LLM-generated text detection on NLPCC Shared Task 1 2025 (test) benchmark leaderboard · SOTA2 Research

LLM-generated text detection on NLPCC Shared Task 1 2025 (test)

99.22Macro F1 (All Lengths)EnsemJudge

Evaluation Results

Method	Links
EnsemJudge2026.03		99.22	100	99.7	98.7	99.9	95.9	97.8	99.4	100
Qwen2.5-7B-Instruct-LoRA2026.03		95.72	96.05	90.64	93.59	96.1	94.7	97	97.9	98.7
Qwen2.5-7B-Instruct-LoRA2026.03		94.09	100	99.7	95.89	99.9	45.44	84.44	98.9	100
Qwen2.5-7B-Instruct-LoRA2026.03		90.57	94.31	65.62	76.41	88.27	92.27	94.79	98.5	99.1
GLM-4-9B-Chat-LoRA2026.03		89.07	98.95	79.84	95.29	99.2	34	61.27	94.28	99.7
HybridFeatureRoBERTa2026.03		85.55	92.14	54.34	85.45	90.2	72.32	82.69	89.01	93.09
ChineseRoBERTa2026.03		83.77	86.94	50.91	85.83	86.45	77.51	82.86	90.57	92.5
Binoculars2026.03		83.17	98.9	48.28	37.77	34.44	83.6	88.8	93.5	98
Fast-DetectGPT2026.03		82.98	93.87	64	71.39	43.55	82.89	84.19	87.48	92.3
HybridFeatureRoBERTa2026.03		82.04	84.01	46.73	79.74	82.74	80.81	85.1	91.6	91.48
Fast-DetectGPT-analytical2026.03		81.46	97.75	44.42	38.08	33.67	80.36	86.8	91.6	96
Fast-DetectGPT-analytical2026.03		81.4	94.14	69.48	59.24	54.85	72.12	81.2	84.55	90.79
ChineseBERT2026.03		77.19	85.92	55.67	85.61	81.97	46.24	56.87	76.36	87.13
ChineseRoBERTa2026.03		77.01	84.26	51.41	70.41	84.76	55.66	68.58	77.73	91.77
CommonToken2026.03		73.27	71.73	59.58	87.57	75.84	68.76	70.65	76.12	77.44
SentenceSegment2026.03		58.99	59.19	62.25	63.61	63.87	48.64	53.73	56.27	62.43
CommonPhrase2026.03		52.99	59.26	51.22	63.06	53.25	38.13	40.2	43.24	46.81
ConsecutivePunctuation2026.03		33.77	33.67	33.95	34	34	33.33	33.56	34	34
SpecialToken2026.03		33.34	33.33	33.38	33.33	33.33	33.33	33.33	33.33	33.33