enviroment problem

[paradox-jiatong]

Hello, I have a problem about environment.
I installed ray, based on pip3 install -r requirements/ray.txt
But I get the following AttributeError: module 'numpy' has no attribute 'bool'.
If I change the version of numpy I get the incompatibility error again.
Specifically as follows：
(RolloutWorker pid=6470) Opening file /tmp/extctrl_6470_act timed out
(RolloutWorker pid=6470) EnergyPlusEnv.step(): Failed to send an action. Quitting.
(RolloutWorker pid=6470) EnergyPlusEnv: (quit)
(RolloutWorker pid=6470) /root/miniconda3/lib/python3.8/site-packages/ray/rllib/utils/pre_checks/env.py:375: FutureWarning: In the future np.bool will be defined as the corresponding NumPy scalar. (This may have returned Python scalars in past versions.
(RolloutWorker pid=6470) elif not isinstance(done, (bool, np.bool, np.bool_)):
(RolloutWorker pid=6470) 2024-10-29 16:30:45,379 ERROR worker.py:449 -- Exception raised in creation task: The actor died because of an error raised in its creation task, ray::RolloutWorker.init() (pid=6470, ip=172.17.0.4, repr=<ray.rllib.evaluation.rollout_worker.RolloutWorker object at 0x7fd4d84d2f70>)
(RolloutWorker pid=6470) File "/root/miniconda3/lib/python3.8/site-packages/ray/rllib/evaluation/rollout_worker.py", line 507, in init
(RolloutWorker pid=6470) check_env(self.env)
(RolloutWorker pid=6470) File "/root/miniconda3/lib/python3.8/site-packages/ray/rllib/utils/pre_checks/env.py", line 65, in check_env
(RolloutWorker pid=6470) check_gym_environments(env)
(RolloutWorker pid=6470) File "/root/miniconda3/lib/python3.8/site-packages/ray/rllib/utils/pre_checks/env.py", line 178, in check_gym_environments
(RolloutWorker pid=6470) _check_done(done)
(RolloutWorker pid=6470) File "/root/miniconda3/lib/python3.8/site-packages/ray/rllib/utils/pre_checks/env.py", line 375, in check_done
(RolloutWorker pid=6470) elif not isinstance(done, (bool, np.bool, np.bool)):
(RolloutWorker pid=6470) File "/root/miniconda3/lib/python3.8/site-packages/numpy/init.py", line 284, in getattr
(RolloutWorker pid=6470) raise AttributeError("module {!r} has no attribute "
(RolloutWorker pid=6470) AttributeError: module 'numpy' has no attribute 'bool'

Thank you very much!

[antoine-galataud]

Hi @paradox-jiatong. Thanks for reporting this issue.
Have you tried downgrading numpy to 1.23.1 (current version is 1.24.1 in requirements/ray.txt)?

[paradox-ljt123]

Let me try！
Thank you！

[paradox-ljt123]

Hello, I have resolved the issue with the environment, but I am encountering a problem when trying to save the training data. The collected_data is [], and I’m not sure why. Could you please help me review the issue?
I would greatly appreciate your assistance.

def custom_callback(lcl, _glb):
    global collected_data
    
    ob = lcl.get('ob', None)                    
    action = lcl.get('ac', None)                
    reward = lcl.get('rew', None)               
    done = lcl.get('done', None)                
    next_ob = lcl.get('next_ob', None)          
    
    if ob is not None and action is not None and reward is not None and next_ob is not None:
        collected_data.append({
            "observation": ob,
            "action": action,
            "reward": reward,
            "next_observation": next_ob,
            "done": done
        })

    if len(collected_data) % 1000 == 0:
        print(f"Saving collected data ({len(collected_data)} entries)...")
        save_data(collected_data, filename="collected_data.pkl")

    return True