TensorFlow 2.0 MNIST example, without IT (#1790)

README.md

@@ -71,7 +71,7 @@ The tool dispatches and runs trial jobs generated by tuning algorithms to search
         <li><b>Examples</b></li>
          <ul>
            <li><a href="examples/trials/mnist-pytorch">MNIST-pytorch</li></a>
-           <li><a href="examples/trials/mnist">MNIST-tensorflow</li></a>
+           <li><a href="examples/trials/mnist-tfv1">MNIST-tensorflow</li></a>
            <li><a href="examples/trials/mnist-keras">MNIST-keras</li></a>
            <li><a href="docs/en_US/TrialExample/GbdtExample.md">Auto-gbdt</a></li>
            <li><a href="docs/en_US/TrialExample/Cifar10Examples.md">Cifar10-pytorch</li></a>
@@ -245,7 +245,7 @@ Linux and MacOS
 * Run the MNIST example.
 
 ```bash
-    nnictl create --config nni/examples/trials/mnist/config.yml
+    nnictl create --config nni/examples/trials/mnist-tfv1/config.yml
 ```
 
 Windows
@@ -253,7 +253,7 @@ Windows
 * Run the MNIST example.
 
 ```bash
-    nnictl create --config nni\examples\trials\mnist\config_windows.yml
+    nnictl create --config nni\examples\trials\mnist-tfv1\config_windows.yml
 ```
 
 * Wait for the message `INFO: Successfully started experiment!` in the command line. This message indicates that your experiment has been successfully started. You can explore the experiment using the `Web UI url`.

docs/en_US/TrialExample/MnistExamples.md

@@ -2,7 +2,8 @@
 
 CNN MNIST classifier for deep learning is similar to `hello world` for programming languages. Thus, we use MNIST as example to introduce different features of NNI. The examples are listed below:
 
- - [MNIST with NNI API](#mnist)
+ - [MNIST with NNI API (TensorFlow v1.x)](#mnist-tfv1)
+ - [MNIST with NNI API (TensorFlow v2.x)](#mnist-tfv2)
  - [MNIST with NNI annotation](#mnist-annotation)
  - [MNIST in keras](#mnist-keras)
  - [MNIST -- tuning with batch tuner](#mnist-batch)
@@ -11,12 +12,19 @@ CNN MNIST classifier for deep learning is similar to `hello world` for programmi
  - [distributed MNIST (tensorflow) using kubeflow](#mnist-kubeflow-tf)
  - [distributed MNIST (pytorch) using kubeflow](#mnist-kubeflow-pytorch)
 
-<a name="mnist"></a>
-**MNIST with NNI API**
+<a name="mnist-tfv1"></a>
+**MNIST with NNI API (TensorFlow v1.x)**
 
 This is a simple network which has two convolutional layers, two pooling layers and a fully connected layer. We tune hyperparameters, such as dropout rate, convolution size, hidden size, etc. It can be tuned with most NNI built-in tuners, such as TPE, SMAC, Random. We also provide an exmaple YAML file which enables assessor.
 
-`code directory: examples/trials/mnist/`
+`code directory: examples/trials/mnist-tfv1/`
+
+<a name="mnist-tfv2"></a>
+**MNIST with NNI API (TensorFlow v2.x)**
+
+Same network to the example above, but written in TensorFlow v2.x Keras API.
+
+`code directory: examples/trials/mnist-tfv2/`
 
 <a name="mnist-annotation"></a>
 **MNIST with NNI annotation**
@@ -65,4 +73,4 @@ This example is to show how to run distributed training on kubeflow through NNI.
 
 Similar to the previous example, the difference is that this example is implemented in pytorch, thus, it uses kubeflow pytorch operator.
 
-`code directory: examples/trials/mnist-distributed-pytorch/`
+`code directory: examples/trials/mnist-distributed-pytorch/`
\ No newline at end of file

examples/trials/mnist-tfv2/config.yml

+authorName: NNI Example
+experimentName: MNIST TF v2.x
+trialConcurrency: 1
+maxExecDuration: 1h
+maxTrialNum: 10
+trainingServicePlatform: local  # choices: local, remote, pai
+searchSpacePath: search_space.json
+useAnnotation: false
+tuner:
+    builtinTunerName: TPE   # choices: TPE, Random, Anneal, Evolution, BatchTuner, MetisTuner,
+                            #          GPTuner, SMAC (SMAC should be installed through nnictl)
+    classArgs:
+        optimize_mode: maximize  # choices: maximize, minimize
+trial:
+  command: python3 mnist.py
+  codeDir: .
+  gpuNum: 0

examples/trials/mnist-tfv2/config_assessor.yml

+authorName: NNI Example
+experimentName: MNIST TF v2.x with assessor
+trialConcurrency: 1
+maxExecDuration: 1h
+maxTrialNum: 50
+#choice: local, remote
+trainingServicePlatform: local
+searchSpacePath: search_space.json
+#choice: true, false
+useAnnotation: false
+tuner:
+  #choice: TPE, Random, Anneal, Evolution, BatchTuner, MetisTuner, GPTuner
+  #SMAC (SMAC should be installed through nnictl)
+  builtinTunerName: TPE
+  classArgs:
+    #choice: maximize, minimize
+    optimize_mode: maximize
+assessor:
+  #choice: Medianstop, Curvefitting
+  builtinAssessorName: Curvefitting
+  classArgs:
+    #choice: maximize, minimize
+    optimize_mode: maximize
+    epoch_num: 20
+    threshold: 0.9
+trial:
+  command: python3 mnist.py
+  codeDir: .
+  gpuNum: 0

examples/trials/mnist-tfv2/config_windows.yml

+authorName: NNI Example
+experimentName: MNIST TF v2.x
+trialConcurrency: 1
+maxExecDuration: 1h
+maxTrialNum: 10
+#choice: local, remote, pai
+trainingServicePlatform: local
+searchSpacePath: search_space.json
+#choice: true, false
+useAnnotation: false
+tuner:
+  #choice: TPE, Random, Anneal, Evolution, BatchTuner, MetisTuner, GPTuner
+  #SMAC (SMAC should be installed through nnictl)
+  builtinTunerName: TPE
+  classArgs:
+    #choice: maximize, minimize
+    optimize_mode: maximize
+trial:
+  command: python mnist.py
+  codeDir: .
+  gpuNum: 0

examples/trials/mnist-tfv2/mnist.py

+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT license.
+
+"""
+NNI example trial code.
+
+- Experiment type: Hyper-parameter Optimization
+- Trial framework: Tensorflow v2.x (Keras API)
+- Model: LeNet-5
+- Dataset: MNIST
+"""
+
+import logging
+
+import tensorflow as tf
+from tensorflow.keras import Model
+from tensorflow.keras.callbacks import Callback
+from tensorflow.keras.layers import (Conv2D, Dense, Dropout, Flatten, MaxPool2D)
+from tensorflow.keras.optimizers import Adam
+
+import nni
+
+_logger = logging.getLogger('mnist_example')
+_logger.setLevel(logging.INFO)
+
+
+class MnistModel(Model):
+    """
+    LeNet-5 Model with customizable hyper-parameters
+    """
+    def __init__(self, conv_size, hidden_size, dropout_rate):
+        """
+        Initialize hyper-parameters.
+
+        Parameters
+        ----------
+        conv_size : int
+            Kernel size of convolutional layers.
+        hidden_size : int
+            Dimensionality of last hidden layer.
+        dropout_rate : float
+            Dropout rate between two fully connected (dense) layers, to prevent co-adaptation.
+        """
+        super().__init__()
+        self.conv1 = Conv2D(filters=32, kernel_size=conv_size, activation='relu')
+        self.pool1 = MaxPool2D(pool_size=2)
+        self.conv2 = Conv2D(filters=64, kernel_size=conv_size, activation='relu')
+        self.pool2 = MaxPool2D(pool_size=2)
+        self.flatten = Flatten()
+        self.fc1 = Dense(units=hidden_size, activation='relu')
+        self.dropout = Dropout(rate=dropout_rate)
+        self.fc2 = Dense(units=10, activation='softmax')
+
+    def call(self, x):
+        """Override ``Model.call`` to build LeNet-5 model."""
+        x = self.conv1(x)
+        x = self.pool1(x)
+        x = self.conv2(x)
+        x = self.pool2(x)
+        x = self.flatten(x)
+        x = self.fc1(x)
+        x = self.dropout(x)
+        return self.fc2(x)
+
+
+class ReportIntermediates(Callback):
+    """
+    Callback class for reporting intermediate accuracy metrics.
+
+    This callback sends accuracy to NNI framework every 100 steps,
+    so you can view the learning curve on web UI.
+
+    If an assessor is configured in experiment's YAML file,
+    it will use these metrics for early stopping.
+    """
+    def on_epoch_end(self, epoch, logs=None):
+        """Reports intermediate accuracy to NNI framework"""
+        # TensorFlow 2.0 API reference claims the key is `val_acc`, but in fact it's `val_accuracy`
+        if 'val_acc' in logs:
+            nni.report_intermediate_result(logs['val_acc'])
+        else:
+            nni.report_intermediate_result(logs['val_accuracy'])
+
+
+def load_dataset():
+    """Download and reformat MNIST dataset"""
+    mnist = tf.keras.datasets.mnist
+    (x_train, y_train), (x_test, y_test) = mnist.load_data()
+    x_train, x_test = x_train / 255.0, x_test / 255.0
+    x_train = x_train[..., tf.newaxis]
+    x_test = x_test[..., tf.newaxis]
+    return (x_train, y_train), (x_test, y_test)
+
+
+def main(params):
+    """
+    Main program:
+      - Build network
+      - Prepare dataset
+      - Train the model
+      - Report accuracy to tuner
+    """
+    model = MnistModel(
+        conv_size=params['conv_size'],
+        hidden_size=params['hidden_size'],
+        dropout_rate=params['dropout_rate']
+    )
+    optimizer = Adam(learning_rate=params['learning_rate'])
+    model.compile(optimizer=optimizer, loss='sparse_categorical_crossentropy', metrics=['accuracy'])
+    _logger.info('Model built')
+
+    (x_train, y_train), (x_test, y_test) = load_dataset()
+    _logger.info('Dataset loaded')
+
+    model.fit(
+        x_train,
+        y_train,
+        batch_size=params['batch_size'],
+        epochs=10,
+        verbose=0,
+        callbacks=[ReportIntermediates()],
+        validation_data=(x_test, y_test)
+    )
+    _logger.info('Training completed')
+
+    loss, accuracy = model.evaluate(x_test, y_test, verbose=0)
+    nni.report_final_result(accuracy)  # send final accuracy to NNI tuner and web UI
+    _logger.info('Final accuracy reported: %s', accuracy)
+
+
+if __name__ == '__main__':
+    params = {
+        'dropout_rate': 0.5,
+        'conv_size': 5,
+        'hidden_size': 1024,
+        'batch_size': 32,
+        'learning_rate': 1e-4,
+    }
+
+    # fetch hyper-parameters from HPO tuner
+    # comment out following two lines to run the code without NNI framework
+    tuned_params = nni.get_next_parameter()
+    params.update(tuned_params)
+
+    _logger.info('Hyper-parameters: %s', params)
+    main(params)

examples/trials/mnist-tfv2/search_space.json

+{
+    "dropout_rate": { "_type": "uniform", "_value": [0.5, 0.9] },
+    "conv_size": { "_type": "choice", "_value": [2, 3, 5, 7] },
+    "hidden_size": { "_type": "choice", "_value": [124, 512, 1024] },
+    "batch_size": { "_type": "choice", "_value": [16, 32] },
+    "learning_rate": { "_type": "choice", "_value": [0.0001, 0.001, 0.01, 0.1] }
+}

test/cli_test.py

@@ -9,7 +9,7 @@ from utils import GREEN, RED, CLEAR, setup_experiment
 def test_nni_cli():
     import nnicli as nc
 
-    config_file = 'config_test/examples/mnist.test.yml'
+    config_file = 'config_test/examples/mnist-tfv1.test.yml'
 
     try:
         # Sleep here to make sure previous stopped exp has enough time to exit to avoid port conflict

test/config_test/examples/mnist.test.yml → test/config_test/examples/mnist-tfv1.test.yml

@@ -12,7 +12,7 @@ assessor:
   classArgs:
     optimize_mode: maximize
 trial:
-  codeDir: ../../../examples/trials/mnist
+  codeDir: ../../../examples/trials/mnist-tfv1
   command: python3 mnist.py --batch_num 100
   gpuNum: 0