3. Runtime, Autograd, and Interop🔗

Once the model exists, the next question is how it runs, how gradients are computed, and how TorchLean talks to workflows that feel familiar from PyTorch. This chapter starts with the everyday runtime choices, then opens the autograd and interop layers that support training, inspection, and verification.

1. 3.1. Execution Modes and Runners
2. 3.2. Autograd Walkthrough
3. 3.3. Runtime Internals and Artifacts
4. 3.4. PyTorch Round-Trip