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

Binary deficiency detection on gold-labeled ICLR (test)

86Accuracy

TADDLE (Qwen3-30B-A3B + 9B integrator)

Updated 7d ago

Evaluation Results

Method	Links
TADDLE (Qwen3-30B-A3B + 9B integrator) 2026.05		86	91.54	74.76	82.3
Longformer-base-4096 2026.05		83.33	88.08	52.86	66.07
BERT-base-uncased 2026.05		77.33	73.08	39.05	50.9
RottenReviews (LR) 2026.05		68	69.23	95.24	80.18
ReviewGuard (Llama-3.1-8B) 2026.05		66	46.92	73.81	57.37
RottenReviews (Llama-3-8B FT) 2026.05		65.33	71.15	82.38	76.36
ReviewGuard (Qwen3-8B) 2026.05		63.33	43.08	57.14	49.12
GLM 4.7 2026.05		61.56	73.46	61.43	66.91
RottenReviews (XGBoost) 2026.05		60.22	74.62	61.9	67.67
RottenReviews (Random Forest) 2026.05		58.89	72.31	66.19	69.11
Qwen3.5-Plus 2026.05		55.33	74.23	45.71	56.58
Qwen3-30B 2026.05		54.44	68.85	51.9	59.19
DeepSeek V3.2 2026.05		49.78	87.31	24.76	38.58
BigBird-RoBERTa-base 2026.05		46	33.08	69.52	44.83
RottenReviews (MLP) 2026.05		31.11	0	0	0
RottenReviews (Qwen3-8B 0-shot) 2026.05		31.11	0	0	0