Instruction Following on WizardLM (test)
6.87ScoreLlama-2-7B-star_instruct
Evaluation Results
| Method | Links | |
|---|---|---|
| Llama-2-7B-star_instructModel Scale=7B2024.11 | 6.87 | |
| Llama-2-7B-evol instructModel Scale=7B2024.11 | 6.57 | |
| Llama-2-7B-alpacaModel Scale=7B2024.11 | 5.08 | |
| JREGModel Size=3.4B, Alpha (α)=1.0, SFT=true2026.01 | 4.23 | |
| BaselineModel Size=3.4B, SFT=true2026.01 | 4.1 | |
| Pythia-1B-star_instructModel Scale=1B2024.11 | 3.9 | |
| Pythia-1B-evol_instructModel Scale=1B2024.11 | 3.55 | |
| Pythia-1B-RandomModel Scale=1B2024.11 | 3.39 | |
| Pythia-1B-IFDModel Scale=1B2024.11 | 3.21 | |
| Pythia-1B-alpacaModel Scale=1B2024.11 | 2.97 | |
| Sheared-LLaMA-1.3BModel Scale=1B2024.11 | 1.86 | |
| OPT-1.3BModel Scale=1B2024.11 | 1.64 | |
| Pythia-1BModel Scale=1B2024.11 | 1.34 | |
| OursBackbone=Mistral-7B-v0.32026.04 | 1.308 | |
| IFDBackbone=Mistral-7B-v0.32026.04 | 1.294 | |
| SelectITBackbone=Mistral-7B-v0.32026.04 | 1.281 | |
| NUGGETSBackbone=Mistral-7B-v0.32026.04 | 1.249 | |
| IFDBackbone=Llama3.1-8B2026.04 | 1.186 | |
| SelectITBackbone=Llama3.1-8B2026.04 | 1.176 | |
| OursBackbone=Llama3.1-8B2026.04 | 1.169 | |
| DEITABackbone=Mistral-7B-v0.32026.04 | 1.14 | |
| NUGGETSBackbone=Llama3.1-8B2026.04 | 1.133 | |
| DEITABackbone=Llama3.1-8B2026.04 | 1.114 | |
| FullBackbone=Llama3.1-8B2026.04 | 1 | |
| FullBackbone=Mistral-7B-v0.32026.04 | 1 |