Installation¶

Requirements¶

• Python 3.10-3.13
• PyTorch 2.0+
• CUDA (optional, for GPU training)

Install from PyPI¶

Core Installation¶

pip install autotimm

This installs the core package with all vision task dependencies:

Core Dependencies:

• torch>=2.0
• torchvision>=0.15
• timm>=1.0
• pytorch-lightning>=2.0
• torchmetrics>=1.0
• numpy>=1.23
• albumentations>=1.3 (included by default)
• opencv-python-headless>=4.8 (included by default)
• pycocotools>=2.0 (included by default)
• huggingface_hub>=0.20
• transformers>=4.30 (included by default)
• matplotlib>=3.7 (included by default)
• watermark>=2.3
• loguru>=0.7
• plotly>=5.0

Optional Dependencies¶

Logger Backends¶

# TensorBoard
pip install autotimm[tensorboard]

# Weights & Biases
pip install autotimm[wandb]

# MLflow
pip install autotimm[mlflow]

Everything¶

pip install autotimm[all]

Includes all optional dependencies: tensorboard, mlflow, and wandb.

Development Installation¶

For contributing or developing locally:

git clone https://github.com/theja-vanka/AutoTimm.git
cd AutoTimm
pip install -e ".[dev,all]"

This installs:

• All optional dependencies
• Development tools: pytest, pytest-cov, ruff, black

Verify Installation¶

import autotimm as at  # recommended alias

# Check available backbones
print(len(at.list_backbones()))  # 1000+ models

# Check version
print(at.__version__)

GPU Support¶

AutoTimm automatically uses GPU when available. Ensure you have CUDA-compatible PyTorch:

# Check PyTorch CUDA support
python -c "import torch; print(torch.cuda.is_available())"

If CUDA is not available, install PyTorch with CUDA support from pytorch.org.

Troubleshooting¶

For common installation issues, see the Troubleshooting - Installation which covers:

• ImportError: No module named 'timm'
• CUDA out of memory
• Albumentations not found
• And more installation-related issues