NAF

[weiliu620]

I like your work of NAF.

Where is the code for this algorithm? Can you also give an example of command to run it?

And, do you have code for both alg. 1 and 2 in the paper?

Thanks,

p.s. I saw rllab/sandbox/rocky has some NAF code, but and not sure that's your implementation or not.

[rlbayes]

Hi, Thank you for your interest. To run NAF, you may run the following command. NAF is effectively the same code as DQN, but for continuous action. I unfortunately do not have the version with model-based acceleration. Also, note that this code is experimental. python algo_gym_stub.py --exp=<exp_name> --algo_name=dqn --env_name=<some_continuous_action_domain>

…

On Fri, Sep 8, 2017 at 6:44 PM, Wei Liu ***@***.***> wrote: I like your work of NAF. Where is the code for this algorithm? Can you also give an example of command to run it? And, do you have code for both alg. 1 and 2 in the paper? Thanks, p.s. I saw rllab/sandbox/rocky has some NAF code, but and not sure that's your implementation or not. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#4>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAU8fZACR-4q7vSKXSCOBiY_DHlzALaFks5sge15gaJpZM4PR2AR> .

-- *Shane Gu*