jprdonnelly/MachineLearningNotebooks
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Update notebooks

Browse Source 
Update notebooks
This commit is contained in:
rastala
2018-09-20 10:02:39 -04:00
parent d926eb03de
commit 90b3bf799f
58 changed files with 2664 additions and 15566 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
01.getting-started/01.train-within-notebook/01.train-within-notebook.ipynb
View File

@@ -447,8 +447,9 @@
},
"outputs": [],
"source": [
"for m in ws.models(name='best_model'):\n",
" print(m.name, m.version)"
"models = ws.models(name='best_model')\n",
"for name, m in models.items():\n",
" print(name, m.version)"
]
},
{

4031
01.getting-started/07.hyperdrive-with-sklearn/.ipynb_checkpoints/07.hyperdrive-with-sklearn-checkpoint.ipynb
View File

File diff suppressed because it is too large Load Diff

4033
01.getting-started/07.hyperdrive-with-sklearn/07.hyperdrive-with-sklearn.ipynb
View File

File diff suppressed because it is too large Load Diff

52
01.getting-started/07.hyperdrive-with-sklearn/diabetes_sklearn.py
View File

@@ -1,52 +0,0 @@
from sklearn.datasets import load_diabetes
from sklearn.linear_model import Ridge
from sklearn.metrics import mean_squared_error
from sklearn.model_selection import train_test_split
from sklearn.externals import joblib
import os
import argparse
# Import Run from azureml.core,
from azureml.core.run import Run
parser = argparse.ArgumentParser()
parser.add_argument('--alpha', type=float, dest='alpha',
default=0.5, help='regularization strength')
args = parser.parse_args()
# Get handle of current run for logging and history purposes
run = Run.get_submitted_run()
X, y = load_diabetes(return_X_y=True)
columns = ['age', 'gender', 'bmi', 'bp', 's1', 's2', 's3', 's4', 's5', 's6']
x_train, x_test, y_train, y_test = train_test_split(
X, y, test_size=0.2, random_state=0)
data = {"train": {"x": x_train, "y": y_train},
"test": {"x": x_test, "y": y_test}}
alpha = args.alpha
print('alpha value is:', alpha)
reg = Ridge(alpha=alpha)
reg.fit(data["train"]["x"], data["train"]["y"])
print('Ridget model fitted.')
preds = reg.predict(data["test"]["x"])
mse = mean_squared_error(preds, data["test"]["y"])
# Log metrics
run.log("alpha", alpha)
run.log("mse", mse)
os.makedirs('./outputs', exist_ok=True)
model_file_name = "model.pkl"
# Save model as part of the run history
with open(model_file_name, "wb") as file:
joblib.dump(reg, 'outputs/' + model_file_name)
print('Mean Squared Error is:', mse)

3448
01.getting-started/08.hyperdrive-with-TensorFlow/.ipynb_checkpoints/08.hyperdrive-with-TensorFlow-checkpoint.ipynb
View File

File diff suppressed because it is too large Load Diff

3448
01.getting-started/08.hyperdrive-with-TensorFlow/08.hyperdrive-with-TensorFlow.ipynb
View File

File diff suppressed because it is too large Load Diff

BIN
01.getting-started/08.hyperdrive-with-TensorFlow/mnist_test_images/img_1.jpg
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 572 B

BIN
01.getting-started/08.hyperdrive-with-TensorFlow/mnist_test_images/img_10.jpg
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 555 B

BIN
01.getting-started/08.hyperdrive-with-TensorFlow/mnist_test_images/img_2.jpg
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 615 B

BIN
01.getting-started/08.hyperdrive-with-TensorFlow/mnist_test_images/img_3.jpg
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 561 B

BIN
01.getting-started/08.hyperdrive-with-TensorFlow/mnist_test_images/img_4.jpg
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 504 B

BIN
01.getting-started/08.hyperdrive-with-TensorFlow/mnist_test_images/img_5.jpg
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 546 B

BIN
01.getting-started/08.hyperdrive-with-TensorFlow/mnist_test_images/img_6.jpg
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 578 B

BIN
01.getting-started/08.hyperdrive-with-TensorFlow/mnist_test_images/img_7.jpg
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 586 B

BIN
01.getting-started/08.hyperdrive-with-TensorFlow/mnist_test_images/img_8.jpg
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 562 B

BIN
01.getting-started/08.hyperdrive-with-TensorFlow/mnist_test_images/img_9.jpg
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 592 B

103
01.getting-started/08.hyperdrive-with-TensorFlow/tf_mnist_score.py
View File

@@ -1,103 +0,0 @@
from __future__ import print_function
import tensorflow as tf
import numpy as np
import os
import json
import base64
from io import BytesIO
from PIL import Image
##############################################
# helper functions
##############################################
def build_model(x, y_, keep_prob):
def weight_variable(shape):
initial = tf.truncated_normal(shape, stddev=0.1)
return tf.Variable(initial)
def bias_variable(shape):
initial = tf.constant(0.1, shape=shape)
return tf.Variable(initial)
def conv2d(x, W):
return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')
def max_pool_2x2(x):
return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
W_conv1 = weight_variable([5, 5, 1, 32])
b_conv1 = bias_variable([32])
x_image = tf.reshape(x, [-1, 28, 28, 1])
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
h_pool1 = max_pool_2x2(h_conv1)
W_conv2 = weight_variable([5, 5, 32, 64])
b_conv2 = bias_variable([64])
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
h_pool2 = max_pool_2x2(h_conv2)
W_fc1 = weight_variable([7 * 7 * 64, 1024])
b_fc1 = bias_variable([1024])
h_pool2_flat = tf.reshape(h_pool2, [-1, 7 * 7 * 64])
h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
W_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])
y_conv = tf.matmul(h_fc1_drop, W_fc2) + b_fc2
return y_conv
def base64ToImg(base64ImgString):
if base64ImgString.startswith('b\''):
base64ImgString = base64ImgString[2:-1]
base64Img = base64ImgString.encode('utf-8')
decoded_img = base64.b64decode(base64Img)
img_buffer = BytesIO(decoded_img)
img = Image.open(img_buffer)
return img
##############################################
# API init() and run() methods
##############################################
def init():
global x, keep_prob, y_conv, sess
g = tf.Graph()
with g.as_default():
x = tf.placeholder(tf.float32, shape=[None, 784])
y_ = tf.placeholder(tf.float32, shape=[None, 10])
keep_prob = tf.placeholder(tf.float32)
y_conv = build_model(x, y_, keep_prob)
saver = tf.train.Saver()
init_op = tf.global_variables_initializer()
model_dir = os.path.join('sample_projects', 'outputs')
saved_model_path = os.path.join(model_dir, 'model.ckpt')
sess = tf.Session(graph=g)
sess.run(init_op)
saver.restore(sess, saved_model_path)
def run(input_data):
img = base64ToImg(json.loads(input_data)['data'])
img_data = np.array(img, dtype=np.float32).flatten()
img_data.resize((1, 784))
y_pred = sess.run(y_conv, feed_dict={x: img_data, keep_prob: 1.0})
predicted_label = np.argmax(y_pred[0])
outJsonString = json.dumps({"label": str(predicted_label)})
return str(outJsonString)

151
01.getting-started/08.hyperdrive-with-TensorFlow/tf_mnist_train.py
View File

@@ -1,151 +0,0 @@
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import tensorflow as tf
# Load MNIST Data
from tensorflow.examples.tutorials.mnist import input_data
import os
import argparse
from azureml.core.run import Run
# the following 10 lines can be removed once BUG# 241943 is fixed
def get_logger():
try:
return Run.get_submitted_run()
except Exception:
return LocalLogger()
class LocalLogger:
def log(self, key, value):
print("AML-Log:", key, value)
def build_model(x, y_, keep_prob):
def weight_variable(shape):
initial = tf.truncated_normal(shape, stddev=0.1)
return tf.Variable(initial)
def bias_variable(shape):
initial = tf.constant(0.1, shape=shape)
return tf.Variable(initial)
def conv2d(x, W):
return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')
def max_pool_2x2(x):
return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
W_conv1 = weight_variable([5, 5, 1, 32])
b_conv1 = bias_variable([32])
x_image = tf.reshape(x, [-1, 28, 28, 1])
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
h_pool1 = max_pool_2x2(h_conv1)
W_conv2 = weight_variable([5, 5, 32, 64])
b_conv2 = bias_variable([64])
h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
h_pool2 = max_pool_2x2(h_conv2)
W_fc1 = weight_variable([7 * 7 * 64, 1024])
b_fc1 = bias_variable([1024])
h_pool2_flat = tf.reshape(h_pool2, [-1, 7 * 7 * 64])
h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
W_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])
y_conv = tf.matmul(h_fc1_drop, W_fc2) + b_fc2
return y_conv
def main():
# Get command-line arguments
parser = argparse.ArgumentParser()
parser.add_argument('--learning_rate', type=float,
default=0.0001, help='learning rate')
parser.add_argument('--minibatch_size', type=int,
default=50, help='minibatch size')
parser.add_argument('--keep_probability', type=float,
default=0.5, help='keep probability for dropout layer')
parser.add_argument('--num_iterations', type=int,
default=1000, help='number of iterations')
parser.add_argument('--output_dir', type=str, default='./outputs',
help='output directory to write checkpoints to')
args = parser.parse_args()
# log parameters
run_logger = get_logger()
run_logger.log("learning_rate", args.learning_rate)
run_logger.log("minibatch_size", args.minibatch_size)
run_logger.log("keep_probability", args.keep_probability)
run_logger.log("num_iterations", args.num_iterations)
# Load MNIST data
mnist = input_data.read_data_sets('MNIST_data', one_hot=True)
sess = tf.InteractiveSession()
x = tf.placeholder(tf.float32, shape=[None, 784])
y_ = tf.placeholder(tf.float32, shape=[None, 10])
keep_prob = tf.placeholder(tf.float32)
y_conv = build_model(x, y_, keep_prob)
cross_entropy = tf.reduce_mean(
tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y_conv))
train_step = tf.train.AdamOptimizer(
args.learning_rate).minimize(cross_entropy)
correct_prediction = tf.equal(tf.argmax(y_conv, 1), tf.argmax(y_, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
sess.run(tf.global_variables_initializer())
for i in range(args.num_iterations):
batch = mnist.train.next_batch(args.minibatch_size)
if i % 100 == 0:
test_acc = accuracy.eval(
feed_dict={x: mnist.test.images, y_: mnist.test.labels, keep_prob: 1.0})
train_accuracy = accuracy.eval(
feed_dict={x: batch[0], y_: batch[1], keep_prob: 1.0})
print("step %d, training accuracy %g, test accuracy, %g" %
(i, train_accuracy, test_acc))
# log test accuracy to AML
run_logger.log("Accuracy", float(test_acc))
run_logger.log("Iterations", i)
sess.run(train_step, feed_dict={
x: batch[0], y_: batch[1], keep_prob: args.keep_probability})
# Save the trained model
model_dir = args.output_dir
model_file = 'model.ckpt'
os.makedirs(model_dir, exist_ok=True)
saver = tf.train.Saver()
saver.save(sess, os.path.join(model_dir, model_file))
final_test_acc = sess.run(accuracy, feed_dict={
x: mnist.test.images, y_: mnist.test.labels, keep_prob: 1.0})
run_logger.log("Accuracy", float(final_test_acc))
run_logger.log("Iterations", args.num_iterations)
print("test accuracy %g" % final_test_acc)
if __name__ == "__main__":
main()

4
01.getting-started/10.register-model-create-image-deploy-service/10.register-model-create-image-deploy-service.ipynb
View File

@@ -130,8 +130,8 @@
"outputs": [],
"source": [
"regression_models = ws.models(tags=['area'])\n",
"for m in regression_models:\n",
" print(\"Name:\", m.name,\"\\tVersion:\", m.version, \"\\tDescription:\", m.description, m.tags)"
"for name, m in regression_models.items():\n",
" print(\"Name:\", name,\"\\tVersion:\", m.version, \"\\tDescription:\", m.description, m.tags)"
]
},
{

173
01.getting-started/12.enable-data-collection-for-models-in-aks/12.enable-data-collection-for-models-in-aks.ipynb
View File

@@ -1,14 +1,5 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Copyright (c) Microsoft Corporation. All rights reserved.\n",
"\n",
"Licensed under the MIT License."
]
},
{
"cell_type": "markdown",
"metadata": {},
@@ -90,8 +81,8 @@
"source": [
"#Register the model\n",
"from azureml.core.model import Model\n",
"model = Model.register(model_path = 'sklearn_regression_model.pkl', # this points to a local file\n",
" model_name = \"best_model\", # this is the name the model is registered as\n",
"model = Model.register(model_path = \"sklearn_regression_model.pkl\", # this points to a local file\n",
" model_name = \"sklearn_regression_model.pkl\", # this is the name the model is registered as\n",
" tags = {'area': \"diabetes\", 'type': \"regression\"},\n",
" description = \"Ridge regression model to predict diabetes\",\n",
" workspace = ws)\n",
@@ -103,22 +94,25 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"## 4. Update your scoring file with Data Collection\n",
"## 4. *Update your scoring file with Data Collection*\n",
"The file below, compared to the file used in notebook 11, has the following changes:\n",
"### a. Import the module\n",
"<code> from azureml.monitoring import ModelDataCollector </code>\n",
"```python \n",
"from azureml.monitoring import ModelDataCollector```\n",
"### b. In your init function add:\n",
"<code> global inputs_dc, prediction_d\n",
" inputs_dc = ModelDataCollector(\"best_model\", identifier=\"inputs\", feature_names=[\"feat1\", \"feat2\", \"feat3\". \"feat4\", \"feat5\", \"Feat6\"])\n",
" prediction_dc = ModelDataCollector(\"best_model\", identifier=\"predictions\", feature_names=[\"prediction1\", \"prediction2\"])</code>\n",
"```python \n",
"global inputs_dc, prediction_d\n",
"inputs_dc = ModelDataCollector(\"best_model\", identifier=\"inputs\", feature_names=[\"feat1\", \"feat2\", \"feat3\". \"feat4\", \"feat5\", \"Feat6\"])\n",
"prediction_dc = ModelDataCollector(\"best_model\", identifier=\"predictions\", feature_names=[\"prediction1\", \"prediction2\"])```\n",
" \n",
"* Identifier: Identifier is later used for building the folder structure in your Blob, it can be used to divide “raw” data versus “processed”.\n",
"* CorrelationId: is an optional parameter, you do not need to set it up if your model doesn’t require it. Having a correlationId in place does help you for easier mapping with other data. (Examples include: LoanNumber, CustomerId, etc.)\n",
"* Feature Names: These need to be set up in the order of your features in order for them to have column names when the .csv is created.\n",
"\n",
"### c. In your run function add:\n",
"<code> inputs_dc.collect(data)\n",
" prediction_dc.collect(result) </code>"
"```python\n",
"inputs_dc.collect(data)\n",
"prediction_dc.collect(result)```"
]
},
{
@@ -130,7 +124,7 @@
"%%writefile score.py\n",
"import pickle\n",
"import json\n",
"import numpy as np\n",
"import numpy \n",
"from sklearn.externals import joblib\n",
"from sklearn.linear_model import Ridge\n",
"from azureml.core.model import Model\n",
@@ -139,31 +133,33 @@
"\n",
"def init():\n",
" global model\n",
" #print (\"model initialized\" + time.strftime(\"%H:%M:%S\"))\n",
" # note here \"best_model\" is the name of the model registered under the workspace\n",
" print (\"model initialized\" + time.strftime(\"%H:%M:%S\"))\n",
" # note here \"sklearn_regression_model.pkl\" is the name of the model registered under the workspace\n",
" # this call should return the path to the model.pkl file on the local disk.\n",
" model_path = Model.get_model_path(model_name = 'best_model')\n",
" model_path = Model.get_model_path(model_name = 'sklearn_regression_model.pkl')\n",
" # deserialize the model file back into a sklearn model\n",
" model = joblib.load(model_path)\n",
" global inputs_dc, prediction_dc\n",
" # this setup will help us save our inputs under the \"inputs\" path in our Azure Blob\n",
" inputs_dc = ModelDataCollector(model_name=\"best_model\", identifier=\"inputs\", feature_names=[\"feat1\", \"feat2\", \"feat3\",\"feat4\", \"feat5\",\"feat6\"]) \n",
" inputs_dc = ModelDataCollector(model_name=\"sklearn_regression_model\", identifier=\"inputs\", feature_names=[\"feat1\", \"feat2\"]) \n",
" # this setup will help us save our ipredictions under the \"predictions\" path in our Azure Blob\n",
" prediction_dc = ModelDataCollector(\"best_model\", identifier=\"predictions\", feature_names=[\"prediction1\", \"prediction2\"]) \n",
" prediction_dc = ModelDataCollector(\"sklearn_regression_model\", identifier=\"predictions\", feature_names=[\"prediction1\", \"prediction2\"]) \n",
" \n",
"# note you can pass in multiple rows for scoring\n",
"def run(raw_data):\n",
" global inputs_dc, prediction_dc\n",
" try:\n",
" data = json.loads(raw_data)['data']\n",
" data = np.array(data)\n",
" data = numpy.array(data)\n",
" result = model.predict(data)\n",
" print (\"saving input data\" + time.strftime(\"%H:%M:%S\"))\n",
" inputs_dc.collect(data) #this call is saving our input data into our blob\n",
" prediction_dc.collect(result)#this call is saving our prediction data into our blob\n",
" print (\"saving prediction data\" + time.strftime(\"%H:%M:%S\"))\n",
" return json.dumps({\"result\": result.tolist()})\n",
" except Exception as e:\n",
" result = str(e)\n",
" #print (result + time.strftime(\"%H:%M:%S\"))\n",
" print (result + time.strftime(\"%H:%M:%S\"))\n",
" return json.dumps({\"error\": result})"
]
},
@@ -171,7 +167,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"## 5. Update your myenv.yml file with the required module"
"## 5. *Update your myenv.yml file with the required module*"
]
},
{
@@ -180,18 +176,13 @@
"metadata": {},
"outputs": [],
"source": [
"%%writefile myenv.yml\n",
"name: myenv\n",
"channels:\n",
" - defaults\n",
"dependencies:\n",
" - pip:\n",
" - numpy\n",
" - scikit-learn\n",
" # Required packages for AzureML execution, history, and data preparation.\n",
" - --extra-index-url https://azuremlsdktestpypi.azureedge.net/sdk-release/Preview/E7501C02541B433786111FE8E140CAA1\n",
" - azureml-core\n",
" - azureml-monitoring"
"from azureml.core.conda_dependencies import CondaDependencies \n",
"\n",
"myenv = CondaDependencies.create(conda_packages=['numpy','scikit-learn'])\n",
"myenv.add_pip_package(\"azureml-monitoring\")\n",
"\n",
"with open(\"myenv.yml\",\"w\") as f:\n",
" f.write(myenv.serialize_to_string())"
]
},
{
@@ -238,9 +229,30 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"## 7. Deploy to AKS service\n",
"For this step you would need to have an AKS cluster setup (Notebook 11).\n",
"In this case we are attaching to a previously created service"
"## 7. Deploy to AKS service"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Create AKS compute if you haven't done so (Notebook 11)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Use the default configuration (can also provide parameters to customize)\n",
"prov_config = AksCompute.provisioning_configuration()\n",
"\n",
"aks_name = 'my-aks-test1' \n",
"# Create the cluster\n",
"aks_target = ComputeTarget.create(workspace = ws, \n",
" name = aks_name, \n",
" provisioning_configuration = prov_config)"
]
},
{
@@ -250,20 +262,40 @@
"outputs": [],
"source": [
"%%time\n",
"resource_id = '/subscriptions/92c76a2f-0e1c-4216-b65e-abf7a3f34c1e/resourcegroups/marthateresource_groupjw/providers/Microsoft.ContainerService/managedClusters/my-aks-colfb348092fd3a760'\n",
"create_name= 'my-existing-aks'\n",
"aks_target = AksCompute.attach(workspace = ws, \n",
" name = create_name, \n",
" resource_id=resource_id)\n",
"# Wait for the operation to complete\n",
"aks_target.wait_for_completion(True)"
"aks_target.wait_for_completion(show_output = True)\n",
"print(aks_target.provisioning_state)\n",
"print(aks_target.provisioning_errors)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### a. Activate Data Collection and App Insights\n",
"If you already have a cluster you can attach the service to it:"
]
},
{
"cell_type": "markdown",
"metadata": {
"scrolled": true
},
"source": [
"```python \n",
" %%time\n",
" resource_id = '/subscriptions/<subscriptionid>/resourcegroups/<resourcegroupname>/providers/Microsoft.ContainerService/managedClusters/<aksservername>'\n",
" create_name= 'myaks4'\n",
" aks_target = AksCompute.attach(workspace = ws, \n",
" name = create_name, \n",
" #esource_id=resource_id)\n",
" ## Wait for the operation to complete\n",
" aks_target.wait_for_provisioning(True)```"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### a. *Activate Data Collection and App Insights through updating AKS Webservice configuration*\n",
"In order to enable Data Collection and App Insights in your service you will need to update your AKS configuration file:"
]
},
@@ -273,7 +305,7 @@
"metadata": {},
"outputs": [],
"source": [
"#Set the web service configuration (using default here)\n",
"#Set the web service configuration\n",
"aks_config = AksWebservice.deploy_configuration(collect_model_data=True, enable_app_insights=True)"
]
},
@@ -291,7 +323,7 @@
"outputs": [],
"source": [
"%%time\n",
"aks_service_name ='aks-w-collv5'\n",
"aks_service_name ='aks-w-dc2'\n",
"\n",
"aks_service = Webservice.deploy_from_image(workspace = ws, \n",
" name = aks_service_name,\n",
@@ -300,7 +332,7 @@
" deployment_target = aks_target\n",
" )\n",
"aks_service.wait_for_deployment(show_output = True)\n",
"print(aks_service.state)\n"
"print(aks_service.state)"
]
},
{
@@ -333,28 +365,12 @@
"print(prediction)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"test_sample = json.dumps({'data': [\n",
" [1,22,3,4,5,68,7,98,95,310], \n",
" [10,92,8,7,6,53,84,23,323,1]\n",
"]})\n",
"test_sample = bytes(test_sample,encoding = 'utf8')\n",
"\n",
"prediction = aks_service.run(input_data = test_sample)\n",
"print(prediction)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 9. Validate you data and analyze it\n",
"You can look into your data following this path format in your Azure Blob:\n",
"You can look into your data following this path format in your Azure Blob (it takes up to 15 minutes for the data to appear):\n",
"\n",
"/modeldata/**subscriptionid>**/**resourcegroupname>**/**workspacename>**/**webservicename>**/**modelname>**/**modelversion>>**/**identifier>**/*year/month/day*/data.csv \n",
"\n",
@@ -365,13 +381,13 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"### a. Create DataBricks cluster and connect it to your blob\n",
"### a. Create DataBricks cluter and connect it to your blob\n",
"https://docs.microsoft.com/en-us/azure/azure-databricks/quickstart-create-databricks-workspace-portal or in your databricks workspace you can look for the template \"Azure Blob Storage Import Example Notebook\".\n",
"\n",
"\n",
"Here is an example for setting up the file location to extract the relevant data:\n",
"\n",
"<code> file_location = \"wasbs://mycontainer@testmartstoragendbblgwy.blob.core.windows.net/unknown/unknown/unknown-bigdataset-unknown/my_iterate_parking_inputs/2018/&deg;/&deg;/data.csv\" \n",
"<code> file_location = \"wasbs://mycontainer@storageaccountname.blob.core.windows.net/unknown/unknown/unknown-bigdataset-unknown/my_iterate_parking_inputs/2018/&deg;/&deg;/data.csv\" \n",
"file_type = \"csv\"</code>\n"
]
},
@@ -405,20 +421,13 @@
"source": [
"aks_service.update(collect_model_data=False)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"display_name": "Python [conda env:myenv3]",
"language": "python",
"name": "python3"
"name": "conda-env-myenv3-py"
},
"language_info": {
"codemirror_mode": {
@@ -430,7 +439,7 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.4"
"version": "3.6.6"
}
},
"nbformat": 4,

5
automl/00.configuration.ipynb
View File

@@ -118,8 +118,7 @@
"* A name for your workspace. You can choose one.\n",
"* Your subscription id. Use *id* value from *az account show* output above. \n",
"* The resource group name. Resource group organizes Azure resources and provides default region for the resources in the group. You can either specify a new one, in which case it gets created for your Workspace, or use an existing one or create a new one from [Azure portal](https://portal.azure.com)\n",
"\n",
"Please these values below. For workspace region, we prefer you use `eastus2` or `eastus2euap` (only if you have access to EUAP) for most scenarios. Other supported regions include `westcentralus`, `southeastasia`, `westeurope`, `australiaeast`, although their support might lag behind `eastus2` and `eastus2euap`."
"* Supported regions include `eastus2`, `eastus`,`westcentralus`, `southeastasia`, `westeurope`, `australiaeast`, `westus2`, `southcentralus`."
]
},
{
@@ -131,7 +130,7 @@
"subscription_id = \"<subscription_id>\"\n",
"resource_group = \"myrg\"\n",
"workspace_name = \"myws\"\n",
"workspace_region = \"eastus2\" # or eastus2euap"
"workspace_region = \"eastus2\""
]
},
{

16
automl/01.auto-ml-classification.ipynb
View File

@@ -15,7 +15,7 @@
"source": [
"# AutoML 01: Classification with local compute\n",
"\n",
"In this example we use the scikit learn's [digit dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_digits.html) to showcase how you can use the AutoML Classifier for a simple classification problem.\n",
"In this example we use the scikit learn's [digit dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_digits.html) to showcase how you can use AutoML for a simple classification problem.\n",
"\n",
"Make sure you have executed the [00.configuration](00.configuration.ipynb) before running this notebook.\n",
"\n",
@@ -143,8 +143,8 @@
"|-|-|\n",
"|**task**|classification or regression|\n",
"|**primary_metric**|This is the metric that you want to optimize.<br> Classification supports the following primary metrics <br><i>accuracy</i><br><i>AUC_weighted</i><br><i>balanced_accuracy</i><br><i>average_precision_score_weighted</i><br><i>precision_score_weighted</i>|\n",
"|**max_time_sec**|Time limit in seconds for each iterations|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML trains the data with a specific pipeline|\n",
"|**max_time_sec**|Time limit in seconds for each iteration|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML trains a specific pipeline with the data |\n",
"|**n_cross_validations**|Number of cross validation splits|\n",
"|**X**|(sparse) array-like, shape = [n_samples, n_features]|\n",
"|**y**|(sparse) array-like, shape = [n_samples, ], [n_samples, n_classes]<br>Multi-class targets. An indicator matrix turns on multilabel classification. This should be an array of integers. |\n",
@@ -160,7 +160,7 @@
"automl_config = AutoMLConfig(task = 'classification',\n",
" debug_log = 'automl_errors.log',\n",
" primary_metric = 'AUC_weighted',\n",
" max_time_sec = 12000,\n",
" max_time_sec = 3600,\n",
" iterations = 50,\n",
" n_cross_validations = 3,\n",
" verbosity = logging.INFO,\n",
@@ -314,7 +314,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Best Model based on any iteration\n",
"#### Model from a specific iteration\n",
"Give me the run and the model from the 3rd iteration:"
]
},
@@ -325,9 +325,9 @@
"outputs": [],
"source": [
"iteration = 3\n",
"best_run, fitted_model = local_run.get_output(iteration = iteration)\n",
"print(best_run)\n",
"print(fitted_model)"
"third_run, third_model = local_run.get_output(iteration = iteration)\n",
"print(third_run)\n",
"print(third_model)"
]
},
{

40
automl/02.auto-ml-regression.ipynb
View File

@@ -15,7 +15,7 @@
"source": [
"# AutoML 02: Regression with local compute\n",
"\n",
"In this example we use the scikit learn's [diabetes dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_diabetes.html) to showcase how you can use the AutoML for a simple regression problem.\n",
"In this example we use the scikit learn's [diabetes dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_diabetes.html) to showcase how you can use AutoML for a simple regression problem.\n",
"\n",
"Make sure you have executed the [00.configuration](00.configuration.ipynb) before running this notebook.\n",
"\n",
@@ -128,7 +128,7 @@
"\n",
"columns = ['age', 'gender', 'bmi', 'bp', 's1', 's2', 's3', 's4', 's5', 's6']\n",
"\n",
"x_train, x_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=0)"
"X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=0)"
]
},
{
@@ -143,8 +143,8 @@
"|-|-|\n",
"|**task**|classification or regression|\n",
"|**primary_metric**|This is the metric that you want to optimize.<br> Regression supports the following primary metrics <br><i>spearman_correlation</i><br><i>normalized_root_mean_squared_error</i><br><i>r2_score</i><br><i>normalized_mean_absolute_error</i><br><i>normalized_root_mean_squared_log_error</i>|\n",
"|**max_time_sec**|Time limit in seconds for each iterations|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML Classifier trains the data with a specific pipeline|\n",
"|**max_time_sec**|Time limit in seconds for each iteration|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML trains a specific pipeline with the data|\n",
"|**n_cross_validations**|Number of cross validation splits|\n",
"|**X**|(sparse) array-like, shape = [n_samples, n_features]|\n",
"|**y**|(sparse) array-like, shape = [n_samples, ], [n_samples, n_classes]<br>Multi-class targets. An indicator matrix turns on multilabel classification. This should be an array of integers. |\n",
@@ -164,7 +164,7 @@
" n_cross_validations = 5,\n",
" debug_log = 'automl.log',\n",
" verbosity = logging.INFO,\n",
" X = x_train, \n",
" X = X_train, \n",
" y = y_train,\n",
" path=project_folder)"
]
@@ -295,7 +295,8 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Best Model based on any iteration\n",
"#### Model from a specific iteration\n",
"\n",
"Simply show the run and model from the 3rd iteration:"
]
},
@@ -311,25 +312,6 @@
"print(third_model)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Register fitted model for deployment"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"description = 'AutoML Model'\n",
"tags = None\n",
"local_run.register_model(description = description, tags = tags)\n",
"print(local_run.model_id) # Use this id to deploy the model as a web service in Azure"
]
},
{
"cell_type": "markdown",
"metadata": {},
@@ -350,10 +332,10 @@
"metadata": {},
"outputs": [],
"source": [
"y_pred_train = fitted_model.predict(x_train)\n",
"y_pred_train = fitted_model.predict(X_train)\n",
"y_residual_train = y_train - y_pred_train\n",
"\n",
"y_pred_test = fitted_model.predict(x_test)\n",
"y_pred_test = fitted_model.predict(X_test)\n",
"y_residual_test = y_test - y_pred_test"
]
},
@@ -380,7 +362,7 @@
"a0.plot(y_residual_train, 'bo', alpha = 0.5)\n",
"a0.plot([-10,360],[0,0], 'r-', lw = 3)\n",
"a0.text(16,170,'RMSE = {0:.2f}'.format(np.sqrt(mean_squared_error(y_train, y_pred_train))), fontsize = 12)\n",
"a0.text(16,140,'Variance = {0:.2f}'.format(r2_score(y_train, y_pred_train)), fontsize = 12)\n",
"a0.text(16,140,'R2 score = {0:.2f}'.format(r2_score(y_train, y_pred_train)), fontsize = 12)\n",
"a0.set_xlabel('Training samples', fontsize = 12)\n",
"a0.set_ylabel('Residual Values', fontsize = 12)\n",
"# plot histogram\n",
@@ -392,7 +374,7 @@
"a1.plot(y_residual_test, 'bo', alpha = 0.5)\n",
"a1.plot([-10,360],[0,0], 'r-', lw = 3)\n",
"a1.text(5,170,'RMSE = {0:.2f}'.format(np.sqrt(mean_squared_error(y_test, y_pred_test))), fontsize = 12)\n",
"a1.text(5,140,'Variance = {0:.2f}'.format(r2_score(y_test, y_pred_test)), fontsize = 12)\n",
"a1.text(5,140,'R2 score = {0:.2f}'.format(r2_score(y_test, y_pred_test)), fontsize = 12)\n",
"a1.set_xlabel('Test samples', fontsize = 12)\n",
"a1.set_yticklabels([])\n",
"# plot histogram\n",

36
automl/03.auto-ml-remote-execution.ipynb
View File

@@ -15,7 +15,7 @@
"source": [
"# AutoML 03: Remote Execution using DSVM (Ubuntu)\n",
"\n",
"In this example we use the scikit learn's [diabetes dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_diabetes.html) to showcase how you can use the AutoML Classifier for a simple classification problem.\n",
"In this example we use the scikit learn's [digit dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_digits.html) to showcase how you can use AutoML for a simple classification problem.\n",
"\n",
"Make sure you have executed the [00.configuration](00.configuration.ipynb) before running this notebook.\n",
"\n",
@@ -194,8 +194,8 @@
"|Property|Description|\n",
"|-|-|\n",
"|**primary_metric**|This is the metric that you want to optimize.<br> Classification supports the following primary metrics <br><i>accuracy</i><br><i>AUC_weighted</i><br><i>balanced_accuracy</i><br><i>average_precision_score_weighted</i><br><i>precision_score_weighted</i>|\n",
"|**max_time_sec**|Time limit in seconds for each iterations|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML Classifier trains the data with a specific pipeline|\n",
"|**max_time_sec**|Time limit in seconds for each iteration|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML trains a specific pipeline with the data|\n",
"|**n_cross_validations**|Number of cross validation splits|\n",
"|**concurrent_iterations**|Max number of iterations that would be executed in parallel. This should be less than the number of cores on the DSVM."
]
@@ -220,7 +220,7 @@
" debug_log = 'automl_errors.log',\n",
" path=project_folder, \n",
" compute_target = dsvm_compute,\n",
" data_script = project_folder + \"./get_data.py\",\n",
" data_script = project_folder + \"/get_data.py\",\n",
" **automl_settings\n",
" )\n"
]
@@ -383,7 +383,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Best Model based on any iteration\n",
"#### Model from a specific iteration\n",
"Show the run and model from the 3rd iteration."
]
},
@@ -399,25 +399,6 @@
"print(third_model)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Register fitted model for deployment"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"description = 'AutoML Model'\n",
"tags = None\n",
"remote_run.register_model(description=description, tags=tags)\n",
"remote_run.model_id # Use this id to deploy the model as a web service in Azure"
]
},
{
"cell_type": "markdown",
"metadata": {},
@@ -464,13 +445,6 @@
" plt.imshow(images[index], cmap=plt.cm.gray_r, interpolation='nearest')\n",
" plt.show()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {

10
automl/03b.auto-ml-remote-batchai.ipynb
View File

@@ -15,7 +15,7 @@
"source": [
"# AutoML 03: Remote Execution using Batch AI\n",
"\n",
"In this example we use the scikit learn's [diabetes dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_diabetes.html) to showcase how you can use the AutoML Classifier for a simple classification problem.\n",
"In this example we use the scikit learn's [diabetes dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_diabetes.html) to showcase how you can use AutoML for a simple classification problem.\n",
"\n",
"Make sure you have executed the [setup](setup.ipynb) before running this notebook.\n",
"\n",
@@ -217,8 +217,8 @@
"|Property|Description|\n",
"|-|-|\n",
"|**primary_metric**|This is the metric that you want to optimize.<br> Classification supports the following primary metrics <br><i>accuracy</i><br><i>AUC_weighted</i><br><i>balanced_accuracy</i><br><i>average_precision_score_weighted</i><br><i>precision_score_weighted</i>|\n",
"|**max_time_sec**|Time limit in seconds for each iterations|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML Classifier trains the data with a specific pipeline|\n",
"|**max_time_sec**|Time limit in seconds for each iteration|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML trains a specific pipeline with the data|\n",
"|**n_cross_validations**|Number of cross validation splits|\n",
"|**concurrent_iterations**|Max number of iterations that would be executed in parallel. This should be less than the number of cores on the DSVM."
]
@@ -243,7 +243,7 @@
" debug_log = 'automl_errors.log',\n",
" path=project_folder,\n",
" compute_target = compute_target,\n",
" data_script = project_folder + \"./get_data.py\",\n",
" data_script = project_folder + \"/get_data.py\",\n",
" **automl_settings\n",
" )\n"
]
@@ -408,7 +408,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Best Model based on any iteration\n",
"#### Model from a specific iteration\n",
"Show the run and model from the 3rd iteration."
]
},

28
automl/04.auto-ml-remote-execution-text-data-blob-store.ipynb
View File

@@ -15,7 +15,7 @@
"source": [
"# Auto ML : Remote Execution with Text data from Blobstorage\n",
"\n",
"In this example we use the [Burning Man 2016 dataset](https://innovate.burningman.org/datasets-page/) to showcase how you can use the AutoML Classifier to handle text data from a Azure blobstorage.\n",
"In this example we use the [Burning Man 2016 dataset](https://innovate.burningman.org/datasets-page/) to showcase how you can use AutoML to handle text data from a Azure blobstorage.\n",
"\n",
"Make sure you have executed the [00.configuration](00.configuration.ipynb) before running this notebook.\n",
"\n",
@@ -227,11 +227,11 @@
"|Property|Description|\n",
"|-|-|\n",
"|**primary_metric**|This is the metric that you want to optimize.<br> Classification supports the following primary metrics <br><i>accuracy</i><br><i>AUC_weighted</i><br><i>balanced_accuracy</i><br><i>average_precision_score_weighted</i><br><i>precision_score_weighted</i>|\n",
"|**max_time_sec**|Time limit in seconds for each iterations|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML Classifier trains the data with a specific pipeline|\n",
"|**max_time_sec**|Time limit in seconds for each iteration|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML trains a specific pipeline with the data|\n",
"|**n_cross_validations**|Number of cross validation splits|\n",
"|**concurrent_iterations**|Max number of iterations that would be executed in parallel. This should be less than the number of cores on the DSVM\n",
"|**preprocess**| *True/False* <br>Setting this to *True* enables Auto ML Classifier to perform preprocessing <br>on the input to handle *missing data*, and perform some common *feature extraction*|\n",
"|**preprocess**| *True/False* <br>Setting this to *True* enables AutoML to perform preprocessing <br>on the input to handle *missing data*, and perform some common *feature extraction*|\n",
"|**max_cores_per_iteration**| Indicates how many cores on the compute target would be used to train a single pipeline.<br> Default is *1*, you can set it to *-1* to use all cores|"
]
},
@@ -242,7 +242,7 @@
"outputs": [],
"source": [
"automl_settings = {\n",
" \"max_time_sec\": 12000,\n",
" \"max_time_sec\": 3600,\n",
" \"iterations\": 10,\n",
" \"n_cross_validations\": 5,\n",
" \"primary_metric\": 'AUC_weighted',\n",
@@ -253,7 +253,7 @@
"automl_config = AutoMLConfig(task = 'classification',\n",
" path=project_folder,\n",
" compute_target = dsvm_compute,\n",
" data_script = project_folder + \"./get_data.py\",\n",
" data_script = project_folder + \"/get_data.py\",\n",
" **automl_settings\n",
" )\n"
]
@@ -264,17 +264,7 @@
"source": [
"## Training the Model <a class=\"anchor\" id=\"Training-the-model-Remote-DSVM\"></a>\n",
"\n",
"You can call the *fit* method on the AutoML instance and pass the dsvm runconfig name. For remote runs the execution is asynchronous, so you will see the iterations get populated as they complete. You can interact with the widgets/models even when the experiment is running to retreive the best model up to that point. Once you are satisfied with the model you can cancel a particular iteration or the whole run.\n",
"\n",
"\n",
"*fit* method on Auto ML Classifier triggers the training of the model. It can be called with the following parameters\n",
"\n",
"**Note**: You cannot pass Numpy arrays directly to the fit method in case of remote executions.\n",
"\n",
"|**Parameter**|**Description**|\n",
"|-|-|\n",
"|**compute_target**|Indicates the compute used for training. <i>local</i> indicates train on the same compute which hosts the jupyter notebook. <br>For DSVM and Batch AI please refer to the relevant notebooks.|\n",
"|**show_output**| True/False to turn on/off console output|"
"For remote runs the execution is asynchronous, so you will see the iterations get populated as they complete. You can interact with the widgets/models even when the experiment is running to retreive the best model up to that point. Once you are satisfied with the model you can cancel a particular iteration or the whole run."
]
},
{
@@ -398,7 +388,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Best Model based on any iteration"
"#### Model from a specific iteration"
]
},
{
@@ -408,7 +398,7 @@
"outputs": [],
"source": [
"iteration = 0\n",
"best_run, fitted_model = remote_run.get_output(iteration=iteration)"
"zero_run, zero_model = remote_run.get_output(iteration=iteration)"
]
},
{

14
automl/05.auto-ml-missing-data-Blacklist-Early-Termination.ipynb
View File

@@ -15,7 +15,7 @@
"source": [
"# AutoML 05 : Blacklisting models, Early termination and handling missing data\n",
"\n",
"In this example we use the scikit learn's [digit dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_digits.html) to showcase how you can use the AutoML Classifier for handling missing values in data. We also provide a stopping metrics indicating a target for the primary metrics so that AutoML can terminate the run without necessarly going through all the iterations. Finally, If you want to avoid a certain pipeline, we allow you to specify a black list of algos that AutoML will ignore for this run.\n",
"In this example we use the scikit learn's [digit dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_digits.html) to showcase how you can use AutoML for handling missing values in data. We also provide a stopping metric indicating a target for the primary metric so that AutoML can terminate the run without necessarly going through all the iterations. Finally, if you want to avoid a certain pipeline, we allow you to specify a black list of algos that AutoML will ignore for this run.\n",
"\n",
"Make sure you have executed the [00.configuration](00.configuration.ipynb) before running this notebook.\n",
"\n",
@@ -158,16 +158,16 @@
"## Instantiate Auto ML Config\n",
"\n",
"\n",
"Instantiate a AutoML Object This creates an Experiment in Azure ML. You can reuse this objects to trigger multiple runs. Each run will be part of the same experiment.\n",
"This defines the settings and data used to run the experiment.\n",
"\n",
"|Property|Description|\n",
"|-|-|\n",
"|**task**|classification or regression|\n",
"|**primary_metric**|This is the metric that you want to optimize.<br> Classification supports the following primary metrics <br><i>accuracy</i><br><i>AUC_weighted</i><br><i>balanced_accuracy</i><br><i>average_precision_score_weighted</i><br><i>precision_score_weighted</i>|\n",
"|**max_time_sec**|Time limit in seconds for each iterations|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML Classifier trains the data with a specific pipeline|\n",
"|**max_time_sec**|Time limit in seconds for each iteration|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML trains the data with a specific pipeline|\n",
"|**n_cross_validations**|Number of cross validation splits|\n",
"|**preprocess**| *True/False* <br>Setting this to *True* enables Auto ML Classifier to perform preprocessing <br>on the input to handle *missing data*, and perform some common *feature extraction*|\n",
"|**preprocess**| *True/False* <br>Setting this to *True* enables Auto ML to perform preprocessing <br>on the input to handle *missing data*, and perform some common *feature extraction*|\n",
"|**exit_score**|*double* value indicating the target for *primary_metric*. <br> Once the target is surpassed the run terminates|\n",
"|**blacklist_algos**|*Array* of *strings* indicating pipelines to ignore for Auto ML.<br><br> Allowed values for **Classification**<br><i>LogisticRegression</i><br><i>SGDClassifierWrapper</i><br><i>NBWrapper</i><br><i>BernoulliNB</i><br><i>SVCWrapper</i><br><i>LinearSVMWrapper</i><br><i>KNeighborsClassifier</i><br><i>DecisionTreeClassifier</i><br><i>RandomForestClassifier</i><br><i>ExtraTreesClassifier</i><br><i>LightGBMClassifier</i><br><br>Allowed values for **Regression**<br><i>ElasticNet<i><br><i>GradientBoostingRegressor<i><br><i>DecisionTreeRegressor<i><br><i>KNeighborsRegressor<i><br><i>LassoLars<i><br><i>SGDRegressor<i><br><i>RandomForestRegressor<i><br><i>ExtraTreesRegressor<i>|\n",
"|**X**|(sparse) array-like, shape = [n_samples, n_features]|\n",
@@ -184,7 +184,7 @@
"automl_config = AutoMLConfig(task = 'classification',\n",
" debug_log = 'automl_errors.log',\n",
" primary_metric = 'AUC_weighted',\n",
" max_time_sec = 12000,\n",
" max_time_sec = 3600,\n",
" iterations = 20,\n",
" n_cross_validations = 5,\n",
" preprocess = True,\n",
@@ -308,7 +308,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Best Model based on any iteration"
"#### Model from a specific iteration"
]
},
{

15
automl/06.auto-ml-sparse-data-custom-cv-split.ipynb
View File

@@ -15,7 +15,7 @@
"source": [
"# AutoML 06: Custom CV splits, handling sparse data\n",
"\n",
"In this example we use the scikit learn's [20newsgroup](In this example we use the scikit learn's [digit dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_digits.html) to showcase how you can use the AutoML Classifier for handling sparse data and specify custom cross validations splits.\n",
"In this example we use the scikit learn's [20newsgroup](In this example we use the scikit learn's [digit dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_digits.html) to showcase how you can use AutoML for handling sparse data and specify custom cross validation splits.\n",
"\n",
"Make sure you have executed the [00.configuration](00.configuration.ipynb) before running this notebook.\n",
"\n",
@@ -157,16 +157,15 @@
"source": [
"## Instantiate Auto ML Config\n",
"\n",
"\n",
"Instantiate a AutoML Object This creates an Experiment in Azure ML. You can reuse this objects to trigger multiple runs. Each run will be part of the same experiment.\n",
"This defines the settings and data used to run the experiment.\n",
"\n",
"|Property|Description|\n",
"|-|-|\n",
"|**task**|classification or regression|\n",
"|**primary_metric**|This is the metric that you want to optimize.<br> Classification supports the following primary metrics <br><i>accuracy</i><br><i>AUC_weighted</i><br><i>balanced_accuracy</i><br><i>average_precision_score_weighted</i><br><i>precision_score_weighted</i>|\n",
"|**max_time_sec**|Time limit in seconds for each iterations|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML Classifier trains the data with a specific pipeline|\n",
"|**preprocess**| *True/False* <br>Setting this to *True* enables Auto ML Classifier to perform preprocessing <br>on the input to handle *missing data*, and perform some common *feature extraction*<br>*Note: If input data is Sparse you cannot use preprocess=True*|\n",
"|**max_time_sec**|Time limit in seconds for each iteration|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML trains a specific pipeline with the data|\n",
"|**preprocess**| *True/False* <br>Setting this to *True* enables Auto ML to perform preprocessing <br>on the input to handle *missing data*, and perform some common *feature extraction*<br>*Note: If input data is Sparse you cannot use preprocess=True*|\n",
"|**X**|(sparse) array-like, shape = [n_samples, n_features]|\n",
"|**y**|(sparse) array-like, shape = [n_samples, ], [n_samples, n_classes]<br>Multi-class targets. An indicator matrix turns on multilabel classification. This should be an array of integers. |\n",
"|**X_valid**|(sparse) array-like, shape = [n_samples, n_features] for the custom Validation set|\n",
@@ -183,7 +182,7 @@
"automl_config = AutoMLConfig(task = 'classification',\n",
" debug_log='automl_errors.log',\n",
" primary_metric='AUC_weighted',\n",
" max_time_sec=12000,\n",
" max_time_sec=3600,\n",
" iterations=5,\n",
" preprocess=False,\n",
" verbosity=logging.INFO,\n",
@@ -313,7 +312,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Best Model based on any iteration"
"#### Model from a specific iteration"
]
},
{

12
automl/07.auto-ml-exploring-previous-runs.ipynb
View File

@@ -21,9 +21,9 @@
"\n",
"In this notebook you would see\n",
"1. List all Experiments for the workspace\n",
"2. List all AutoML Runs for a given project\n",
"2. List AutoML runs for an Experiment\n",
"3. Get details for a AutoML Run. (Automl settings, run widget & all metrics)\n",
"4. Downlaod fitted pipeline for any iteration\n"
"4. Download fitted pipeline for any iteration\n"
]
},
{
@@ -105,7 +105,7 @@
"metadata": {},
"source": [
"# List AutoML runs for an Experiment\n",
"You can set <i>Experiment</i> name with any experiment name from the result of the previous cell to load the AutoML runs."
"You can set <i>Experiment</i> name with any experiment name from the result of the Experiment.list cell to load the AutoML runs."
]
},
{
@@ -154,7 +154,7 @@
"metadata": {},
"outputs": [],
"source": [
"run_id = 'AutoML_b7c4076b-181d-4ef4-ab9f-36bb44c1e36c'#<<run_id>> # Replace with a Auto ML Run ID like 'AutoML_a5a90841-8915-4fba-91a4-bbbfc07dd132'\n",
"run_id = 'AutoML_b7c4076b-181d-4ef4-ab9f-36bb44c1e36c'\n",
"\n",
"from azureml.train.widgets import RunDetails\n",
"\n",
@@ -237,7 +237,7 @@
"metadata": {},
"outputs": [],
"source": [
"iteration = 4 # Replace with a interation number\n",
"iteration = 4 # Replace with an interation number\n",
"best_run, fitted_model = ml_run.get_output(iteration=iteration)\n",
"fitted_model"
]
@@ -294,7 +294,7 @@
"metadata": {},
"outputs": [],
"source": [
"iteration = 4 # Replace with a interation number\n",
"iteration = 4 # Replace with an interation number\n",
"description = 'AutoML Model'\n",
"tags = None\n",
"ml_run.register_model(description=description, tags=tags, iteration=iteration)\n",

24
automl/08.auto-ml-remote-execution-with-text-file-on-DSVM.ipynb
View File

@@ -215,11 +215,11 @@
"|Property|Description|\n",
"|-|-|\n",
"|**primary_metric**|This is the metric that you want to optimize.<br> Classification supports the following primary metrics <br><i>accuracy</i><br><i>AUC_weighted</i><br><i>balanced_accuracy</i><br><i>average_precision_score_weighted</i><br><i>precision_score_weighted</i>|\n",
"|**max_time_sec**|Time limit in seconds for each iterations|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML Classifier trains the data with a specific pipeline|\n",
"|**max_time_sec**|Time limit in seconds for each iteration|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML trains a specific pipeline with the data|\n",
"|**n_cross_validations**|Number of cross validation splits|\n",
"|**concurrent_iterations**|Max number of iterations that would be executed in parallel. This should be less than the number of cores on the DSVM\n",
"|**preprocess**| *True/False* <br>Setting this to *True* enables Auto ML Classifier to perform preprocessing <br>on the input to handle *missing data*, and perform some common *feature extraction*|\n",
"|**preprocess**| *True/False* <br>Setting this to *True* enables Auto ML to perform preprocessing <br>on the input to handle *missing data*, and perform some common *feature extraction*|\n",
"|**max_cores_per_iteration**| Indicates how many cores on the compute target would be used to train a single pipeline.<br> Default is *1*, you can set it to *-1* to use all cores|"
]
},
@@ -230,7 +230,7 @@
"outputs": [],
"source": [
"automl_settings = {\n",
" \"max_time_sec\": 12000,\n",
" \"max_time_sec\": 3600,\n",
" \"iterations\": 10,\n",
" \"n_cross_validations\": 5,\n",
" \"primary_metric\": 'AUC_weighted',\n",
@@ -242,7 +242,7 @@
" debug_log = 'automl_errors.log',\n",
" path=project_folder,\n",
" compute_target = dsvm_compute,\n",
" data_script = project_folder + \"./get_data.py\",\n",
" data_script = project_folder + \"/get_data.py\",\n",
" **automl_settings\n",
" )"
]
@@ -253,17 +253,7 @@
"source": [
"## Training the Model <a class=\"anchor\" id=\"Training-the-model-Remote-DSVM\"></a>\n",
"\n",
"You can call the *fit* method on the AutoML instance and pass the dsvm runconfig name. For remote runs the execution is asynchronous, so you will see the iterations get populated as they complete. You can interact with the widgets/models even when the experiment is running to retreive the best model up to that point. Once you are satisfied with the model you can cancel a particular iteration or the whole run.\n",
"\n",
"\n",
"*fit* method on Auto ML Classifier triggers the training of the model. It can be called with the following parameters\n",
"\n",
"**Note**: You cannot pass Numpy arrays directly to the fit method in case of remote executions.\n",
"\n",
"|**Parameter**|**Description**|\n",
"|-|-|\n",
"|**compute_target**|Indicates the compute used for training. <i>local</i> indicates train on the same compute which hosts the jupyter notebook. <br>For DSVM and Batch AI please refer to the relevant notebooks.|\n",
"|**show_output**| True/False to turn on/off console output|"
"For remote runs the execution is asynchronous, so you will see the iterations get populated as they complete. You can interact with the widgets/models even when the experiment is running to retreive the best model up to that point. Once you are satisfied with the model you can cancel a particular iteration or the whole run."
]
},
{
@@ -385,7 +375,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Best Model based on any iteration"
"#### Model from a specific iteration"
]
},
{

6
automl/09.auto-ml-classification-with-deployment.ipynb
View File

@@ -15,7 +15,7 @@
"source": [
"# AutoML 09: Classification with deployment\n",
"\n",
"In this example we use the scikit learn's [digit dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_digits.html) to showcase how you can use the AutoML Classifier for a simple classification problem.\n",
"In this example we use the scikit learn's [digit dataset](http://scikit-learn.org/stable/modules/generated/sklearn.datasets.load_digits.html) to showcase how you can use AutoML for a simple classification problem.\n",
"\n",
"Make sure you have executed the [00.configuration](00.configuration.ipynb) before running this notebook.\n",
"\n",
@@ -120,8 +120,8 @@
"|-|-|\n",
"|**task**|classification or regression|\n",
"|**primary_metric**|This is the metric that you want to optimize.<br> Classification supports the following primary metrics <br><i>accuracy</i><br><i>AUC_weighted</i><br><i>balanced_accuracy</i><br><i>average_precision_score_weighted</i><br><i>precision_score_weighted</i>|\n",
"|**max_time_sec**|Time limit in seconds for each iterations|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML trains the data with a specific pipeline|\n",
"|**max_time_sec**|Time limit in seconds for each iteration|\n",
"|**iterations**|Number of iterations. In each iteration Auto ML trains a specific pipeline with the data|\n",
"|**n_cross_validations**|Number of cross validation splits|\n",
"|**X**|(sparse) array-like, shape = [n_samples, n_features]|\n",
"|**y**|(sparse) array-like, shape = [n_samples, ], [n_samples, n_classes]<br>Multi-class targets. An indicator matrix turns on multilabel classification. This should be an array of integers. |\n",

22
automl/11.auto-ml-sample-weight.ipynb
View File

@@ -133,7 +133,7 @@
"automl_classifier = AutoMLConfig(task = 'classification',\n",
" debug_log = 'automl_errors.log',\n",
" primary_metric = 'AUC_weighted',\n",
" max_time_sec = 12000,\n",
" max_time_sec = 3600,\n",
" iterations = 10,\n",
" n_cross_validations = 2,\n",
" verbosity = logging.INFO,\n",
@@ -144,7 +144,7 @@
"automl_sample_weight = AutoMLConfig(task = 'classification',\n",
" debug_log = 'automl_errors.log',\n",
" primary_metric = 'AUC_weighted',\n",
" max_time_sec = 12000,\n",
" max_time_sec = 3600,\n",
" iterations = 10,\n",
" n_cross_validations = 2,\n",
" verbosity = logging.INFO,\n",
@@ -160,18 +160,8 @@
"source": [
"## Training the Models\n",
"\n",
"You can call the fit method on the AutoML instance and pass the run configuration. For Local runs the execution is synchronous. Depending on the data and number of iterations this can run for while.\n",
"You will see the currently running iterations printing to the console.\n",
"\n",
"*fit* method on Auto ML Classifier triggers the training of the model. It can be called with the following parameters\n",
"\n",
"|**Parameter**|**Description**|\n",
"|-|-|\n",
"|**X**|(sparse) array-like, shape = [n_samples, n_features]|\n",
"|**y**|(sparse) array-like, shape = [n_samples, ], [n_samples, n_classes]<br>Multi-class targets. An indicator matrix turns on multilabel classification. This should be an array of integers. |\n",
"|**sample_weight**|(sparse) array-like, shape must be the same as **y**.<br>A weight value for each label. Higher values indicate that the sample is more important.|\n",
"|**compute_target**|Indicates the compute used for training. <i>local</i> indicates train on the same compute which hosts the jupyter notebook. <br>For DSVM and Batch AI please refer to the relevant notebooks.|\n",
"|**show_output**| True/False to turn on/off console output|"
"Call the submit method on the experiment and pass the configuration. For Local runs the execution is synchronous. Depending on the data and number of iterations this can run for while.\n",
"You will see the currently running iterations printing to the console."
]
},
{
@@ -213,7 +203,7 @@
"metadata": {},
"source": [
"#### Compare the pipelines\n",
"The prediction from the sample weight model correctly predicts all 4's including one that the model without sample weights does not. However, it also predicts 4 for two images that are not labelled as 4."
"The prediction from the sample weight model is more likely to correctly predict 4's. However, it is also more likely to predict 4 for some images that are not labelled as 4."
]
},
{
@@ -253,7 +243,7 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.6"
"version": "3.6.5"
}
},
"nbformat": 4,

28
automl/README.md
View File

@@ -7,21 +7,21 @@
6. [Troubleshooting](#troubleshooting)
# Auto ML Introduction <a name="introduction"></a>
AutoML builds high quality Machine Learning model for you by automating model selection and hyper parameter selection for you. Bring a labelled dataset that you want to build a model for, AutoML will give you a high quality machine learning model that you can use for predictions.
AutoML builds high quality Machine Learning models for you by automating model and hyperparameter selection. Bring a labelled dataset that you want to build a model for, AutoML will give you a high quality machine learning model that you can use for predictions.
If you are new to Data Science, AutoML will help you get jumpstarted by simplifying machine learning model building. It abstracts you from needing to perform model selection, hyper parameter selection and in one step creates a high quality trained model for you to use.
If you are new to Data Science, AutoML will help you get jumpstarted by simplifying machine learning model building. It abstracts you from needing to perform model selection, hyperparameter selection and in one step creates a high quality trained model for you to use.
If you are an experienced data scientist, AutoML will help increase your productivity by intelligently performing the model selection, hyper parameter selection for your training and generates high quality models much quicker than manually specifying several combinations of the parameters and running training jobs. AutoML provides visibility and access to all the training jobs and the performance characteristics of the models and help you further tune the pipeline if you desire.
If you are an experienced data scientist, AutoML will help increase your productivity by intelligently performing the model and hyperparameter selection for your training and generates high quality models much quicker than manually specifying several combinations of the parameters and running training jobs. AutoML provides visibility and access to all the training jobs and the performance characteristics of the models to help you further tune the pipeline if you desire.
# Running samples in a Local Conda environment <a name="localconda"></a>
It is best if you create a new conda environment locally to try this SDK, so it doesn't mess up with your existing Python environment.
### 1. Install mini-conda from [here](https://conda.io/miniconda.html), choose Python 3.7 or higher.
- **Note**: if you already have conda installed, you can keep using it but it must be version 5.2 or later. If you have an previous version installed, you can update it using the command: conda update conda.
- **Note**: if you already have conda installed, you can keep using it but it should be version 4.4.10 or later (as shown by: conda -V). If you have a previous version installed, you can update it using the command: conda update conda.
There's no need to install mini-conda specifically.
### 2. Dowloading the sample notebooks
### 2. Downloading the sample notebooks
- Download the sample notebooks from [GitHub](https://github.com/Azure/MachineLearningNotebooks) as zip and extract the contents to a local directory. The AutoML sample notebooks are in the "automl" folder.
### 3. Setup a new conda environment
@@ -127,6 +127,12 @@ cd to the "automl" folder where the sample notebooks were extracted and then run
- [13.auto-ml-dataprep.ipynb](13.auto-ml-dataprep.ipynb)
- Using DataPrep for reading data
- [14a.auto-ml-classification-ensemble.ipynb](14a.auto-ml-classification-ensemble.ipynb)
- Classification with ensembling
- [14b.auto-ml-regression-ensemble.ipynb](14b.auto-ml-regression-ensemble.ipynb)
- Regression with ensembling
# Documentation <a name="documentation"></a>
## Table of Contents
1. [Auto ML Settings ](#automlsettings)
@@ -137,12 +143,12 @@ cd to the "automl" folder where the sample notebooks were extracted and then run
|Property|Description|Default|
|-|-|-|
|**primary_metric**|This is the metric that you want to optimize.<br><br> Classification supports the following primary metrics <br><i>accuracy</i><br><i>AUC_weighted</i><br><i>balanced_accuracy</i><br><i>average_precision_score_weighted</i><br><i>precision_score_weighted</i><br><br> Regression supports the following primary metrics <br><i>spearman_correlation</i><br><i>normalized_root_mean_squared_error</i><br><i>r2_score</i><br><i>normalized_mean_absolute_error</i><br><i>normalized_root_mean_squared_log_error</i>| Classification: accuracy <br><br> Regression: spearman_correlation
|**max_time_sec**|Time limit in seconds for each iterations|None|
|**iterations**|Number of iterations. In each iteration trains the data with a specific pipeline. To get the best result, use at least 100. |25|
|**max_time_sec**|Time limit in seconds for each iteration|None|
|**iterations**|Number of iterations. In each iteration trains the data with a specific pipeline. To get the best result, use at least 100. |100|
|**n_cross_validations**|Number of cross validation splits|None|
|**validation_size**|Size of validation set as percentage of all training samples|None|
|**concurrent_iterations**|Max number of iterations that would be executed in parallel|1|
|**preprocess**|*True/False* <br>Setting this to *True* enables preprocessing <br>on the input to handle *missing data*, and perform some common *feature extraction*<br>*Note: If input data is Sparse you cannot use preprocess=True*|False|
|**preprocess**|*True/False* <br>Setting this to *True* enables preprocessing <br>on the input to handle missing data, and perform some common feature extraction<br>*Note: If input data is Sparse you cannot use preprocess=True*|False|
|**max_cores_per_iteration**| Indicates how many cores on the compute target would be used to train a single pipeline.<br> You can set it to *-1* to use all cores|1|
|**exit_score**|*double* value indicating the target for *primary_metric*. <br> Once the target is surpassed the run terminates|None|
|**blacklist_algos**|*Array* of *strings* indicating pipelines to ignore for Auto ML.<br><br> Allowed values for **Classification**<br><i>LogisticRegression</i><br><i>SGDClassifierWrapper</i><br><i>NBWrapper</i><br><i>BernoulliNB</i><br><i>SVCWrapper</i><br><i>LinearSVMWrapper</i><br><i>KNeighborsClassifier</i><br><i>DecisionTreeClassifier</i><br><i>RandomForestClassifier</i><br><i>ExtraTreesClassifier</i><br><i>gradient boosting</i><br><i>LightGBMClassifier</i><br><br>Allowed values for **Regression**<br><i>ElasticNet</i><br><i>GradientBoostingRegressor</i><br><i>DecisionTreeRegressor</i><br><i>KNeighborsRegressor</i><br><i>LassoLars</i><br><i>SGDRegressor</i><br><i>RandomForestRegressor</i><br><i>ExtraTreesRegressor</i>|None|
@@ -166,8 +172,8 @@ The *get_data()* function can be used to return a dictionary with these values:
|y|Pandas Dataframe or Numpy Array|X|label|Label data to train with. For classification, this should be an array of integers. |
|X_valid|Pandas Dataframe or Numpy Array|X, y, y_valid|data_train, label|*Optional* All features to validate with. If this is not specified, X is split between train and validate|
|y_valid|Pandas Dataframe or Numpy Array|X, y, X_valid|data_train, label|*Optional* The label data to validate with. If this is not specified, y is split between train and validate|
|sample_weight|Pandas Dataframe or Numpy Array|y|data_train, label, columns|*Optional*A weight value for each label. Higher values indicate that the sample is more important.|
|sample_weight_valid|Pandas Dataframe or Numpy Array|y_valid|data_train, label, columns|*Optional*A weight value for each validation label. Higher values indicate that the sample is more important. If this is not specified, sample_weight is split between train and validate|
|sample_weight|Pandas Dataframe or Numpy Array|y|data_train, label, columns|*Optional* A weight value for each label. Higher values indicate that the sample is more important.|
|sample_weight_valid|Pandas Dataframe or Numpy Array|y_valid|data_train, label, columns|*Optional* A weight value for each validation label. Higher values indicate that the sample is more important. If this is not specified, sample_weight is split between train and validate|
|data_train|Pandas Dataframe|label|X, y, X_valid, y_valid|All data (features+label) to train with|
|label|string|data_train|X, y, X_valid, y_valid|Which column in data_train represents the label|
|columns|Array of strings|data_train||*Optional* Whitelist of columns to use for features|
@@ -193,5 +199,3 @@ To resolve this issue, allocate a DSVM with more memory or reduce the value spec
This can be caused by too many concurrent iterations for a remote DSVM. Each concurrent iteration usually takes 100% of a core when it is running. Some iterations can use multiple cores. So, the concurrent_iterations setting should always be less than the number of cores of the DSVM.
To resolve this issue, try reducing the value specified for the concurrent_iterations setting.
## Workspace.create gives the error "The resource type could not be found in the namespace 'Microsoft.MachineLearningServices' for api version '2018-03-01-preview'."
This can indicate that the Azure Subscription has not been whitelisted for AutoML.

1
databricks/01.Installation_and_Configuration.ipynb Normal file
View File

@@ -0,0 +1 @@
{"cells":[{"cell_type":"markdown","source":["Azure ML & Azure Databricks notebooks by Parashar Shah.\n\nCopyright (c) Microsoft Corporation. All rights reserved.\n\nLicensed under the MIT License."],"metadata":{}},{"cell_type":"markdown","source":["Please ensure you have run this notebook before proceeding."],"metadata":{}},{"cell_type":"markdown","source":["Now we support installing AML SDK as library from GUI. When attaching a library follow this https://docs.databricks.com/user-guide/libraries.html and add the below string as your PyPi package (during private preview). You can select the option to attach the library to all clusters or just one cluster.\n\nProvide this full string to install the SDK:\n\nazureml-sdk[databricks]"],"metadata":{}},{"cell_type":"code","source":["import azureml.core\n\n# Check core SDK version number - based on build number of preview/master.\nprint(\"SDK version:\", azureml.core.VERSION)"],"metadata":{},"outputs":[],"execution_count":4},{"cell_type":"code","source":["subscription_id = \"<your-subscription-id>\"\nresource_group = \"<your-existing-resource-group>\"\nworkspace_name = \"<a-new-or-existing-workspace; it is unrelated to Databricks workspace>\"\nworkspace_region = \"<your-resource group-region>\""],"metadata":{},"outputs":[],"execution_count":5},{"cell_type":"code","source":["# import the Workspace class and check the azureml SDK version\n# exist_ok checks if workspace exists or not.\n\nfrom azureml.core import Workspace\n\nws = Workspace.create(name = workspace_name,\n subscription_id = subscription_id,\n resource_group = resource_group, \n location = workspace_region,\n exist_ok=True)\n\nws.get_details()"],"metadata":{},"outputs":[],"execution_count":6},{"cell_type":"code","source":["ws = Workspace(workspace_name = workspace_name,\n subscription_id = subscription_id,\n resource_group = resource_group)\n\n# persist the subscription id, resource group name, and workspace name in aml_config/config.json.\nws.write_config()"],"metadata":{},"outputs":[],"execution_count":7},{"cell_type":"code","source":["%sh\ncat /databricks/driver/aml_config/config.json"],"metadata":{},"outputs":[],"execution_count":8},{"cell_type":"code","source":["# import the Workspace class and check the azureml SDK version\nfrom azureml.core import Workspace\n\nws = Workspace.from_config()\nprint('Workspace name: ' + ws.name, \n 'Azure region: ' + ws.location, \n 'Subscription id: ' + ws.subscription_id, \n 'Resource group: ' + ws.resource_group, sep = '\\n')"],"metadata":{},"outputs":[],"execution_count":9},{"cell_type":"code","source":["dbutils.notebook.exit(\"success\")"],"metadata":{},"outputs":[],"execution_count":10},{"cell_type":"code","source":[""],"metadata":{},"outputs":[],"execution_count":11}],"metadata":{"name":"01.Installation_and_Configuration","notebookId":3874566296719377},"nbformat":4,"nbformat_minor":0}

1
databricks/02.Ingest_data.ipynb Normal file
View File

@@ -0,0 +1 @@
{"cells":[{"cell_type":"markdown","source":["Azure ML & Azure Databricks notebooks by Parashar Shah.\n\nCopyright (c) Microsoft Corporation. All rights reserved.\n\nLicensed under the MIT License."],"metadata":{}},{"cell_type":"markdown","source":["Please ensure you have run all previous notebooks in sequence before running this."],"metadata":{}},{"cell_type":"markdown","source":["#Data Ingestion"],"metadata":{}},{"cell_type":"code","source":["import os\nimport urllib"],"metadata":{},"outputs":[],"execution_count":4},{"cell_type":"code","source":["# Download AdultCensusIncome.csv from Azure CDN. This file has 32,561 rows.\nbasedataurl = \"https://amldockerdatasets.azureedge.net\"\ndatafile = \"AdultCensusIncome.csv\"\ndatafile_dbfs = os.path.join(\"/dbfs\", datafile)\n\nif os.path.isfile(datafile_dbfs):\n print(\"found {} at {}\".format(datafile, datafile_dbfs))\nelse:\n print(\"downloading {} to {}\".format(datafile, datafile_dbfs))\n urllib.request.urlretrieve(os.path.join(basedataurl, datafile), datafile_dbfs)"],"metadata":{},"outputs":[],"execution_count":5},{"cell_type":"code","source":["# Create a Spark dataframe out of the csv file.\ndata_all = sqlContext.read.format('csv').options(header='true', inferSchema='true', ignoreLeadingWhiteSpace='true', ignoreTrailingWhiteSpace='true').load(datafile)\nprint(\"({}, {})\".format(data_all.count(), len(data_all.columns)))\ndata_all.printSchema()"],"metadata":{},"outputs":[],"execution_count":6},{"cell_type":"code","source":["#renaming columns\ncolumns_new = [col.replace(\"-\", \"_\") for col in data_all.columns]\ndata_all = data_all.toDF(*columns_new)\ndata_all.printSchema()"],"metadata":{},"outputs":[],"execution_count":7},{"cell_type":"code","source":["display(data_all.limit(5))"],"metadata":{},"outputs":[],"execution_count":8},{"cell_type":"markdown","source":["#Data Preparation"],"metadata":{}},{"cell_type":"code","source":["# Choose feature columns and the label column.\nlabel = \"income\"\nxvals_all = set(data_all.columns) - {label}\n\n#dbutils.widgets.remove(\"xvars_multiselect\")\ndbutils.widgets.removeAll()\n\ndbutils.widgets.multiselect('xvars_multiselect', 'hours_per_week', xvals_all)\nxvars_multiselect = dbutils.widgets.get(\"xvars_multiselect\")\nxvars = xvars_multiselect.split(\",\")\n\nprint(\"label = {}\".format(label))\nprint(\"features = {}\".format(xvars))\n\ndata = data_all.select([*xvars, label])\n\n# Split data into train and test.\ntrain, test = data.randomSplit([0.75, 0.25], seed=123)\n\nprint(\"train ({}, {})\".format(train.count(), len(train.columns)))\nprint(\"test ({}, {})\".format(test.count(), len(test.columns)))"],"metadata":{},"outputs":[],"execution_count":10},{"cell_type":"markdown","source":["#Data Persistence"],"metadata":{}},{"cell_type":"code","source":["# Write the train and test data sets to intermediate storage\ntrain_data_path = \"AdultCensusIncomeTrain\"\ntest_data_path = \"AdultCensusIncomeTest\"\n\ntrain_data_path_dbfs = os.path.join(\"/dbfs\", \"AdultCensusIncomeTrain\")\ntest_data_path_dbfs = os.path.join(\"/dbfs\", \"AdultCensusIncomeTest\")\n\ntrain.write.mode('overwrite').parquet(train_data_path)\ntest.write.mode('overwrite').parquet(test_data_path)\nprint(\"train and test datasets saved to {} and {}\".format(train_data_path_dbfs, test_data_path_dbfs))"],"metadata":{},"outputs":[],"execution_count":12},{"cell_type":"code","source":["dbutils.notebook.exit(\"success\")"],"metadata":{},"outputs":[],"execution_count":13},{"cell_type":"code","source":[""],"metadata":{},"outputs":[],"execution_count":14}],"metadata":{"name":"02.Ingest_data","notebookId":3874566296719393},"nbformat":4,"nbformat_minor":0}

1
databricks/03a.Build_model.ipynb Normal file
View File

File diff suppressed because one or more lines are too long

1
databricks/03b.Build_model_runHistory.ipynb Normal file
View File

File diff suppressed because one or more lines are too long

1
databricks/04.Deploy_to_ACI.ipynb Normal file
View File

File diff suppressed because one or more lines are too long

1
databricks/05.Deploy_to_AKS_existingImage.ipynb Normal file
View File

@@ -0,0 +1 @@
{"cells":[{"cell_type":"markdown","source":["Azure ML & Azure Databricks notebooks by Parashar Shah.\n\nCopyright (c) Microsoft Corporation. All rights reserved.\n\nLicensed under the MIT License."],"metadata":{}},{"cell_type":"markdown","source":["Please ensure you have run all previous notebooks in sequence before running this. This notebook uses image from ACI notebook for deploying to AKS."],"metadata":{}},{"cell_type":"code","source":["from azureml.core import Workspace\nimport azureml.core\n\n# Check core SDK version number\nprint(\"SDK version:\", azureml.core.VERSION)\n\n#'''\nws = Workspace.from_config()\nprint('Workspace name: ' + ws.name, \n 'Azure region: ' + ws.location, \n 'Subscription id: ' + ws.subscription_id, \n 'Resource group: ' + ws.resource_group, sep = '\\n')\n#'''"],"metadata":{},"outputs":[],"execution_count":3},{"cell_type":"code","source":["# List images by ws\n\nfrom azureml.core.image import ContainerImage\nfor i in ContainerImage.list(workspace = ws):\n print('{}(v.{} [{}]) stored at {} with build log {}'.format(i.name, i.version, i.creation_state, i.image_location, i.image_build_log_uri))"],"metadata":{},"outputs":[],"execution_count":4},{"cell_type":"code","source":["from azureml.core.image import Image\nmyimage = Image(workspace=ws, id=\"aciws:25\")"],"metadata":{},"outputs":[],"execution_count":5},{"cell_type":"code","source":["#create AKS compute\n#it may take 20-25 minutes to create a new cluster\n\nfrom azureml.core.compute import AksCompute, ComputeTarget\n\n# Use the default configuration (can also provide parameters to customize)\nprov_config = AksCompute.provisioning_configuration()\n\naks_name = 'ps-aks-clus2' \n\n# Create the cluster\naks_target = ComputeTarget.create(workspace = ws, \n name = aks_name, \n provisioning_configuration = prov_config)\n\naks_target.wait_for_completion(show_output = True)\n\nprint(aks_target.provisioning_state)\nprint(aks_target.provisioning_errors)"],"metadata":{},"outputs":[],"execution_count":6},{"cell_type":"code","source":["from azureml.core.webservice import Webservice\nhelp( Webservice.deploy_from_image)"],"metadata":{},"outputs":[],"execution_count":7},{"cell_type":"code","source":["from azureml.core.webservice import Webservice, AksWebservice\nfrom azureml.core.image import ContainerImage\n\n#Set the web service configuration (using default here)\naks_config = AksWebservice.deploy_configuration()\n\n#unique service name\nservice_name ='ps-aks-service'\n\n# Webservice creation using single command, there is a variant to use image directly as well.\naks_service = Webservice.deploy_from_image(\n workspace=ws, \n name=service_name,\n deployment_config = aks_config,\n image = myimage,\n deployment_target = aks_target\n )\n\naks_service.wait_for_deployment(show_output=True)"],"metadata":{},"outputs":[],"execution_count":8},{"cell_type":"code","source":["#for using the Web HTTP API \nprint(aks_service.scoring_uri)\nprint(aks_service.get_keys())"],"metadata":{},"outputs":[],"execution_count":9},{"cell_type":"code","source":["import json\n\n#get the some sample data\ntest_data_path = \"AdultCensusIncomeTest\"\ntest = spark.read.parquet(test_data_path).limit(5)\n\ntest_json = json.dumps(test.toJSON().collect())\n\nprint(test_json)"],"metadata":{},"outputs":[],"execution_count":10},{"cell_type":"code","source":["#using data defined above predict if income is >50K (1) or <=50K (0)\naks_service.run(input_data=test_json)"],"metadata":{},"outputs":[],"execution_count":11},{"cell_type":"code","source":["#comment to not delete the web service\naks_service.delete()\n#image.delete()\n#model.delete()\n#aks_target.delete()"],"metadata":{},"outputs":[],"execution_count":12},{"cell_type":"code","source":[""],"metadata":{},"outputs":[],"execution_count":13}],"metadata":{"name":"04.DeploytoACI","notebookId":3874566296719318},"nbformat":4,"nbformat_minor":0}

BIN
databricks/Databricks_AMLSDK_github.dbc Normal file
View File

Binary file not shown.

26
databricks/readme.md Normal file
View File

@@ -0,0 +1,26 @@
# Azure Databricks - Azure ML SDK Sample Notebooks
**NOTE**: With the latest version of our AML SDK, there are some API changes due to which previous version of notebooks will not work.
Kindly use this v4 notebooks (updated Sep 18)– if you had installed the AML SDK in your Databricks cluster please update to latest SDK version by installing azureml-sdk[databricks] as a library from GUI.
**NOTE**: Please create your Azure Databricks cluster as v4.x (high concurrency preferred) with **Python 3** (dropdown). We are extending it to more runtimes asap.
**NOTE**: Some packages like psutil upgrade libs that can cause a conflict, please install such packages by freezing lib version. Eg. "pstuil **cryptography==1.5 pyopenssl==16.0.0 ipython=2.2.0**" to avoid install error. This issue is related to Databricks and not related to AML SDK.
**NOTE**: You should at least have contributor access to your Azure subcription to run some of the notebooks.
The iPython Notebooks have to be run sequentially after making changes based on your subscription. The corresponding DBC archive contains all the notebooks and can be imported into your Databricks workspace. You can the run notebooks after importing .dbc instead of downloading individually.
This set of notebooks are related to Income prediction experiment based on this [dataset](https://archive.ics.uci.edu/ml/datasets/adult) and demonstrate how to data prep, train and operationalize a Spark ML model with Azure ML Python SDK from within Azure Databricks. For details on SDK concepts, please refer to [Private preview notebooks](https://github.com/Azure/ViennaDocs/tree/master/PrivatePreview/notebooks)
(Recommended) [Azure Databricks AML SDK notebooks](Databricks_AMLSDK_github.dbc) A single DBC package to import all notebooks in your Databricks workspace.
01. [Installation and Configuration](01.Installation_and_Configuration.ipynb): Install the Azure ML Python SDK and Initialize an Azure ML Workspace and save the Workspace configuration file.
02. [Ingest data](02.Ingest_data.ipynb): Download the Adult Census Income dataset and split it into train and test sets.
03. [Build model](03a.Build_model.ipynb): Train a binary classification model in Azure Databricks with a Spark ML Pipeline.
04. [Build model with Run History](03b.Build_model_runHistory.ipynb): Train model and also capture run history (tracking) with Azure ML Python SDK.
05. [Deploy to ACI](04.Deploy_to_ACI.ipynb): Deploy model to Azure Container Instance (ACI) with Azure ML Python SDK.
06. [Deploy to AKS](04.Deploy_to_AKS_existingImage.ipynb): Deploy model to Azure Kubernetis Service (AKS) with Azure ML Python SDK from an existing Image with model, conda and score file.
Copyright (c) Microsoft Corporation. All rights reserved.
All notebooks in this folder are licensed under the MIT License.

30
onnx/release.json Normal file
View File

@@ -0,0 +1,30 @@
{
"channels": {
"master": [
"sample-01",
"sample-02"
],
"candidate": [
"sample-01",
"sample-02"
],
"preview": [
"sample-01",
"sample-02"
]
},
"notebooks": {
"sample-01": {
"name": "onnx-inference-mnist.ipynb",
"widgets": [ "azureml.train.widgets" ],
"dependencies": [],
"requirements": [ "matplotlib", "numpy", "onnx"]
},
"sample-02": {
"name": "onnx-inference-emotion-recognition.ipynb",
"widgets": ["azureml.train.widgets"],
"dependencies": [],
"requirements": [ "matplotlib", "numpy", "onnx"]
}
}
}

76
pipeline/00.pipeline-setup.ipynb Normal file
View File

@@ -0,0 +1,76 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Copyright (c) Microsoft Corporation. All rights reserved.\n",
"\n",
"Licensed under the MIT License."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Packages"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"!pip install pandas\n",
"!pip install requests"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Widgets\n",
"Install the following widgets to see the status of each run"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"!jupyter nbextension install --py --user azureml.train.widgets\n",
"!jupyter nbextension enable --py --user azureml.train.widgets"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.3"
}
},
"nbformat": 4,
"nbformat_minor": 2
}

36
pipeline/pipeline-batch-scoring.ipynb
View File

@@ -314,6 +314,23 @@
"# Steps to run"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"A subset of the parameters to the python script can be given as input when we re-run a `PublishedPipeline`. In the current example, we define `batch_size` taken by the script as such parameter."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.pipeline.core.graph import PipelineParameter\n",
"batch_size_param = PipelineParameter(name=\"param_batch_size\", default_value=20)"
]
},
{
"cell_type": "code",
"execution_count": null,
@@ -323,7 +340,10 @@
"step = PythonScriptStep(\n",
" name=\"batch ai scoring\",\n",
" script_name=\"batchai_score.py\",\n",
" arguments=[\"--dataset_path\", input_images, \"--model_dir\", model_dir, \"--output_dir\", output_dir, \"--batch_size\", 20],\n",
" arguments=[\"--dataset_path\", input_images, \n",
" \"--model_dir\", model_dir, \n",
" \"--output_dir\", output_dir, \n",
" \"--batch_size\", batch_size_param],\n",
" target=cluster,\n",
" inputs=[input_images, model_dir],\n",
" outputs=[output_dir],\n",
@@ -340,7 +360,7 @@
"outputs": [],
"source": [
"pipeline = Pipeline(workspace=ws, steps=[step])\n",
"pipeline_run = Experiment(ws, 'batch_scoring').submit(pipeline)"
"pipeline_run = Experiment(ws, 'batch_scoring').submit(pipeline, pipeline_params={\"param_batch_size\": 20})"
]
},
{
@@ -475,8 +495,8 @@
"from azureml.pipeline.core import PublishedPipeline\n",
"\n",
"rest_endpoint = PublishedPipeline.get_endpoint(published_id, ws)\n",
"#response = requests.post(rest_endpoint, headers=aad_token, json={})\n",
"#run_id = response.json()[\"Id\"]"
"response = requests.post(rest_endpoint, headers=aad_token, json={\"param_batch_size\": 50})\n",
"run_id = response.json()[\"Id\"]"
]
},
{
@@ -492,10 +512,10 @@
"metadata": {},
"outputs": [],
"source": [
"#from azureml.pipeline.core.run import PipelineRun\n",
"#published_pipeline_run = PipelineRun(ws.experiments()[\"batch_scoring\"], run_id)\n",
"from azureml.pipeline.core.run import PipelineRun\n",
"published_pipeline_run = PipelineRun(ws.experiments()[\"batch_scoring\"], run_id)\n",
"\n",
"#RunDetails(published_pipeline_run).show()"
"RunDetails(published_pipeline_run).show()"
]
},
{
@@ -522,7 +542,7 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.3"
"version": "3.6.6"
}
},
"nbformat": 4,

92
release.json Normal file
View File

@@ -0,0 +1,92 @@
{
"help": {
"notebooks": {
"id": "define every notebook under the ID in a case to be added into a test suite",
"path": "path to the notebooks",
"name": "name of the notebook in the same folder",
"dependencies": "notebook dependencies needed for execution, these will be uploaded into Git repo with the notebook",
"widget": "list widgets to be installed for notebook execution",
"requirements": "additional packages to pip install for the execution"
},
"uploads": "TODO: list files that are not notebooks or notebook dependencies to be updated in the Git repo",
"include": "Nested release.json files id: nested folder. Same logic applied"
},
"include": {
"intro": "01.getting-started",
"automl-intro": "00.AutoML.Getting Started",
"onnx": "onnx",
"pipeline": "pipeline",
"training": "training",
"fr": "functionalreadiness"
},
"channels": {
"master": [
"sample-00",
"tutorial-01",
"tutorial-02",
"tutorial-03"
],
"candidate": [
"sample-00",
"tutorial-01",
"tutorial-02",
"tutorial-03"
],
"preview": [
"sample-00",
"tutorial-01",
"tutorial-02",
"tutorial-03"
],
"automl-test": [
"tutorial-03"
]
},
"notebooks": {
"sample-00": {
"name": "00.configuration.ipynb"
},
"tutorial-01": {
"path": "tutorials",
"name": "01.train-models.ipynb",
"widgets": [ "azureml.train.widgets" ],
"dependencies": [ "utils.py" ],
"conda": [ "matplotlib", "scikit-learn" ],
"requirements": ["azureml-train-widgets"]
},
"tutorial-02": {
"path": "tutorials",
"name": "02.deploy-models.ipynb",
"widgets": [],
"dependencies": [ "utils.py", "sklearn_mnist_model.pkl" ],
"conda": [ "matplotlib", "scikit-learn" ]
},
"tutorial-03": {
"path": "tutorials",
"name": "03.auto-train-models.ipynb",
"widgets": [ "azureml.train.widgets" ],
"dependencies": [ ],
"requirements": [ "matplotlib", "numpy", "scipy", "scikit-learn", "pandas", "azureml-train-automl", "azureml-train-widgets" ]
},
"azurenotebook-01": {
"path": "azurenotebooks",
"name": "01.run-experiment.ipynb",
"widgets": [],
"dependencies": [],
"requirements": []
},
"azurenotebook-02": {
"path": "azurenotebooks",
"name": "02.deploy-web-service.ipynb",
"widgets": [],
"dependencies": [],
"requirements": []
}
},
"uploads": [
"README.md"
]
}

150
training/01.train-tune-deploy-pytorch/01.train-tune-deploy-pytorch.ipynb → training/01.train-hyperparameter-tune-deploy-with-pytorch/01.train-hyperparameter-tune-deploy-with-pytorch.ipynb
View File

@@ -13,7 +13,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"# 01. Train and deploy with PyTorch\n",
"# 01. Train, hyperparameter tune, and deploy with PyTorch\n",
"\n",
"In this tutorial, you will train, hyperparameter tune, and deploy a PyTorch model using the Azure Machine Learning (AML) Python SDK.\n",
"\n",
@@ -119,13 +119,8 @@
"metadata": {},
"source": [
"## Upload training data\n",
"The dataset we will use consists of about 120 training images each for ants and bees, with 75 validation images for each class."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The dataset we will use consists of about 120 training images each for ants and bees, with 75 validation images for each class.\n",
"\n",
"First, download the dataset (located [here](https://download.pytorch.org/tutorial/hymenoptera_data.zip) as a zip file) locally to your current directory and extract the files. This will create a folder called `hymenoptera_data` with two subfolders `train` and `val` that contain the training and validation images, respectively. [Hymenoptera](https://en.wikipedia.org/wiki/Hymenoptera) is the order of insects that includes ants and bees."
]
},
@@ -244,14 +239,28 @@
"### Prepare training script\n",
"Now you will need to create your training script. In this tutorial, the training script is already provided for you at `pytorch_train.py`. In practice, you should be able to take any custom training script as is and run it with AML without having to modify your code.\n",
"\n",
"However, if you would like to use AML's [tracking and metrics](https://docs.microsoft.com/azure/machine-learning/service/concept-azure-machine-learning-architecture#metrics) capabilities, you will have to add a small amount of AML code inside your training script. "
"However, if you would like to use AML's [tracking and metrics](https://docs.microsoft.com/azure/machine-learning/service/concept-azure-machine-learning-architecture#metrics) capabilities, you will have to add a small amount of AML code inside your training script. \n",
"\n",
"In `pytorch_train.py`, we will log some metrics to our AML run. To do so, we will access the AML run object within the script:\n",
"```Python\n",
"from azureml.core.run import Run\n",
"run = Run.get_submitted_run()\n",
"```\n",
"Further within `pytorch_train.py`, we log the learning rate and momentum parameters, and the best validation accuracy the model achieves:\n",
"```Python\n",
"run.log('lr', np.float(learning_rate))\n",
"run.log('momentum', np.float(momentum))\n",
"\n",
"run.log('best_val_acc', np.float(best_acc))\n",
"```\n",
"These run metrics will become particularly important when we begin hyperparameter tuning our model in the \"Tune model hyperparameters\" section."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Copy the training script `pytorch_train.py` into your project directory."
"Once your script is ready, copy the training script `pytorch_train.py` into your project directory."
]
},
{
@@ -364,8 +373,19 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"### Register the trained model\n",
"Finally, register the trained model from your run to your workspace. The `model_path` parameter takes in the relative path on the remote VM to the model in your `outputs` directory. In the next section, we will deploy this registered model as a web service."
"## Tune model hyperparameters\n",
"Now that we've seen how to do a simple PyTorch training run using the SDK, let's see if we can further improve the accuracy of our model. We can optimize our model's hyperparameters using Azure Machine Learning's hyperparameter tuning capabilities."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Start a hyperparameter sweep\n",
"First, we will define the hyperparameter space to sweep over. Since our training script uses a learning rate schedule to decay the learning rate every several epochs, let's tune the initial learning rate and the momentum parameters. In this example we will use random sampling to try different configuration sets of hyperparameters to maximize our primary metric, the best validation accuracy (`best_val_acc`).\n",
"\n",
"Then, we specify the early termination policy to use to early terminate poorly performing runs. Here we use the `BanditPolicy`, which will terminate any run that doesn't fall within the slack factor of our primary evaluation metric. In this tutorial, we will apply this policy every epoch (since we report our `best_val_acc` metric every epoch and `evaluation_interval=1`). Notice we will delay the first policy evaluation until after the first `10` epochs (`delay_evaluation=10`).\n",
"Refer [here](https://docs.microsoft.com/azure/machine-learning/service/how-to-tune-hyperparameters#specify-an-early-termination-policy) for more information on the BanditPolicy and other policies available."
]
},
{
@@ -374,7 +394,107 @@
"metadata": {},
"outputs": [],
"source": [
"model = run.register_model(model_name = 'pytorch-hymenoptera', model_path = 'outputs/model.pt')\n",
"from azureml.train.hyperdrive import *\n",
"\n",
"param_sampling = RandomParameterSampling( {\n",
" 'learning_rate': uniform(0.0005, 0.005),\n",
" 'momentum': uniform(0.9, 0.99)\n",
" }\n",
")\n",
"\n",
"early_termination_policy = BanditPolicy(slack_factor=0.15, evaluation_interval=1, delay_evaluation=10)\n",
"\n",
"hyperdrive_run_config = HyperDriveRunConfig(estimator=estimator,\n",
" hyperparameter_sampling=param_sampling, \n",
" policy=early_termination_policy,\n",
" primary_metric_name='best_val_acc',\n",
" primary_metric_goal=PrimaryMetricGoal.MAXIMIZE,\n",
" max_total_runs=20,\n",
" max_concurrent_runs=4)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Finally, lauch the hyperparameter tuning job."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# start the HyperDrive run\n",
"hyperdrive_run = experiment.submit(hyperdrive_run_config)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Monitor HyperDrive runs\n",
"You can monitor the progress of the runs with the following Jupyter widget. "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.train.widgets import RunDetails\n",
"\n",
"RunDetails(hyperdrive_run).show()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Find and register the best model\n",
"Once all the runs complete, we can find the run that produced the model with the highest accuracy."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"best_run = hyperdrive_run.get_best_run_by_primary_metric()\n",
"best_run_metrics = best_run.get_metrics()\n",
"print(best_run)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"print('Best Run is:\\n Validation accuracy: {0:.5f} \\n Learning rate: {1:.5f} \\n Momentum: {2:.5f}'.format(\n",
" best_run_metrics['best_val_acc'][-1],\n",
" best_run_metrics['lr'],\n",
" best_run_metrics['momentum'])\n",
" )"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Finally, register the model from your best-performing run to your workspace. The `model_path` parameter takes in the relative path on the remote VM to the model file in your `outputs` directory. In the next section, we will deploy this registered model as a web service."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"model = best_run.register_model(model_name = 'pytorch-hymenoptera', model_path = 'outputs/model.pt')\n",
"print(model.name, model.id, model.version, sep = '\\t')"
]
},
@@ -422,8 +542,6 @@
" - torch\n",
" - torchvision\n",
" - pillow\n",
" # Required packages for AzureML execution, history, and data preparation.\n",
" - --extra-index-url https://azuremlsdktestpypi.azureedge.net/sdk-release/Preview/E7501C02541B433786111FE8E140CAA1\n",
" - azureml-core"
]
},
@@ -528,7 +646,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"To get the logs from the deployment process, run the following command:"
"**Tip: If something goes wrong with the deployment, the first thing to look at is the logs from the service by running the following command:**"
]
},
{

2
training/01.train-tune-deploy-pytorch/pytorch_score.py → training/01.train-hyperparameter-tune-deploy-with-pytorch/pytorch_score.py
View File

@@ -3,9 +3,7 @@
import torch
import torch.nn as nn
import torchvision
from torchvision import transforms
import os
import json
import base64
from io import BytesIO

27
training/01.train-tune-deploy-pytorch/pytorch_train.py → training/01.train-hyperparameter-tune-deploy-with-pytorch/pytorch_train.py
View File

@@ -1,15 +1,13 @@
# Copyright (c) 2017, PyTorch contributors
# Modifications copyright (C) Microsoft Corporation
# Licensed under the BSD license
# Adapted from https://pytorch.org/tutorials/beginner/transfer_learning_tutorial.html
from __future__ import print_function, division
import torch
import torch.nn as nn
import torch.optim as optim
from torch.optim import lr_scheduler
import torchvision
from torchvision import datasets, models, transforms
import numpy as np
import time
@@ -17,6 +15,10 @@ import os
import copy
import argparse
from azureml.core.run import Run
# get the Azure ML run object
run = Run.get_submitted_run()
def load_data(data_dir):
"""Load the train/val data."""
@@ -112,6 +114,9 @@ def train_model(model, criterion, optimizer, scheduler, num_epochs, data_dir):
best_acc = epoch_acc
best_model_wts = copy.deepcopy(model.state_dict())
# log the best val accuracy to AML run
run.log('best_val_acc', np.float(best_acc))
print()
time_elapsed = time.time() - since
@@ -124,9 +129,13 @@ def train_model(model, criterion, optimizer, scheduler, num_epochs, data_dir):
return model
def fine_tune_model(num_epochs, data_dir):
def fine_tune_model(num_epochs, data_dir, learning_rate, momentum):
"""Load a pretrained model and reset the final fully connected layer."""
# log the hyperparameter metrics to the AML run
run.log('lr', np.float(learning_rate))
run.log('momentum', np.float(momentum))
model_ft = models.resnet18(pretrained=True)
num_ftrs = model_ft.fc.in_features
model_ft.fc = nn.Linear(num_ftrs, 2) # only 2 classes to predict
@@ -137,7 +146,7 @@ def fine_tune_model(num_epochs, data_dir):
criterion = nn.CrossEntropyLoss()
# Observe that all parameters are being optimized
optimizer_ft = optim.SGD(model_ft.parameters(), lr=0.001, momentum=0.9)
optimizer_ft = optim.SGD(model_ft.parameters(), lr=learning_rate, momentum=momentum)
# Decay LR by a factor of 0.1 every 7 epochs
exp_lr_scheduler = lr_scheduler.StepLR(optimizer_ft, step_size=7, gamma=0.1)
@@ -148,19 +157,17 @@ def fine_tune_model(num_epochs, data_dir):
def main():
for root, dirs, files in os.walk("."):
print(root)
print(dirs)
# get command-line arguments
parser = argparse.ArgumentParser()
parser.add_argument('--data_dir', type=str, help='directory of training data')
parser.add_argument('--num_epochs', type=int, default=25, help='number of epochs to train')
parser.add_argument('--output_dir', type=str, help='output directory')
parser.add_argument('--learning_rate', type=float, help='learning rate')
parser.add_argument('--momentum', type=float, help='momentum')
args = parser.parse_args()
print("data directory is: " + args.data_dir)
model = fine_tune_model(args.num_epochs, args.data_dir)
model = fine_tune_model(args.num_epochs, args.data_dir, args.learning_rate, args.momentum)
os.makedirs(args.output_dir, exist_ok=True)
torch.save(model, os.path.join(args.output_dir, 'model.pt'))

1624
training/03.train-hyperparameter-tune-deploy-with-tensorflow/03.train-hyperparameter-tune-deploy-with-tensorflow.ipynb Normal file
View File

File diff suppressed because it is too large Load Diff

BIN
training/03.train-hyperparameter-tune-deploy-with-tensorflow/nn.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 119 KiB

106
training/03.train-hyperparameter-tune-deploy-with-tensorflow/tf_mnist.py Normal file
View File

@@ -0,0 +1,106 @@
# Copyright (c) Microsoft Corporation. All rights reserved.
# Licensed under the MIT License.
import numpy as np
import argparse
import os
import tensorflow as tf
from azureml.core import Run
from utils import load_data
print("TensorFlow version:", tf.VERSION)
parser = argparse.ArgumentParser()
parser.add_argument('--data-folder', type=str, dest='data_folder', help='data folder mounting point')
parser.add_argument('--batch-size', type=int, dest='batch_size', default=50, help='mini batch size for training')
parser.add_argument('--first-layer-neurons', type=int, dest='n_hidden_1', default=100,
help='# of neurons in the first layer')
parser.add_argument('--second-layer-neurons', type=int, dest='n_hidden_2', default=100,
help='# of neurons in the second layer')
parser.add_argument('--learning-rate', type=float, dest='learning_rate', default=0.01, help='learning rate')
args = parser.parse_args()
data_folder = os.path.join(args.data_folder, 'mnist')
print('training dataset is stored here:', data_folder)
X_train = load_data(os.path.join(data_folder, 'train-images.gz'), False) / 255.0
X_test = load_data(os.path.join(data_folder, 'test-images.gz'), False) / 255.0
y_train = load_data(os.path.join(data_folder, 'train-labels.gz'), True).reshape(-1)
y_test = load_data(os.path.join(data_folder, 'test-labels.gz'), True).reshape(-1)
print(X_train.shape, y_train.shape, X_test.shape, y_test.shape, sep='\n')
training_set_size = X_train.shape[0]
n_inputs = 28 * 28
n_h1 = args.n_hidden_1
n_h2 = args.n_hidden_2
n_outputs = 10
learning_rate = args.learning_rate
n_epochs = 50
batch_size = args.batch_size
with tf.name_scope('network'):
# construct the DNN
X = tf.placeholder(tf.float32, shape=(None, n_inputs), name='X')
y = tf.placeholder(tf.int64, shape=(None), name='y')
h1 = tf.layers.dense(X, n_h1, activation=tf.nn.relu, name='h1')
h2 = tf.layers.dense(h1, n_h2, activation=tf.nn.relu, name='h2')
output = tf.layers.dense(h2, n_outputs, name='output')
with tf.name_scope('train'):
cross_entropy = tf.nn.sparse_softmax_cross_entropy_with_logits(labels=y, logits=output)
loss = tf.reduce_mean(cross_entropy, name='loss')
optimizer = tf.train.GradientDescentOptimizer(learning_rate)
train_op = optimizer.minimize(loss)
with tf.name_scope('eval'):
correct = tf.nn.in_top_k(output, y, 1)
acc_op = tf.reduce_mean(tf.cast(correct, tf.float32))
init = tf.global_variables_initializer()
saver = tf.train.Saver()
# start an Azure ML run
run = Run.get_submitted_run()
with tf.Session() as sess:
init.run()
for epoch in range(n_epochs):
# randomly shuffle training set
indices = np.random.permutation(training_set_size)
X_train = X_train[indices]
y_train = y_train[indices]
# batch index
b_start = 0
b_end = b_start + batch_size
for _ in range(training_set_size // batch_size):
# get a batch
X_batch, y_batch = X_train[b_start: b_end], y_train[b_start: b_end]
# update batch index for the next batch
b_start = b_start + batch_size
b_end = min(b_start + batch_size, training_set_size)
# 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
run.log('training_acc', np.float(acc_train))
run.log('validation_acc', np.float(acc_val))
print(epoch, '-- Training accuracy:', acc_train, '\b Validation accuracy:', acc_val)
y_hat = np.argmax(output.eval(feed_dict={X: X_test}), axis=1)
run.log('final_acc', np.float(acc_val))
os.makedirs('./outputs/model', exist_ok=True)
# files saved in the "./outputs" folder are automatically uploaded into run history
saver.save(sess, './outputs/model/mnist-tf.model')

27
training/03.train-hyperparameter-tune-deploy-with-tensorflow/utils.py Normal file
View File

@@ -0,0 +1,27 @@
# Copyright (c) Microsoft Corporation. All rights reserved.
# Licensed under the MIT License.
import gzip
import numpy as np
import struct
# load compressed MNIST gz files and return numpy arrays
def load_data(filename, label=False):
with gzip.open(filename) as gz:
struct.unpack('I', gz.read(4))
n_items = struct.unpack('>I', gz.read(4))
if not label:
n_rows = struct.unpack('>I', gz.read(4))[0]
n_cols = struct.unpack('>I', gz.read(4))[0]
res = np.frombuffer(gz.read(n_items[0] * n_rows * n_cols), dtype=np.uint8)
res = res.reshape(n_items[0], n_rows * n_cols)
else:
res = np.frombuffer(gz.read(n_items[0]), dtype=np.uint8)
res = res.reshape(n_items[0], 1)
return res
# one-hot encode a 1-D array
def one_hot_encode(array, num_of_classes):
return np.eye(num_of_classes)[array.reshape(-1)]

99
training/06.distributed-cntk-with-custom-docker/06.distributed-cntk-with-custom-docker.ipynb
View File

@@ -22,11 +22,10 @@
"metadata": {},
"source": [
"## Prerequisites\n",
"* Understand the [architecture and terms](https://docs.microsoft.com/azure/machine-learning/service/concept-azure-machine-learning-architecture) introduced by Azure Machine Learning services\n",
"* Understand the [architecture and terms](https://docs.microsoft.com/azure/machine-learning/service/concept-azure-machine-learning-architecture) introduced by Azure Machine Learning\n",
"* Go through the [00.configuration.ipynb]() notebook to:\n",
" * install the AML SDK\n",
" * create a workspace and its configuration file (`config.json`)\n",
"* Review the [tutorial]() on single-node PyTorch training using the SDK"
" * create a workspace and its configuration file (`config.json`)"
]
},
{
@@ -106,6 +105,80 @@
" print(compute_target.status.serialize())"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Upload training data\n",
"For this tutorial, we will be using the MNIST dataset.\n",
"\n",
"First, let's download the dataset. We've included the `install_mnist.py` script to download the data and convert it to a CNTK-supported format. Our data files will get written to a directory named `'mnist'`."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import install_mnist\n",
"\n",
"install_mnist.main('mnist')"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"To make the data accessible for remote training, you will need to upload the data from your local machine to the cloud. AML provides a convenient way to do so via a [Datastore](https://docs.microsoft.com/azure/machine-learning/service/how-to-access-data). The datastore provides a mechanism for you to upload/download data, and interact with it from your remote compute targets. \n",
"\n",
"Each workspace is associated with a default datastore. In this tutorial, we will upload the training data to this default datastore, which we will then mount on the remote compute for training in the next section."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"ds = ws.get_default_datastore()\n",
"print(ds.datastore_type, ds.account_name, ds.container_name)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The following code will upload the training data to the path `./mnist` on the default datastore."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"ds.upload(src_dir='./mnist', target_path='./mnist')"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now let's get a reference to the path on the datastore with the training data. We can do so using the `path` method. In the next section, we can then pass this reference to our training script's `--data_dir` argument. "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"path_on_datastore = 'mnist'\n",
"ds_data = ds.path(path_on_datastore)\n",
"print(ds_data)"
]
},
{
"cell_type": "markdown",
"metadata": {},
@@ -138,7 +211,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"Copy the training script `tf_mnist_replica.py` into this project directory."
"Copy the training script `cntk_distr_mnist.py` into this project directory."
]
},
{
@@ -148,7 +221,7 @@
"outputs": [],
"source": [
"import shutil\n",
"shutil.copy('cntk_mnist.py', project_folder)"
"shutil.copy('cntk_distr_mnist.py', project_folder)"
]
},
{
@@ -187,21 +260,29 @@
"source": [
"from azureml.train.estimator import *\n",
"\n",
"script_params = {\n",
" '--num_epochs': 50,\n",
" '--data_dir': ds_data.as_mount(),\n",
" '--output_dir': './outputs'\n",
"}\n",
"\n",
"estimator = Estimator(source_directory=project_folder,\n",
" compute_target=compute_target,\n",
" entry_script='cntk_mnist.py',\n",
" entry_script='cntk_distr_mnist.py',\n",
" script_params=script_params,\n",
" node_count=2,\n",
" process_count_per_node=1,\n",
" distributed_backend='mpi', \n",
" pip_packages=['cntk==2.5.1'],\n",
" custom_docker_base_image='microsoft/mmlspark:0.12')"
" pip_packages=['cntk-gpu==2.6'],\n",
" custom_docker_base_image='microsoft/mmlspark:gpu-0.12',\n",
" use_gpu=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We would like to train our model using a [pre-built Docker container](https://hub.docker.com/r/microsoft/mmlspark/). To do so, we specify the name of the docker image to the argument `custom_docker_base_image`. You can only provide images available in public docker repositories such as Docker Hub using this argument. To use an image from a private docker repository, use the constructor's `environment_definition` parameter instead. Finally, we provide the `cntk` package to `pip_packages` to install CNTK 2.5.1 on our custom image.\n",
"We would like to train our model using a [pre-built Docker container](https://hub.docker.com/r/microsoft/mmlspark/). To do so, specify the name of the docker image to the argument `custom_docker_base_image`. You can only provide images available in public docker repositories such as Docker Hub using this argument. To use an image from a private docker repository, use the constructor's `environment_definition` parameter instead. Finally, we provide the `cntk` package to `pip_packages` to install CNTK 2.6 on our custom image.\n",
"\n",
"The above code specifies that we will run our training script on `2` nodes, with one worker per node. In order to run distributed CNTK, which uses MPI, you must provide the argument `distributed_backend='mpi'`."
]

117
training/06.distributed-cntk-with-custom-docker/cntk_distr_mnist.py Normal file
View File

@@ -0,0 +1,117 @@
# Copyright (c) Microsoft. All rights reserved.
# Licensed under the MIT license.
# Adapted from:
# https://github.com/Microsoft/CNTK/blob/master/Examples/Image/Classification/ConvNet/Python/ConvNet_MNIST.py
# ====================================================================
"""Train a CNN model on the MNIST dataset via distributed training."""
from __future__ import print_function
import numpy as np
import os
import cntk as C
import argparse
from cntk.train.training_session import CheckpointConfig, TestConfig
def create_reader(path, is_training, input_dim, label_dim, total_number_of_samples):
"""Define the reader for both training and evaluation action."""
return C.io.MinibatchSource(C.io.CTFDeserializer(path, C.io.StreamDefs(
features=C.io.StreamDef(field='features', shape=input_dim),
labels=C.io.StreamDef(field='labels', shape=label_dim)
)), randomize=is_training, max_samples=total_number_of_samples)
def convnet_mnist(max_epochs, output_dir, data_dir, debug_output=False, epoch_size=60000, minibatch_size=64):
"""Creates and trains a feedforward classification model for MNIST images."""
image_height = 28
image_width = 28
num_channels = 1
input_dim = image_height * image_width * num_channels
num_output_classes = 10
# Input variables denoting the features and label data
input_var = C.ops.input_variable((num_channels, image_height, image_width), np.float32)
label_var = C.ops.input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = C.ops.element_times(C.ops.constant(0.00390625), input_var)
with C.layers.default_options(activation=C.ops.relu, pad=False):
conv1 = C.layers.Convolution2D((5, 5), 32, pad=True)(scaled_input)
pool1 = C.layers.MaxPooling((3, 3), (2, 2))(conv1)
conv2 = C.layers.Convolution2D((3, 3), 48)(pool1)
pool2 = C.layers.MaxPooling((3, 3), (2, 2))(conv2)
conv3 = C.layers.Convolution2D((3, 3), 64)(pool2)
f4 = C.layers.Dense(96)(conv3)
drop4 = C.layers.Dropout(0.5)(f4)
z = C.layers.Dense(num_output_classes, activation=None)(drop4)
ce = C.losses.cross_entropy_with_softmax(z, label_var)
pe = C.metrics.classification_error(z, label_var)
# Load train data
reader_train = create_reader(os.path.join(data_dir, 'Train-28x28_cntk_text.txt'), True,
input_dim, num_output_classes, max_epochs * epoch_size)
# Load test data
reader_test = create_reader(os.path.join(data_dir, 'Test-28x28_cntk_text.txt'), False,
input_dim, num_output_classes, C.io.FULL_DATA_SWEEP)
# Set learning parameters
lr_per_sample = [0.001] * 10 + [0.0005] * 10 + [0.0001]
lr_schedule = C.learning_parameter_schedule_per_sample(lr_per_sample, epoch_size=epoch_size)
mms = [0] * 5 + [0.9990239141819757]
mm_schedule = C.learners.momentum_schedule_per_sample(mms, epoch_size=epoch_size)
# Instantiate the trainer object to drive the model training
local_learner = C.learners.momentum_sgd(z.parameters, lr_schedule, mm_schedule)
progress_printer = C.logging.ProgressPrinter(
tag='Training',
rank=C.train.distributed.Communicator.rank(),
num_epochs=max_epochs,
)
learner = C.train.distributed.data_parallel_distributed_learner(local_learner)
trainer = C.Trainer(z, (ce, pe), learner, progress_printer)
# define mapping from reader streams to network inputs
input_map_train = {
input_var: reader_train.streams.features,
label_var: reader_train.streams.labels
}
input_map_test = {
input_var: reader_test.streams.features,
label_var: reader_test.streams.labels
}
C.logging.log_number_of_parameters(z)
print()
C.train.training_session(
trainer=trainer,
mb_source=reader_train,
model_inputs_to_streams=input_map_train,
mb_size=minibatch_size,
progress_frequency=epoch_size,
checkpoint_config=CheckpointConfig(frequency=epoch_size,
filename=os.path.join(output_dir, "ConvNet_MNIST")),
test_config=TestConfig(reader_test, minibatch_size=minibatch_size,
model_inputs_to_streams=input_map_test)
).train()
return
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--num_epochs', help='Total number of epochs to train', type=int, default='40')
parser.add_argument('--output_dir', help='Output directory', required=False, default='outputs')
parser.add_argument('--data_dir', help='Directory with training data')
args = parser.parse_args()
os.makedirs(args.output_dir, exist_ok=True)
convnet_mnist(args.num_epochs, args.output_dir, args.data_dir)
# Must call MPI finalize when process exit without exceptions
C.train.distributed.Communicator.finalize()

96
training/06.distributed-cntk-with-custom-docker/install_mnist.py Normal file
View File

@@ -0,0 +1,96 @@
# Copyright (c) Microsoft. All rights reserved.
# Licensed under the MIT license.
# Script from:
# https://github.com/Microsoft/CNTK/blob/master/Examples/Image/DataSets/MNIST/install_mnist.py
from __future__ import print_function
try:
from urllib.request import urlretrieve
except ImportError:
from urllib import urlretrieve
import gzip
import os
import struct
import numpy as np
def loadData(src, cimg):
print('Downloading ' + src)
gzfname, h = urlretrieve(src, './delete.me')
print('Done.')
try:
with gzip.open(gzfname) as gz:
n = struct.unpack('I', gz.read(4))
# Read magic number.
if n[0] != 0x3080000:
raise Exception('Invalid file: unexpected magic number.')
# Read number of entries.
n = struct.unpack('>I', gz.read(4))[0]
if n != cimg:
raise Exception('Invalid file: expected {0} entries.'.format(cimg))
crow = struct.unpack('>I', gz.read(4))[0]
ccol = struct.unpack('>I', gz.read(4))[0]
if crow != 28 or ccol != 28:
raise Exception('Invalid file: expected 28 rows/cols per image.')
# Read data.
res = np.fromstring(gz.read(cimg * crow * ccol), dtype=np.uint8)
finally:
os.remove(gzfname)
return res.reshape((cimg, crow * ccol))
def loadLabels(src, cimg):
print('Downloading ' + src)
gzfname, h = urlretrieve(src, './delete.me')
print('Done.')
try:
with gzip.open(gzfname) as gz:
n = struct.unpack('I', gz.read(4))
# Read magic number.
if n[0] != 0x1080000:
raise Exception('Invalid file: unexpected magic number.')
# Read number of entries.
n = struct.unpack('>I', gz.read(4))
if n[0] != cimg:
raise Exception('Invalid file: expected {0} rows.'.format(cimg))
# Read labels.
res = np.fromstring(gz.read(cimg), dtype=np.uint8)
finally:
os.remove(gzfname)
return res.reshape((cimg, 1))
def load(dataSrc, labelsSrc, cimg):
data = loadData(dataSrc, cimg)
labels = loadLabels(labelsSrc, cimg)
return np.hstack((data, labels))
def savetxt(filename, ndarray):
with open(filename, 'w') as f:
labels = list(map(' '.join, np.eye(10, dtype=np.uint).astype(str)))
for row in ndarray:
row_str = row.astype(str)
label_str = labels[row[-1]]
feature_str = ' '.join(row_str[:-1])
f.write('|labels {} |features {}\n'.format(label_str, feature_str))
def main(data_dir):
os.makedirs(data_dir, exist_ok=True)
train = load('http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz',
'http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz', 60000)
print('Writing train text file...')
train_txt = os.path.join(data_dir, 'Train-28x28_cntk_text.txt')
savetxt(train_txt, train)
print('Done.')
test = load('http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz',
'http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz', 10000)
print('Writing test text file...')
test_txt = os.path.join(data_dir, 'Test-28x28_cntk_text.txt')
savetxt(test_txt, test)
print('Done.')
if __name__ == "__main__":
main('mnist')

4
training/07.tensorboard/07.tensorboard.ipynb
View File

@@ -351,7 +351,9 @@
"\n",
"try:\n",
" # If you already have a cluster named this, we don't need to make a new one.\n",
" compute_target = [ct for ct in ws.compute_targets() if ct.name == clust_name and ct.type == 'BatchAI'][0]\n",
" cts = ws.compute_targets() \n",
" compute_target = cts[clust_name]\n",
" assert compute_target.type == 'BatchAI'\n",
"except:\n",
" # Let's make a new one here.\n",
" provisioning_config = BatchAiCompute.provisioning_configuration(cluster_max_nodes=2, \n",

2
tutorials/01.train-models.ipynb
View File

@@ -664,7 +664,7 @@
"\n",
"You are ready to deploy this registered model using the instructions in the next part of the tutorial series:\n",
"\n",
"> [Tutorial 2 - Deploy models](deploy-models.ipynb)"
"> [Tutorial 2 - Deploy models](02.deploy-models.ipynb)"
]
}
],

6
tutorials/03.auto-train-models.ipynb
View File

@@ -15,11 +15,11 @@
"source": [
"# Tutorial: Automatically train a classification model with Azure Automated Machine Learning\n",
"\n",
"In this tutorial, you'll learn how to automatically generate a machine learning model. This model can then be deployed following the workflow in the [Deploy a model](tutorial-deploy-models-with-aml.md) tutorial.\n",
"In this tutorial, you'll learn how to automatically generate a machine learning model. This model can then be deployed following the workflow in the [Deploy a model](02.deploy-models.ipynb) tutorial.\n",
"\n",
"[flow diagram](./imgs/nn.png)\n",
"[flow diagram](./imgs/flow2.png)\n",
"\n",
"Similar to the [train models tutorial](tutorial-train-models-with-aml.md), this tutorial classifies handwritten images of digits (0-9) from the [MNIST](http://yann.lecun.com/exdb/mnist/) dataset.\n",
"Similar to the [train models tutorial](01.train-models.ipynb), this tutorial classifies handwritten images of digits (0-9) from the [MNIST](http://yann.lecun.com/exdb/mnist/) dataset.\n",
"\n",
"You'll learn how to:\n",
"\n",