Best feasible prompt identification on CNN/DailyMail (test)
0.172Average Soft Constrained RewardMLP-CSR
Evaluation Results
| Method | Links | |
|---|---|---|
| MLP-CSRBackbone=Gemma-7B, K=50, b=82026.05 | 0.172 | |
| CSRBackbone=Gemma-7B, K=50, b=102026.05 | 0.167 | |
| CSRBackbone=Gemma-7B, K=30, b=102026.05 | 0.166 | |
| CSRBackbone=Llama3-8B, K=50, b=102026.05 | 0.165 | |
| MLP-CSRBackbone=Gemma-7B, K=30, b=52026.05 | 0.164 | |
| MLP-CSRBackbone=Gemma-7B, K=30, b=82026.05 | 0.164 | |
| MLP-CSRBackbone=Gemma-7B, K=30, b=102026.05 | 0.163 | |
| MLP-CSRBackbone=Gemma-7B, K=50, b=52026.05 | 0.162 | |
| CSRBackbone=Gemma-7B, K=50, b=52026.05 | 0.161 | |
| CSRBackbone=Gemma-7B, K=50, b=82026.05 | 0.161 | |
| MLP-CSRBackbone=Llama3-8B, K=50, b=82026.05 | 0.16 | |
| MLP-CSRBackbone=Llama3-8B, K=50, b=102026.05 | 0.16 | |
| CSRBackbone=Llama3-8B, K=100, b=102026.05 | 0.16 | |
| MLP-CSRBackbone=Llama3-8B, K=100, b=102026.05 | 0.16 | |
| CSRBackbone=Llama3-8B, K=30, b=32026.05 | 0.157 | |
| CSRBackbone=Llama3-8B, K=50, b=52026.05 | 0.157 | |
| CSRBackbone=Llama3-8B, K=50, b=82026.05 | 0.155 | |
| MLP-CSRBackbone=Llama3-8B, K=50, b=52026.05 | 0.155 | |
| CSRBackbone=Llama3-8B, K=30, b=102026.05 | 0.154 | |
| MLP-CSRBackbone=Llama3-8B, K=30, b=102026.05 | 0.154 | |
| CSRBackbone=Llama3-8B, K=50, b=32026.05 | 0.154 | |
| MLP-CSRBackbone=Llama3-8B, K=100, b=52026.05 | 0.154 | |
| MLP-CSRBackbone=Llama3-8B, K=100, b=82026.05 | 0.154 | |
| CSRBackbone=Gemma-7B, K=30, b=82026.05 | 0.153 | |
| MLP-CSRBackbone=Gemma-7B, K=50, b=102026.05 | 0.153 | |
| CSRBackbone=Llama3-8B, K=100, b=82026.05 | 0.151 | |
| CSRBackbone=Llama3-8B, K=30, b=82026.05 | 0.148 | |
| MLP-CSRBackbone=Gemma-7B, K=50, b=32026.05 | 0.146 | |
| MLP-CSRBackbone=Llama3-8B, K=50, b=32026.05 | 0.144 | |
| MLP-CSRBackbone=Llama3-8B, K=30, b=32026.05 | 0.142 | |
| MLP-CSRBackbone=Llama3-8B, K=30, b=82026.05 | 0.142 | |
| MLP-CSRBackbone=Gemma-7B, K=30, b=32026.05 | 0.138 | |
| CSRBackbone=Llama3-8B, K=100, b=52026.05 | 0.138 | |
| CSRBackbone=Llama3-8B, K=30, b=52026.05 | 0.136 | |
| CSRBackbone=Gemma-7B, K=30, b=52026.05 | 0.134 | |
| MLP-CSRBackbone=Llama3-8B, K=100, b=32026.05 | 0.134 | |
| MLP-CSRBackbone=Llama3-8B, K=30, b=52026.05 | 0.126 | |
| CSRBackbone=Llama3-8B, K=100, b=32026.05 | 0.123 | |
| CSRBackbone=Gemma-7B, K=50, b=32026.05 | 0.091 | |
| CSRBackbone=Gemma-7B, K=100, b=32026.05 | 0.088 | |
| CSRBackbone=Gemma-7B, K=30, b=32026.05 | 0.081 | |
| MLP-CSRBackbone=Gemma-7B, K=100, b=102026.05 | 0.072 | |
| CSRBackbone=Gemma-7B, K=100, b=82026.05 | 0.062 | |
| UniformBackbone=Gemma-7B, K=100, b=52026.05 | 0.045 | |
| MLP-CSRBackbone=Gemma-7B, K=100, b=52026.05 | 0.045 | |
| UniformBackbone=Llama3-8B, K=100, b=52026.05 | 0.044 | |
| UniformBackbone=Gemma-7B, K=30, b=52026.05 | 0.032 | |
| CSRBackbone=Gemma-7B, K=100, b=102026.05 | 0.032 | |
| MLP-CSRBackbone=Gemma-7B, K=100, b=32026.05 | 0.032 | |
| MLP-CSRBackbone=Gemma-7B, K=100, b=82026.05 | 0.032 | |
| UniformBackbone=Gemma-7B, K=30, b=32026.05 | 0.021 | |
| UniformBackbone=Gemma-7B, K=30, b=82026.05 | 0.021 | |
| UniformBackbone=Gemma-7B, K=30, b=102026.05 | 0.021 | |
| UniformBackbone=Gemma-7B, K=50, b=52026.05 | 0.02 | |
| UniformBackbone=Gemma-7B, K=50, b=82026.05 | 0.02 | |
| UniformBackbone=Llama3-8B, K=100, b=32026.05 | 0.019 | |
| UniformBackbone=Llama3-8B, K=100, b=82026.05 | 0.018 | |
| UniformBackbone=Llama3-8B, K=100, b=102026.05 | 0.018 | |
| UniformBackbone=Llama3-8B, K=50, b=52026.05 | 0.016 | |
| UniformBackbone=Gemma-7B, K=100, b=32026.05 | 0.015 | |
| UniformBackbone=Gemma-7B, K=100, b=102026.05 | 0.015 | |
| CSRBackbone=Gemma-7B, K=100, b=52026.05 | 0.015 | |
| UniformBackbone=Llama3-8B, K=50, b=32026.05 | 0.015 | |
| UniformBackbone=Llama3-8B, K=50, b=102026.05 | 0.01 | |
| UniformBackbone=Llama3-8B, K=30, b=82026.05 | 0.008 | |
| UniformBackbone=Llama3-8B, K=50, b=82026.05 | 0.008 | |
| UniformBackbone=Gemma-7B, K=50, b=32026.05 | 0.007 | |
| UniformBackbone=Gemma-7B, K=50, b=102026.05 | 0 | |
| UniformBackbone=Gemma-7B, K=100, b=82026.05 | 0 | |
| UniformBackbone=Llama3-8B, K=30, b=32026.05 | 0 | |
| UniformBackbone=Llama3-8B, K=30, b=52026.05 | 0 | |
| UniformBackbone=Llama3-8B, K=30, b=102026.05 | 0 |