Dheeraj Peri
1) The resnet model from timm is trained on ImageNet dataset with an image size of 224x224. 2) You are using CIFAR 10 for calibration which has image size of...
>What is the loss of test accuracy after you finish torch-TRT? Is the loss of precision after quantization as expected? We didn't perform RN50 PTQ test on ImageNet (in Torch-TRT)...
I've tried setting the model in eval mode by `scripted_module = torch.jit.script(encoder.eval())` and this issue no longer exists. However, I see a segfault after this log which is a different...
Merging this PR as this doesn't affect the library. Please re-open incase you see any issues
@frank-wei @narendasan It looks like FX tests test-py-fx-x86_64-pyt-release are failing with the following error. Is this a known issue ? ``` RuntimeError: CUDA error: an illegal memory access was encountered...
We currently don't support dynamic shape + fallback. During fallback, we generate random inputs of static shape to perform shape analysis and it fails here since the shape has -1...
This error is from [aten_to_trt_type mapping](https://github.com/NVIDIA/TRTorch/blob/master/core/util/trt_util.cpp#L226). The pytorch tensor in your graph has a double datatype and TRT doesn't support beyond float datatype. My guess: Looking at your log, it...
@inocsin Can you try with torch.fx https://pytorch.org/docs/stable/fx.html#direct-graph-manipulation ? I'm not sure if that would work but we can basically modify layers/add new layers etc. Hopefully we can modify these constant...
@lhai37 I don't think we support optional tensors at the moment. cc @narendasan. We expect inputs and outputs of a module to be torch::Tensors. Can you share how your torchscript...
@lixiaolx Do you still see the same error with the latest main ?