# Download and install everything
from huggingface_hub import snapshot_download
import subprocess
import os

# Download all wheels
repo_path = snapshot_download(repo_id="RDHub/pytorch_python_310")
wheel_path = os.path.join(repo_path, "lib_wheel")

# Install all wheels
subprocess.run(["pip", "install"] + [f"{wheel_path}/*.whl"], shell=True)

Or manually:

bash

# 1. Download repository
git clone https://huggingface.co/RDHub/pytorch_python_310

# 2. Install everything with requirements file for correct versions
cd pytorch_python_310
pip install -r lib_wheel/requirements.txt --find-links lib_wheel --no-index

# 3. Set up CUDA libraries (for conda environments)
# Create activation script for automatic library path setup
mkdir -p $CONDA_PREFIX/etc/conda/activate.d
cat > $CONDA_PREFIX/etc/conda/activate.d/pytorch_cuda_libs.sh << 'EOF'
#!/bin/bash
# Set up NVIDIA CUDA library paths for PyTorch
NVIDIA_LIB_PATH=$(find $CONDA_PREFIX -path "*/nvidia/*/lib" -type d 2>/dev/null | tr '\n' ':')
CUSPARSELT_LIB_PATH=$(find $CONDA_PREFIX -path "*/cusparselt/lib" -type d 2>/dev/null | tr '\n' ':')
export LD_LIBRARY_PATH="${NVIDIA_LIB_PATH}${CUSPARSELT_LIB_PATH}${LD_LIBRARY_PATH}"
EOF
chmod +x $CONDA_PREFIX/etc/conda/activate.d/pytorch_cuda_libs.sh

# 4. Reactivate environment and test
conda deactivate && conda activate your_env_name
python -c "import torch; print(f'PyTorch {torch.__version__} - CUDA: {torch.cuda.is_available()}')"

✅ Key Versions

Package	Version	Python
PyTorch	2.7.1	3.10
Transformers	4.52.3	3.10
NumPy	2.0.2	3.10
CUDA	12.6	-

🎯 Use Cases

Perfect for:

Machine Learning projects
Large Language Model training
Computer Vision
Audio processing
Research environments

📝 Notes

No dependency conflicts - all versions tested together
Offline ready - no internet needed after download
CUDA included - ready for GPU training with library path setup
Linux x86_64 compatible
Requires conda environment - for automatic CUDA library path management

Repository Size: ~2GB
Total Packages: 80+ wheels
Tested: Ubuntu 22.04, Python 3.10