Personalized Preference Modeling on Summarize from Human Feedback (test)
66.01Mean AccuracyEXACT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| EXACTBase Model=Gemma-2-9B-it2026.02 | 66.01 | 0.13 | |
| EXACTBase Model=Llama-3.1-8B2026.02 | 65.12 | 0.14 | |
| BaseBase Model=Gemma-2-9B-it2026.02 | 63.58 | 0.18 | |
| RewardBase Model=Qwen2.5-7B-Instruct2026.02 | 60.75 | 0.21 | |
| EXACTBase Model=Qwen2.5-7B-Instruct2026.02 | 60.28 | 0.15 | |
| DriftBase Model=Llama-3.1-8B2026.02 | 58.68 | 0.13 | |
| DriftBase Model=Gemma-2-9B-it2026.02 | 58.21 | 0.12 | |
| BaseBase Model=Llama-3.1-8B2026.02 | 57.54 | 0.18 | |
| DriftBase Model=Qwen2.5-7B-Instruct2026.02 | 55.63 | 0.16 | |
| RewardBase Model=Gemma-2-9B-it2026.02 | 54.33 | 0.13 | |
| RewardBase Model=Llama-3.1-8B2026.02 | 54.21 | 0.15 | |
| BaseBase Model=Qwen2.5-7B-Instruct2026.02 | 54.04 | 0.12 |