serving icon indicating copy to clipboard operation
serving copied to clipboard

A flexible, high-performance serving system for machine learning models

Results 219 serving issues
Sort by recently updated
recently updated
newest added

We have to do a fair amount of text preprocessing of our data before feeding it into tensorflow. Since the text manipulation abilities of tensorflow and tensorflow tranform are still...

type:feature
stat:awaiting tensorflower
needs prio
type:performance

Question as above~

type:feature
stat:contributions welcome

This PR adds a command-line flag to expose total model memory limit parameter to user. For ease of use limit is expressed in MB units and converted to bytes internally...

cla: yes

In specific version policy mode, when the filesystem crashed or the version directory was deleted by mistake, the serving version may be unloaded. Which is a disaster for online service....

cla: yes

The Dockerfile entrypoint /usr/bin/tf_serving_entrypoint.sh calls tensorflow_model_server with "$@" which results in an error message from tensorflow_model_server that "/usr/bin/tf_serving_entrypoint.sh" is not a supported option. This is because "$@" passes the first...

cla: yes