Towards Learning a Generalist Model for Embodied Navigation

Benchmark Model Rank Results
3d-question-answering-3d-qa-on-scanqa-test-wNaviLLM#3Exact Match: 26.27BLEU-1: 39.73BLEU-4: 13.90ROUGE: 40.23
visual-navigation-on-cooperative-vision-and-1NaviLLM#1dist_to_end_reduction: 7.90spl: 0.09
visual-navigation-on-room-to-room-1NaviLLM#3spl: 0.60
visual-navigation-on-soon-testNaviLLM#2Nav-SPL: 26.26SR: 35.04