggingFace Open LLM Leaderboard Old (test) benchmark leaderboard · SOTA2 Research

LLM Evaluation on HuggingFace Open LLM Leaderboard Old (test)

92.08GSM8K ScoreQwen+SFT+UNA

Evaluation Results

Method	Links
Qwen+SFT+UNA2024.10		92.08	62.74	82.4	71.5	85.37	83.36	79.58
Qwen+UFT2024.10		90.68	66.7	83.58	71.84	85.58	83.35	80.29
Qwen+UFT2024.10		88.29	58.96	82.64	70.9	85	83.34	78.19
Qwen+SFT2024.10		87.38	58.81	82	70.9	84.99	83.39	77.91
Qwen2024.10		85.79	57.75	82	70.31	85.22	83.19	77.38
Qwen+SFT+DPO2024.10		85.67	58.54	82	70.14	85.24	83.31	77.48
Qwen+SFT+KTO2024.10		85.56	59.77	82.16	70.14	85.32	83.32	77.71
Mistral+UFT2024.10		45.57	51.18	—	63.82	83.54	62.44	64.25
Mistral+SFT+KTO2024.10		42.57	49.67	—	61.86	83.83	62.06	63.23
Mistral+SFT+DPO2024.10		42.19	47.83	—	62.16	84.03	62.38	62.84
Mistral+UFT2024.10		41.59	54.05	—	63.82	84.44	62.33	64.34
Mistral+SFT+UNA2024.10		39.99	49.54	—	62.46	84.08	62.3	63.02
Mistral+SFT2024.10		39.65	51.06	—	63.99	83.78	61.99	63.17
Mistral2024.10		38.02	42.58	—	61.43	83.44	62.51	60.93