Multi-task Language Understanding on MMLU-Redux (unseen categories)
72AccuracyAFlow
Evaluation Results
| Method | Links | |
|---|---|---|
| AFlowBase LLM=GPT-5-Mini2026.02 | 72 | |
| GDesignerBase LLM=GPT-5-Mini2026.02 | 72 | |
| HieraMASBase LLM=Qwen3-80B2026.02 | 72 | |
| SC+CoTBase LLM=GPT-5-Mini2026.02 | 68 | |
| AFlowBase LLM=Qwen3-80B2026.02 | 68 | |
| GDesignerBase LLM=Qwen3-80B2026.02 | 68 | |
| MASRouterBase LLM=Qwen3-80B2026.02 | 68 | |
| HieraMASBase LLM=GPT-5-Mini2026.02 | 68 | |
| SC+CoTBase LLM=Qwen3-80B2026.02 | 64 | |
| MASRouterBase LLM=GPT-5-Mini2026.02 | 56 |