DanielTakeshi/gist:fec9a5cd957eb05b04b6d06a16cc88ae

## gistfile1.txt
import gym
import numpy as np
import imageio

# Make the environment and the initial observation `o`.
env_name = 'Walker2d-v3'
env = gym.make(env_name)
obs_dim = env.observation_space.shape
act_dim = env.action_space.shape[0]
act_limit = env.action_space.high[0]
o = env.reset()

# Book-keeping.
num_episodes = 3
ep_ret = 0
ep_len = 0
ep_done = 0
ep_obs = []

# For video / GIF.
dur = 0.01
width = 250
height = 200

while ep_done < num_episodes:
    obs = env.render(mode='rgb_array', width=width, height=height)
    assert obs.shape == (height, width, 3), obs.shape  # height first!
    ep_obs.append(obs)

    # Take action, step into environment, etc.
    a = np.random.randn(act_dim)
    a = np.clip(a, -act_limit, act_limit)
    o, r, d, _ = env.step(a)
    ep_ret += r
    ep_len += 1

    if d:
        # Form GIF. imageio should read from numpy: https://imageio.github.io/
        print(f'Episode {ep_done}, cum. return: {ep_ret:0.1f}, length: {ep_len}.')
        ep_name = f'ep_{env_name}_{str(ep_done).zfill(2)}_dur_{dur}_len_{str(ep_len).zfill(3)}.gif'
        with imageio.get_writer(ep_name, mode='I', duration=dur) as writer:
            for obs_np in ep_obs:
                writer.append_data(obs_np)

        # Reset information.
        o = env.reset()
        ep_ret = 0
        ep_len = 0
        ep_done += 1
        ep_obs = []
	import gym
	import numpy as np
	import imageio

	# Make the environment and the initial observation `o`.
	env_name = 'Walker2d-v3'
	env = gym.make(env_name)
	obs_dim = env.observation_space.shape
	act_dim = env.action_space.shape[0]
	act_limit = env.action_space.high[0]
	o = env.reset()

	# Book-keeping.
	num_episodes = 3
	ep_ret = 0
	ep_len = 0
	ep_done = 0
	ep_obs = []

	# For video / GIF.
	dur = 0.01
	width = 250
	height = 200

	while ep_done < num_episodes:
	obs = env.render(mode='rgb_array', width=width, height=height)
	assert obs.shape == (height, width, 3), obs.shape # height first!
	ep_obs.append(obs)

	# Take action, step into environment, etc.
	a = np.random.randn(act_dim)
	a = np.clip(a, -act_limit, act_limit)
	o, r, d, _ = env.step(a)
	ep_ret += r
	ep_len += 1

	if d:
	# Form GIF. imageio should read from numpy: https://imageio.github.io/
	print(f'Episode {ep_done}, cum. return: {ep_ret:0.1f}, length: {ep_len}.')
	ep_name = f'ep_{env_name}_{str(ep_done).zfill(2)}_dur_{dur}_len_{str(ep_len).zfill(3)}.gif'
	with imageio.get_writer(ep_name, mode='I', duration=dur) as writer:
	for obs_np in ep_obs:
	writer.append_data(obs_np)

	# Reset information.
	o = env.reset()
	ep_ret = 0
	ep_len = 0
	ep_done += 1
	ep_obs = []