Mathematical Reasoning on SVAMP (Exact Accuracy, Avg Tokens)
55.9Exact-match AccuracyTILR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TILRBackbone=Qwen2.5-Math-1.5B, LoRA rank=16, Internalization curriculum=Stepwise, Reasoning method=TILR2026.06 | 55.9 | 4 | |
| RefinementBackbone=Qwen2.5-Math-1.5B, LoRA rank=16, Internalization curriculum=Stepwise, Reasoning method=Refinement2026.06 | 52.6 | 4 | |
| CoconutBackbone=Qwen2.5-Math-1.5B, LoRA rank=16, Internalization curriculum=Stepwise, Reasoning method=Coconut2026.06 | 51 | 4 | |
| CoTBackbone=Qwen2.5-Math-1.5B, LoRA rank=16, Internalization curriculum=Stepwise, Reasoning method=CoT2026.06 | 47.4 | 42 | |
| AdaAnchorBackbone=Qwen2.5-Math-1.5B, LoRA rank=16, Internalization curriculum=Stepwise, Reasoning method=AdaAnchor2026.06 | 43.8 | 4 | |
| No-CoTBackbone=Qwen2.5-Math-1.5B, LoRA rank=16, Internalization curriculum=Stepwise, Reasoning method=No-CoT2026.06 | 32.6 | 3 |