Prerequisites: NVIDIA GPU(s), Docker with the NVIDIA Container ToolkitOn GPU cloud platforms like Vast.ai and RunPod, you just need to create a custom template with the image ghcr.io/eduardoslonski/telescope:latest — they handle the rest. On Lambda, CoreWeave, and similar VM-based platforms, Docker comes preinstalled so you can pull and run the image directly.
Pull the image
Start the container
--ipc=host and --shm-size=16g are required for NCCL shared memory across GPUs. --ulimit memlock=-1 unlocks GPU memory pinning for efficient transfers. Platforms like Vast.ai and RunPod handle these flags automatically when using their template system.
/root/telescope inside the container.Set up Weights & Biases
Telescope logs training data to Weights & Biases, which the UI Visualization uses to display metrics and rollouts. Log in before starting training:Run training
Inside the container, run training with any of the example configs: Prerequisites: NVIDIA GPU(s), Python 3.11+, uvInstall uv (if not already installed):Set up Weights & Biases
Telescope logs training data to Weights & Biases, which the UI Visualization uses to display metrics and rollouts. Log in before starting training:Run training
Run training with any of the example configs:Installing from source does not include some performance libraries (Transformer Engine, NVIDIA Apex) that are pre-built in the Docker image and require complex compilation from source. Training will still work fine, just without the accelerated performance those provide.
