[tune] Initial Commit for Tune CLI (#3983)

This introduces a light CLI for Tune.

python/ray/tune/suggest/bayesopt.py

@@ -4,13 +4,16 @@ from __future__ import print_function
 
 import copy
 
-try:
-    import bayes_opt as byo
-except Exception:
-    byo = None
-
 from ray.tune.suggest.suggestion import SuggestionAlgorithm
 
+byo = None
+
+
+def _import_bayesopt():
+    global byo
+    import bayes_opt
+    byo = bayes_opt
+
 
 class BayesOptSearch(SuggestionAlgorithm):
     """A wrapper around BayesOpt to provide trial suggestions.
@@ -56,6 +59,7 @@ class BayesOptSearch(SuggestionAlgorithm):
                  random_state=1,
                  verbose=0,
                  **kwargs):
+        _import_bayesopt()
         assert byo is not None, (
             "BayesOpt must be installed!. You can install BayesOpt with"
             " the command: `pip install bayesian-optimization`.")

python/ray/tune/suggest/hyperopt.py

@@ -6,17 +6,19 @@ import numpy as np
 import copy
 import logging
 
-try:
-    hyperopt_logger = logging.getLogger("hyperopt")
-    hyperopt_logger.setLevel(logging.WARNING)
-    import hyperopt as hpo
-    from hyperopt.fmin import generate_trials_to_calculate
-except Exception:
-    hpo = None
-
 from ray.tune.error import TuneError
 from ray.tune.suggest.suggestion import SuggestionAlgorithm
 
+hpo = None
+
+
+def _import_hyperopt():
+    global hpo
+    hyperopt_logger = logging.getLogger("hyperopt")
+    hyperopt_logger.setLevel(logging.WARNING)
+    import hyperopt
+    hpo = hyperopt
+
 
 class HyperOptSearch(SuggestionAlgorithm):
     """A wrapper around HyperOpt to provide trial suggestions.
@@ -73,7 +75,9 @@ class HyperOptSearch(SuggestionAlgorithm):
                  reward_attr="episode_reward_mean",
                  points_to_evaluate=None,
                  **kwargs):
+        _import_hyperopt()
         assert hpo is not None, "HyperOpt must be installed!"
+        from hyperopt.fmin import generate_trials_to_calculate
         assert type(max_concurrent) is int and max_concurrent > 0
         self._max_concurrent = max_concurrent
         self._reward_attr = reward_attr