Skip to content

System & Hardware Requirements

This section outlines hardware recommendations and software dependencies required to run training and inference workloads reproducibly.

Software Prerequisites

  • Python: >= 3.11
  • Package & Environment Manager: uv (recommended) or pip
  • Container Runtime (Optional): Docker + nvidia-docker (NVIDIA Container Toolkit)

🐍 Changing Python Version with uv

uv allows seamless switching to any lower, higher, or specific Python release without manually installing external Python distributions:

# 1. Download and pin a specific Python version (e.g., 3.10, 3.12, or 3.13)
uv python pin 3.12

# 2. Re-sync virtual environment with the target Python version
uv sync --all-groups --python 3.12

[!NOTE] If switching to a lower Python version (e.g., Python 3.10), update the requires-python constraint in pyproject.toml: 

[project]
requires-python = ">=3.10"

Hardware Recommendations

Component Minimum Recommended
CPU 4 Cores 16+ Cores
RAM 16 GB 64 GB+
GPU NVIDIA GPU (8GB VRAM) NVIDIA A100 / H100 (40GB+ VRAM)
CUDA 11.8+ / 12.1+ 12.2+

CUDA & PyTorch GPU Environment

To containerize GPU workloads reproducibly, inspect the provided GPU Docker Tutorial:

# Build CUDA container
docker build -t dl-template:latest docker/