Knowledge Question Answering on HLE
4.36AccuracyKTAE
Evaluation Results
| Method | Links | |
|---|---|---|
| KTAEModel Scale=Qwen3-1.7B2025.08 | 4.36 | |
| VRPOModel Scale=Qwen3-1.7B2025.08 | 4.22 | |
| GRPOModel Scale=Qwen3-1.7B2025.08 | 3.94 | |
| PPOModel Scale=Qwen3-1.7B2025.08 | 3.94 | |
| λ-GRPOModel Scale=Qwen3-1.7B2025.08 | 3.89 | |
| VRPOModel Scale=Qwen3-8B2025.08 | 3.89 | |
| λ-GRPOModel Scale=Qwen3-8B2025.08 | 3.57 | |
| Reinforce++Model Scale=Qwen3-1.7B2025.08 | 3.52 | |
| PPOModel Scale=Qwen3-8B2025.08 | 3.29 | |
| Reinforce++Model Scale=Qwen3-8B2025.08 | 3.29 | |
| GRPOModel Scale=Qwen3-8B2025.08 | 3.24 | |
| Dr.GRPOModel Scale=Qwen3-1.7B2025.08 | 3.1 | |
| Dr.GRPOModel Scale=Qwen3-8B2025.08 | 3.1 | |
| BaseModel Scale=Qwen3-8B2025.08 | 2.89 | |
| BaseModel Scale=Qwen3-1.7B2025.08 | 2.87 | |
| KTAEModel Scale=Qwen3-8B2025.08 | 2.64 |