SpatialClaw 不用额外训练，靠写代码搞定复杂视觉任务，在 20 个基准上平均提升 11.2 分，还兼容多种模型。
NVIDIA Research 推出 SpatialClaw，一个无需训练的智能体，通过编写 Python 代码作为动作接口。它在持久内核中动态组合感知模块，检查中间结果并跨步骤调整策略。感知输出作为普通变量，可结合 NumPy、SciPy 等库复用。SpatialClaw 在 20 个基准上平均比先前方法高 11.2 分，在 6 种不同模型骨干上表现稳定。
Code is the right action interface for spatial reasoning agents. New from NVIDIA Research: SpatialC...
Code is the right action interface for spatial reasoning agents. New from NVIDIA Research: SpatialClaw, a training-free agent that uses code as its action interface for complex visual tasks. Instead of calling a fixed set of pre-defined tools, the agent writes Python inside a persistent kernel, so it can compose perception modules, inspect intermediate results, and revise its strategy across steps. Perception outputs become ordinary variables it can reuse and combine with libraries like NumPy and SciPy. With no benchmark-specific or model-specific tuning, it beats a recent prior agent by 11.2 points across 20 benchmarks and holds up consistently across six different model backbones. You can check out SpatialClaw here: nvda.ws/4esHxr9 💬 8 🔄 17 ❤️ 95 👀 5473 📊 29 ⚡