In Jupyter Notebook error "ppo module not found"

[elanrosen]

When using the Jupyter Notebook and running the PPO algorithm.
The error I received was ppo module not found. My first thought would be to add something like sys.path.append('C:\Users\elanr\Desktop\ml-agents-master\python'). I am unfamiliar with the jupyter repository. Does anyone more experienced know the quick fix to this.

Also should i be connected my tensor-flow GPU environment to this or does tensor-flow automatically use it?

Thanks in advance to anyone who responds

[Maomao110]

I met the same problem with you .. Hope someone could help us solve the problem

[liuyihunter]

The easiest solution is that making sure executable and PPO.ipynb under same folder. if you are trying 3DBall,just put executable file into "/ml-agents-master/python/" and then change "env_name" to the name of your exe file.

[Maomao110]

Now , the exe file are in the same folder of the PPO.ipynb.
and I have checked the env_name, it is the name of my exe file .. But it still doesn't work.

[Maomao110]

[liuyihunter]

Literally, The error was No module name "ppo", Maybe check the file name or are there any files inside ppo folder?

[Maomao110]

@liuyihunter thanks for your help. I have solve the problem..Because I did not set the work folder in the jupyter . After I set the default folder as the python, it works.

[elanrosen]

@Maomao110 I probally have the exact same issue how do you do that

[Maomao110]

find this content in the PPO.ipynb , and then set the value . the value is where your ppo lib located . mine is :E:\StudyProject\ml-agents-master_unity2017\python\ppo

you can try , if it doesn't work , I will show you another way.

[elanrosen]

@Maomao110 Didn't work whats the other way?

I would think the work folder would need to be established before you import not after.

[vincentpierre]

The run_path variable is the name of the folder that will contain the tensorflow graph checkpoints. When you will run your model, it will store the checkpoints into python/models/ppo. The fact that the ppo folder containing the scripts for training and the run_path have the same name is coincidence.
Please make sure you start jupyter notebook from the python folder and also that the ppo folder inside python folder contains __init__.py, history.py, models.py and trainer.py
Let me know if this helps.

[elanrosen]

Wrong start folder
anyone who had my problem just look up how to change it on google. Pretty straightforward
Thanks

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.