Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.
And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.
Check torchlib torch version
See original GitHub issueIm running nvcr.io/nvidia/tritonserver:22.07-pyt-python-py3 on aarch64 and on x86, so i have a completely different output from PyTorch model.
I guess that it is not a triton bug actually, probably different torchlib versions are used (at least one of them built for arm), so how i can check the version of torchlib, to compate them on x86 and arm biuld? And probably reproduce this bug separately from triton.
I see shared library in backends/pytorch/libtorch.so, but actually that has no version in the file like backends/pytorch/libopencv_imgcodecs.so.3.4.
Issue Analytics
- State:
- Created a year ago
- Comments:5 (5 by maintainers)
Top Results From Across the Web
Top Related Medium Post
Top Related StackOverflow Question
Troubleshoot Live Code
Top Related Reddit Thread
Top Related Hackernoon Post
Top Related Tweet
Top Related Dev.to Post
Top Related Hashnode Post
@alxmamaev Triton just takes the PyTorch version built for the corresponding
nvcr.io/nvidia/pytorch:YY.MM-py3release so we are in alignment with the other DL framework containers.ex:
nvcr.io/nvidia/pytorch:22.07-py3should generally have the same pytorch version asnvcr.io/nvidia/tritonserver:22.07-py3@rmccorm4 ok, thank you. It’s it is what I wanted. Will try to reproduce bug separately.