Merge pull request #282 from microsoft/v2.0

V2.0

docs/en_US/TrainingService/RemoteMachineMode.rst

@@ -217,3 +217,11 @@ If you want multiple commands to be executed, you can use ``&&`` to connect thes
    preCommand: command1 && command2 && command3
 
 **Note**\ : Because **preCommand** will execute before other commands each time, it is strongly not recommended to set **preCommand** that will make changes to system, i.e. ``mkdir`` or ``touch``.
+
+Remote machine supports running experiment in reuse mode. In this mode, NNI will reuse remote machine jobs to run as many as possible trials. It can save time of creating new jobs. User needs to make sure each trial can run independent in the same job, for example, avoid loading checkpoint from previous trials.  
+Follow the setting to enable reuse mode:
+
+.. code-block:: yaml
+
+   remoteConfig:
+     reuse: true

docs/en_US/Tutorial/HowToLaunchFromPython.rst

@@ -0,0 +1,97 @@
+**How to Launch an experiment from Python**
+===========================================
+
+Overview
+--------
+Since ``nni v2.0``, we provide a new way to launch experiments. Before that, you need to configure the experiment in the yaml configuration file and then use the experiment ``nnictl`` command to launch the experiment. Now, you can also configure and run experiments directly in python file. If you are familiar with python programming, this will undoubtedly bring you more convenience.
+
+How to Use
+----------
+After successfully installing ``nni``, you can start the experiment with a python script in the following 3 steps.
+
+..
+
+    Step 1 - Initialize a tuner you want to use
+
+
+.. code-block:: python
+
+    from nni.algorithms.hpo.hyperopt_tuner import HyperoptTuner
+    tuner = HyperoptTuner('tpe')
+
+Very simple, you have successfully initialized a ``HyperoptTuner`` instance called ``tuner``.
+
+See all real `builtin tuners <../builtin_tuner.rst>`__ supported in NNI.
+
+..
+
+    Step 2 - Initialize an experiment instance and configure it
+
+.. code-block:: python
+
+    experiment = Experiment(tuner=tuner, training_service='local')
+
+Now, you have a ``Experiment`` instance with ``tuner`` you have initialized in the previous step, and this experiment will launch trials on your local machine due to ``training_service='local'``.
+
+See all `training services <../training_services.rst>`__ supported in NNI.
+
+.. code-block:: python
+
+    experiment.config.experiment_name = 'test'
+    experiment.config.trial_concurrency = 2
+    experiment.config.max_trial_number = 5
+    experiment.config.search_space = search_space
+    experiment.config.trial_command = 'python3 mnist.py'
+    experiment.config.trial_code_directory = Path(__file__).parent
+    experiment.config.training_service.use_active_gpu = True
+
+Use the form like ``experiment.config.foo = 'bar'`` to configure your experiment.
+
+See `parameter configuration <../reference/experiment_config.rst>`__ required by different training services.
+
+..
+
+    Step 3 - Just run
+
+.. code-block:: python
+
+    experiment.run(port=8081)
+
+Now, you have successfully launched an NNI experiment. And you can type ``localhost:8081`` in your browser to observe your experiment in real time.
+
+Example
+-------
+Below is an example for this new launching approach. You can also find this code in :githublink:`mnist-tfv2/launch.py <examples/trials/mnist-tfv2/launch.py>` .
+
+.. code-block:: python
+
+    from pathlib import Path
+    from nni.experiment import Experiment
+    from nni.algorithms.hpo.hyperopt_tuner import HyperoptTuner
+
+    tuner = HyperoptTuner('tpe')
+
+    search_space = {
+        "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] }
+    }
+
+    experiment = Experiment(tuner, 'local')
+    experiment.config.experiment_name = 'test'
+    experiment.config.trial_concurrency = 2
+    experiment.config.max_trial_number = 5
+    experiment.config.search_space = search_space
+    experiment.config.trial_command = 'python3 mnist.py'
+    experiment.config.trial_code_directory = Path(__file__).parent
+    experiment.config.training_service.use_active_gpu = True
+
+    experiment.run(8081)
+
+API
+---
+
+..  autoclass:: nni.experiment.Experiment
+    :members:

docs/en_US/nnicli_ref.rst

@@ -25,17 +25,3 @@ NNI client is a python API of ``nnictl``\ , which implements the most commonly u
 
    # stop the experiment, then disconnect the instance from the experiment.
    exp.stop_experiment()
-
-References
-----------
-
-..  autoclass:: nni.experiment.Experiment
-    :members:
-..  autoclass:: nni.experiment.TrialJob
-    :members:
-..  autoclass:: nni.experiment.TrialHyperParameters
-    :members:
-..  autoclass:: nni.experiment.TrialMetricData
-    :members:
-..  autoclass:: nni.experiment.TrialResult
-    :members:

docs/en_US/reference.rst

@@ -11,3 +11,4 @@ References
     NNI Annotation <Tutorial/AnnotationSpec>
     SDK API References <sdk_reference>
     Supported Framework Library <SupportedFramework_Library>
+    Launch from python <Tutorial/HowToLaunchFromPython>

docs/en_US/training_services.rst

@@ -6,7 +6,6 @@ Introduction to NNI Training Services
     Local<./TrainingService/LocalMode>
     Remote<./TrainingService/RemoteMachineMode>
     OpenPAI<./TrainingService/PaiMode>
-    OpenPAI Yarn Mode<./TrainingService/PaiYarnMode>
     Kubeflow<./TrainingService/KubeflowMode>
     AdaptDL<./TrainingService/AdaptDLMode>
     FrameworkController<./TrainingService/FrameworkControllerMode>