Mathematical Reasoning on AIME '25 (Accuracy, Tokens, SuperBPE, Deltas)
84.1AccuracyShorthand for Thought (SFT)
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Shorthand for Thought (SFT)Base Model=QWEN-3-30B-A3B, Training Condition=SFT2026.04 | 84.1 | 18,517 | 11.1 | 0.1 | -5 | |
| QWEN-3-30B-A3B BaselineBase Model=QWEN-3-30B-A3B, Training Condition=Baseline2026.04 | 84 | 19,491 | — | — | — | |
| Shorthand for Thought (SFT)Base Model=QwQ-32B, Training Condition=SFT2026.04 | 72.7 | 14,880 | 7.7 | 5.5 | -10.6 | |
| QwQ-32B BaselineBase Model=QwQ-32B, Training Condition=Baseline2026.04 | 69.2 | 16,651 | — | — | — | |
| DeepSeek-R1-Llama-70B-Distill BaselineBase Model=DeepSeek-R1-Llama-70B-Distill, Training Condition=Baseline2026.04 | 54.5 | 11,332 | — | — | — | |
| Shorthand for Thought (SFT)Base Model=DeepSeek-R1-Llama-70B-Distill, Training Condition=SFT2026.04 | 53.4 | 9,390 | 11.6 | -2 | -17 |