Michael A. Alcorn

Thanks for this. This behavior was driving me crazy.

Yes, if you add: ```python print(path) ``` to `load_test_set` in `tf_utils.py` it prints: `/home/michael/mv3d/../data` Changing: ```python path = os.path.join(script_path, '../data') ``` to: ```python path = os.path.join(script_path, 'data') ``` makes it...

Notably, the below code works: ```python3 import torch import torch_tensorrt class Model(torch.nn.Module): def __init__(self): super().__init__() def forward(self, topk_ind): gather_index = topk_ind.unsqueeze(-1) return gather_index def main(): model = Model().cuda() model.eval() topk_ind...

Adding `output_format="torchscript"` seems to solve the issue: ```python3 import torch import torch_tensorrt class Model(torch.nn.Module): def __init__(self): super().__init__() self.num_queries = 300 def forward(self, enc_outputs_class): _, topk_ind = torch.topk( enc_outputs_class.max(-1).values, self.num_queries, dim=1...

Re: this error: ```bash /usr/bin/ld: cannot find -ltorchtrt ``` The failing command was: ```bash aarch64-linux-gnu-g++ -pthread -shared -Wl,-O1 -Wl,-Bsymbolic-functions -Wl,-Bsymbolic-functions -Wl,-z,relro -g -fwrapv -O2 /TensorRT/build/temp.linux-aarch64-cpython-38/py/torch_tensorrt/csrc/register_tensorrt_classes.o /TensorRT/build/temp.linux-aarch64-cpython-38/py/torch_tensorrt/csrc/tensorrt_backend.o /TensorRT/build/temp.linux-aarch64-cpython-38/py/torch_tensorrt/csrc/tensorrt_classes.o /TensorRT/build/temp.linux-aarch64-cpython-38/py/torch_tensorrt/csrc/torch_tensorrt_py.o -L/TensorRT/py/torch_tensorrt/lib/ -L/opt/conda/lib/python3.6/config-3.6m-x86_64-linux-gnu...

I went back to using the current version of `WORKSPACE.jp50` and made a couple changes, including commenting out the `pip_install` as suggested in the comment above that step. Using the...

I followed the instructions under "Build from Source" [here](https://forums.developer.nvidia.com/t/pytorch-for-jetson/72048) and that seemed to work. I used the attached Dockerfile and `WORKSPACE`. Once `torch_tensorrt` is installed, you have to use: ```bash...

Confirmed fixed for me using PyTorch 2.3.1 and Torch-TensorRT 2.3.0.

Thanks for the workarounds, @gs-olive! I went with: ```python3 pseudo_images = torch.zeros(N, self.side_cells, self.side_cells, C).to(pn_feats.device) batch_idxs = torch.arange(N).repeat_interleave(P).to(pillar_pixels.device) ``` for consistency's sake and that worked for me. For the Dynamo...

That newly discovered error in the Dynamo path led me to [this issue](https://github.com/pytorch/pytorch/issues/94414) and [this issue](https://github.com/pytorch/pytorch/issues/108195), which was fixed [here](https://github.com/pytorch/pytorch/pull/108763) according to the comments.