search

tkn-tub / ns3-gym

ns3-gym - The Playground for Reinforcement Learning in Networking Research
GNU General Public License v2.0
540 stars 199 forks source link

Basic example for connnecting ns3-gym with stable-baselines #11

Closed H-Park closed 3 years ago

H-Park commented 5 years ago

It would be very convenient to be able to use stable-baselines out of the box on the gym-esque environments in this project. Although the environment's API access mirrors OpenAI's gym, under the hood there are discrepancies that prevent integration with stable-baselines as is.

rlan commented 5 years ago

It would be very convenient to be able to use stable-baselines out of the box on the gym-esque environments in this project. Although the environment's API access mirrors OpenAI's gym, under the hood there are discrepancies that prevent integration with stable-baselines as is.

@H-Park I'm also looking into integrating them. Would you mind sharing what discrepancies you have found so the community could improve it? Thanks!

H-Park commented 5 years ago

Now that I think about it, I think this isn't an error on ns3-gym's end, it was an update to gym that caused the breakage, mostly their spaces.

Ns3-gym just has to update some stuff about spaces to start. I don't remember if there are other inconsistencies

sheila-janota commented 2 years ago

Hi @H-Park I am trying to integrate ns3gym with stable baseline PPO algorithm, but I'm facing some issues when trying to check the new custom environment which I'm creating. Did had any success when you tried to integrate ns3gym with stable baseline?

Thanks in advance.