Mujoco-py: numpy.ndarray size changed

[bebbo203]

Describe the bug
As for the title, I got this error message:
ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 88 from C header, got 80 from PyObject

It's solvable by updating numpy from 1.19.2 to 1.20.0 but for other dependecies (tensorflow), I can't do this

To Reproduce
Just run Mujoco with numpy 1.19.2 installed

Error Messages
Traceback (most recent call last): File "/home/bebbo203/anaconda3/envs/rl2/lib/python3.7/multiprocessing/process.py", line 297, in _bootstrap self.run() File "/home/bebbo203/anaconda3/envs/rl2/lib/python3.7/multiprocessing/process.py", line 99, in run self._target(*self._args, **self._kwargs) File "/home/bebbo203/Scrivania/Tesi/rnd_algo/RND/vec_env.py", line 210, in worker env = env_fn_wrapper.x() File "/home/bebbo203/Scrivania/Tesi/rnd_algo/RND/cmd_util.py", line 26, in _thunk env = gym.make(env_id) File "/home/bebbo203/anaconda3/envs/rl2/lib/python3.7/site-packages/gym/envs/registration.py", line 156, in make return registry.make(id, **kwargs) File "/home/bebbo203/anaconda3/envs/rl2/lib/python3.7/site-packages/gym/envs/registration.py", line 101, in make env = spec.make(**kwargs) File "/home/bebbo203/anaconda3/envs/rl2/lib/python3.7/site-packages/gym/envs/registration.py", line 72, in make cls = load(self.entry_point) File "/home/bebbo203/anaconda3/envs/rl2/lib/python3.7/site-packages/gym/envs/registration.py", line 17, in load mod = importlib.import_module(mod_name) File "/home/bebbo203/anaconda3/envs/rl2/lib/python3.7/importlib/__init__.py", line 127, in import_module return _bootstrap._gcd_import(name[level:], package, level) File "<frozen importlib._bootstrap>", line 1006, in _gcd_import File "<frozen importlib._bootstrap>", line 983, in _find_and_load File "<frozen importlib._bootstrap>", line 953, in _find_and_load_unlocked File "<frozen importlib._bootstrap>", line 219, in _call_with_frames_removed File "<frozen importlib._bootstrap>", line 1006, in _gcd_import File "<frozen importlib._bootstrap>", line 983, in _find_and_load File "<frozen importlib._bootstrap>", line 967, in _find_and_load_unlocked File "<frozen importlib._bootstrap>", line 677, in _load_unlocked File "<frozen importlib._bootstrap_external>", line 728, in exec_module File "<frozen importlib._bootstrap>", line 219, in _call_with_frames_removed File "/home/bebbo203/anaconda3/envs/rl2/lib/python3.7/site-packages/gym/envs/mujoco/__init__.py", line 1, in <module> from gym.envs.mujoco.mujoco_env import MujocoEnv File "/home/bebbo203/anaconda3/envs/rl2/lib/python3.7/site-packages/gym/envs/mujoco/mujoco_env.py", line 12, in <module> import mujoco_py File "/home/bebbo203/anaconda3/envs/rl2/lib/python3.7/site-packages/mujoco_py/__init__.py", line 3, in <module> from mujoco_py.builder import cymj, ignore_mujoco_warnings, functions, MujocoException File "/home/bebbo203/anaconda3/envs/rl2/lib/python3.7/site-packages/mujoco_py/builder.py", line 510, in <module> cymj = load_cython_ext(mujoco_path) File "/home/bebbo203/anaconda3/envs/rl2/lib/python3.7/site-packages/mujoco_py/builder.py", line 101, in load_cython_ext mod = load_dynamic_ext('cymj', cext_so_path) File "/home/bebbo203/anaconda3/envs/rl2/lib/python3.7/site-packages/mujoco_py/builder.py", line 125, in load_dynamic_ext return loader.load_module() File "cymj.pyx", line 1, in init mujoco_py.cymj ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 88 from C header, got 80 from PyObject

Desktop (please complete the following information):

• OS: Linux Mint 20.1
• Python Version 3.7.9
• mujoco-py version 2.0.2.9

Environment

• output of: echo $LD_LIBRARY_PATH :/usr/lib/nvidia-460:/home/bebbo203/.mujoco/mujoco200/bin
• output of: echo $HOME /home/bebbo203
• output of: echo $USER bebbo203

[Jiayuan-Gu]

I encountered the same issue. However, I thought the error message was quite clear, and led to this answer. Installing the latest numpy works, but I wonder what has changed in fact. I find that only numpy 1.20 and <=1.17 can work.

[bebbo203]

Thanks, the solution to make it works for numpy 1.19 was to use mujoco_py==2.0.2.5 or earlier

[samarth-robo]

If you need latest mujoco_py (e.g. 2.0.2.9 because of https://robotsuite.ai), with numpy 1.19.2 (because tensorflow does not accept numpy 1.20), you can use the following command to build the mujoco-py wheel locally after installing numpy

pip install mujoco-py==2.0.2.9 --no-cache-dir --no-binary :all: --no-build-isolation

with help from https://github.com/scikit-learn-contrib/hdbscan/issues/457#issuecomment-774033215