DL_Permutations

Visualizing permutation spaces using deep learning techniques.

Papers: https://drive.google.com/drive/folders/12XPfBZJrhvgZ8Cdq9mIjcBwNCj0IWOCO?usp=sharing

Setup

Requires uv.

uv sync

PyTorch is installed from the CUDA 13.0 build (index pinned in pyproject.toml). GPU use needs an NVIDIA driver with CUDA 13.0+.

Usage

uv run jupyter lab

Then open a notebook from notebooks/.

Contrastive trajectory experiment:

uv run python -m dl_permutations.contrastive_trajectory_learning

The same through make: make sync, make lab, make train, make notebooks (executes all notebooks into build/notebooks/), make clean.

Structure

src/dl_permutations/   # source package (genetic algorithm, models)
notebooks/             # Jupyter notebooks
outputs/               # generated artifacts (git-ignored, except stn_pca.png)