Math on GSM8K (Accuracy)
96.06AccuracyLLaDA-2.0-flash
Evaluation Results
| Method | Links | |
|---|---|---|
| LLaDA-2.0-flashParams=100B-A5B, Sampling mode=Diffusion2026.06 | 96.06 | |
| LLaDA-2.0-miniParams=16B-A1B, Sampling mode=Diffusion2026.06 | 94.24 | |
| FLARE-9BParams=9B, Sampling mode=AR-Trust2026.06 | 93.33 | |
| FLARE-9BParams=9B, Sampling mode=Diffusion-Trust2026.06 | 93.1 | |
| FLAREParams=4B, Sampling mode=Diffusion-Trust2026.06 | 91.58 | |
| SDAR 30B-A3BParams=30B-A3B, Sampling mode=Diffusion2026.06 | 91.4 | |
| SDARParams=8B, Sampling mode=Diffusion2026.06 | 91.3 | |
| FLAREParams=4B, Sampling mode=AR-Trust2026.06 | 91.05 | |
| Mercury-2Params=Commercial, Sampling mode=Diffusion2026.06 | 90.62 | |
| SDARParams=4B, Sampling mode=Diffusion2026.06 | 89.9 | |
| Qwen3.5-9BParams=9B, Sampling mode=AR2026.06 | 89.16 | |
| Qwen3.5Params=4B, Sampling mode=AR2026.06 | 89.16 | |
| FLAREParams=2B, Sampling mode=AR-Trust2026.06 | 84.46 | |
| FLAREParams=2B, Sampling mode=Diffusion-Trust2026.06 | 82.79 | |
| SDARParams=1.7B, Sampling mode=Diffusion2026.06 | 80.1 | |
| Qwen3.5Params=2B, Sampling mode=AR2026.06 | 77.63 |