Mathematical Reasoning on AMC23 (TPF, AVG@8)
93.1AVG@8NPR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| NPRData=orz-8k, Train=Parallel RL, Base=NPR-BETA2025.12 | 93.1 | — | |
| Qwen3-4B-Instruct-2507Data=N/A, Train=N/A, Base=N/A2025.12 | 92.2 | — | |
| SR-BETAData=orz-8k, Train=Sequential SFT, Base=Q3-4B-Inst.2025.12 | 91.6 | — | |
| SRData=orz-8k, Train=Sequential RL, Base=NPR-BETA2025.12 | 90.9 | — | |
| NPR (Variant)Data=orz-8k, Train=Parallel RL, Base=NPR-BETA2025.12 | 89.7 | — | |
| Qwen3-8B-Thinking†Model Type=Autoregressive2026.02 | 88.8 | 100 | |
| NPR-BETAData=orz-8k, Train=Parallel SFT, Base=Q3-4B-Inst.2025.12 | 85.9 | — | |
| NPR-BETA (Variant)Data=orz-8k, Train=Parallel SFT, Base=Q3-4B2025.12 | 85.3 | — | |
| TDAR-8B-thinking + BACD + TCCFModel Type=Block Diffusion, Block Size (B)=16, Decoding=BACD + TCCF2026.02 | 80 | 268 | |
| Multiverse-4BData=s1.1-8k, Train=S→P SFT, Base=Q3-4B-Inst.2025.12 | 75 | — | |
| TraDo-8B-Thinking + BACD + TCCFModel Type=Block Diffusion, Block Size (B)=4, Decoding=BACD + TCCF2026.02 | 74.1 | 136 | |
| TraDo-8B-Thinking + BACDModel Type=Block Diffusion, Block Size (B)=4, Decoding=BACD2026.02 | 73.8 | 145 | |
| TraDo-8B-ThinkingModel Type=Block Diffusion, Block Size (B)=42026.02 | 72.8 | 137 | |
| Multiverse-32BData=s1.1-8k, Train=S→P SFT, Base=Q2.5-32B-Inst.2025.12 | 72.5 | — | |
| TDAR-8B-thinking + BACDModel Type=Block Diffusion, Block Size (B)=16, Decoding=BACD2026.02 | 71.3 | 559 | |
| TDAR-8B-thinkingModel Type=Block Diffusion, Block Size (B)=162026.02 | 69.1 | 503 | |
| DiRL-8B-InstructModel Type=Block Diffusion, Block Size (B)=42026.02 | 65.6 | 205 | |
| Qwen3-4B (Non-Thinking)Data=N/A, Train=N/A, Base=N/A2025.12 | 65.6 | — | |
| Qwen2.5-32B-InstructData=N/A, Train=N/A, Base=N/A2025.12 | 62.8 | — | |
| TraDo-8B-InstructModel Type=Block Diffusion, Block Size (B)=42026.02 | 55.3 | 223 | |
| LLaDA-MoEModel Type=Masked Diffusion2026.02 | 32.5 | 316 | |
| Fast-dLLM-v2Model Type=Block Diffusion, Block Size (B)=322026.02 | 25 | 277 | |
| SDAR-8B-ChatModel Type=Block Diffusion, Block Size (B)=42026.02 | 22.5 | 283 | |
| LLaDAModel Type=Masked Diffusion2026.02 | 12.5 | 407 | |
| LLaDA-1.5Model Type=Masked Diffusion2026.02 | 10 | 401 |