Helpfulness Evaluation on InstructionFollow
63.2AccuracyNone
Evaluation Results
| Method | Links | |
|---|---|---|
| NoneTarget LLM=GPT-3.5-turbo2026.05 | 63.2 | |
| Perplexity FilterTarget LLM=GPT-3.5-turbo2026.05 | 62.5 | |
| NoneTarget LLM=Qwen2.52026.05 | 61.6 | |
| DR-Smoothing-PolicyTarget LLM=GPT-3.5-turbo2026.05 | 61.2 | |
| Perplexity FilterTarget LLM=Qwen2.52026.05 | 61 | |
| SemanticSmooth-PolicyTarget LLM=GPT-3.5-turbo2026.05 | 58.7 | |
| DR-Smoothing-PolicyTarget LLM=Qwen2.52026.05 | 58 | |
| DR-Smoothing-UniformTarget LLM=GPT-3.5-turbo2026.05 | 56.4 | |
| SemanticSmooth-PolicyTarget LLM=Qwen2.52026.05 | 56.2 | |
| DR-Smoothing-UniformTarget LLM=Qwen2.52026.05 | 52.4 | |
| SemanticSmooth-UniformTarget LLM=GPT-3.5-turbo2026.05 | 52.2 | |
| Erase-and-CheckTarget LLM=GPT-3.5-turbo2026.05 | 47.2 | |
| SmoothLLMTarget LLM=GPT-3.5-turbo2026.05 | 44 | |
| SemanticSmooth-UniformTarget LLM=Qwen2.52026.05 | 42.5 | |
| NoneTarget LLM=Vicuna2026.05 | 38.1 | |
| Perplexity FilterTarget LLM=Vicuna2026.05 | 37.7 | |
| DR-Smoothing-PolicyTarget LLM=Vicuna2026.05 | 36.7 | |
| NoneTarget LLM=Llama-22026.05 | 36.6 | |
| Perplexity FilterTarget LLM=Llama-22026.05 | 36.2 | |
| SemanticSmooth-PolicyTarget LLM=Vicuna2026.05 | 35.4 | |
| Erase-and-CheckTarget LLM=Qwen2.52026.05 | 34.4 | |
| DR-Smoothing-PolicyTarget LLM=Llama-22026.05 | 33.6 | |
| SmoothLLMTarget LLM=Qwen2.52026.05 | 31.3 | |
| SemanticSmooth-PolicyTarget LLM=Llama-22026.05 | 30.9 | |
| DR-Smoothing-UniformTarget LLM=Vicuna2026.05 | 30.5 | |
| DR-Smoothing-UniformTarget LLM=Llama-22026.05 | 28.7 | |
| SemanticSmooth-UniformTarget LLM=Vicuna2026.05 | 25.2 | |
| SemanticSmooth-UniformTarget LLM=Llama-22026.05 | 21.3 | |
| SmoothLLMTarget LLM=Vicuna2026.05 | 19.8 | |
| Erase-and-CheckTarget LLM=Vicuna2026.05 | 19.3 | |
| Erase-and-CheckTarget LLM=Llama-22026.05 | 17.2 | |
| SmoothLLMTarget LLM=Llama-22026.05 | 17 |