Merge branch 'master' into travis-job

Author: Tomcli

community/FfDL-Seldon/pytorch-model/README.md

@@ -15,7 +15,7 @@ You can skip this step if you are happy to use the already packaged image ```sel
 The runtime MNIST scrorer is contained within a standalone [python class PyMnist.py](./PyMnist.py). This needs to be packaged in a Docker container to run within Seldon. For this we use [Redhat's Source-to-image](https://github.com/openshift/source-to-image).
 
  * Install [S2I](https://github.com/openshift/source-to-image#installation)
- * From the pytorch-model folder run the following s2i build. You will need to change *seldonio* to your Docker repo. You will need at least 8GB for your local Docker.
+ * From the pytorch-model folder run the following s2i build. You will need to change *seldonio* to your Docker repo. **You will need at least 8GB for your local Docker.**
 ```
 s2i build . seldonio/seldon-core-s2i-python2 seldonio/ffdl-pymnist:0.1
 ```
@@ -51,4 +51,3 @@ To test the running model with example MNIST images you can run either of two no
 
  * [Ambassador Example](serving_ambassador.ipynb)
  * [Seldon OAuth Example](serving_oauth.ipynb)
- 

community/FfDL-Seldon/tf-model/README.md

@@ -14,7 +14,7 @@ You can skip this step if you are happy to use the already packaged image ```sel
 The runtime MNIST scrorer is contained within a standalone [python class TFMnist.py](./TFMnist.py). This needs to be packaged in a Docker container to run within Seldon. For this we use [Redhat's Source-to-image](https://github.com/openshift/source-to-image).
 
  * Install [S2I](https://github.com/openshift/source-to-image#installation)
- * From the tf-model folder run, (*change seldonio to your Docker repo*):
+ * From the tf-model folder run, (*change seldonio to your Docker repo*) **You will need at least 8GB for your local Docker.** :
 ```
 s2i build . seldonio/seldon-core-s2i-python2 seldonio/ffdl-mnist:0.1
 ```
@@ -50,4 +50,3 @@ To test the running model with example MNIST images you can run either of two no
 
  * [Ambassador Example](serving_ambassador.ipynb)
  * [Seldon OAuth Example](serving_oauth.ipynb)
- 

demos/fashion-mnist-training/fashion-mnist-webapp/README.md

@@ -10,7 +10,7 @@ The webapp is designed to take images that are uploaded, display them on the web
   ```
 
 2. Modify deployment resource from the template `fashion-mnist-webapp.yaml`. You need to set:
-  * `MODEL_ENDPOINT`: Your Seldon model endpoint. (e.g. http://<AMBASSADOR_API_IP>/seldon/<modelDeploymentName>/api/v0.1/predictions")
+  * `MODEL_ENDPOINT`: Your Seldon model endpoint. (e.g. http://<AMBASSADOR_API_IP>/seldon/<Model_Deployment_Name>/api/v0.1/predictions) The `AMBASSADOR_API_IP` is your `seldon-core-ambassador`'s service endpoint which by default is exposed with NodePort.
   * `image` : Your web app image at DockerHub
 
 3. Congratulations, your web app should be running now. You can use the following commands to check where your web app is hosted.

demos/fashion-mnist-training/fashion-mnist-webapp/fashion-mnist-webapp.yaml

@@ -20,7 +20,7 @@ spec:
         env:
           - name: MODEL_ENDPOINT
             # Put down your Seldon model endpoint
-            value: <your Seldon model endpoint hosted on AMBASSADOR_API_IP>
+            value: http://<PUBLIC_IP:AMBASSADOR_API_NODEPORT>/seldon/<Model_Deployment_Name>/api/v0.1/predictions
         ports:
         - containerPort: 8088
 ---

demos/fashion-mnist-training/seldon-deployment/fashion-seldon.json

@@ -1,5 +1,5 @@
 {
-  "apiVersion": "machinelearning.seldon.io/v1alpha1",
+  "apiVersion": "machinelearning.seldon.io/v1alpha2",
   "kind": "SeldonDeployment",
   "metadata": {
     "labels": {
@@ -17,7 +17,7 @@
     "oauth_secret": "oauth-secret",
     "predictors": [
       {
-        "componentSpec": {
+        "componentSpecs": [{
           "spec": {
             "containers": [
               {
@@ -76,7 +76,7 @@
             ],
             "terminationGracePeriodSeconds": 20
           }
-        },
+        }],
         "graph": {
           "children": [],
           "name": "classifier",

docs/user-guide.md

@@ -25,15 +25,15 @@ Currently, Fabric for Deep Learning supports following community frameworks
 
 | Framework     | Versions      | Processing Unit |
 | ------------- | ------------- | --------------- |
-| [tensorflow](https://hub.docker.com/r/tensorflow/tensorflow/)    | 1.3.0, 1.3.0-py3, 1.4.0, 1.4.0-py3, 1.5.0, 1.5.0-py3, 1.5.1, 1.5.1-py3, 1.6.0, 1.6.0-py3, 1.7.0, 1.7.0-py3, 1.8.0, 1.8.0-py3, latest, latest-py3 | CPU |
-| [tensorflow](https://hub.docker.com/r/tensorflow/tensorflow/)    | 1.3.0-gpu, 1.3.0-gpu-py3, 1.4.0-gpu, 1.4.0-gpu-py3, 1.5.0-gpu, 1.5.0-gpu-py3, 1.5.1-gpu, 1.5.1-gpu-py3, 1.6.0-gpu, 1.6.0-gpu-py3, 1.7.0-gpu, 1.7.0-gpu-py3, 1.8.0-gpu, 1.8.0-gpu-py3, latest-gpu, latest-gpu-py3 | GPU |
+| [tensorflow](https://hub.docker.com/r/tensorflow/tensorflow/)    | 1.4.0, 1.4.0-py3, 1.5.0, 1.5.0-py3, 1.5.1, 1.5.1-py3, 1.6.0, 1.6.0-py3, 1.7.0, 1.7.0-py3, 1.8.0, 1.8.0-py3, 1.9.0, 1.9.0-py3, latest, latest-py3 | CPU |
+| [tensorflow](https://hub.docker.com/r/tensorflow/tensorflow/)    | 1.4.0-gpu, 1.4.0-gpu-py3, 1.5.0-gpu, 1.5.0-gpu-py3, 1.5.1-gpu, 1.5.1-gpu-py3, 1.6.0-gpu, 1.6.0-gpu-py3, 1.7.0-gpu, 1.7.0-gpu-py3, 1.8.0-gpu, 1.8.0-gpu-py3, 1.9.0-gpu, 1.9.0-gpu-py3, latest-gpu, latest-gpu-py3 | GPU |
 | [caffe](https://hub.docker.com/r/bvlc/caffe/)         | cpu, intel   | CPU |
 | [caffe](https://hub.docker.com/r/bvlc/caffe/)         | gpu           | GPU |
 | [pytorch](https://hub.docker.com/r/pytorch/pytorch/)       | v0.2, latest | CPU, GPU |
 | [caffe2](https://hub.docker.com/r/caffe2ai/caffe2/)        | c2v0.8.1.cpu.full.ubuntu14.04, c2v0.8.0.cpu.full.ubuntu16.04 | CPU |
 | [caffe2](https://hub.docker.com/r/caffe2ai/caffe2/)        | c2v0.8.1.cuda8.cudnn7.ubuntu16.04, latest | GPU |
 | [h2o3](https://hub.docker.com/r/opsh2oai/h2o3-ffdl/)    | latest | CPU |
-| [horovod](https://hub.docker.com/r/uber/horovod/)       | 0.13.4-tf1.8.0-torch0.4.0-py3.5, 0.13.4-tf1.8.0-torch0.4.0-py2.7 | CPU, GPU |
+| [horovod](https://hub.docker.com/r/uber/horovod/)       | 0.13.10-tf1.9.0-torch0.4.0-py2.7, 0.13.10-tf1.9.0-torch0.4.0-py3.5 | CPU, GPU |
 
 You can deploy models based on these frameworks and then train your models using the FfDL CLI or FfDL UI.
 

etc/examples/horovod/manifest_pytorchmnist.yml

@@ -20,5 +20,5 @@ data_stores:
 
 framework:
   name: horovod
-  version: "0.13.4-tf1.8.0-torch0.4.0-py3.5"
+  version: "0.13.10-tf1.9.0-torch0.4.0-py3.5"
   command: python pytorch_mnist.py

etc/examples/horovod/manifest_tfmnist.yml

@@ -20,6 +20,6 @@ data_stores:
 
 framework:
   name: horovod
-  version: "0.13.4-tf1.8.0-torch0.4.0-py3.5"
+  version: "0.13.10-tf1.9.0-torch0.4.0-py3.5"
   command: python tensorflow_mnist.py
   # the command is basically running the above command via openmpi, feel free to remove  -x NCCL_DEBUG=INFO

templates/services/learner-configmap.yml

@@ -31,10 +31,10 @@ data:
   tensorflow_gpu_1.4.0-gpu_CURRENT: manual
   tensorflow_cpu_1.4.0_CURRENT: manual
   tensorflow_cpu_1.4.0-py3_CURRENT: manual
-  tensorflow_gpu_1.3.0-gpu-py3_CURRENT: manual
-  tensorflow_gpu_1.3.0-gpu_CURRENT: manual
-  tensorflow_cpu_1.3.0_CURRENT: manual
-  tensorflow_cpu_1.3.0-py3_CURRENT: manual
+  tensorflow_gpu_1.9.0-gpu-py3_CURRENT: manual
+  tensorflow_gpu_1.9.0-gpu_CURRENT: manual
+  tensorflow_cpu_1.9.0_CURRENT: manual
+  tensorflow_cpu_1.9.0-py3_CURRENT: manual
   h2o3_cpu_latest_CURRENT: manual
   caffe_cpu_cpu_CURRENT: master-39
   caffe_gpu_gpu_CURRENT: master-39
@@ -45,8 +45,8 @@ data:
   caffe2_gpu_c2v0.8.1.cuda8.cudnn7.ubuntu16.04_CURRENT: master-39
   caffe2_cpu_c2v0.8.0.cpu.full.ubuntu16.04_CURRENT: master-39
   caffe2_gpu_latest_CURRENT: master-39
-  horovod_gpu_0.13.4-tf1.8.0-torch0.4.0-py3.5_CURRENT: manual
-  horovod_gpu_0.13.4-tf1.8.0-torch0.4.0-py2.7_CURRENT: manual
+  horovod_gpu_0.13.10-tf1.9.0-torch0.4.0-py3.5_CURRENT: manual
+  horovod_gpu_0.13.10-tf1.9.0-torch0.4.0-py2.7_CURRENT: manual
 ---
 
 apiVersion: v1