Skip to content

CPU_SETUP.md

Latest commit

 

History

History
80 lines (55 loc) · 2.36 KB

CPU_SETUP.md

File metadata and controls

80 lines (55 loc) · 2.36 KB

TakuNet CPU-Only Setup

This document explains how to install the CPU-only version of TakuNet's dependencies.

Contents

  1. requirements_cpu.txt - CPU-compatible requirements file
  2. install_cpu_requirements.py - Python script for installing dependencies
  3. install_cpu_requirements.bat - Windows batch script for easy execution

Installation Options

Option 1: Using the Python Script (Recommended)

This method uses optimized installation options and handles timeouts for large packages:

# For Windows:
install_cpu_requirements.bat

# For Linux/Mac:
python install_cpu_requirements.py

The script performs the following steps:

  1. Installs PyTorch CPU packages first with an extended timeout
  2. Installs all remaining dependencies
  3. Uses --no-cache-dir to ensure fresh packages are downloaded

Option 2: Install PyTorch manually first

If you encounter download timeouts, you can install PyTorch separately:

# Step 1: Install PyTorch CPU packages manually
pip install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/cpu --timeout 180

# Step 2: Install all other requirements 
pip install --no-cache-dir --upgrade-strategy only-if-needed -r src/requirements/requirements_cpu.txt

Option 3: Manual Conda Setup

If you prefer using Conda:

# Create a new conda environment
conda create -n TakuNet python=3.10 -y
conda activate TakuNet

# Install CPU-only PyTorch with extended timeout
pip install torch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 --index-url https://download.pytorch.org/whl/cpu --timeout 180

# Install remaining requirements
pip install --no-cache-dir --upgrade-strategy only-if-needed -r src/requirements/requirements_cpu.txt

Troubleshooting

Handling timeout errors

If you see timeout errors like this:

HTTPSConnectionPool(...): Read timed out.

Try the following:

  1. Use a more stable internet connection
  2. Install PyTorch packages manually first with the command in Option 2
  3. If using a proxy, ensure it allows large downloads (PyTorch is ~200MB)

Verification

To verify CPU-only installation, run:

import torch
print(f"PyTorch version: {torch.__version__}")
print(f"CUDA available: {torch.cuda.is_available()}")  # Should print False