import argparse import os import ray from ray import air, tune from ray.tune.registry import get_trainable_cls parser = argparse.ArgumentParser() parser.add_argument( "--run", type=str, default="PPO", help="The RLlib-registered algorithm to use." ) parser.add_argument("--num-cpus", type=int, default=0) parser.add_argument( "--framework", choices=["tf", "tf2", "torch"], default="torch", help="The DL framework specifier.", ) parser.add_argument("--stop-iters", type=int, default=200) parser.add_argument("--stop-timesteps", type=int, default=100000) parser.add_argument("--stop-reward", type=float, default=150.0) parser.add_argument( "--local-mode", action="store_true", help="Init Ray in local mode for easier debugging.", ) if __name__ == "__main__": args = parser.parse_args() ray.init(num_cpus=args.num_cpus or None, local_mode=args.local_mode) # Simple generic config. config = ( get_trainable_cls(args.run) .get_default_config() .environment("CartPole-v1") # Run with tracing enabled for tf2. .framework(args.framework) # Run 3 trials. .training( lr=tune.grid_search([0.01, 0.001, 0.0001]), train_batch_size=2341 ) # TEST # Use GPUs iff `RLLIB_NUM_GPUS` env var set to > 0. .resources(num_gpus=int(os.environ.get("RLLIB_NUM_GPUS", "0"))) ) stop = { "training_iteration": args.stop_iters, "timesteps_total": args.stop_timesteps, "episode_reward_mean": args.stop_reward, } # Run tune for some iterations and generate checkpoints. tuner = tune.Tuner( args.run, param_space=config.to_dict(), run_config=air.RunConfig( stop=stop, checkpoint_config=air.CheckpointConfig(checkpoint_frequency=1) ), ) results = tuner.fit() # Get the best of the 3 trials by using some metric. # NOTE: Choosing the min `episodes_this_iter` automatically picks the trial # with the best performance (over the entire run (scope="all")): # The fewer episodes, the longer each episode lasted, the more reward we # got each episode. # Setting scope to "last", "last-5-avg", or "last-10-avg" will only compare # (using `mode=min|max`) the average values of the last 1, 5, or 10 # iterations with each other, respectively. # Setting scope to "avg" will compare (using `mode`=min|max) the average # values over the entire run. metric = "episodes_this_iter" # notice here `scope` is `all`, meaning for each trial, # all results (not just the last one) will be examined. best_result = results.get_best_result(metric=metric, mode="min", scope="all") value_best_metric = best_result.metrics_dataframe[metric].min() print( "Best trial's lowest episode length (over all " "iterations): {}".format(value_best_metric) ) # Confirm, we picked the right trial. assert value_best_metric <= results.get_dataframe()[metric].min() # Get the best checkpoints from the trial, based on different metrics. # Checkpoint with the lowest policy loss value: if config._enable_learner_api: policy_loss_key = "info/learner/default_policy/policy_loss" else: policy_loss_key = "info/learner/default_policy/learner_stats/policy_loss" ckpt = results.get_best_result(metric=policy_loss_key, mode="min").checkpoint print("Lowest pol-loss: {}".format(ckpt)) # Checkpoint with the highest value-function loss: if config._enable_learner_api: vf_loss_key = "info/learner/default_policy/vf_loss" else: vf_loss_key = "info/learner/default_policy/learner_stats/vf_loss" ckpt = results.get_best_result(metric=vf_loss_key, mode="max").checkpoint print("Highest vf-loss: {}".format(ckpt)) ray.shutdown()