Mathematical Reasoning on MATH 500 (Acc, Avg. Tok., SuperBPE)
86.8AccuracyShorthand for Thought (SFT)
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Shorthand for Thought (SFT)Base Model=QWEN-3-30B-A3B, Training Condition=SFT2026.04 | 86.8 | 4,849 | 12.9 | 1.2 | -11.6 | |
| QWEN-3-30B-A3B BaselineBase Model=QWEN-3-30B-A3B, Training Condition=Baseline2026.04 | 85.7 | 5,488 | — | — | — | |
| QwQ-32B BaselineBase Model=QwQ-32B, Training Condition=Baseline2026.04 | 80.6 | 4,441 | — | — | — | |
| Shorthand for Thought (SFT)Base Model=QwQ-32B, Training Condition=SFT2026.04 | 80.5 | 4,183 | 10.7 | -0.1 | -5.8 | |
| DeepSeek-R1-Llama-70B-Distill BaselineBase Model=DeepSeek-R1-Llama-70B-Distill, Training Condition=Baseline2026.04 | 78.1 | 2,701 | — | — | — | |
| Shorthand for Thought (SFT)Base Model=DeepSeek-R1-Llama-70B-Distill, Training Condition=SFT2026.04 | 74.8 | 2,541 | 12.2 | -3.1 | -5.9 |