Merge pull request #341 from luisquintanilla/make-runnable-on-azure

Make mnist-vscode-docs-sample run on Azure by default

.gitignore

@@ -286,3 +286,7 @@ __pycache__/
 *.btm.cs
 *.odx.cs
 *.xsd.cs
+
+# AML
+**/aml_config/*
+**/azureml_outputs/*

mnist-vscode-docs-sample/amlrun.py

@@ -0,0 +1,12 @@
+import azureml
+from azureml.core import Run
+
+# Access the Azure ML run
+# Init run param to check if running within AML
+def get_AMLRun():
+    try:
+        run = Run.get_submitted_run()
+        return run
+    except Exception as e:
+        print("Caught = {}".format(e.message))
+        return None

mnist-vscode-docs-sample/env.yml

@@ -0,0 +1,9 @@
+name: vs-code-azure-ml-tutorial
+channels:
+  - defaults
+dependencies:
+  - python=3.6.2
+  - tensorflow=1.15.0
+  - pip
+  - pip:
+    - azureml-defaults

mnist-vscode-docs-sample/score.py

@@ -10,7 +10,7 @@ def init():
     global X, output, sess
     tf.reset_default_graph()
     # retreive the local path to the model using the model name
-    model_root = Model.get_model_path('mnist_tf_model')
+    model_root = Model.get_model_path('MNIST-TensorFlow-model')
     saver = tf.train.import_meta_graph(os.path.join(model_root, 'mnist-tf.model.meta'))
     X = tf.get_default_graph().get_tensor_by_name("network/X:0")
     output = tf.get_default_graph().get_tensor_by_name("network/output/MatMul:0")

mnist-vscode-docs-sample/train.py

@@ -7,6 +7,7 @@ import sys
 import gzip
 import struct
 from utils import prepare_data
+from amlrun import get_AMLRun
 
 # ## Download MNIST dataset
 # In order to train on the MNIST dataset we will first need to download
@@ -43,6 +44,7 @@ with tf.name_scope('eval'):
 
 init = tf.global_variables_initializer()
 saver = tf.train.Saver()
+run = get_AMLRun()
 
 with tf.Session() as sess:
     init.run()
@@ -66,11 +68,17 @@ with tf.Session() as sess:
 
             # train
             sess.run(train_op, feed_dict = {X: X_batch, y: y_batch})
+
         # evaluate training set
         acc_train = acc_op.eval(feed_dict = {X: X_batch, y: y_batch})
         # evaluate validation set
         acc_val = acc_op.eval(feed_dict = {X: X_test, y: y_test})
 
+        # Log accuracies to AML logger if using AML
+        if run != None:
+            run.log('Validation Accuracy', np.float(acc_val))
+            run.log('Training Accuracy', np.float(acc_train))
+
         # print out training and validation accuracy
         print(epoch, '-- Training accuracy:', acc_train, '\b Validation accuracy:', acc_val)
         y_hat = np.argmax(output.eval(feed_dict = {X: X_test}), axis = 1)