NNICTL set classArgs as optional (#374)

In nnictl, classArgs is not required, now set it as optional for some kind of tuner and assessor may not require classArgs.

azure-pipelines.yml

 jobs:
 
-- job: 'Install_through_pip'
-  pool:
-    vmImage: 'Ubuntu 16.04'
-  strategy:
-    matrix:
-      Python36:
-        PYTHON_VERSION: '3.6'
-
-  steps:
-  - script: python3 -m pip install --upgrade pip setuptools
-    displayName: 'Install python tools'
-  - script: |
-      python3 -m pip install nni --user
-    displayName: 'Install nni toolkit via pip'
-  - script: |
-      cd test
-      PATH=$HOME/.local/bin:$PATH python3 naive_test.py
-    displayName: 'Integration tests'
-  - script: |
-      cd test
-      PATH=$HOME/.local/bin:$PATH python3 sdk_test.py
-    displayName: 'Built-in tuner tests'
-
 - job: 'Install_through_source_code'
   pool:
     vmImage: 'Ubuntu 16.04'

docs/ExperimentConfig.md

@@ -168,10 +168,10 @@ machineList:
   * __builtinTunerName__ and __classArgs__
     * __builtinTunerName__
     
-	  __builtinTunerName__ specifies the name of system tuner you want to use, nni sdk provides four kinds of tuner, including {__TPE__, __Random__, __Anneal__, __Evolution__}
+	  __builtinTunerName__ specifies the name of system tuner you want to use, nni sdk provides four kinds of tuner, including {__TPE__, __Random__, __Anneal__, __Evolution__, __BatchTuner__, __GridSearch__}
 	 * __classArgs__
 	
-	   __classArgs__ specifies the arguments of tuner algorithm
+	   __classArgs__ specifies the arguments of tuner algorithm. If the __builtinTunerName__ is in {__TPE__, __Random__, __Anneal__, __Evolution__}, you should set __optimize_mode__.
   * __codeDir__, __classFileName__, __className__ and __classArgs__
       * __codeDir__
         

src/sdk/pynni/nni/batch_tuner/batch_tuner.py

@@ -47,7 +47,7 @@ class BatchTuner(Tuner):
     }
     '''
     
-    def __init__(self, optimize_mode):
+    def __init__(self):
         self.count = -1
         self.values = []
 

src/sdk/pynni/nni/gridsearch_tuner/gridsearch_tuner.py

@@ -48,7 +48,7 @@ class GridSearchTuner(Tuner):
     and sample and then change the sampled value back.
     '''
 
-    def __init__(self, optimize_mode):
+    def __init__(self):
         self.count = -1
         self.expanded_search_space = []
 

test/sdk_test.py

@@ -29,7 +29,7 @@ GREEN = '\33[32m'
 RED = '\33[31m'
 CLEAR = '\33[0m'
 
-TUNER_LIST = ['BatchTuner', 'TPE', 'Random', 'Anneal', 'Evolution']
+TUNER_LIST = ['GridSearch', 'BatchTuner', 'TPE', 'Random', 'Anneal', 'Evolution']
 ASSESSOR_LIST = ['Medianstop']
 EXPERIMENT_URL = 'http://localhost:8080/api/v1/nni/experiment'
 
@@ -38,6 +38,11 @@ def switch(dispatch_type, dispatch_name):
     '''Change dispatch in config.yml'''
     config_path = 'sdk_test/local.yml'
     experiment_config = get_yml_content(config_path)
+    if dispatch_name in ['GridSearch', 'BatchTuner']:
+        experiment_config[dispatch_type.lower()] = {
+            'builtin' + dispatch_type + 'Name': dispatch_name
+        }
+    else:
         experiment_config[dispatch_type.lower()] = {
             'builtin' + dispatch_type + 'Name': dispatch_name,
             'classArgs': {

tools/nni_cmd/config_schema.py

@@ -34,12 +34,14 @@ Optional('multiPhase'): bool,
 Optional('multiThread'): bool,
 'useAnnotation': bool,
 'tuner': Or({
-    'builtinTunerName': Or('TPE', 'Random', 'Anneal', 'Evolution', 'SMAC', 'BatchTuner', 'GridSearch'),
-    'classArgs': {
-        'optimize_mode': Or('maximize', 'minimize'),
-        Optional('speed'): int
+    'builtinTunerName': Or('TPE', 'Random', 'Anneal', 'SMAC', 'Evolution'),
+    Optional('classArgs'): {
+        'optimize_mode': Or('maximize', 'minimize')
     },
     Optional('gpuNum'): And(int, lambda x: 0 <= x <= 99999),
+},{
+    'builtinTunerName': Or('BatchTuner', 'GridSearch'),
+    Optional('gpuNum'): And(int, lambda x: 0 <= x <= 99999),
 },{
     'codeDir': os.path.exists,
     'classFileName': str,
@@ -49,8 +51,10 @@ Optional('multiThread'): bool,
 }),
 Optional('assessor'): Or({
     'builtinAssessorName': lambda x: x in ['Medianstop'],
-    'classArgs': {
-        'optimize_mode': lambda x: x in ['maximize', 'minimize']},
+    Optional('classArgs'): {
+        Optional('optimize_mode'): Or('maximize', 'minimize'),
+        Optional('start_step'): And(int, lambda x: 0 <= x <= 9999)
+    },
     Optional('gpuNum'): And(int, lambda x: 0 <= x <= 99999)
 },{
     'codeDir': os.path.exists,