PyTorch Setup: Installation, Devices, Seeds and Reproducibility

Recommended Project Layout

A clean PyTorch setup prevents many frustrating bugs. You need matching package versions, correct device handling, fixed seeds for experiments, and a project layout that separates data, model, training, evaluation, and inference code.

Device management is explicit in PyTorch. Tensors and models must live on the same device. Many beginner errors come from moving the model to GPU but leaving the batch on CPU, or the reverse.

A production-friendly layout keeps model definition independent from training scripts. This lets you test the model, run inference separately, and reuse data pipelines.

`src/data.py` for datasets, transforms, and dataloaders.
`src/model.py` for neural network modules.
`src/train.py` for training and validation loops.
`src/infer.py` for loading checkpoints and predicting.
`configs/` for hyperparameters and experiment settings.

Reproducibility

Machine learning is not perfectly deterministic across every hardware and kernel combination, but seeds and consistent configuration make experiments much easier to compare.

Set Python, NumPy, and PyTorch seeds.
Log model version, dataset version, hyperparameters, and metrics.
Save checkpoints with epoch, model state, optimizer state, and validation score.

Device and Seed Setup

Use one helper to keep device and seed behavior consistent across notebooks, scripts, and tests.

Device and Seed Setup

import random
import numpy as np
import torch

def setup_experiment(seed: int = 42):
    random.seed(seed)
    np.random.seed(seed)
    torch.manual_seed(seed)
    torch.cuda.manual_seed_all(seed)

    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    print(f"Using device: {device}")
    return device

device = setup_experiment()

x = torch.randn(4, 3).to(device)
model = torch.nn.Linear(3, 2).to(device)
out = model(x)

print(out.device)
print(out.shape)

Every tensor batch and the model must be on the same device.
Print shapes and devices early while building a training script.

Report the PyTorch Runtime and Selected Device

import torch

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
print('torch:', torch.__version__)
print('device:', device.type)
print('cuda devices:', torch.cuda.device_count())

Record this output with the Python version and installation command when diagnosing environment-specific failures.

Before you move on