Scale Robot Reinforcement Learning with NVIDIA Isaac Lab on Amazon SageMaker AI
Esta publicación demuestra cómo entrenar políticas de robots para el humanoide Unitree H1 utilizando NVIDIA Isaac Lab en Amazon SageMaker AI. Explora dos opciones de cómputo: Amazon SageMaker HyperPod y Amazon SageMaker Training Jobs.
![Trained a Qwen2.5-0.5B-Instruct bf16 model on Reddit post summarization task with GRPO written from scratch in PyTorch - updates! [P]](/cdn-cgi/image/width=3840,quality=75,format=webp/https://preview.redd.it/7nrsulwdkbvg1.png?width=140&height=69&auto=webp&s=7c61d2f68d6b094614b5dff0cb9347873885e226)