Hi, I notice from your README file that torch.compile provides a great speedup. However, I didn't see where you implemented torch.compile for your train_acc.py. I tried to add torch.compile(model) for models containing Mamba blocks, but it causes some errors. May I know how you implemented torch.compile for your zigzag model? Thanks!
Hi, I notice from your README file that torch.compile provides a great speedup. However, I didn't see where you implemented torch.compile for your train_acc.py. I tried to add torch.compile(model) for models containing Mamba blocks, but it causes some errors. May I know how you implemented torch.compile for your zigzag model? Thanks!