Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

improve gym speedup #210

Merged
merged 16 commits into from
Nov 1, 2022
Merged

improve gym speedup #210

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
16 commits
Select commit Hold shift + click to select a range
3fbad8e
improve gym speedup
wangsiping97 Oct 27, 2022
d6329cf
fix lint
wangsiping97 Oct 27, 2022
46aa42d
fix test
wangsiping97 Oct 28, 2022
5089233
fix test
wangsiping97 Oct 28, 2022
4340e49
fix test
wangsiping97 Oct 28, 2022
8a4f167
Merge branch 'main' into gym-speedup
wangsiping97 Oct 28, 2022
555e575
refactor test env entry point call
Trinkle23897 Oct 31, 2022
8a4d5da
Merge branch 'main' into gym-speedup
Trinkle23897 Oct 31, 2022
cc081c9
partly finished
wangsiping97 Oct 31, 2022
1663a72
gym mjc
Trinkle23897 Oct 31, 2022
642eb04
vizdoom
Trinkle23897 Oct 31, 2022
c6e8b73
atari
Trinkle23897 Oct 31, 2022
826c076
fix lint
Trinkle23897 Oct 31, 2022
6222cc6
remove duplicated no_type_check
Trinkle23897 Nov 1, 2022
3c677c0
Merge branch 'main' into gym-speedup
Trinkle23897 Nov 1, 2022
7724565
update doc
Trinkle23897 Nov 1, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
vizdoom
  • Loading branch information
@Trinkle23897
Trinkle23897 committed Oct 31, 2022
commit 642eb04f75b2fa5b61dcc2fac23cbe2f3279c97e
55 changes: 12 additions & 43 deletions envpool/mujoco/gym/mujoco_gym_align_test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -80,7 +80,7 @@ def run_align_check(
def test_ant(self) -> None:
assert version.parse(gym.__version__) >= version.parse("0.26.0")
env0 = gym.make("Ant-v4")
env1 = make_gym("Ant-v4", gym_reset_return_info=True)
env1 = make_gym("Ant-v4")
self.run_space_check(env0, env1)
self.run_align_check(env0, env1)
env0 = gym.make(
Expand All @@ -93,14 +93,13 @@ def test_ant(self) -> None:
terminate_when_unhealthy=False,
exclude_current_positions_from_observation=False,
max_episode_steps=100,
gym_reset_return_info=True,
)
self.run_space_check(env0, env1)
self.run_align_check(env0, env1, no_time_limit=True)

def test_half_cheetah(self) -> None:
env0 = gym.make("HalfCheetah-v4")
env1 = make_gym("HalfCheetah-v4", gym_reset_return_info=True)
env1 = make_gym("HalfCheetah-v4")
self.run_space_check(env0, env1)
self.run_align_check(env0, env1, no_time_limit=True)
env0 = gym.make(
Expand All @@ -109,16 +108,12 @@ def test_half_cheetah(self) -> None:
env1 = make_gym(
"HalfCheetah-v4",
exclude_current_positions_from_observation=True,
gym_reset_return_info=True,
max_episode_steps=1000,
)
self.run_space_check(env0, env1)

def test_hopper(self) -> None:
env0 = gym.make("Hopper-v4")
env1 = make_gym(
"Hopper-v4", gym_reset_return_info=True, max_episode_steps=1000
)
env1 = make_gym("Hopper-v4")
self.run_space_check(env0, env1)
self.run_align_check(env0, env1)
env0 = gym.make(
Expand All @@ -127,19 +122,16 @@ def test_hopper(self) -> None:
exclude_current_positions_from_observation=False,
)
env1 = make_gym(
"Hopper-v4",
terminate_when_unhealthy=False,
exclude_current_positions_from_observation=False,
gym_reset_return_info=True,
max_episode_steps=1000,
)
self.run_space_check(env0, env1)
self.run_align_check(env0, env1, no_time_limit=True)

def test_humanoid(self) -> None:
env0 = gym.make("Humanoid-v4")
env1 = make_gym(
"Humanoid-v4", gym_reset_return_info=True, max_episode_steps=1000
)
env1 = make_gym("Humanoid-v4")
self.run_space_check(env0, env1)
self.run_align_check(env0, env1)
env0 = gym.make(
Expand All @@ -151,61 +143,43 @@ def test_humanoid(self) -> None:
"Humanoid-v4",
terminate_when_unhealthy=False,
exclude_current_positions_from_observation=False,
gym_reset_return_info=True,
max_episode_steps=1000,
)
self.run_space_check(env0, env1)
self.run_align_check(env0, env1, no_time_limit=True)

def test_humanoid_standup(self) -> None:
env0 = gym.make("HumanoidStandup-v4")
env1 = make_gym(
"HumanoidStandup-v4", gym_reset_return_info=True, max_episode_steps=1000
)
env1 = make_gym("HumanoidStandup-v4")
self.run_space_check(env0, env1)
self.run_align_check(env0, env1, no_time_limit=True)

def test_inverted_double_pendulum(self) -> None:
env0 = gym.make("InvertedDoublePendulum-v4")
env1 = make_gym(
"InvertedDoublePendulum-v4",
gym_reset_return_info=True,
max_episode_steps=1000,
)
env1 = make_gym("InvertedDoublePendulum-v4",)
self.run_space_check(env0, env1)
self.run_align_check(env0, env1)

def test_inverted_pendulum(self) -> None:
env0 = gym.make("InvertedPendulum-v4")
env1 = make_gym(
"InvertedPendulum-v4",
gym_reset_return_info=True,
max_episode_steps=1000
)
env1 = make_gym("InvertedPendulum-v4",)
self.run_space_check(env0, env1)
self.run_align_check(env0, env1)

def test_pusher(self) -> None:
env0 = gym.make("Pusher-v4")
env1 = make_gym(
"Pusher-v4", gym_reset_return_info=True, max_episode_steps=100
)
env1 = make_gym("Pusher-v4")
self.run_space_check(env0, env1)
self.run_align_check(env0, env1, no_time_limit=True)

def test_reacher(self) -> None:
env0 = gym.make("Reacher-v4")
env1 = make_gym(
"Reacher-v4", gym_reset_return_info=True, max_episode_steps=50
)
env1 = make_gym("Reacher-v4")
self.run_space_check(env0, env1)
self.run_align_check(env0, env1, no_time_limit=True)

def test_swimmer(self) -> None:
env0 = gym.make("Swimmer-v4")
env1 = make_gym(
"Swimmer-v4", gym_reset_return_info=True, max_episode_steps=1000
)
env1 = make_gym("Swimmer-v4")
self.run_space_check(env0, env1)
self.run_align_check(env0, env1, no_time_limit=True)
env0 = gym.make(
Expand All @@ -214,16 +188,12 @@ def test_swimmer(self) -> None:
env1 = make_gym(
"Swimmer-v4",
exclude_current_positions_from_observation=False,
gym_reset_return_info=True,
max_episode_steps=1000,
)
self.run_space_check(env0, env1)

def test_walker2d(self) -> None:
env0 = gym.make("Walker2d-v4")
env1 = make_gym(
"Walker2d-v4", gym_reset_return_info=True, max_episode_steps=1000
)
env1 = make_gym("Walker2d-v4")
self.run_space_check(env0, env1)
self.run_align_check(env0, env1)
env0 = gym.make(
Expand All @@ -236,7 +206,6 @@ def test_walker2d(self) -> None:
terminate_when_unhealthy=False,
exclude_current_positions_from_observation=False,
max_episode_steps=100,
gym_reset_return_info=True,
)
self.run_space_check(env0, env1)
self.run_align_check(env0, env1, no_time_limit=True)
Expand Down
14 changes: 8 additions & 6 deletions envpool/vizdoom/BUILD
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,11 +66,18 @@ py_library(
deps = ["//envpool/python:api"],
)

py_library(
name = "vizdoom_registration",
srcs = ["registration.py"],
deps = ["//envpool:registration"],
)

py_test(
name = "vizdoom_test",
srcs = ["vizdoom_test.py"],
deps = [
":vizdoom",
":vizdoom_registration",
requirement("numpy"),
requirement("absl-py"),
requirement("opencv-python-headless"),
Expand All @@ -84,16 +91,11 @@ py_test(
data = [":gen_pretrain_weight"],
deps = [
":vizdoom",
":vizdoom_registration",
"//envpool/atari:atari_network",
requirement("numpy"),
requirement("absl-py"),
requirement("tianshou"),
requirement("torch"),
],
)

py_library(
name = "vizdoom_registration",
srcs = ["registration.py"],
deps = ["//envpool:registration"],
)
9 changes: 5 additions & 4 deletions envpool/vizdoom/vizdoom_pretrain_test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,7 +24,9 @@
from tianshou.policy import C51Policy

from envpool.atari.atari_network import C51
from envpool.vizdoom import VizdoomEnvSpec, VizdoomGymEnvPool

import envpool.vizdoom.registration # noqa: F401
from envpool.registration import make_gym

# try:
# import cv2
Expand All @@ -47,6 +49,7 @@ def eval_c51(
cfg_path: Optional[str] = None,
reward_config: Optional[dict] = None,
) -> Tuple[np.ndarray, np.ndarray]:
task_id = "".join([g.capitalize() for g in task.split("_")]) + "-v1"
kwargs = {
"num_envs": num_envs,
"seed": seed,
Expand All @@ -59,9 +62,7 @@ def eval_c51(
kwargs.update(cfg_path=cfg_path)
if reward_config is not None:
kwargs.update(reward_config=reward_config)
env = VizdoomGymEnvPool(
VizdoomEnvSpec(VizdoomEnvSpec.gen_config(**kwargs))
)
env = make_gym(task_id, **kwargs)

state_shape = env.observation_space.shape
action_shape = env.action_space.n
Expand Down
Loading