Cannot run launch_experiments.py

[xlws1]

Hello, I'm a first-year graduate student. When running the launch_experiments.py file in the multiagent_particles/experiments folder, I encountered the following error: WARNING:tensorflow: The TensorFlow contrib module will not be included in TensorFlow 2.0. For more information, please see:

• https://github.com/tensorflow/community/blob/master/rfcs/20180907-contrib-sunset.md
• https://github.com/tensorflow/addons
• https://github.com/tensorflow/io (for I/O related ops) If you depend on functionality not listed there, please file an issue. Then I noticed an error in the fourth line of the def save_goal_agents(N, folder_name, agent_speeds): function: exp_name = f"{foldername}/{run}{behaviour}{str(speeds)}{n}" The error message indicated unresolved references to behaviour, speeds, and n. After modifying the code to: python def save_goal_agents(N, folder_name, agent_speeds): runs = ["run_10", "run_11", "run_12", "run_13", "run_14"] missing_agents_behaviours = ["idle", "random_player", "random"] # Ensure this list includes all required behaviors for n in range(N): for run in runs: for behaviour in missing_agents_behaviours: for speeds in agent_speeds:

  Convert each element in the speeds list to a string and separate them with commas

              speeds_str = ','.join(map(str, speeds))
              exp_name = f"{folder_name}/{run}_{behaviour}_{speeds_str}_{n}"
              fname = f"{exp_name}.csv"
              if not os.path.exists(fname):
                  command = f'python run.py --load-dir "saves/{run}/episode_200000/model" --exp-name {exp_name} --save-dir {folder_name} --rollout --num-episodes {N} --agent-speeds'
                  for speed in speeds:
                      command += f" {speed}"
                  subprocess.run(command, shell=True)

  There are no more error messages, but I still receive the warning: WARNING:tensorflow: The TensorFlow contrib module will not be included in TensorFlow 2.0. For more information, please see:

• https://github.com/tensorflow/community/blob/master/rfcs/20180907-contrib-sunset.md
• https://github.com/tensorflow/addons
• https://github.com/tensorflow/io (for I/O related ops) If you depend on functionality not listed there, please file an issue.

[Fabien-Couthouis]

Hi! The tensorflow version used in this work is tensorflow==1.15.0, please refer to the conda environment.yml file available here: https://github.com/Fabien-Couthouis/XAI-in-RL/blob/master/multiagent_particles/environment.yml for packages version.

Hope it helps. Tell me if you have other issues.

[xlws1]

Thank you very much for your reply, I directly used the command conda env create -n mparticles environment.yml to import the environment.yml file you provided to create the experimental environment for multi-agent particles. I believe the environment should not have any issues (Figure 1).

I used: python run.py --load-dir "saves/run_10/episode_200000/model" --missing-agents-behaviour "idle" --exp-name run_10_idle_0 --save-dir rewards2-ht/1 --shapley-M 1000 --num-episodes 1 to reproduce your code by adjusting the parameters --load-dir and --missing-agents-behaviour.

After saving the experimental results to the folder I created, /rewards2-ht/1, I wanted to plot the results using the command python plots.py rewards --plot_type your_plot_type.

When I entered python plots.py --result_dir rewards2-ht/1 --plot-type model_rewards --model-dir saves/run_10/, the resulting graph did not match the one you provided (Figure 2).

Why did the rewards\exp2 path appear? I suspect that I modified the parameters in the command line, but the code did not execute as expected.

[Fabien-Couthouis]

If you want to reproduce the shapley values results, use the launch_experiment script:

python launch_experiments.py

[xlws1]

I recreated the virtual environment and then ran the launch.experiment script. A new error has occurred:Traceback (most recent call last): File "E:\project\XAI-in-RL-master\multiagent_particles\experiments\launch_experiments.py", line 4, in from rollout import rollout File "E:\project\XAI-in-RL-master\multiagent_particles\experiments\rollout.py", line 8, in from maddpg.trainer.maddpg import MADDPGAgentTrainer File "E:\project\XAI-in-RL-master\multiagent_particles\experiments\maddpg\trainer\maddpg.py", line 6, in from maddpg.common.distributions import make_pdtype File "E:\project\XAI-in-RL-master\multiagent_particles\experiments\maddpg\common\distributions.py", line 5, in from multiagent.multi_discrete import MultiDiscrete ModuleNotFoundError: No module named 'multiagent'
What is the reason for this?

[xlws1]

I added multi-agent article env master to the system environment variable and resolved an error message that when I executed python launch.exe. py, it reported a series of Warnings but still did not run the experiment. I checked the files in the XAI in RL master \ multiagent_marticles directory and found three more places where an error message appeared before running pycharm. Does this affect the experiment? I am unable to resolve these error messages. Can you please help me?