[tune] Ray Tune API cleanup

doc/source/rllib.rst

@@ -262,6 +262,7 @@ in the ``config`` section of the experiments.
 
 .. code-block:: python
 
+    import ray
     from ray.tune.tune import run_experiments
     from ray.tune.variant_generator import grid_search
 
@@ -286,6 +287,7 @@ in the ``config`` section of the experiments.
         # put additional experiments to run concurrently here
     }
 
+    ray.init()
     run_experiments(experiment)
 
 Contributing to RLlib

doc/source/tune.rst

@@ -18,6 +18,7 @@ Getting Started
 
 ::
 
+    import ray
     from ray.tune import register_trainable, grid_search, run_experiments
 
     def my_func(config, reporter):
@@ -30,6 +31,7 @@ Getting Started
 
     register_trainable("my_func", my_func)
 
+    ray.init()
     run_experiments({
         "my_experiment": {
             "run": "my_func",
@@ -154,8 +156,3 @@ The JSON config passed to ``run_experiments`` can also be put in a JSON or YAML
 
 
 For more examples of experiments described by YAML files, see `RLlib tuned examples <https://github.com/ray-project/ray/tree/master/python/ray/rllib/tuned_examples>`__.
-
-Running in a large cluster
---------------------------
-
-The ``run_experiments`` also takes any arguments that ``ray.init()`` does. This can be used to pass in the redis address of a multi-node Ray cluster. For more details, check out the `tune.py script <https://github.com/ray-project/ray/blob/master/python/ray/tune/tune.py>`__.

examples/carla/a3c_lane_keep.py

@@ -2,6 +2,7 @@
 from __future__ import division
 from __future__ import print_function
 
+import ray
 from ray.tune import register_env, run_experiments
 
 from env import CarlaEnv, ENV_CONFIG
@@ -25,6 +26,7 @@
 register_env(env_name, lambda env_config: CarlaEnv(env_config))
 register_carla_model()
 
+ray.init()
 run_experiments({
     "carla-a3c": {
         "run": "A3C",

examples/carla/dqn_lane_keep.py

@@ -2,6 +2,7 @@
 from __future__ import division
 from __future__ import print_function
 
+import ray
 from ray.tune import register_env, run_experiments
 
 from env import CarlaEnv, ENV_CONFIG
@@ -25,6 +26,7 @@
 register_env(env_name, lambda env_config: CarlaEnv(env_config))
 register_carla_model()
 
+ray.init()
 run_experiments({
     "carla-dqn": {
         "run": "DQN",

examples/carla/ppo_lane_keep.py

@@ -2,6 +2,7 @@
 from __future__ import division
 from __future__ import print_function
 
+import ray
 from ray.tune import register_env, run_experiments
 
 from env import CarlaEnv, ENV_CONFIG
@@ -25,6 +26,7 @@
 register_env(env_name, lambda env_config: CarlaEnv(env_config))
 register_carla_model()
 
+ray.init()
 run_experiments({
     "carla-ppo": {
         "run": "PPO",

examples/carla/train_a3c.py

@@ -27,6 +27,7 @@
 register_carla_model()
 redis_address = ray.services.get_node_ip_address() + ":6379"
 
+ray.init(redis_address=redis_address)
 run_experiments({
     "carla-a3c": {
         "run": "A3C",
@@ -50,4 +51,4 @@
             "num_workers": 2,
         },
     },
-}, redis_address=redis_address)
+}

examples/carla/train_dqn.py

@@ -2,6 +2,7 @@
 from __future__ import division
 from __future__ import print_function
 
+import ray
 from ray.tune import register_env, run_experiments
 
 from env import CarlaEnv, ENV_CONFIG
@@ -23,6 +24,7 @@
 register_env(env_name, lambda env_config: CarlaEnv(env_config))
 register_carla_model()
 
+ray.init()
 run_experiments({
     "carla-dqn": {
         "run": "DQN",

examples/carla/train_ppo.py

@@ -2,6 +2,7 @@
 from __future__ import division
 from __future__ import print_function
 
+import ray
 from ray.tune import register_env, run_experiments
 
 from env import CarlaEnv, ENV_CONFIG
@@ -22,6 +23,7 @@
 register_env(env_name, lambda env_config: CarlaEnv(env_config))
 register_carla_model()
 
+ray.init(redirect_output=True)
 run_experiments({
     "carla": {
         "run": "PPO",
@@ -55,4 +57,4 @@
             }
         },
     },
-}, redirect_output=True)
+})

python/ray/rllib/train.py

@@ -8,6 +8,7 @@
 import sys
 import yaml
 
+import ray
 from ray.tune.config_parser import make_parser, resources_to_json
 from ray.tune.tune import _make_scheduler, run_experiments
 
@@ -76,7 +77,7 @@
         if not exp.get("env") and not exp.get("config", {}).get("env"):
             parser.error("the following arguments are required: --env")
 
-    run_experiments(
-        experiments, scheduler=_make_scheduler(args),
+    ray.init(
         redis_address=args.redis_address,
         num_cpus=args.num_cpus, num_gpus=args.num_gpus)
+    run_experiments(experiments, scheduler=_make_scheduler(args))

python/ray/tune/examples/tune_mnist_ray.py

@@ -33,6 +33,7 @@
 import tempfile
 import time
 
+import ray
 from ray.tune import grid_search, run_experiments, register_trainable
 
 from tensorflow.examples.tutorials.mnist import input_data
@@ -222,4 +223,5 @@ def train(config={'activation': 'relu'}, reporter=None):
     if args.fast:
         mnist_spec['stop']['training_iteration'] = 2
 
+    ray.init()
     run_experiments({'tune_mnist_test': mnist_spec})

python/ray/tune/tune.py

@@ -61,7 +61,7 @@ def _make_scheduler(args):
                 args.scheduler, _SCHEDULERS.keys()))
 
 
-def run_experiments(experiments, scheduler=None, **ray_args):
+def run_experiments(experiments, scheduler=None):
     if scheduler is None:
         scheduler = FIFOScheduler()
     runner = TrialRunner(scheduler)
@@ -71,8 +71,6 @@ def run_experiments(experiments, scheduler=None, **ray_args):
             runner.add_trial(trial)
     print(runner.debug_string())
 
-    ray.init(**ray_args)
-
     while not runner.is_finished():
         runner.step()
         print(runner.debug_string())
@@ -89,6 +87,7 @@ def run_experiments(experiments, scheduler=None, **ray_args):
     args = parser.parse_args(sys.argv[1:])
     with open(args.config_file) as f:
         experiments = yaml.load(f)
-    run_experiments(
-        experiments, _make_scheduler(args), redis_address=args.redis_address,
+    ray.init(
+        redis_address=args.redis_address,
         num_cpus=args.num_cpus, num_gpus=args.num_gpus)
+    run_experiments(experiments, _make_scheduler(args))

test/trial_runner_test.py

@@ -20,6 +20,9 @@
 
 
 class TrainableFunctionApiTest(unittest.TestCase):
+    def setUp(self):
+        ray.init()
+
     def tearDown(self):
         ray.worker.cleanup()
         _register_all()  # re-register the evicted objects