Mathematical Reasoning on AIME24 (Pass@1 Avg@16)
51.3Pass@1 (Avg@16)DLER-R1-7B-Research
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DLER-R1-7B-ResearchBackbone=DeepSeek-R1-Distill-Qwen-7B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 51.3 | 3,209 | 1.6 | |
| Laser-DE-L4096-7BBackbone=DeepSeek-R1-Distill-Qwen-7B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 47.7 | 4,642 | 1.1 | |
| AdaptThink-7B-delta0.05Backbone=DeepSeek-R1-Distill-Qwen-7B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 45.6 | 6,255 | 0.7 | |
| SB DS7B alpha 2Backbone=DeepSeek-R1-Distill-Qwen-7B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 45.4 | 3,955 | 1.02 | |
| L1-Qwen-7B-MaxBackbone=DeepSeek-R1-Distill-Qwen-7B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 44.8 | 4,041 | 0.96 | |
| APR-7B (Ours)Backbone=DeepSeek-R1-Distill-Qwen-7B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 44 | 3,051 | 1.04 | |
| TrainingEfficient DS-7BBackbone=DeepSeek-R1-Distill-Qwen-7B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 42.1 | 6,230 | 0.42 | |
| Original ModelBackbone=DeepSeek-R1-Distill-Qwen-7B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 37.7 | 6,707 | — | |
| DLER-R1-1.5B-ResearchBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 32.1 | 3,376 | 2.55 | |
| DS-1.5B-thinkprune-iter2kBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 30 | 4,670 | 2.05 | |
| TrainingEfficient DS-1.5BBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 29.2 | 6,245 | 1.7 | |
| APR-1.5B (Ours)Backbone=DeepSeek-R1-Distill-Qwen-1.5B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 29 | 3,767 | 2.01 | |
| L1-Qwen-1.5B-MaxBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 28.8 | 2,893 | 2.1 | |
| Laser-L8192-1.5BBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 27.3 | 6,139 | 1.42 | |
| AdaptThink-1.5B-delta0.05Backbone=DeepSeek-R1-Distill-Qwen-1.5B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 26.2 | 6,172 | 1.25 | |
| Laser-DE-L4096-1.5BBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 25.6 | 5,173 | 1.29 | |
| Original ModelBackbone=DeepSeek-R1-Distill-Qwen-1.5B, Sampling Temperature=0.6, Max Response Length=8192, Rollout samples=162026.01 | 19.2 | 7,278 | — |