Emotion-Guided Content Generation (EGCG) on AICA-Bench 1.0 (test)
0.26MSEAICA-Bench Scoring Model
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| AICA-Bench Scoring Modelbase_model=QwenVL2.5-7B, training_regime=fine-tuned with human-labeled data2026.04 | 0.26 | 0.24 | 0.9 | |
| ChatGPT-4o2026.04 | 0.58 | 0.47 | 0.775 | |
| Qwen2.5VL-7B2026.04 | 0.845 | 0.645 | 0.528 | |
| Gemini2.5-Pro2026.04 | 0.855 | 0.62 | 0.735 |