Instruction Following Evaluation on SelfInst Out-of-Distribution
51.6GPT-4o ScoreSODA
Evaluation Results
| Method | Links | |
|---|---|---|
| SODAModel=Llama-3.1-8B-Instruct2026.04 | 51.6 | |
| SODAModel=Qwen2.5-7B-Instruct2026.04 | 50.8 | |
| SODAModel=Llama-3.2-3B-Instruct2026.04 | 50.5 | |
| GADModel=Qwen2.5-7B-Instruct2026.04 | 50.1 | |
| TeacherModel=GPT-5-Chat2026.04 | 49.7 | |
| GADModel=Llama-3.1-8B-Instruct2026.04 | 49.5 | |
| GADModel=Llama-3.2-3B-Instruct2026.04 | 49.1 | |
| SeqKDModel=Llama-3.1-8B-Instruct2026.04 | 48.7 | |
| BaseModel=Llama-3.1-8B-Instruct2026.04 | 48.4 | |
| BaseModel=Qwen2.5-7B-Instruct2026.04 | 48.3 | |
| SeqKDModel=Qwen2.5-7B-Instruct2026.04 | 48.3 | |
| SODAModel=Qwen2.5-3B-Instruct2026.04 | 48.2 | |
| GADModel=Qwen2.5-3B-Instruct2026.04 | 47.7 | |
| SeqKDModel=Llama-3.2-3B-Instruct2026.04 | 47.1 | |
| BaseModel=Llama-3.2-3B-Instruct2026.04 | 47 | |
| SeqKDModel=Qwen2.5-3B-Instruct2026.04 | 45.7 | |
| BaseModel=Qwen2.5-3B-Instruct2026.04 | 45.6 |