Paper-to-slides generation on PSP 1.0 (50 sampled papers)
74.47Preference Coverage ScoreSlideTailor
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| SlideTailorBackbone=GPT-4.1, Evaluation Protocol=Zero-shot2025.12 | 74.47 | 66.65 | 72.8 | 98 | 67.64 | 75.24 | 75.8 | |
| AutoPresentBackbone=GPT-4.1, Evaluation Protocol=Zero-shot2025.12 | 72.84 | 59.58 | 49.6 | 22.4 | 28.05 | 60.2 | 48.78 | |
| SlideTailorBackbone=Qwen2.5+Qwen2.5VL, Evaluation Protocol=Zero-shot2025.12 | 70.19 | 62.16 | 68.41 | 92.8 | 48.84 | 72.84 | 69.21 | |
| PPTAgentBackbone=GPT-4.1, Evaluation Protocol=Zero-shot2025.12 | 64.41 | 54.24 | 57.6 | 97.2 | 58.36 | 71.96 | 67.3 | |
| ChatGPTBackbone=ChatGPT, Evaluation Protocol=Zero-shot2025.12 | 62.62 | 56.84 | 61.6 | 80.8 | 47 | 68.32 | 62.86 |