Visual Question Answering on Road Reasoning Dataset (val)
50.1Total Accuracygemma-4-26b-a4b-it
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| gemma-4-26b-a4b-itModel=gemma-4-26b-a4b-it2026.05 | 50.1 | 52.9 | 38.5 | |
| qwen3-vl-8b-instructModel=qwen3-vl-8b-instruct2026.05 | 49.4 | 50.9 | 43.5 | |
| qwen3-vl-30b-a3b-instructModel=qwen3-vl-30b-a3b-instruct2026.05 | 46 | 47.8 | 38.5 | |
| gemma-4-E2B-itModel=gemma-4-E2B-it2026.05 | 38.4 | 41.5 | 26 |