Dheeraj Peri

- How does the merge adjacent segments work ? What's the criteria to merge ?

Hello @mfeliz-cruise , I removed [this snippet ](https://github.com/pytorch/TensorRT/blob/master/core/partitioning/partitioning.cpp#L121-L133)and ran the tests and they ran successfully. I'm wondering what is the usecase behind this snippet ? From my understanding this is...

Yes, we need a plugin. Group norm plugin is shipped with TensorRT package (as a part of libnvinfer_plugin.so). The source code is available here https://github.com/NVIDIA/TensorRT/tree/main/plugin/groupNormalizationPlugin. An implementation would look like...

@zsef123 Can you share your model or your debug logs? The debug logs can be generated via ``` with torch_tensorrt.logging.debug(): model_new_trt = trt.compile( model_new, inputs=[trt.Input( min_shape=[1, 1, 210, 748, 748],...

Can you share some more information ? Is the error coming from torch.jit.script () call or from trtorch.compile ? If it's a TRTorch error, can you enable debug log by...

End to end compilation: `at::Int` is a bit tricky to implement and we are looking into it. This op just converts an integer scalar to an integer tensor which can...

@narendasan the fix seems ok to me. The graph for inceptionv3 scripted has the following ``` %self.Conv2d_1a_3x3.conv.bias.1 : NoneType = prim::Constant() %1611 : str = prim::Constant[value="Scripted Inception3 always returns Inception3...

@void-main @narendasan Looks like this is a special case where they are using `aux` for training which is by default optional null tensor https://github.com/pytorch/vision/blob/main/torchvision/models/inception.py#L135-L155

Also add this as a test case for this PR https://github.com/pytorch/TensorRT/pull/1067 ``` class ModifyingOpError(nn.Module): def __init__(self): super(ModifyingOpError, self).__init__() def forward(self, x, y): mod_list = [x] if x.sum() > y.sum(): mod_list.append(y)...

@batrlatom @joshuafc We had our focus on transformers and dynamic shape models in the recent times. We couldn't get to finish an example based on detectron @bowang007 Can you share...