NN.Runtime.PyTorch.Export

Reusable PyTorch export/adaptation surface.

Use this umbrella when you want the runtime bridge, not the example models:

• Export.Core provides shared Python string-generation utilities.
• Export.IRPyTorch lowers a TorchLean NN.IR.Graph plus parameters into readable PyTorch nn.Module source.
• Export.ONNX emits a conservative ONNX-to-torchlean.ir.v1 adapter for static graph fragments, including expanded graph lowerings for common Conv/Gemm/BatchNorm patterns.
• Export.StateDict emits the general checkpoint-to-JSON adapter for PyTorch state_dict artifacts.
• Export.TorchExport emits the Python graph-capture adapter for PyTorch nn.Module → TorchLean IR JSON.

For ONNX workflows, the architecture is the same: the Python-side adapter reads ONNX and emits torchlean.ir.v1; Lean then accepts or rejects the result through NN.Runtime.PyTorch.Import.TorchExport. That keeps graph import tied to the same checked IR contracts used by torch.export/FX capture. The adapter validates graph structure and shapes; runtime execution of imported parameterized nodes still needs the matching payload store.

Example-specific MLP/CNN/Transformer code lives beside its reference artifacts under NN.Examples.Interop.PyTorch.{MLP,CNN,Transformer}.*.