Posts by Sarunas Kalade

Fine-tuning Robotics Vision Language Action Models with AMD ROCm and LeRobot

This blog showcases training and deploying robotics policy models on AMD Instinct™ GPUs using ROCm with Hugging Face’s LeRobot framework. Recent advancements in Vision Language Action Models (VLAs) represent a breakthrough in robotics AI, combining computer vision, language understanding, and robotic control into unified architectures that can process visual observations, understand task descriptions, and generate precise motor commands.

Read more ...