some problem in scripts directory related to two launch files

[ataeiamirhosein]

hello;
I am a master student of telecom in Politecnico di Milano under the supervision of prof. Giacomo Verticale and we work on this platform and edited some section that I described...

in file launch_agent.py I change some lines as below:

add from sacred.observers import MongoObserver after from sacred import Experiment for running like MongoDB observer.

in file launch_q_learn_umts_olpc.py I changed some lines as below:

• also, add from sacred.observers import MongoObserver after from sacred import Experiment for running like MongoDB observer.
• we have a problem with the done parameter in loop so change;

if done:
    break

to

if s == num_episodes:
    break

in the section of # Collect progress because done not execute and we ran into out dimension.

• in the section of # Plot results we actually do not need the first plot.figure()

thanks

[alvarovalcarce]

Hi,
yes, you can add MongoDB observers if you want.

What is exactly the problem with the done parameter?

[ataeiamirhosein]

actually in the file of launch_q_learn_umts_olpc.py when running the code

if done:
    break

not execute and not break the code. so, I get an error about the dimension control

[alvarovalcarce]

Seems to be related to max_steps_per_episode in that script. I've changed that. Can you please pull and try again?

[ataeiamirhosein]

sure
i will tell you asap

[ataeiamirhosein]

I try again and this time problem was solved
perfect
can you please describe about your changes?

[alvarovalcarce]

I just read avoided that t_max is read from the json configuration file by default (see the commit). You can, of course, change this if you need.

[ataeiamirhosein]

nice
yes, i got it
i wanna work on some new environment and agent
thanks a lot for your help