Skip to content

Setup and System Requirements

This guide covers the setup requirements for running Cosmos Reason 1 for wafer map classification task and post-training workflows.

System Requirements

Minimum Hardware Requirements

  • GPU: 1 or more GPUs (A100, H100, or later recommended)
  • Memory: Sufficient VRAM for model inference
  • Storage: Adequate disk space for model weights

Software Requirements

The setup requires the Cosmos Reason 1 repository and model to be properly installed and configured.

Installation

Install Cosmos Reason 1 Dependencies

  • Install uv and Hugging Face CLI
curl -LsSf https://astral.sh/uv/install.sh | sh
source $HOME/.local/bin/env
uv tool install -U "huggingface_hub[cli]"
hf auth login

Clone repositories

git clone https://github.com/nvidia-cosmos/cosmos-reason1
cd cosmos-reason1
git clone https://github.com/nvidia-cosmos/cosmos-rl

Install the cosmos-rl Dependencies

  • Prerequisites : Ensure your system meets the minimum versions:
  • Python ≥ 3.9
  • CUDA ≥ 12.2
  • PyTorch ≥ 2.6.0
  • vLLM ≥ 0.8.5
  • Install Redis and other Python dependencies :
apt-get update && apt-get install redis-server
cd cosmos-rl
pip install -r requirements.txt
pip install -e .

Next Steps

Once setup is complete, proceed to the post-training tutorial to learn how to use Cosmos Reason 1 for physical plausibility augmentation of ITS images.