Share your thoughts, 1 month free Claude Pro on usSee more

Role-playing Dialogue Evaluation on FURINA-Bench English

42.99Context Reliance

o3

Updated 1mo ago

Evaluation Results

Method	Links
o3 2025.10		42.99	35.31	43	42.68	55.93	43.98
Deepseek-R1 2025.10		36.31	36.18	33.39	32.5	44.14	36.5
Claude-4-Sonnet 2025.10		32.62	27.98	48.3	30.41	41.74	36.21
Claude-4-Sonnet 2025.10		30.8	29.72	53.45	27.71	40.9	36.52
Qwen3-32B 2025.10		28.49	27.47	13.9	27.72	38.78	27.27
Qwen3-235B 2025.10		27.37	35.45	21.88	28.08	45.5	31.66
GPT-4o 2025.10		25.7	22.94	19.53	26.57	28.7	24.69
Deepseek-V3 2025.10		22.51	21.27	14.44	19.51	29.87	21.52
Qwen3-235B 2025.10		19.61	18.99	19.88	23.16	32.23	22.77
Qwen3-32B 2025.10		15.98	13.44	9.08	15.29	20.21	14.8
Llama3.1-70B 2025.10		14.62	12.01	11.74	12.48	10.84	12.34
Qwen3-8B 2025.10		14.28	13.39	7.67	11.2	23.27	13.96
Qwen3-8B 2025.10		12.78	12	7.53	10.55	24.07	13.39
CoSER-70B 2025.10		11.16	10.42	5.89	9.89	9.9	9.45
Llama3.1-8B 2025.10		10.85	11.92	10.12	8.63	8.42	9.99