Physics Problem Solving on Interphyre Down to Earth level 1.0 (test)
72AccuracyQwen-2.5-7B HEXA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Qwen-2.5-7B HEXAModel=Qwen-2.5-7B, Agent Strategy=HEXA, Evolving Configuration=Online-Evolving (k=3, 17 rounds)2026.06 | 72 | 12.6 | |
| Qwen-2.5-7B REACTModel=Qwen-2.5-7B, Agent Strategy=REACT, Evolving Configuration=Online-Evolving (k=3, 17 rounds)2026.06 | 62 | 12.5 | |
| Qwen-2.5-3B HEXAModel=Qwen-2.5-3B, Agent Strategy=HEXA, Evolving Configuration=Online-Evolving (k=3, 17 rounds)2026.06 | 24 | 18.6 | |
| Qwen-2.5-3B REACTModel=Qwen-2.5-3B, Agent Strategy=REACT, Evolving Configuration=Online-Evolving (k=3, 17 rounds)2026.06 | 8 | 23.8 |