Chinese Spelling Check on LEMON benchmark leaderboard · SOTA2 Research

Chinese Spelling Check on LEMON

63.28CARQwen3-32B-RL

Evaluation Results

Method	Links
Qwen3-32B-RL2025.12		63.28	66.89	61.3	44.29	74.87	79.91	51.29
Qwen3-14B-RL2025.12		60.32	66.71	59.77	42.43	68.02	73.39	48.96
ReLM-D2C2025.12		58.6	75.5	53.7	65.5	58.4	63	50
C-LLM2025.12		57.54	60.4	56.48	38.02	65.31	64.49	43.92
Gmini 2.52025.12		56.01	61.27	55.8	40.12	54.89	61.04	41.97
Doubao2025.12		55.81	63.03	56.23	39.89	57.34	55.89	42.31
DeepSeek-32B2025.12		55.57	63.52	55.03	39.29	56.63	55.93	44.77
Claude 3.72025.12		55.32	64.19	54.05	37.86	53.58	58.95	46.78
GPT-42025.12		54.44	62.82	55.12	36.27	56.36	56.09	45.64
ReLM2025.12		53.1	66.8	49.2	33	54	58.5	37.8
DeepSeek-14B2025.12		53.07	56.85	55.89	38.95	55.19	53.04	43.1
Qwen3-32B2025.12		52.97	57.45	55.12	36.27	56.36	56.09	45.64
Qwen3-14B2025.12		46.88	56.95	55.37	35.39	53.71	51.99	40.12
ChatGPT2025.12		44.88	57.11	54.46	28.78	49.85	44.4	31.77
PGT (BERT)2025.12		42.82	48.04	39.8	29.57	32.51	34.05	24.93
SCOPE2025.12		40.71	43.89	35.23	24.74	38.12	48.72	33.17
MDCSpell+ARM2025.12		37.1	52.7	35.2	15.3	33	36.4	15.6
MDCSpell2025.12		34.1	49.2	32.8	14.8	29.5	34.4	14.3
SoftMask2025.12		31.6	44.2	31.7	12.1	29.8	32.3	15.5
SMBERT2025.12		29.91	34.85	29.33	16.18	26.91	29.16	19.56
BERT2025.12		25.14	17.3	13.6	14.3	12.6	16.6	15.1