Text Generation on PAVE
41.97CIDErWalkGPT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| WalkGPTModel Scale=7B, Evaluation Protocol=Fine-tuned2026.03 | 41.97 | 42.36 | |
| WalkGPTModel Scale=13B, Evaluation Protocol=Fine-tuned2026.03 | 41.17 | 43.01 | |
| Sa2VA-FTEvaluation Protocol=Fine-tuned2026.03 | 38.82 | 39.66 | |
| OMG-LLaVA-FTEvaluation Protocol=Fine-tuned2026.03 | 38.01 | 38.96 | |
| GLAMM-FTEvaluation Protocol=Fine-tuned2026.03 | 37.96 | 39.12 | |
| PixelLM-FTEvaluation Protocol=Fine-tuned2026.03 | 37.49 | 38.02 | |
| GSVA-FTEvaluation Protocol=Fine-tuned2026.03 | 35.78 | 38.15 | |
| LISA-FTEvaluation Protocol=Fine-tuned2026.03 | 35.14 | 36.17 | |
| GLAMMEvaluation Protocol=Zero-shot2026.03 | 1.32 | 21.98 | |
| Sa2VAEvaluation Protocol=Zero-shot2026.03 | 1.28 | 21.02 | |
| PixelLMEvaluation Protocol=Zero-shot2026.03 | 1.08 | 21.87 | |
| LISAEvaluation Protocol=Zero-shot2026.03 | 0.97 | 17.75 | |
| OMG-LLaVAEvaluation Protocol=Zero-shot2026.03 | 0.97 | 19.99 | |
| GSVAEvaluation Protocol=Zero-shot2026.03 | 0.78 | 20.74 |