Mathematical Reasoning on AIME 2025 (Acc, Tok #N, CR, Think #R)
83.3AccDAMI-Pref
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DAMI-PrefModel Base Pair=Qwen3-4B-2507-Thinking/Qwen3-4B-2507-Instruct, Sampling Rounds=8x2026.01 | 83.3 | 17,488 | 82 | 92 | |
| DAMI-ConfModel Base Pair=Qwen3-4B-2507-Thinking/Qwen3-4B-2507-Instruct, Sampling Rounds=8x2026.01 | 79.2 | 16,412 | 77 | 94 | |
| DEERModel Base Pair=Qwen3-4B-2507-Thinking/Qwen3-4B-2507-Instruct, Sampling Rounds=8x2026.01 | 76.7 | 19,287 | 90 | 100 | |
| MI-07Model Base Pair=Qwen3-4B-2507-Thinking/Qwen3-4B-2507-Instruct, Sampling Rounds=8x2026.01 | 75 | 16,086 | 75 | 98 | |
| DAMI-PromptModel Base Pair=Qwen3-4B-2507-Thinking/Qwen3-4B-2507-Instruct, Sampling Rounds=8x2026.01 | 75 | 13,009 | 61 | 88 | |
| TAModel Base Pair=Qwen3-4B-2507-Thinking/Qwen3-4B-2507-Instruct, Sampling Rounds=8x2026.01 | 73.3 | 15,546 | 73 | 98 | |
| RoutingModel Base Pair=Qwen3-4B-2507-Thinking/Qwen3-4B-2507-Instruct, Sampling Rounds=8x2026.01 | 73.3 | 21,035 | 99 | 100 | |
| ThinkingModel Base Pair=Qwen3-4B-2507-Thinking/Qwen3-4B-2507-Instruct, Sampling Rounds=8x2026.01 | 71.7 | 21,340 | 100 | 100 | |
| MI-05Model Base Pair=Qwen3-4B-2507-Thinking/Qwen3-4B-2507-Instruct, Sampling Rounds=8x2026.01 | 65.8 | 10,533 | 49 | 54 | |
| DAMI-PredModel Base Pair=Qwen3-4B-2507-Thinking/Qwen3-4B-2507-Instruct, Sampling Rounds=8x2026.01 | 53.3 | 8,776 | 41 | 0 | |
| TIESModel Base Pair=Qwen3-4B-2507-Thinking/Qwen3-4B-2507-Instruct, Sampling Rounds=8x2026.01 | 51.7 | 8,419 | 39 | 0 | |
| MI-03Model Base Pair=Qwen3-4B-2507-Thinking/Qwen3-4B-2507-Instruct, Sampling Rounds=8x2026.01 | 51.7 | 9,280 | 43 | 0 | |
| InstructModel Base Pair=Qwen3-4B-2507-Thinking/Qwen3-4B-2507-Instruct, Sampling Rounds=8x2026.01 | 49.2 | 8,384 | 39.3 | 0 | |
| DAMI-PrefModel Base Pair=DeepSeek-R1-Distill-Qwen-7B/Qwen2.5-Math-7B, Sampling Rounds=8x2026.01 | 32.5 | 16,837 | 78 | 94 | |
| ThinkingModel Base Pair=DeepSeek-R1-Distill-Qwen-7B/Qwen2.5-Math-7B, Sampling Rounds=8x2026.01 | 30 | 21,681 | 100 | 100 | |
| DEERModel Base Pair=DeepSeek-R1-Distill-Qwen-7B/Qwen2.5-Math-7B, Sampling Rounds=8x2026.01 | 30 | 15,941 | 74 | 100 | |
| MI-07Model Base Pair=DeepSeek-R1-Distill-Qwen-7B/Qwen2.5-Math-7B, Sampling Rounds=8x2026.01 | 30 | 17,169 | 79 | 60 | |
| DAMI-ConfModel Base Pair=DeepSeek-R1-Distill-Qwen-7B/Qwen2.5-Math-7B, Sampling Rounds=8x2026.01 | 30 | 16,174 | 75 | 90 | |
| TAModel Base Pair=DeepSeek-R1-Distill-Qwen-7B/Qwen2.5-Math-7B, Sampling Rounds=8x2026.01 | 25.8 | 11,189 | 52 | 83 | |
| MI-05Model Base Pair=DeepSeek-R1-Distill-Qwen-7B/Qwen2.5-Math-7B, Sampling Rounds=8x2026.01 | 20 | 10,665 | 49 | 0 | |
| MI-03Model Base Pair=DeepSeek-R1-Distill-Qwen-7B/Qwen2.5-Math-7B, Sampling Rounds=8x2026.01 | 16.7 | 4,485 | 21 | 0 | |
| DAMI-PredModel Base Pair=DeepSeek-R1-Distill-Qwen-7B/Qwen2.5-Math-7B, Sampling Rounds=8x2026.01 | 13.3 | 13,813 | 64 | 26 | |
| TIESModel Base Pair=DeepSeek-R1-Distill-Qwen-7B/Qwen2.5-Math-7B, Sampling Rounds=8x2026.01 | 10 | 1,827 | 8 | 0 | |
| DAMI-PromptModel Base Pair=DeepSeek-R1-Distill-Qwen-7B/Qwen2.5-Math-7B, Sampling Rounds=8x2026.01 | 10 | 4,962 | 23 | 0 | |
| RoutingModel Base Pair=DeepSeek-R1-Distill-Qwen-7B/Qwen2.5-Math-7B, Sampling Rounds=8x2026.01 | 6.7 | 2,851 | 13 | 10 | |
| InstructModel Base Pair=DeepSeek-R1-Distill-Qwen-7B/Qwen2.5-Math-7B, Sampling Rounds=8x2026.01 | 3.3 | 1,104 | — | 0 |