Task Progress Estimation on Ego4D
19.25PMAER2VLM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| R2VLMSize=7B2026.03 | 19.25 | 3.88 | 31.8 | 76.1 | |
| R2VLM-SFTSize=7B2026.03 | 23.03 | 3.98 | 26.6 | 68.5 | |
| R2VLM-ZeroSize=7B2026.03 | 23.58 | 10.97 | 25.6 | 67.6 | |
| GPT-52026.03 | 25.04 | 15.89 | 25.9 | 74.9 | |
| GVL-SFTSize=7B2026.03 | 26.8 | 13.26 | 25.5 | 66.5 | |
| Qwen2.5-VLSize=72B2026.03 | 26.88 | 11.11 | 25.4 | 62.4 | |
| InternVL3Size=78B2026.03 | 28.16 | 13.37 | 20.4 | 58.6 | |
| Qwen3-VL-Plus2026.03 | 28.2 | 10.14 | 25.1 | 65.2 | |
| Gemini-2.5-Pro2026.03 | 28.22 | 16.71 | 21.7 | 71.3 | |
| Qwen2.5-VLSize=7B2026.03 | 28.32 | 10.11 | 20.6 | 48.5 | |
| InternVL3Size=8B2026.03 | 29.43 | 16.8 | 18.7 | 51.1 | |
| MiniCPM-V-2.6Size=8B2026.03 | 30.17 | 18.03 | 20.8 | 48.2 | |
| GLM-4.1V-ThinkingSize=9B2026.03 | 30.45 | 8.21 | 19.7 | 48.8 | |
| LIVSize=<1B2026.03 | 34.57 | 24.51 | 15.1 | 48.5 | |
| ROVER2026.03 | 39.83 | 6.48 | 17.4 | 54.8 |