Created
October 13, 2016 12:47
-
-
Save ddugovic/1d3a63790495c78f3fc7619ee4b976e8 to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import gym | |
import numpy as np | |
def run_episode(env, parameters): | |
observation = env.reset() | |
totalreward = 0 | |
#for 200 timesteps | |
for _ in xrange(200): | |
#env.render() | |
#initalize random weights | |
action = 0 if np.matmul(parameters, observation)<0 else 1 | |
observation, reward, done, info = env.step(action) | |
totalreward += reward | |
if done: | |
break | |
return totalreward | |
#hill climbing algo training | |
def train(submit): | |
np.random.seed(0) | |
outdir = '/tmp/gym/hill-climbing-agent-results' | |
env = gym.make('CartPole-v0') | |
env.monitor.start(outdir, force=True, seed=0) | |
episodes_per_update = 5 | |
noise_scaling = 0.1 | |
parameters = np.zeros(env.observation_space.shape[0]) | |
bestreward = 0 | |
#100 episodes | |
for x in xrange(100): | |
newparams = parameters + (np.random.rand(len(parameters)) * 2 - 1) * noise_scaling | |
reward = run_episode(env, newparams) | |
print "reward %d best %d" % (reward, bestreward) | |
if reward > bestreward: | |
bestreward = reward | |
parameters = newparams | |
if reward == 200: | |
noise_scaling /= 2.0 | |
# Dump result info to disk | |
env.monitor.close() | |
# Upload to the scoreboard | |
if submit: | |
gym.upload(outdir) | |
r = train(submit=True) | |
print r |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment