Instruction Following on MT-bench and AlpacaEval
1.55Aggregated PNovelSelect
Evaluation results
| Method | Links | |
|---|---|---|
| NovelSelectBase Model=LLaMA-3-8B, Data Budget=10k samples2025.02 | 1.55 | |
| K-meansBase Model=LLaMA-3-8B, Data Budget=10k samples2025.02 | 1.32 | |
| K-Center-GreedyBase Model=LLaMA-3-8B, Data Budget=10k samples2025.02 | 1.31 | |
| QDITBase Model=LLaMA-3-8B, Data Budget=10k samples2025.02 | 1.25 | |
| RandomBase Model=LLaMA-3-8B, Data Budget=10k samples2025.02 | 1.2 | |
| Repr FilterBase Model=LLaMA-3-8B, Data Budget=10k samples2025.02 | 1.05 |