fix meta_rl evironmental setup (#76)

* fix meta_rl evironmental setup

* Update makefile

* Update torch version

* modify torch-cpu version requirement

* Update torch version

* ..

* ..

* ..

* ..

* ..

* ..

* ..

Co-authored-by: dongminlee94 <[email protected]>

Makefile

@@ -1,10 +1,3 @@
-GPU := $(shell which nvcc)
-ifdef GPU
- DEVICE="gpu"
-else
- DEVICE="cpu"
-endif
-
 STAGED := $(shell git diff --cached --name-only --diff-filter=ACMR -- 'src/***.py' | sed 's| |\\ |g')
 
 all: format lint
@@ -34,10 +27,13 @@ endif
 
 init:
  pip install -U pip
- pip install -U setuptools
  pip install -e .
- pip install -r requirements-common.txt
- pip install -r requirements-$(DEVICE).txt
+ pip install -r requirements.txt
+ python3 ./scripts/download-torch.py
+ conda install -y tensorboard
+ jupyter contrib nbextension install --user
+ jupyter nbextensions_configurator enable --user
+ python3 -m ipykernel install --user
  bash ./hooks/install.sh
 
 init-dev:

requirements.txt

@@ -0,0 +1,9 @@
+autopep8==1.5.0
+GPUtil==1.4.0
+gym>=0.24.1
+jupyter==1.0.0
+jupyter-contrib-nbextensions==0.5.1
+jupyter-nbextensions-configurator==0.4.1
+mujoco>=2.2.0
+torchmeta>=1.8.0
+tqdm>=4.62.3

scripts/download-torch.py

@@ -0,0 +1,21 @@
+# -*- coding: utf-8 -*-
+import os
+import sys
+
+import GPUtil
+
+
+def main() -> None:
+ if sys.platform == "win32" or sys.platform == "linux":
+ if GPUtil.getAvailable():
+ cli = "pip install torch==1.8.1+cu111 torchvision==0.9.1+cu111 -f https://download.pytorch.org/whl/torch_stable.html"
+ else:
+ cli = "pip install torch==1.8.1+cpu torchvision==0.9.1+cpu -f https://download.pytorch.org/whl/torch_stable.html"
+ elif sys.platform == "darwin":
+ cli = "pip install torch==1.8.1 torchvision==0.9.1 -f https://download.pytorch.org/whl/torch_stable.html"
+ print(cli)
+ os.system(cli)
+
+
+if __name__ == "__main__":
+ main()

src/meta_rl/envs/__init__.py

@@ -26,4 +26,4 @@ def register_env_fn(filename):
 for file in os.listdir(os.path.dirname(__file__)):
  if file.endswith(".py") and not file.startswith("_"):
  module = file[: file.find(".py")]
- importlib.import_module("src.envs." + module)
+ importlib.import_module("meta_rl.envs." + module)

src/meta_rl/envs/half_cheetah.py

@@ -5,19 +5,25 @@
  https://github.com/katerakelly/oyster/blob/master/rlkit/envs/half_cheetah.py
 """
 
-from typing import List, Union
+from typing import List, Optional, Union
 
 import numpy as np
+from gym import utils
 from gym.envs.mujoco import HalfCheetahEnv as HalfCheetahEnv_
+from gym.envs.mujoco import mujoco_env
 
 
 class HalfCheetahEnv(HalfCheetahEnv_):
+ def __init__(self):
+ mujoco_env.MujocoEnv.__init__(self, "half_cheetah.xml", 5)
+ utils.EzPickle.__init__(self)
+
  def _get_obs(self) -> np.ndarray:
  return (
  np.concatenate(
  [
- self.sim.data.qpos.flat[1:],
- self.sim.data.qvel.flat,
+ self.data.qpos.flat[1:],
+ self.data.qvel.flat,
  self.get_body_com("torso").flat,
  ],
  )

src/meta_rl/envs/half_cheetah_dir.py

@@ -42,9 +42,9 @@ def __init__(self, num_tasks: int) -> None:
  super().__init__()
 
  def step(self, action: np.ndarray) -> Tuple[np.ndarray, np.float64, bool, Dict[str, Any]]:
- xposbefore = self.sim.data.qpos[0]
+ xposbefore = self.data.qpos[0]
  self.do_simulation(action, self.frame_skip)
- xposafter = self.sim.data.qpos[0]
+ xposafter = self.data.qpos[0]
 
  progress = (xposafter - xposbefore) / self.dt
  run_cost = self._goal_dir * progress

src/meta_rl/envs/half_cheetah_vel.py

@@ -42,9 +42,9 @@ def __init__(self, num_tasks: int) -> None:
  super().__init__()
 
  def step(self, action: np.ndarray) -> Tuple[np.ndarray, np.float64, bool, Dict[str, Any]]:
- xposbefore = self.sim.data.qpos[0]
+ xposbefore = self.data.qpos[0]
  self.do_simulation(action, self.frame_skip)
- xposafter = self.sim.data.qpos[0]
+ xposafter = self.data.qpos[0]
 
  progress = (xposafter - xposbefore) / self.dt
  run_cost = progress - self._goal_vel

src/meta_rl/rl2/rl2_trainer.py

@@ -33,7 +33,7 @@
  tasks: List[int] = env.get_all_task_idx()
 
  # Set a random seed
- env.seed(experiment_config["seed"])
+ env.reset(seed=experiment_config["seed"])
  np.random.seed(experiment_config["seed"])
  torch.manual_seed(experiment_config["seed"])