All Apps and Add-ons

How can I add model.binary_nn_classifier module for DeepLearning toolkit

davietch
Path Finder

Hello,

I set up the Deeplearning toolkit and started the Tensorflow CPU container. On the containers dashboards, on the panel listing the containers I've got an error throwed by the indexer:

[myIndexer] Failed to fetch REST endpoint uri=https://127.0.0.1:8089/services/mltk-container/status?count=0 from server https://127.0.0.1:8089. Check that the URI path provided exists in the REST API.

Does the DeepLearning toolkit needs to be also installed on the indexers?

When I try to run the "Neural Network Classifier Example" dashboard, I also get an error, even though the dashboard is returning some results:

MLTKC error: /fit: ERROR: unable to load algo code from module. Ended with exception: No module named 'model.binary_nn_classifier'

Any help would be much appreciated.

0 Karma
1 Solution

pdrieger_splunk
Splunk Employee
Splunk Employee

Hi @davietch ,
thanks for your question. Let me answer all 3 questions that I identified in your post:

  1. That's indeed an unintended behaviour and will be fixed in the upcoming version 3.1. As a workaround you can modify the relevant | rest ... searches on your dashboard, e.g. with | rest splunk_server=local services/mltk-container/status. By adding splunk_server=local the search does not propagate to the indexers and will resolve this issue. As a pointer: please also update other saved searches like the MLTK Container Sync, too.
  2. DLTK currently does not support execution on the indexers. So the short answer here is that there is no benefit in installing DLTK on your indexers.
  3. It seems that there was no module code generated from the Jupyter Notebook in the container. Simply resolve by opening the binary_nn_classifier.ipynb in Jupyter Lab and save it. This automatically generates a python module in the container that you can verify as file in the /app/model/binary_nn_classifier.py - you should see the example dashboard working as soon as this file exists.

Hope this answers all your questions 🙂

View solution in original post

0 Karma

pdrieger_splunk
Splunk Employee
Splunk Employee

Hi @davietch ,
thanks for your question. Let me answer all 3 questions that I identified in your post:

  1. That's indeed an unintended behaviour and will be fixed in the upcoming version 3.1. As a workaround you can modify the relevant | rest ... searches on your dashboard, e.g. with | rest splunk_server=local services/mltk-container/status. By adding splunk_server=local the search does not propagate to the indexers and will resolve this issue. As a pointer: please also update other saved searches like the MLTK Container Sync, too.
  2. DLTK currently does not support execution on the indexers. So the short answer here is that there is no benefit in installing DLTK on your indexers.
  3. It seems that there was no module code generated from the Jupyter Notebook in the container. Simply resolve by opening the binary_nn_classifier.ipynb in Jupyter Lab and save it. This automatically generates a python module in the container that you can verify as file in the /app/model/binary_nn_classifier.py - you should see the example dashboard working as soon as this file exists.

Hope this answers all your questions 🙂

View solution in original post

0 Karma

davietch
Path Finder

Hi,

Thanks for your answer! Sorry for the delay in my response, I had trouble connecting to the JupyterLab with our Firewall rules...
I now can go the Jupyter page but there is a password... Any clue?

0 Karma

pdrieger_splunk
Splunk Employee
Splunk Employee

Hi @davietch , sure that's easy 🙂
Q: What is the password for Jupyter Lab?
A: Please have a look at the Model Development Guide page in the Deep Learning Toolkit app.
https://splunkbase.splunk.com/app/4607/#/details
Hope this helps. Best, Philipp

0 Karma
Did you miss .conf21 Virtual?

Good news! The event's keynotes and many of its breakout sessions are now available online, and still totally FREE!