Multi-class classification on PROMISE original (10-fold cross-validation) benchmark leaderboard · SOTA2 Research

Multi-class classification on PROMISE original (10-fold cross-validation)

100F1 Score (A)DeepSeek-V3

Evaluation Results

Method	Links
DeepSeek-V32025.09		100	94	100	100	92	94	93	91	89	90	91	91	91	83	100	91	86	86	86	100	93	96	89	0.89	89	95	89	92	93	0.81	87	—	—	92
PRC-RoBERTa-L2025.09		100	93	—	—	—	—	100	—	—	96	—	—	93	—	—	94	—	—	94	—	—	96	—	—	91	—	—	98	—	—	97	—	—	96
GPT-3.5-turbo2025.09		94	89	91	95	100	60	75	83	95	89	86	79	83	71	59	65	83	73	78	70	74	72	76	0.62	68	90	71	80	92	69	79	82	—	82
Mistral-7B-ins.2025.09		93	90	91	95	100	60	75	83	95	89	86	79	82	71	59	65	83	75	78	70	74	72	76	0.62	68	90	71	80	92	69	79	—	—	81
PRC-BERT-B2025.09		93	91	—	—	—	—	74	—	—	68	—	—	90	—	—	86	—	—	93	—	—	91	—	—	87	—	—	98	—	—	94	—	—	91
NB (DP)2025.09		90	—	90	90	90	97	93	100	75	86	94	94	94	82	90	86	91	78	84	100	90	95	83	0.83	83	100	97	98	77	0.97	86	—	—	90
LLaMA-3.1-8B-ins.2025.09		81	94	73	90	100	70	82	94	91	92	76	89	82	86	35	50	81	81	81	82	91	86	77	0.81	79	83	88	85	92	0.82	87	—	—	86
NoR-BERT-B2025.09		79	90	77	81	60	30	40	91	77	83	80	74	77	70	41	52	83	84	83	94	87	90	64	0.67	65	78	92	85	78	0.85	81	—	—	80
NoR-BERT-L2025.09		78	90	80	76	60	60	60	91	77	83	81	79	80	62	47	53	78	84	81	92	87	90	76	0.76	76	90	92	91	83	0.88	86	—	—	82
LDA (DP)2025.09		74	—	60	95	2	10	3	20	47	28	85	60	70	52	70	60	70	35	47	95	70	81	57	0.81	70	87	87	87	76	0.61	68	—	—	62
GPT-4o2025.09		68	92	51	100	100	40	57	90	83	87	73	79	76	40	59	48	55	39	45	83	56	67	69	43	53	88	80	84	78	0.87	82	76	—	76
Gemma-3-4B2025.09		67	91	100	50	90	90	90	87	88	88	0	0	0	40	80	53	86	86	86	82	100	90	100	0.5	67	100	100	100	93	0.72	81	—	—	77
LLaMA-3.2-3B-ins.2025.09		55	64	50	62	50	10	17	91	83	85	82	61	70	0	0	0	71	51	61	83	37	51	100	0.38	55	76	77	77	90	0.69	78	—	—	71