Mathematical Reasoning on AQUA-RAT (Exact Match Accuracy & Token Usage)
47.6Accuracy (Exact Match)TILR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TILRBackbone=Qwen2.5-Math-1.5B, LoRA rank=16, Internalization curriculum=Stepwise, Reasoning method=TILR2026.06 | 47.6 | 9 | |
| RefinementBackbone=Qwen2.5-Math-1.5B, LoRA rank=16, Internalization curriculum=Stepwise, Reasoning method=Refinement2026.06 | 45.1 | 9 | |
| CoTBackbone=Qwen2.5-Math-1.5B, LoRA rank=16, Internalization curriculum=Stepwise, Reasoning method=CoT2026.06 | 44.8 | 82 | |
| CoconutBackbone=Qwen2.5-Math-1.5B, LoRA rank=16, Internalization curriculum=Stepwise, Reasoning method=Coconut2026.06 | 42.3 | 9 | |
| AdaAnchorBackbone=Qwen2.5-Math-1.5B, LoRA rank=16, Internalization curriculum=Stepwise, Reasoning method=AdaAnchor2026.06 | 40.5 | 9 | |
| No-CoTBackbone=Qwen2.5-Math-1.5B, LoRA rank=16, Internalization curriculum=Stepwise, Reasoning method=No-CoT2026.06 | 37.1 | 4 |