Closed liuqi8827 closed 3 years ago
Hey,
thank you for the detailed description!
Just to check first, are you passing --fullObs
to visualize.py as well?
Thanks for your quick reply
You are right. I did not pass --fullObs
to visualize.py.
When I run python3 -m scripts.visualize --env MiniGrid-MultiRoom-N3r-v0 --model N3r-vib1e6 --fullObs
, it works well.
Thanks a lot!
Great! I'm closing this, but let me know if you have any other questions!
Hi Thanks your great work. I'm facing a problem of the visualization performence of the agent in Grid-world environment. I train the agent successfully. However, I found the agent learn nothing in Grid-world environment by visualize.py. The agent chose it's action rundamly. It never success!
python -m scripts.train --frames 100000000 --algo ppo --env MiniGrid-MultiRoom-N3r-v0 --model N3r-vib1e6 --save-interval 100 --tb --fullObs --model_type default2 --use_bottleneck --beta 0.000001
And I got the model parameter in the folder /storage/N3r-vib1e6python -m scripts.train --frames 100000000 --algo ppo --env MiniGrid-MultiRoom-N3r-v0 --model N3r-vibS1e6 --save-interval 100 --tb --fullObs --model_type default2 --use_bottleneck --beta 0.000001 --sni_type vib
And I got the model parameter in the folder /storage/N3r-vibS1e6plots.py
And I got the picture below:visualize.py
for N3r-vib1e6 and N3r-vibS1e6 respectively, I found the agent learn nothing in Grid-world environment. The agent chose it's action rundamly. It never success!Can you give some suggestions?
Thanks a lot!