# Run TorchCode exercises locally

The static library contains 86 exercise descriptions, starter files, hints, and reference solutions. It does not execute Python or grade answers. A reviewed checkmark is only your own progress marker.

1. Install Python and create a virtual environment.
2. Install PyTorch using the command appropriate to your operating system and CPU/GPU at https://pytorch.org/get-started/locally/.
3. Read the exercise requirements. Some exercises require additional dependencies such as torchvision, external data, or a GPU.
4. Replace the unfinished starter sections with your implementation. Download the file from the browser editor.
5. Run it with `python exercise-name.py`, or open it in your preferred editor or notebook environment.

The source exercises sometimes expect notebook variables, fixtures, or additional setup. A reference solution is preserved teaching material, not a standalone test harness or a promise that the file runs without adaptation. Consult the exercise description and original TorchCode/TorchLeet source when required.

## Save your work

Drafts and personal review marks are stored only in the browser at the current website address. They do not sync to an account or another device. Export progress regularly. On another browser or address, use Import. Import keeps any existing drafts unchanged.

Existing account data on the retired server is not included in this static site. This archive contains public teaching content only.