Text Classification on TREC (test)
97.2AccuracySupervised GT
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Supervised GTNoise Level=0%2025.04 | 97.2 | — | — | — | — | |
| NoiseALNoise Type=Symmetric, Noise Level=20%2025.04 | 97.16 | — | — | — | — | |
| NoiseALNoise Type=Asymmetric, Noise Level=20%2025.04 | 97.12 | — | — | — | — | |
| SyntacticAttack Type=Hidden Killer2022.05 | 97.1 | — | — | — | — | |
| BenignAttack Status=Clean/No Attack2022.05 | 96.9 | — | — | — | — | |
| BITE (Subset)Data Access=Target-label instances only2022.05 | 96.9 | — | — | — | — | |
| NoiseALNoise Type=Asymmetric, Noise Level=20%2025.04 | 96.8 | — | — | — | — | |
| NoiseALNoise Type=Asymmetric, Noise Level=40%2025.04 | 96.8 | — | — | — | — | |
| BITE (Full)Data Access=Full training set access2022.05 | 96.7 | — | — | — | — | |
| StyleAttack Type=StyleBkd2022.05 | 96.5 | — | — | — | — | |
| NoiseALNoise Type=Symmetric, Noise Level=40%2025.04 | 96.4 | — | — | — | — | |
| NoiseALNoise Type=Symmetric, Noise Level=20%2025.04 | 96.36 | — | — | — | — | |
| DyGenNoise Type=Symmetric, Noise Level=20%2025.04 | 95.88 | — | — | — | — | |
| CLNoise Type=Symmetric, Noise Level=20%2025.04 | 95.88 | — | — | — | — | |
| NoiseALNoise Type=Asymmetric, Noise Level=40%2025.04 | 95.8 | — | — | — | — | |
| NoiseALNoise Type=Symmetric, Noise Level=40%2025.04 | 95.72 | — | — | — | — | |
| CLNoise Type=Symmetric, Noise Level=20%2025.04 | 95.64 | — | — | — | — | |
| CLNoise Type=Asymmetric, Noise Level=20%2025.04 | 95.52 | — | — | — | — | |
| Co-TeachingNoise Type=Symmetric, Noise Level=20%2025.04 | 95.36 | — | — | — | — | |
| ELRNoise Type=Symmetric, Noise Level=20%2025.04 | 95.36 | — | — | — | — | |
| CLNoise Type=Asymmetric, Noise Level=20%2025.04 | 95.32 | — | — | — | — | |
| ELRNoise Type=Asymmetric, Noise Level=20%2025.04 | 95.24 | — | — | — | — | |
| SCENoise Type=Symmetric, Noise Level=20%2025.04 | 95.23 | — | — | — | — | |
| SelfMixNoise Type=Symmetric, Noise Level=20%2025.04 | 95.2 | — | — | — | — | |
| SelfMixNoise Type=Asymmetric, Noise Level=20%2025.04 | 95.16 | — | — | — | — | |
| Co-TeachingNoise Type=Symmetric, Noise Level=40%2025.04 | 95.16 | — | — | — | — | |
| ELRNoise Type=Asymmetric, Noise Level=20%2025.04 | 95.14 | — | — | — | — | |
| SCENoise Type=Asymmetric, Noise Level=20%2025.04 | 95.12 | — | — | — | — | |
| Co-TeachingNoise Type=Symmetric, Noise Level=20%2025.04 | 95.08 | — | — | — | — | |
| Co-TeachingNoise Type=Asymmetric, Noise Level=20%2025.04 | 95.08 | — | — | — | — | |
| CLNoise Type=Symmetric, Noise Level=40%2025.04 | 95.04 | — | — | — | — | |
| DyGenNoise Type=Symmetric, Noise Level=20%2025.04 | 95 | — | — | — | — | |
| DyGenNoise Type=Asymmetric, Noise Level=20%2025.04 | 94.96 | — | — | — | — | |
| ELRNoise Type=Symmetric, Noise Level=20%2025.04 | 94.92 | — | — | — | — | |
| SCENoise Type=Symmetric, Noise Level=40%2025.04 | 94.92 | — | — | — | — | |
| Co-TeachingNoise Type=Asymmetric, Noise Level=20%2025.04 | 94.88 | — | — | — | — | |
| ELRNoise Type=Symmetric, Noise Level=40%2025.04 | 94.72 | — | — | — | — | |
| SelfMixNoise Type=Asymmetric, Noise Level=20%2025.04 | 94.72 | — | — | — | — | |
| BERTNoise Type=Symmetric, Noise Level=20%2025.04 | 94.64 | — | — | — | — | |
| SCENoise Type=Symmetric, Noise Level=20%2025.04 | 94.44 | — | — | — | — | |
| SelfMixNoise Type=Symmetric, Noise Level=20%2025.04 | 94.44 | — | — | — | — | |
| DyGenNoise Type=Asymmetric, Noise Level=20%2025.04 | 94.4 | — | — | — | — | |
| KNN-Promptingprivacy budget (ε)=Non-private, number of hyperplanes=24, number of hash tables=42026.05 | 94.4 | — | — | — | — | |
| RoBERTaNoise Type=Asymmetric, Noise Level=20%2025.04 | 94.16 | — | — | — | — | |
| SCENoise Type=Asymmetric, Noise Level=20%2025.04 | 94.12 | — | — | — | — | |
| SelfMixNoise Type=Symmetric, Noise Level=40%2025.04 | 93.88 | — | — | — | — | |
| BERTNoise Type=Asymmetric, Noise Level=20%2025.04 | 93.6 | — | — | — | — | |
| RoBERTaNoise Type=Symmetric, Noise Level=20%2025.04 | 93.48 | — | — | — | — | |
| DyGenNoise Type=Symmetric, Noise Level=40%2025.04 | 93.28 | — | — | — | — | |
| SelfMixNoise Type=Asymmetric, Noise Level=40%2025.04 | 92.44 | — | — | — | — | |
| FewshotQAmeta-trained=true2023.06 | 92.4 | — | — | — | — | |
| RoBERTaNoise Type=Symmetric, Noise Level=40%2025.04 | 92.32 | — | — | — | — | |
| DyGenNoise Type=Asymmetric, Noise Level=40%2025.04 | 92.24 | — | — | — | — | |
| Meta-training with Demonstration Retrievalmeta-trained=true2023.06 | 91.7 | — | — | — | — | |
| ELRNoise Type=Asymmetric, Noise Level=40%2025.04 | 91.64 | — | — | — | — | |
| Co-TeachingNoise Type=Asymmetric, Noise Level=40%2025.04 | 91.56 | — | — | — | — | |
| FewshotQAmeta-trained=false2023.06 | 91 | — | — | — | — | |
| DP SimHash Bucketingprivacy budget (ε)=8, number of hyperplanes=24, number of hash tables=42026.05 | 90 | — | — | — | — | |
| SelfMixNoise Type=Symmetric, Noise Level=40%2025.04 | 89.8 | — | — | — | — | |
| CLNoise Type=Symmetric, Noise Level=40%2025.04 | 89.72 | — | — | — | — | |
| DP SimHash Bucketingprivacy budget (ε)=7, number of hyperplanes=24, number of hash tables=42026.05 | 89.6 | — | — | — | — | |
| LM-BFF2023.06 | 89.4 | — | — | — | — | |
| Co-TeachingNoise Type=Symmetric, Noise Level=40%2025.04 | 89.3 | — | — | — | — | |
| ELRNoise Type=Symmetric, Noise Level=40%2025.04 | 89.28 | — | — | — | — | |
| SCENoise Type=Symmetric, Noise Level=40%2025.04 | 89.1 | — | — | — | — | |
| SelfMixNoise Type=Asymmetric, Noise Level=40%2025.04 | 89 | — | — | — | — | |
| DyGenNoise Type=Symmetric, Noise Level=40%2025.04 | 89 | — | — | — | — | |
| RoBERTa2023.06 | 88.8 | — | — | — | — | |
| DyGenNoise Type=Asymmetric, Noise Level=40%2025.04 | 88.56 | — | — | — | — | |
| CLNoise Type=Asymmetric, Noise Level=40%2025.04 | 88.36 | — | — | — | — | |
| RAGmeta-trained=true2023.06 | 87.8 | — | — | — | — | |
| BERTNoise Type=Symmetric, Noise Level=40%2025.04 | 87.45 | — | — | — | — | |
| DP SimHash Bucketingprivacy budget (ε)=5, number of hyperplanes=24, number of hash tables=42026.05 | 87.4 | — | — | — | — | |
| SCENoise Type=Asymmetric, Noise Level=40%2025.04 | 87.3 | — | — | — | — | |
| Co-TeachingNoise Type=Asymmetric, Noise Level=40%2025.04 | 87.16 | — | — | — | — | |
| ELRNoise Type=Asymmetric, Noise Level=40%2025.04 | 86.9 | — | — | — | — | |
| CLNoise Type=Asymmetric, Noise Level=40%2025.04 | 86.24 | — | — | — | — | |
| BERTNoise Type=Asymmetric, Noise Level=40%2025.04 | 85.72 | — | — | — | — | |
| TextGrad-MOptimization Strategy=Blockwise, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.05 | 85.44 | — | — | 74.59 | 2.2 | |
| iPET2023.06 | 85 | — | — | — | — | |
| AdalFlowLMforward=GPT-4o-mini, LMbackward=GPT-4o2025.05 | 85 | — | — | 71.83 | — | |
| AdalFlow-MOptimization Strategy=Promptwise, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.05 | 85 | — | — | 72.63 | 0.8 | |
| GEPALMforward=GPT-4o-mini, LMbackward=GPT-4o2025.05 | 84.33 | — | — | 72.11 | — | |
| DP SimHash Bucketingprivacy budget (ε)=3, number of hyperplanes=24, number of hash tables=42026.05 | 84.2 | — | — | — | — | |
| TextGrad-MOptimization Strategy=Promptwise, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.05 | 83.36 | — | — | 74.12 | 1.73 | |
| COPRO-MOptimization Strategy=Promptwise, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.05 | 83.36 | — | — | 68.79 | 0.94 | |
| SCENoise Type=Asymmetric, Noise Level=40%2025.04 | 82.6 | — | — | — | — | |
| COPRO-MOptimization Strategy=Blockwise, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.05 | 82.52 | — | — | 68.9 | 1.05 | |
| Upper BoundModel=Claude, Evaluation Protocol=non-private baseline, epsilon=infinity2023.05 | 82.2 | — | — | — | — | |
| TextGradValidation Protocol=w/o val revert, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.05 | 81.92 | — | — | 72.96 | — | |
| RAGmeta-trained=false2023.06 | 81.1 | — | — | — | — | |
| MoLAMMethod Category=Others2026.01 | 80.4 | — | — | — | — | |
| COPROLMforward=GPT-4o-mini, LMbackward=GPT-4o2025.05 | 80.28 | — | — | 67.84 | — | |
| Ens. Acc.Model=Claude, Evaluation Protocol=ensemble, epsilon=infinity2023.05 | 79.9 | — | — | — | — | |
| PromptPATEModel=Claude, epsilon=0.0682023.05 | 79.3 | — | — | — | — | |
| MoLMethod Category=Others2026.01 | 79.24 | — | — | — | — | |
| GA2025.12 | 79.2 | — | — | — | — | |
| PIAST (E)variant=Ensemble2025.12 | 78.4 | — | — | — | — | |
| TextGradValidation Protocol=w/ val revert, LMforward=GPT-4o-mini, LMbackward=GPT-4o2025.05 | 77.4 | — | — | 72.39 | — | |
| APE2025.12 | 77.07 | — | — | — | — |