Math Reasoning on Overall Across Five Math Reasoning Datasets
45.8Overall AccuracyED-GRPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ED-GRPOBase Model=Qwen2.5-7B-Instruct-SFT, Self-Consistency=true2026.05 | 45.8 | 11.5 | |
| ED-iDPOBase Model=Qwen2.5-7B-Instruct-SFT, Self-Consistency=true2026.05 | 45 | 10.8 | |
| ETOBase Model=Qwen2.5-7B-Instruct-SFT, Self-Consistency=true2026.05 | 44.3 | 10 | |
| GRPOBase Model=Qwen2.5-7B-Instruct-SFT, Self-Consistency=true2026.05 | 43.4 | 9.1 | |
| DAPOBase Model=Qwen2.5-7B-Instruct-SFT, Self-Consistency=true2026.05 | 43 | 8.7 | |
| CoTBase Model=Qwen2.5-7B-Instruct-SFT, Self-Consistency=true2026.05 | 41.7 | 7.4 | |
| ED-GRPOBase Model=Qwen2.5-7B-Instruct-SFT, Self-Consistency=false2026.05 | 40.5 | 6.3 | |
| GRPOBase Model=Qwen2.5-7B-Instruct-SFT, Self-Consistency=false2026.05 | 37.8 | 3.5 | |
| ED-iDPOBase Model=Qwen2.5-7B-Instruct-SFT, Self-Consistency=false2026.05 | 37.6 | 3.4 | |
| ETOBase Model=Qwen2.5-7B-Instruct-SFT, Self-Consistency=false2026.05 | 36.5 | 2.3 | |
| DAPOBase Model=Qwen2.5-7B-Instruct-SFT, Self-Consistency=false2026.05 | 36.5 | 2.2 | |
| CoTBase Model=Qwen2.5-7B-Instruct-SFT, Self-Consistency=false2026.05 | 34.3 | — | |
| ED-iDPOBase Model=LLaMA3.1-8B-Instruct-SFT, Self-Consistency=true2026.05 | 25.8 | 9.2 | |
| ED-GRPOBase Model=LLaMA3.1-8B-Instruct-SFT, Self-Consistency=true2026.05 | 25.5 | 8.9 | |
| ETOBase Model=LLaMA3.1-8B-Instruct-SFT, Self-Consistency=true2026.05 | 24.7 | 8.2 | |
| GRPOBase Model=LLaMA3.1-8B-Instruct-SFT, Self-Consistency=true2026.05 | 23.1 | 6.6 | |
| CoTBase Model=LLaMA3.1-8B-Instruct-SFT, Self-Consistency=true2026.05 | 22.9 | 6.4 | |
| DAPOBase Model=LLaMA3.1-8B-Instruct-SFT, Self-Consistency=true2026.05 | 21.8 | 5.3 | |
| ED-GRPOBase Model=LLaMA3.1-8B-Instruct-SFT, Self-Consistency=false2026.05 | 20.6 | 4.1 | |
| ED-iDPOBase Model=LLaMA3.1-8B-Instruct-SFT, Self-Consistency=false2026.05 | 20 | 3.4 | |
| ETOBase Model=LLaMA3.1-8B-Instruct-SFT, Self-Consistency=false2026.05 | 18.2 | 1.6 | |
| GRPOBase Model=LLaMA3.1-8B-Instruct-SFT, Self-Consistency=false2026.05 | 17.9 | 1.4 | |
| DAPOBase Model=LLaMA3.1-8B-Instruct-SFT, Self-Consistency=false2026.05 | 17.6 | 1 | |
| CoTBase Model=LLaMA3.1-8B-Instruct-SFT, Self-Consistency=false2026.05 | 16.5 | — |