multi-view-reasoning

Tag

Cards List
#multi-view-reasoning

Learning to Think and Explore in 3D World

Hugging Face Daily Papers ↗ · 2026-10-15 Cached

ACT3D 训练视觉-语言模型通过强化学习在 3D 环境中主动探索,仅凭单一第一人称视角即可在六个 3D 基准上达到 53.2% 平均准确率,较基座 Qwen 提升 22.0% 并超越 Gemini 3.6 Flash。

0 favorites 0 likes
← Back to home

Submit Feedback