Tag
LingBot-Video, a 30B parameter MoE-based video foundation model for embodied intelligence, has been released on Hugging Face with only 3B active parameters at inference, augmented with 70K hours of embodied data.