question-mark
Stuck on an issue?

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.

Tensorflow.js looks for CUDA 10.0 with libtensorflow recompiled against CUDA 10.1

See original GitHub issue

To get help from the community, we encourage using Stack Overflow and the tensorflow.js tag.

TensorFlow.js version

2.0.0

Browser version

n/a, Node.js v14.2.0

Describe the problem or feature request

If I recompile libtensorflow against CUDA 10.1.168 according to the instructions here and substitute it in as instructed into my Tensorflow.js install, it is observed that Tensorflow.js continues to fail to load the CUDA 10.0 binaries.

For the exact steps I followed to compile libtensorflow, see this comment I made on a different issue. Following that, I did the following:

cp bazel-bin/tensorflow/tools/lib_package/libtensorflow.tar.gz ~/myproject/node_modules/@tensorflow/tfjs-node/deps
cp bazel-bin/tensorflow/tools/lib_package/libtensorflow.tar.gz ~/myproject/node_modules/@tensorflow/tfjs-node-gpu/deps
cd ~/path-to-my-project/node_modules/@tensorflow/tfjs-node/deps
tar -xf libtensorflow.tar.gz
cd ~/path-to-my-project/node_modules/@tensorflow/tfjs-node-gpu/deps
tar -xf libtensorflow.tar.gz

…as instructed by your README (link above).

Specifically, I’m getting this error: https://ybin.me/p/2007f2550ecf75a4#m91xA9jifKJumfGEs1YOpANyeuRKrtTI4m7SnfgdM6I=

(it’s a lot of output, so I used a pastebin).

To be clear, I recompiled libtensorflow against CUDA 10.1, but even after substituting it in Tensorflow.js complains it can’t find CUDA 10.0.

Code to reproduce the bug / link to feature request

Any example Tensorflow.js program for Node.js (it doesn’t even have to do anything).

Issue Analytics

  • State:closed
  • Created 3 years ago
  • Comments:6 (1 by maintainers)

github_iconTop GitHub Comments

1reaction
spazmodiuscommented, Sep 20, 2020

I have found (I’m on Windows) that I have to replace the binary in the lib/napi folder, not deps.

0reactions
google-ml-butler[bot]commented, Aug 5, 2020

Are you satisfied with the resolution of your issue? Yes No

Read more comments on GitHub >

github_iconTop Results From Across the Web

Why is tensorflow looking for cuda10.1 while cuda 10.0 ...
According to Tensorflow tested build configuration from TF 2.1 to TF 2.3 , it required CUDA 10.1 version, hence you received above error....
Read more >
Build from source - TensorFlow
Build a TensorFlow pip package from source and install it on Ubuntu Linux and macOS. While the instructions might work for other systems,...
Read more >
Environments for running AI in Node.js - IBM Developer
We look at customizing and optimizing TensorFlow.js in Node.js ... The build process requires the libtensorflow and CUDA shared libraries.
Read more >
How do the super resolution filters in FFmpeg work?
sudo apt-get install --no-install-recommends \ cuda-10-0 ... wget https://storage.googleapis.com/tensorflow/libtensorflow/libtensorflow-gpu- ...
Read more >
Use Coral Edge TPUs to run TFlite models in Node with ...
Run TensorFlow Lite models in Node.js and accelerate them with Coral Edge ... an Intel x86-64 machine running Ubuntu 20.04 or Windows 10....
Read more >

github_iconTop Related Medium Post

No results found

github_iconTop Related StackOverflow Question

No results found

github_iconTroubleshoot Live Code

Lightrun enables developers to add logs, metrics and snapshots to live code - no restarts or redeploys required.
Start Free

github_iconTop Related Reddit Thread

No results found

github_iconTop Related Hackernoon Post

No results found

github_iconTop Related Tweet

No results found

github_iconTop Related Dev.to Post

No results found

github_iconTop Related Hashnode Post

No results found