Question Answering on HeadQA (pass@1)
78.12Pass@1Llama3.1-8B-Instruct
Evaluation Results
| Method | Links | |
|---|---|---|
| Llama3.1-8B-InstructTraining Pipeline=PSFT → GRPO2025.08 | 78.12 | |
| Llama3.1-8B-InstructTraining Pipeline=PSFT2025.08 | 77.9 | |
| Qwen2.5-7B-InstructTraining Pipeline=PSFT → GRPO2025.08 | 75.89 | |
| Qwen2.5-7B-InstructTraining Pipeline=PSFT2025.08 | 75.82 | |
| Qwen2.5-7B-InstructTraining Pipeline=SFT → GRPO2025.08 | 75.42 | |
| Qwen2.5-7B-InstructTraining Pipeline=SFT2025.08 | 74.65 | |
| Llama3.1-8B-InstructTraining Pipeline=SFT → GRPO2025.08 | 73.81 | |
| Llama3.1-8B-InstructTraining Pipeline=SFT2025.08 | 73.52 |