Captioning on W3D Normal Driving original (test)
44BLEULLada*
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| LLada*Training Regime=Full-data (W3D) [~70k samples]2025.11 | 44 | 36 | 58 | 96 | |
| FSDAMTraining Regime=Few-shot2025.11 | 42 | 37 | 48 | 83 | |
| GazeXplain*Training Regime=Full-data (W3D) [~70k samples]2025.11 | 31 | 30 | 22 | 42 | |
| Qwen-VLTraining Regime=In-context learning (no fine-tuning)2025.11 | 13 | 18 | 22 | 36 | |
| MLNet + LLaVATraining Regime=Few-shot2025.11 | 13 | 19 | 27 | 31 | |
| LLaVATraining Regime=Zero-shot (no training)2025.11 | 12 | 14 | 23 | 35 | |
| DeepGazeI + LLaVATraining Regime=Few-shot2025.11 | 12 | 23 | 28 | 14 | |
| DeepGazeIIE + LLaVATraining Regime=Few-shot2025.11 | 11 | 18 | 26 | 17 | |
| Qwen-VLTraining Regime=Zero-shot (no training)2025.11 | 10 | 19 | 28 | 34 |