Question Answering on GSM8K 10 (test)
97.92F1 ScoreOrchMAS
Evaluation results
| Method | Links | ||
|---|---|---|---|
| OrchMASCategory=Ours2026.03 | 97.92 | 97.92 | |
| Qwen3-4BCategory=GRPO2026.03 | 92.71 | 92.71 | |
| GEPACategory=MAS, Base Model=GPT-4o-mini2026.03 | 91.1 | 84.38 | |
| GPT-4o-miniCategory=CoT Reasoning2026.03 | 87.02 | 84.38 | |
| GPT-4o-miniCategory=Baseline2026.03 | 86.61 | 83.33 | |
| Qwen3-4BCategory=Baseline2026.03 | 84.38 | 84.38 | |
| OPROCategory=MAS, Base Model=GPT-4o-mini2026.03 | 82.11 | 63.54 | |
| Qwen3-4BCategory=CoT Reasoning2026.03 | 81.25 | 81.25 | |
| TextGradCategory=MAS, Base Model=GPT-4o-mini2026.03 | 81.21 | 70.83 | |
| Qwen3-4BCategory=SFT2026.03 | 32.29 | 32.29 |