torch2trt

torch2trt copied to clipboard

Tried with max_batch_size or defining min and max shapes but only one batch is possible.

5

dynamic batch is not possible even though I defined min and max shapes or defined max_batch_size. ``` x = torch.rand([1,1,5,5,2000]).cuda().half() model = model.half() model = torch2trt(model, [x], fp16_mode=True, max_batch_size=10) y...

YOONAHLEE

Lower accuracy when converted to fp32 trt model

4

Hello,Great work! I have a pytorch segmentation model,which has been trained on a custom dataset.When I convert this model to the tensorrt version,I get a decrease in miou score from...

poornimajd

AttributeError: 'tensorrt.tensorrt.Builder' object has no attribute 'max_workspace_size'

28

When I run the Usage demo ``` import torch from torch2trt import torch2trt from torchvision.models.alexnet import alexnet # create some regular pytorch model... model = alexnet(pretrained=True).eval().cuda() # create example data...

weihaosky

[`torch.Tensor.__getitem__`] Add advanced indexing (GatherND) support to `torch.Tensor.__getitem__`

8

The following documentation also appears at the top of the file: ``` Our conversion of __getitem__ needs to handle basic and advanced indexing (specifically GatherND). See the numpy description for...

chaoz-dev

[torch2trt/converters] Add element support to `torch.Tensor.__getitem__` converter

2

Resolves #768. This PR adds support for single element arguments to `torch.Tensor.__getitem__` converter. We convert the input into a tuple if it is not already a tuple, which can occur...

chaoz-dev

08/06/2022-13:26:57] [TRT] [E] 1: [virtualMemoryBuffer.cpp::nvinfer1::StdVirtualMemoryBufferImpl::resizePhysical::148] Error Code 1: Cuda Driver (invalid argument) [08/06/2022-13:26:57] [TRT] [E] 2: [globWriter.cpp::nvinfer1::builder::HybridGlobWriter::HybridGlobWriter::193] Error Code 2: OutOfMemory (no further information)

RomanticQq

Hi, I am experiencing an issue where I load a Torch2trt model and am only getting NaN as a return value. Here is how I am saving the model after...

kneatco

- adds shape tracking with wrapper classes Fixed: - [x] flatten - [x] squeeze - [x] unsqueeze - [x] view - [x] reshape - [x] interpolate - [x] __getitem__ -...

jaybdub

libcublas.so.11

1

ImportError: libcublas.so.11: cannot open shared object file: No such file or directory cuda-11.0 cudnn8.0.1

wafaer

[Bug/Feature Request] `torch.Tensor.__getitem__` doesn't support list or `torch.Tensor` arguments for indexing/slicing when TRT converting

5

`torch.Tensor.__getitem__` supports list or `torch.Tensor` inputs for indexing, which appears to not convert correctly currently in `torch2trt`. In other words, we'd like to perform the following ops, which currently fails...

chaoz-dev