Mathematical Reasoning on GSM8K (Pass@1, Token count)
95Pass@1 AccuracyRED
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| REDModel=Qwen3-8B-thinking2026.04 | 95 | 465 | |
| REDModel=DeepSeek-R1-Distill-Llama-70B2026.04 | 94.9 | 269 | |
| RL + LPModel=Qwen3-32B-thinking2026.04 | 94.7 | 822 | |
| RL + LPModel=DpSk-R1-Distill-Qwen-32B2026.04 | 94.7 | 239 | |
| S-GRPOModel=DpSk-R1-Distill-Qwen-32B2026.04 | 94.7 | 269 | |
| GRPOModel=DeepSeek-R1-Distill-Llama-70B2026.04 | 94.7 | 473 | |
| S-GRPOModel=Qwen3-8B-thinking2026.04 | 94.6 | 775 | |
| S-GRPOModel=Qwen3-32B-thinking2026.04 | 94.6 | 809 | |
| REDModel=Qwen3-32B-thinking2026.04 | 94.6 | 443 | |
| REDModel=DpSk-R1-Distill-Qwen-32B2026.04 | 94.6 | 209 | |
| AlphaOneModel=Qwen3-8B-thinking2026.04 | 94.5 | 1,095 | |
| Think or NotModel=Qwen3-32B-thinking2026.04 | 94.5 | 641 | |
| GRPOModel=Qwen3-32B-thinking2026.04 | 94.5 | 1,558 | |
| S-GRPOModel=DeepSeek-R1-Distill-Llama-70B2026.04 | 94.5 | 249 | |
| GRPOModel=Qwen3-8B-thinking2026.04 | 94.4 | 1,373 | |
| AlphaOneModel=Qwen3-32B-thinking2026.04 | 94.4 | 1,090 | |
| GRPOModel=DpSk-R1-Distill-Qwen-32B2026.04 | 94.4 | 420 | |
| RL + LPModel=DeepSeek-R1-Distill-Llama-70B2026.04 | 94.4 | 255 | |
| DEERModel=Qwen3-8B-thinking2026.04 | 94.3 | 711 | |
| VanillaModel=Qwen3-32B-thinking2026.04 | 94.3 | 1,435 | |
| Think or NotModel=Qwen3-8B-thinking2026.04 | 94.2 | 599 | |
| DASTModel=Qwen3-32B-thinking2026.04 | 94.2 | 609 | |
| AlphaOneModel=DeepSeek-R1-Distill-Llama-70B2026.04 | 94.2 | 427 | |
| VanillaModel=Qwen3-8B-thinking2026.04 | 94.1 | 1,573 | |
| DASTModel=Qwen3-8B-thinking2026.04 | 94.1 | 483 | |
| RL + LPModel=Qwen3-8B-thinking2026.04 | 94.1 | 823 | |
| DEERModel=Qwen3-32B-thinking2026.04 | 94.1 | 792 | |
| REDModel=DpSk-R1-Distill-Qwen-7B2026.04 | 94.1 | 271 | |
| AlphaOneModel=DpSk-R1-Distill-Qwen-32B2026.04 | 94.1 | 433 | |
| VanillaModel=DeepSeek-R1-Distill-Llama-70B2026.04 | 94 | 432 | |
| VanillaModel=DpSk-R1-Distill-Qwen-32B2026.04 | 93.9 | 438 | |
| DASTModel=DpSk-R1-Distill-Qwen-32B2026.04 | 93.9 | 266 | |
| DASTModel=DeepSeek-R1-Distill-Llama-70B2026.04 | 93.9 | 239 | |
| S-GRPOModel=DpSk-R1-Distill-Qwen-7B2026.04 | 93.7 | 297 | |
| Think or NotModel=DeepSeek-R1-Distill-Llama-70B2026.04 | 93.7 | 241 | |
| AlphaOneModel=DpSk-R1-Distill-Qwen-7B2026.04 | 93.4 | 459 | |
| Think or NotModel=DpSk-R1-Distill-Qwen-32B2026.04 | 93.3 | 247 | |
| DEERModel=DeepSeek-R1-Distill-Llama-70B2026.04 | 93.3 | 277 | |
| S-GRPOModel=DeepSeek-R1-Distill-Llama-8B2026.04 | 93.2 | 432 | |
| AlphaOneModel=DeepSeek-R1-Distill-Llama-8B2026.04 | 93.1 | 598 | |
| REDModel=DeepSeek-R1-Distill-Llama-8B2026.04 | 93.1 | 283 | |
| Think or NotModel=DpSk-R1-Distill-Qwen-7B2026.04 | 92.9 | 264 | |
| GRPOModel=DeepSeek-R1-Distill-Llama-8B2026.04 | 92.9 | 571 | |
| RL + LPModel=DpSk-R1-Distill-Qwen-7B2026.04 | 92.5 | 291 | |
| DEERModel=DpSk-R1-Distill-Qwen-32B2026.04 | 92.5 | 290 | |
| Think or NotModel=DeepSeek-R1-Distill-Llama-8B2026.04 | 92.5 | 257 | |
| VanillaModel=DpSk-R1-Distill-Qwen-7B2026.04 | 92.4 | 442 | |
| VanillaModel=DeepSeek-R1-Distill-Llama-8B2026.04 | 92.3 | 606 | |
| RL + LPModel=DeepSeek-R1-Distill-Llama-8B2026.04 | 92.3 | 446 | |
| GRPOModel=DpSk-R1-Distill-Qwen-7B2026.04 | 92.1 | 275 | |
| DASTModel=DeepSeek-R1-Distill-Llama-8B2026.04 | 91.9 | 388 | |
| DASTModel=DpSk-R1-Distill-Qwen-7B2026.04 | 91.8 | 301 | |
| DEERModel=DpSk-R1-Distill-Qwen-7B2026.04 | 89.7 | 297 | |
| DEERModel=DeepSeek-R1-Distill-Llama-8B2026.04 | 89.6 | 394 |