Dheeraj Peri
Ah it looks like the same issue has been filed https://github.com/NVIDIA/TensorRT/issues/3666#issuecomment-1944178324 and building from source works.
Related pytorch issue : https://github.com/pytorch/pytorch/issues/116831
This is fixed by https://github.com/pytorch/TensorRT/pull/2575
`OptimizedModule` is an artifact of `torch.compile` compilation. We don't have a way to save those models. Instead, `ir=dynamo` is the way to go if you want to serialize models. The...
@AvivSham If you use torch.compile, the control flow would cause graph breaks. If you use torch.export, there are some ways to get the control flow captured eg: https://pytorch.org/tutorials/intermediate/torch_export_tutorial.html#control-flow-ops. If you...
Closing this issue as the PR is merged
> > I've implemented your fix (refactored it a bit) as a part of https://github.com/pytorch/TensorRT/pull/2756/files. Can you take a look and make these changes? We need to add this pass...
> Hi @peri044 I checked your PR and found you made a lot of changes like `pre_export_lowering` and `post_lowering` with the help of PassManager. I will rebase on your PR...
@zidong-onepiece1 Which version of Torch-TRT are you using ? As @gs-olive mentioned, please try with `allow_shape_tensors=True`. Also, it does seem like you're using Torchscript backend. Can you please try compiling...
@arunikayadav42 Check out the extract_image_features.py script in the STT repo. It has the necessary calls for preprocessing input images. The backbone specific preprocessor implementations are in the preprocessing directory