RadixRootMind/RadixQuantVLA

RadixQuantVLA is a unified, Lego-like framework for post-training quantization, benchmarking, and deployment preparation of Vision-Language-Action models, integrating QuantVLA, Omega-QVLA, OpenVLA, UniVLA, StarVLA, GR00T, and Pi0.5 routes with standardized launchers and validation on LIBERO benchmarks.

语言：Jupyter Notebook · GitHub 星标：90 · Fork：10

在 GitHub 上查看原仓库