Kuo, Megan

Hello @JibAxelera and @amemov I believe that under the hood, torch_mlir.fx relies on FX symbolic tracing of a nn.Module. According to the [documentation](https://pytorch.org/docs/stable/fx.html#limitations-of-symbolic-tracing), FX only handles static control flow, it...

Hi @cathyzhyi, Thanks for the clear problem statement! Even though this comment is a few years old, I’m interested in picking this up. I’m new to Torch-MLIR and wanted to...