Text Classification on SST2 (ACC)
97.36AccuracyLS-unLLaMA-2-7B
Evaluation Results
| Method | Links | |
|---|---|---|
| LS-unLLaMA-2-7BModel Size=7B2023.10 | 97.36 | |
| PlaintextBase Model=Qwen2.5-14B-Instruct2026.03 | 97.25 | |
| AloePriBase Model=Qwen2.5-14B-Instruct2026.03 | 97.13 | |
| LS-LLaMA-2-13BModel Size=13B2023.10 | 96.9 | |
| LS-LLaMA-2-7BModel Size=7B2023.10 | 96.67 | |
| RoBERTa-LargeModel Architecture=Discriminant baseline2023.10 | 96.1 | |
| SCModel=Llama, k-shot=8, Seeds=52025.05 | 95.39 | |
| SCModel=Mistral, Shots=42025.05 | 95.23 | |
| CCModel=Mistral, Shots=42025.05 | 94.92 | |
| DCModel=Mistral, k-shot=8, Seeds=52025.05 | 94.84 | |
| CCModel=Llama, k-shot=8, Seeds=52025.05 | 94.77 | |
| RoBERTa-BaseModel Architecture=Discriminant baseline2023.10 | 94.61 | |
| DCModel=Llama, k-shot=8, Seeds=52025.05 | 94.61 | |
| Base LLMModel=Mistral, k-shot=8, Seeds=52025.05 | 94.53 | |
| SCModel=Llama, Shots=42025.05 | 94.45 | |
| CCModel=Mistral, k-shot=8, Seeds=52025.05 | 94.38 | |
| DCModel=Mistral, Shots=42025.05 | 94.38 | |
| Base LLMModel=Llama, k-shot=8, Seeds=52025.05 | 94.3 | |
| BCModel=Mistral, k-shot=8, Seeds=52025.05 | 94.3 | |
| BCModel=Llama, k-shot=8, Seeds=52025.05 | 94.22 | |
| Base LLMModel=Mistral, Shots=42025.05 | 94.22 | |
| BCModel=Mistral, Shots=42025.05 | 94.14 | |
| DCModel=Llama, Shots=42025.05 | 93.83 | |
| BCModel=Llama, Shots=42025.05 | 93.75 | |
| Base LLMModel=Llama, Shots=42025.05 | 93.63 | |
| DP-ForwardBase Model=Qwen2.5-14B-Instruct2026.03 | 93.6 | |
| SCModel=Mistral, k-shot=8, Seeds=52025.05 | 93.59 | |
| EspeWAttack=No Attack2025.12 | 93.54 | |
| WARDENAttack=Detect2025.12 | 93.5 | |
| CCModel=Llama, Shots=42025.05 | 93.48 | |
| EspeWAttack=Detect2025.12 | 93.43 | |
| SemMarkAttack=Dim2025.12 | 93.43 | |
| GPT-3 175BEvaluation Protocol=few-shot, Model Size=175B2023.10 | 93.4 | |
| OriginalAttack=None2025.12 | 93.39 | |
| EmbMarkerAttack=No Attack2025.12 | 93.35 | |
| WETAttack=Dim2025.12 | 93.35 | |
| SemMarkAttack=No Attack2025.12 | 93.31 | |
| SemMarkAttack=Detect2025.12 | 93.27 | |
| AGNModel Architecture=Discriminant baseline2023.10 | 93.27 | |
| WARDENAttack=No Attack2025.12 | 93.2 | |
| WETAttack=Detect2025.12 | 93.2 | |
| EmbMarkerAttack=Dim2025.12 | 93.12 | |
| WETAttack=No Attack2025.12 | 93.08 | |
| EmbMarkerAttack=Detect2025.12 | 93 | |
| BERT-LargeModel Architecture=Discriminant baseline2023.10 | 92.86 | |
| BERT-BaseModel Architecture=Discriminant baseline2023.10 | 92.78 | |
| WARDENAttack=Dim2025.12 | 92.77 | |
| LS-unLLaMA-2-13BModel Size=13B2023.10 | 92.77 | |
| SGTBase Model=Qwen2.5-14B-Instruct2026.03 | 92.43 | |
| EspeWAttack=Dim2025.12 | 92.35 | |
| CCModel=Qwen, Shots=42025.05 | 91.99 | |
| LLaMA-2-7BEvaluation Protocol=instruction-tuning, Model Size=7B2023.10 | 91.97 | |
| SCModel=Qwen, k-shot=8, Seeds=52025.05 | 91.88 | |
| SCModel=Qwen, Shots=42025.05 | 90.82 | |
| EmbMarkerAttack=CSE2025.12 | 90.48 | |
| CCModel=Qwen, k-shot=8, Seeds=52025.05 | 89.53 | |
| SemMarkAttack=CSE2025.12 | 89.41 | |
| BCModel=Qwen, Shots=42025.05 | 89.06 | |
| WARDENAttack=CSE2025.12 | 88.34 | |
| Base LLMModel=Qwen, Shots=42025.05 | 87.93 | |
| RANTEXTBase Model=Qwen2.5-14B-Instruct2026.03 | 87.5 | |
| DCModel=Qwen, Shots=42025.05 | 87.3 | |
| EspeWAttack=CSE2025.12 | 86.81 | |
| WETAttack=CSE2025.12 | 86.81 | |
| BCModel=Qwen, k-shot=8, Seeds=52025.05 | 86.52 | |
| SANTEXTBase Model=Qwen2.5-14B-Instruct2026.03 | 81.31 | |
| LLaMA-2-7BEvaluation Protocol=zero-shot, Model Size=7B2023.10 | 76.26 | |
| Base LLMModel=Qwen, k-shot=8, Seeds=52025.05 | 76.09 | |
| DCModel=Qwen, k-shot=8, Seeds=52025.05 | 71.72 | |
| LLaMA-2-13BEvaluation Protocol=zero-shot, Model Size=13B2023.10 | 69.9 | |
| GPT-3 175BEvaluation Protocol=zero-shot, Model Size=175B2023.10 | 54.3 |