Closed
Description
❓ Question
How to solve aten::floor converter not found?
What you have already tried
I am trying to convert a jit trace of a Fast SCNN network into TensorRT. I've confirmed that the trace was created in python3.6 using PyTorch 1.6.0. When printing the trace graph I do not even see the aten::floor operator. I also cannot locate a torch.floor operator in the original PyTorch model structure so I'm not sure what is even calling this operator? Here is the resulting error:
RuntimeError: [enforce fail at core/conversion/conversion.cpp:112] Expected converter to be true but got false
Unable to convert node: %376 : Tensor = aten::floor(%324) # /home/nmonhollen/tensorrt/venv/lib/python3.6/site-packages/torch/nn/functional.py:3010:0 (conversion.AddLayer)
Schema: aten::floor.int(int a) -> (int)
Converter for aten::floor requested, but no such converter was found.
If you need a converter for this operator, you can try implementing one yourself
or request a converter: https://www.github.com/NVIDIA/TRTorch/issues
Environment
Build information about the TRTorch compiler can be found by turning on debug messages
- PyTorch Version (e.g., 1.0): 1.6.0
- CPU Architecture: x86-64
- OS (e.g., Linux): Ubuntu 18.04
- How you installed PyTorch (
conda,pip,libtorch, source): pip - Build command you used (if compiling from source):
- Are you using local sources or building from archives: local sources (cuDNN=7.6.5, TensorRT=7.0.0.11)
- Python version: 3.6.9
- CUDA version: 10.2
- GPU models and configuration:
- Any other relevant information: