George S

Thanks for the update. I do not have `torchaudio` installed, and my versions of `torch` and `torchvision` are: ``` torch 2.3.0.dev20240116+cu121 torchvision 0.18.0.dev20240116+cu121 ``` I installed these using something very...

Thanks for the details - the error at the end is not expected and I want to determine whether the issue is in `torch.compile` or `tensorrt`. Does an error appear...

Hi - thanks for the report. Are you encountering this bug when compiling a specific model? ## Fix Outline Ensure all `np.array` tensors have nonzero rank before calling `add_constant` here:...

Hello - for models with tensors that have Dynamic dimensions, you can specify the minimum, optimal, and maximum shapes via the `torch_tensorrt.Input` API. For instance, when compiling via `torch_tensorrt.compile`, you...

Hello - I tried the following minimal example to reproduce the error: - Compile `resnet18` on GPU 0 - Load two instances of the same saved model (one on GPU...

Hi - thank you for the question. Currently, there is not a way to export/serialize an artifact from `torch.compile`. Our `ir="dynamo"` path does have serialization capabilities, however, and could be...

Thank you both for the follow-up. After corresponding with @narendasan on this, the reason for which compiling the model on A100 and instantiating on 3090 is an issue is due...

Hello - as an update on this issue, we recently added #2325 to `main` which addresses compilation of the model on one GPU and loading on a different (or multiple)...

Hello - we recently added #2445 which enables the `hardware_compatibility` feature for TRT Engines generated with `ir="torch_compile"` or `ir="dynamo"`. If you are able to test out multi-GPU usage with `hardware_compatible=True`...

Hi @BDHU - thanks for the report. Our `torch.compile` backend currently does not support dynamic shape compilation, so the `dynamic=True` specification is likely causing the issue here. If you specify...