Does Caffe/DIGITS work with Python3

AI & Data Science Deep Learning (Training & Inference)
1

I installed Caffe and DIGITs on python 3 and it seems to work partially, however I still get this Protobuf error about version 2.6.1 versus 3.5.

So I reinstalled Protobuf and was then following through with reinstalling Caffe. I have an Anaconda python3.6 environment and I was recompiling Caffe in that environment. During the “cmake …” step, I saw that the version of python that is detected is Python2.7 instead of Python3–even though I am in the conda python 3.6 environment. Even the version of numpy detected is for 2.7.:

Found PythonLibs: /usr/lib/x86_64-linux-gnu/libpython2.7.so (found suitable version "2.7.12", minimum required is "2.7") 
-- Found NumPy: /home/krishnab/.local/lib/python2.7/site-packages/numpy/core/include (found suitable version "1.11.0", minimum required is "1.7.1") 
-- NumPy ver. 1.11.0 found (include: /home/krishnab/.local/lib/python2.7/site-packages/numpy/core/include)

So my real question is whether DIGITS works with Caffe in python 3? I could not find any clear indication in the documentation except for an opaque reference to an issue:

https://github.com/NVIDIA/DIGITS/issues/511

Can anyone clarify?

2

We created a new “Deep Learning Training and Inference” section in Devtalk to improve the experience for deep learning and accelerated computing, and HPC users:
https://devtalk.nvidia.com/default/board/301/deep-learning-training-and-inference-/

We are moving active deep learning threads to the new section.

URLs for topics will not change with the re-categorization. So your bookmarks and links will continue to work as earlier.

-Siddharth

3

@siddharth I have the same question. please give us answer directly. why you just give us a link to another forum. it is pointless and leaves us awful experience.

4

I have the similar issue here is my post https://devtalk.nvidia.com/default/topic/1071363/container-digits/how-to-make-nvidia-digits-work-with-caffe-and-tensorflow-installed-with-python-3-/

Has anyone able to solve this issue yet?