State Transition Graph on STG High/Huge (OOD)
53.4AccuracyQwen2.5-1.5B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2.5-1.5BArchitecture=AR2025.11 | 53.4 | |
| Llama-3.1-8BArchitecture=AR2025.11 | 49.6 | |
| C2DLMArchitecture=DLM2025.11 | 32.4 | |
| Llama-3.2-1BArchitecture=AR2025.11 | 27.9 | |
| LLaDA-8B-InstructArchitecture=DLM, Training=SFT2025.11 | 25.6 |