Contains Bazel builds for both Tensorflow-Serving Server and Tensorflow-serving client - mnist model
789
Run the Docker image:
docker run --name=tensorflow-serving -it tensorflow-serving /bin/bash
Note the IP Address first(Tools are installed):
ifconfig
Look for inet addr and make a note.
Run the Model Server:
cd serving bazel-bin/tensorflow_serving/model_servers/tensorflow_model_server --port=9000
--model_name=mnist --model_base_path=/tmp/mnist_model
Python or Bazel Client Run the same Docker image for running the client
docker run --name=tensorflow-serving-client -it tensorflow-serving /bin/bash cd serving python tensorflow_serving/example/mnist_client.py --num_tests=1000
--server=:9000
If everything goes well, it should print "Inference error rate: 10.5%"
***Java Based Client ***
docker pull svnp10/tf-serving-java-client docker run -it --name tf-client svnp10/tf-serving-java-client /bin/bash cd /home java -jar tensorflowserving-1.0-SNAPSHOT-jar-with-dependencies.jar
*OutPut should look like: INFO: [ManagedChannelImpl@340f438e] Created with target 172.17.0.3:9000 label: 7, predicted : [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0] label: 2, predicted : [0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0] label: 1, predicted : [0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0] label: 0, predicted : [1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0] label: 4, predicted : [0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 0.0] label: 1, predicted : [0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0] label: 4, predicted : [0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 0.0] label: 9, predicted : [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0] label: 5, predicted : [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0] label: 9, predicted : [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0]
Content type
Image
Digest
Size
4.2 GB
Last updated
over 8 years ago
docker pull svnp10/tensorflow-serving