Difficulty-controllable item generation on Reading Comprehension Item Generation (Human Evaluation)
61.9DASMAFIG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MAFIGBackbone=Qwen3-32B2026.05 | 61.9 | 76.19 | |
| Direct PromptingBackbone=GPT-52026.05 | 47.22 | 57.14 | |
| Direct PromptingBackbone=Qwen3-32B2026.05 | 28.17 | 42.86 |