ActQuant: Sub-4-bit Action-Guided Quantization for Vision-Language-Action Models

Published in NeurIPS 2026, 2026