LLM-as-a-Judge Performance on MT-Bench (test)
2.24StdDevDI
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DIBase Model=GPT4o2024.10 | 2.24 | 82.26 | |
| CalibraEvalBase Model=GPT4o2024.10 | 2.89 | 82.29 | |
| GPT4oCalibration=None2024.10 | 3.23 | 82.27 | |
| DCBase Model=GPT4o2024.10 | 3.57 | 81.93 | |
| PrideBase Model=GPT4o2024.10 | 3.84 | 82.1 | |
| CCBase Model=Qwen-72B2024.10 | 4.72 | 81.05 | |
| CCBase Model=GPT4o2024.10 | 4.84 | 81.23 | |
| CalibraEvalBase Model=Qwen-72B2024.10 | 4.99 | 81.25 | |
| CalibraEvalBackbone=ChatGPT2024.10 | 5.2 | 72.98 | |
| CalibraEvalBase Model=ChatGPT2024.10 | 5.2 | 72.98 | |
| CalibraEvalBackbone=Llama-3-8B2024.10 | 5.22 | 70.63 | |
| CalibraEvalBase Model=Llama-3-8B2024.10 | 5.22 | 70.63 | |
| PrideBase Model=Qwen-72B2024.10 | 5.24 | 81.23 | |
| Qwen-72BCalibration=None2024.10 | 5.76 | 81.32 | |
| DIBase Model=Qwen-72B2024.10 | 5.79 | 80.82 | |
| PrideBackbone=ChatGPT2024.10 | 6.01 | 72.86 | |
| PrideBase Model=ChatGPT2024.10 | 6.01 | 72.86 | |
| CalibraEvalBackbone=Qwen-14B2024.10 | 6.26 | 68.64 | |
| CalibraEvalBase Model=Qwen-14B2024.10 | 6.26 | 68.64 | |
| CalibraEvalBase Model=Llama-3.1-8B2024.10 | 6.67 | 67.86 | |
| CCBase Model=Llama-3.1-8B2024.10 | 6.75 | 65.09 | |
| DCBase Model=Qwen-72B2024.10 | 6.76 | 80.52 | |
| DIBase Model=ChatGPT2024.10 | 7.01 | 69.84 | |
| ChatGPTBackbone=ChatGPT, Configuration=Vanilla2024.10 | 7.66 | 72.67 | |
| ChatGPTCalibration=None2024.10 | 7.66 | 72.67 | |
| DCBase Model=Qwen-14B2024.10 | 8.23 | 68.48 | |
| CCBase Model=Llama-3-8B2024.10 | 8.7 | 69.09 | |
| DIBase Model=Llama-3-8B2024.10 | 9.42 | 66.79 | |
| DCBase Model=Llama-3.1-8B2024.10 | 9.72 | 67.74 | |
| PrideBackbone=Llama-3-8B2024.10 | 11.64 | 70.63 | |
| PrideBase Model=Llama-3-8B2024.10 | 11.64 | 70.63 | |
| PrideBase Model=Llama-3.1-8B2024.10 | 12.63 | 67.56 | |
| DIBase Model=Llama-3.1-8B2024.10 | 13.72 | 67.47 | |
| Llama-3.1-8BCalibration=None2024.10 | 14.73 | 67.58 | |
| PrideBackbone=Qwen-14B2024.10 | 16.31 | 65.29 | |
| PrideBase Model=Qwen-14B2024.10 | 16.31 | 65.29 | |
| CCBase Model=ChatGPT2024.10 | 16.31 | 70.49 | |
| Llama-3-8BBackbone=Llama-3-8B, Configuration=Vanilla2024.10 | 16.42 | 67.08 | |
| Llama-3-8BCalibration=None2024.10 | 16.42 | 67.08 | |
| Qwen-14BBackbone=Qwen-14B, Configuration=Vanilla2024.10 | 17.24 | 65.61 | |
| Qwen-14BCalibration=None2024.10 | 17.24 | 65.61 | |
| DIBase Model=Qwen-14B2024.10 | 19.09 | 62.18 | |
| DCBase Model=ChatGPT2024.10 | 20.23 | 68.67 | |
| DCBase Model=Llama-3-8B2024.10 | 20.86 | 64.6 | |
| CCBase Model=Qwen-14B2024.10 | 26.47 | 58.21 |