Question Answering on AI2SciE Far OOD
96.2AccuracyFew-shot*
Evaluation Results
| Method | Links | |
|---|---|---|
| Few-shot*Model=Llama3.1-70B2025.09 | 96.2 | |
| ICRModel=Llama3.1-70B2025.09 | 92 | |
| ICRModel=Qwen3-32B2025.09 | 88.6 | |
| Zero-shotModel=Qwen3-32B2025.09 | 85.8 | |
| I2CLModel=Qwen3-32B2025.09 | 84.6 | |
| I2CLModel=Llama3.1-70B2025.09 | 84.4 | |
| FVModel=Llama3.1-70B2025.09 | 80.8 | |
| Few-shot*Model=Qwen3-32B2025.09 | 80.3 | |
| FVModel=Qwen3-32B2025.09 | 79.4 | |
| Zero-shotModel=Llama3.1-70B2025.09 | 68.6 |