Setting Up Your AI Development Environment

# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
# AI Development Environment Setup
# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

# OPTION 1: Conda (recommended - handles CUDA + Python together)
# conda create -n ai-course python=3.11 -y
# conda activate ai-course
# conda install pytorch torchvision torchaudio pytorch-cuda=12.1 -c pytorch -c nvidia

# OPTION 2: pip + venv
# python -m venv venv
# source venv/bin/activate  (Linux/Mac) or venv\Scripts\activate (Windows)
# pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121

# Core AI packages
# pip install transformers datasets accelerate peft  # HuggingFace stack
# pip install openai langchain langchain-community   # LLM apps
# pip install diffusers                              # Stable Diffusion
# pip install scikit-learn matplotlib seaborn        # ML + visualization
# pip install jupyter jupyterlab ipywidgets          # notebooks
# pip install fastapi uvicorn                        # model serving

# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
# VERIFY YOUR SETUP - run this script to confirm everything works
# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

import sys
import torch

def check_environment():
    print("=" * 50)
    print("AI Environment Health Check")
    print("=" * 50)

    # Python version
    print(f"Python: {sys.version.split()[0]}")

    # PyTorch
    print(f"PyTorch: {torch.__version__}")

    # GPU / CUDA
    if torch.cuda.is_available():
        print(f"GPU: {torch.cuda.get_device_name(0)}")
        print(f"CUDA: {torch.version.cuda}")
        print(f"VRAM: {torch.cuda.get_device_properties(0).total_memory / 1e9:.1f} GB")
    elif hasattr(torch.backends, 'mps') and torch.backends.mps.is_available():
        print("GPU: Apple Silicon MPS (Metal Performance Shaders)")
    else:
        print("GPU: Not available - using CPU (slower but works)")

    # Test tensor computation on GPU/MPS or CPU
    device = (
        "cuda" if torch.cuda.is_available()
        else "mps" if hasattr(torch.backends, "mps") and torch.backends.mps.is_available()
        else "cpu"
    )
    print(f"\nActive compute device: {device.upper()}")

    # Quick GPU test
    a = torch.randn(1000, 1000, device=device)
    b = torch.randn(1000, 1000, device=device)
    c = a @ b  # matrix multiply - GPU should be 100x faster
    print(f"Test matmul (1000x1000): OK - result shape {tuple(c.shape)}")

    # HuggingFace
    try:
        import transformers
        print(f"Transformers: {transformers.__version__}")
    except ImportError:
        print("Transformers: NOT installed - run: pip install transformers")

    # OpenAI
    try:
        import openai
        print(f"OpenAI SDK: {openai.__version__}")
    except ImportError:
        print("OpenAI: NOT installed - run: pip install openai")

    print("\n✅ Environment check complete!")
    return device

device = check_environment()

# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
# Google Colab - free GPU (T4, A100 with Colab Pro)
# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
# Open: colab.research.google.com
# Runtime → Change runtime type → GPU (T4)
# Free tier: ~12h sessions, resets state between sessions
# Colab Pro ($10/mo): A100 access, longer sessions, more RAM

# GPU VRAM requirements for different tasks:
gpu_requirements = {
    "Train small CNN (CIFAR-10)":             "4 GB VRAM",
    "Fine-tune DistilBERT":                   "4-8 GB VRAM",
    "Fine-tune Llama 3 8B (4-bit QLoRA)":    "6-8 GB VRAM",
    "Fine-tune Llama 3 70B (4-bit QLoRA)":   "40 GB VRAM (A100)",
    "Run Stable Diffusion XL inference":      "8-12 GB VRAM",
    "Train GPT-2 from scratch":               "8 GB VRAM",
}

print("\nGPU VRAM requirements:")
for task, vram in gpu_requirements.items():
    print(f"  {task:45s} → {vram}")