Deep Learning AMI with Source Code (CUDA 8, Ubuntu)
Amazon Web Services | 2.5_Jan2018Linux/Unix, Ubuntu 16.04 - 64-bit Amazon Machine Image (AMI)
Keras, tensorflow etc. not installed
Like others have noted, Keras, Tensorflow are missing and Amazon support wants me to pay to get it sorted out. No thanks!
- Leave a Comment |
- Mark review as helpful
Great AMI with Python3
As others pointed out, use python3.
Most of the major frameworks are installed with GPU enabled.
Of course we can instantly use the Gluon.
Great AMI
Saves lots of time in searching for CUDA drivers, cuDNN, and building DL frameworks. New Keras support for MxNet is great too!
Deep Learning AMI is just an Ubuntu image with nothing installed :)
Two weeks ago I got excited at a presentation from two AWS architects in Cambridge telling us how great it is to use Amazon cloud services because they come with everything pre-installed and you could focus on doing your research.
I did choose the p2.xlarge AMI in Ireland to do deep learning on image data. I tried following several tutorials I found on Amazon websites and smart IT blogs to be successful. The disappointment started with Jupyter not being installed. Then trying to do simple Python import commands resulted in Tensorflow not found, MXnet not found and including other core packages required.
After installing Jupyter and lots of other libraries needed to get tensorflow-gpu support to work it turned out that the Cuda library is not properly installed. See also other reviews.
I have spent now a full day installing stuff from scratch which cost me a lot of time and 10$ in cloud services without much progress. I am not going to give up and will soon be a deep learning professional IT architect. Thank you Amazon for your enticing marketing and despicable software services! Lesson learned :)
Getting started REALLY required
I need to get the following:
- CUDA (8.0)
- cuDNN (6)
- Tensorflow (latest)
It is extremely unclear how to get started activating just that.
Also - I may be mistaken, but only tensorflow was installed, not tensorflow-gpu, so I really don't see the point in the pre-installed environments here...
Deep Learning AMI Ubuntu: Free tier or not Free tier
Tried to launch the instance in subj, which is marked as 'free tier eligible' and got the message - '...not eligible for free usage tier'. Bug? User error? else?
Got Tensorflow working via Tensorflow3
Per the other reviews, the default tensorflow in ~/src/tensorflow was problematic.
However, we've had no trouble using the Python3+ compilation, even with Python2.7:
~/src/tensorflow3/
It's a reasonable work-around until an updated AMI is released.
Tensorflow does not work with Jupyter
import tensorflow fails from a jupyter notebook in both python2 and python3. It was successful on the python command line, however.
Cannot load tensorflow
Got it up and running and ran "import tensorflow" -- no luck though.
Traceback (most recent call last):
File "/home/ubuntu/deployments/data-pipelines/scratch/function_test.py", line 1, in
import tensorflow as tf
File "/usr/local/lib/python2.7/dist-packages/tensorflow/__init__.py", line 24, in
from tensorflow.python import *
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/__init__.py", line 49, in
from tensorflow.python import pywrap_tensorflow
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/pywrap_tensorflow.py", line 52, in
raise ImportError(msg)
ImportError: Traceback (most recent call last):
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/pywrap_tensorflow.py", line 41, in
from tensorflow.python.pywrap_tensorflow_internal import *
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 28, in
_pywrap_tensorflow_internal = swig_import_helper()
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 24, in swig_import_helper
_mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description)
ImportError: libcusolver.so.8.0: cannot open shared object file: No such file or directory
Failed to load the native TensorFlow runtime.
See https://www.tensorflow.org/install/install_sources#common_installation_problems
for some common reasons and solutions. Include the entire stack trace
above this error message when asking for help.
Process finished with exit code 1
Tensorflow doesn't work.
I am using this AMI with a p2 xlarge instance. When I try to test tensorflow, I got the following message:
ImportError: libcusolver.so.8.0: cannot open shared object file: No such file or directory
Failed to load the native TensorFlow runtime.