I've successfully deployed my Keras model to R Studio Connect (using both the newest versions of RSC and rsconnect), however, after checking the deployed application in RSC I'm seeing the following message:
Does anyone know what the issue is and point me to the right resource which describes the entire deployment process end-to-end (including making predictions against it)?
This documentation section outlines a few things to check to see why Tensorflow content is not being run:
https://docs.rstudio.com/connect/admin/getting-started.html#getting-started-tensorflow
sudo /opt/rstudio-connect/bin/license-manager status to view the capabilities for your licenseOnce you are able to deploy and serve Tensorflow models, this documentation section describes how to query the models:
https://docs.rstudio.com/connect/user/publishing.html#publishing-tensorflow
Also, to see if the Tensorflow dependencies are installed correctly, you can check for the following message in your RSC logs at startup:
2019/01/21 06:31:44 Disabling TensorFlow Model API support:
rstudio-connect cannot run TensorFlow Model APIs because
'libtensorflow.so' is unavailable
which are located in /var/log/rstudio-connect.log by default. Note that the logs might be rotated, so be sure to check the logs at RSC startup time, or restart the RSC server and check the most recent logs.
This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.
If you have a query related to it or one of the replies, start a new topic and refer back with a link.