Optical computing basic model

Minimal PyTorch image classification training script.

Setup

python -m venv .venv

Activate the venv:

• Windows PowerShell:

.\.venv\Scripts\Activate.ps1

Install dependencies:

pip install -r requirements.txt

Run

python -m src.main

Use your own dataset (no MNIST)

Put images into data/ using one of these layouts:

Option A (recommended): train/test folders

data/
  train/
    class_a/
      img1.png
    class_b/
      img2.jpg
  test/            # or `val/`
    class_a/
    class_b/

Run:

python -m src.main --data-dir data --epochs 3 --batch-size 64 --img-size 28

Option B: single folder (auto 80/20 split)

data/
  class_a/
  class_b/

Run the same command as above.

Built With

• optical
• python
• pytorch