Chang to randint to support lower bound (#1096)

* update readme in ga_squad

* update readme

* fix typo

* Update README.md

* Update README.md

* Update README.md

* update readme

* update

* fix path

* update reference

* fix bug in config file

* update nni_arch_overview.png

* update

* update

* update

* update home page

* update default value of metis tuner

* update randint to quniform

* update metis

* update doc about chosen value

* update util

* update smac/bohb/hyperband

* update

* optimize the code

docs/en_US/AnnotationSpec.md

@@ -36,8 +36,8 @@ There are 10 types to express your search space as follows:
 
 * `@nni.variable(nni.choice(option1,option2,...,optionN),name=variable)`
   Which means the variable value is one of the options, which should be a list The elements of options can themselves be stochastic expressions
-* `@nni.variable(nni.randint(upper),name=variable)`
-  Which means the variable value is a random integer in the range [0, upper).
+* `@nni.variable(nni.randint(lower, upper),name=variable)`
+  Which means the variable value is a value like round(uniform(low, high)). For now, the type of chosen value is float. If you want to use integer value, please convert it explicitly.
 * `@nni.variable(nni.uniform(low, high),name=variable)`
   Which means the variable value is a value uniformly between low and high.
 * `@nni.variable(nni.quniform(low, high, q),name=variable)`

docs/en_US/SearchSpaceSpec.md

@@ -36,9 +36,9 @@ All types of sampling strategies and their parameter are listed here:
     - Anneal
     - Evolution
 
-* {"_type":"randint","_value":[upper]}
+* {"_type":"randint","_value":[lower, upper]}
 
-  * Which means the variable value is a random integer in the range [0, upper). The semantics of this distribution is that there is no more correlation in the loss function between nearby integer values, as compared with more distant integer values. This is an appropriate distribution for describing random seeds for example. If the loss function is probably more correlated for nearby integer values, then you should probably use one of the "quantized" continuous distributions, such as either quniform, qloguniform, qnormal or qlognormal. Note that if you want to change lower bound, you can use `quniform` for now.
+  * For now, we implment the "randint" distribution with "quniform", which means the variable value is a value like round(uniform(lower, upper)). The type of chosen value is float. If you want to use integer value, please convert it explicitly.
 
 * {"_type":"uniform","_value":[low, high]}
   * Which means the variable value is a value uniformly between low and high.

examples/trials/auto-gbdt/main.py

@@ -74,6 +74,8 @@ def load_data(train_path='./data/regression.train', test_path='./data/regression
 def run(lgb_train, lgb_eval, params, X_test, y_test):
     print('Start training...')
 
+    params['num_leaves'] = int(params['num_leaves'])
+
     # train
     gbm = lgb.train(params,
                     lgb_train,

examples/trials/auto-gbdt/search_space.json

 {
-    "num_leaves":{"_type":"choice","_value":[31, 28, 24, 20]},
+    "num_leaves":{"_type":"randint","_value":[20, 31]},
     "learning_rate":{"_type":"choice","_value":[0.01, 0.05, 0.1, 0.2]},
     "bagging_fraction":{"_type":"uniform","_value":[0.7, 1.0]},
     "bagging_freq":{"_type":"choice","_value":[1, 2, 4, 8, 10]}

src/sdk/pynni/nni/bohb_advisor/bohb_advisor.py

@@ -31,7 +31,7 @@ import ConfigSpace.hyperparameters as CSH
 
 from nni.protocol import CommandType, send
 from nni.msg_dispatcher_base import MsgDispatcherBase
-from nni.utils import OptimizeMode, extract_scalar_reward
+from nni.utils import OptimizeMode, extract_scalar_reward, randint_to_quniform
 
 from .config_generator import CG_BOHB
 
@@ -443,6 +443,7 @@ class BOHB(MsgDispatcherBase):
             search space of this experiment
         """
         search_space = data
+        randint_to_quniform(search_space)
         cs = CS.ConfigurationSpace()
         for var in search_space:
             _type = str(search_space[var]["_type"])

src/sdk/pynni/nni/evolution_tuner/evolution_tuner.py

@@ -26,7 +26,7 @@ import random
 
 import numpy as np
 from nni.tuner import Tuner
-from nni.utils import NodeType, OptimizeMode, extract_scalar_reward, split_index
+from nni.utils import NodeType, OptimizeMode, extract_scalar_reward, split_index, randint_to_quniform
 
 import nni.parameter_expressions as parameter_expressions
 
@@ -175,6 +175,7 @@ class EvolutionTuner(Tuner):
         search_space : dict
         """
         self.searchspace_json = search_space
+        randint_to_quniform(self.searchspace_json)
         self.space = json2space(self.searchspace_json)
 
         self.random_state = np.random.RandomState()

src/sdk/pynni/nni/hyperband_advisor/hyperband_advisor.py

@@ -31,7 +31,7 @@ import json_tricks
 from nni.protocol import CommandType, send
 from nni.msg_dispatcher_base import MsgDispatcherBase
 from nni.common import init_logger
-from nni.utils import NodeType, OptimizeMode, extract_scalar_reward
+from nni.utils import NodeType, OptimizeMode, extract_scalar_reward, randint_to_quniform
 import nni.parameter_expressions as parameter_expressions
 
 _logger = logging.getLogger(__name__)
@@ -357,6 +357,7 @@ class Hyperband(MsgDispatcherBase):
             number of trial jobs
         """
         self.searchspace_json = data
+        randint_to_quniform(self.searchspace_json)
         self.random_state = np.random.RandomState()
 
     def handle_trial_end(self, data):

src/sdk/pynni/nni/hyperopt_tuner/hyperopt_tuner.py

@@ -27,7 +27,7 @@ import logging
 import hyperopt as hp
 import numpy as np
 from nni.tuner import Tuner
-from nni.utils import NodeType, OptimizeMode, extract_scalar_reward, split_index
+from nni.utils import NodeType, OptimizeMode, extract_scalar_reward, split_index, randint_to_quniform
 
 logger = logging.getLogger('hyperopt_AutoML')
 
@@ -231,6 +231,8 @@ class HyperoptTuner(Tuner):
         search_space : dict
         """
         self.json = search_space
+        randint_to_quniform(self.json)
+
         search_space_instance = json2space(self.json)
         rstate = np.random.RandomState()
         trials = hp.Trials()

src/sdk/pynni/nni/metis_tuner/metis_tuner.py

@@ -133,7 +133,7 @@ class MetisTuner(Tuner):
                         self.x_bounds[idx] = bounds
                         self.x_types[idx] = 'discrete_int'
                 elif key_type == 'randint':
-                    self.x_bounds[idx] = [0, key_range[0]]
+                    self.x_bounds[idx] = [key_range[0], key_range[1]]
                     self.x_types[idx] = 'range_int'
                 elif key_type == 'uniform':
                     self.x_bounds[idx] = [key_range[0], key_range[1]]

src/sdk/pynni/nni/smac_tuner/smac_tuner.py

@@ -37,6 +37,9 @@ from ConfigSpaceNNI import Configuration
 
 from .convert_ss_to_scenario import generate_scenario
 
+from nni.tuner import Tuner
+from nni.utils import OptimizeMode, extract_scalar_reward, randint_to_quniform
+
 
 class SMACTuner(Tuner):
     """
@@ -136,6 +139,7 @@ class SMACTuner(Tuner):
         search_space:
             search space
         """
+        randint_to_quniform(search_space)
         if not self.update_ss_done:
             self.categorical_dict = generate_scenario(search_space)
             if self.categorical_dict is None:

src/sdk/pynni/nni/utils.py

@@ -40,6 +40,7 @@ class OptimizeMode(Enum):
     Minimize = 'minimize'
     Maximize = 'maximize'
 
+
 class NodeType:
     """Node Type class
     """
@@ -83,6 +84,7 @@ def extract_scalar_reward(value, scalar_key='default'):
         raise RuntimeError('Incorrect final result: the final result should be float/int, or a dict which has a key named "default" whose value is float/int.')
     return reward
 
+
 def convert_dict2tuple(value):
     """
     convert dict type to tuple to solve unhashable problem.
@@ -94,9 +96,30 @@ def convert_dict2tuple(value):
     else:
         return value
 
+
 def init_dispatcher_logger():
     """ Initialize dispatcher logging configuration"""
     logger_file_path = 'dispatcher.log'
     if dispatcher_env_vars.NNI_LOG_DIRECTORY is not None:
         logger_file_path = os.path.join(dispatcher_env_vars.NNI_LOG_DIRECTORY, logger_file_path)
     init_logger(logger_file_path, dispatcher_env_vars.NNI_LOG_LEVEL)
+
+
+def randint_to_quniform(in_x):
+    if isinstance(in_x, dict):
+        if NodeType.TYPE in in_x.keys():
+            if in_x[NodeType.TYPE] == 'randint':
+                value = in_x[NodeType.VALUE]
+                value.append(1)
+
+                in_x[NodeType.TYPE] = 'quniform'
+                in_x[NodeType.VALUE] = value
+ 
+            elif in_x[NodeType.TYPE] == 'choice':
+                randint_to_quniform(in_x[NodeType.VALUE])
+        else:
+            for key in in_x.keys():
+                randint_to_quniform(in_x[key])
+    elif isinstance(in_x, list):
+        for temp in in_x:
+            randint_to_quniform(temp)