Mathematical Reasoning on GSM8K (Parameter & Training Efficiency)
509.7Parameters (Full) (M)Hot-25% (MoE-Sieve)
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Hot-25% (MoE-Sieve)Backbone Model=Qwen2026.03 | 509.7 | 151.3 | 70.3 | 2.04 | 606 | 71 | 3 | 1 | 49 | |
| Hot-25% (MoE-Sieve)Backbone Model=OLMoE2026.03 | 311.5 | 85 | 72.7 | 1.25 | 340 | 73.4 | 1 | 54 | 50 |