search

avivt / VIN

Value Iteration Networks
Other
288 stars 69 forks source link

why l_q equals 10 rather than 8? #13

Open husha1993 opened 6 years ago

husha1993 commented 6 years ago

Hi, https://github.com/avivt/VIN/blob/fe11bb1ae8ad9bcb3a02e2cc5e21b9499ccf0db4/vin.py#L30 why l_q equals 10 rather than 8? Thanks

avivt commented 6 years ago

It can be any number, 10 worked well. As we explain in the paper, there's no requirement that the number of actions in the MDP that we learn (and do VI over) matches the number of actions in the real MDP.

On Mon, Apr 2, 2018 at 12:32 AM, Sha Hu notifications@github.com wrote:

Hi, https://github.com/avivt/VIN/blob/fe11bb1ae8ad9bcb3a02e2cc5e21b9 499ccf0db4/vin.py#L30 why l_q equals 10 rather than 8? Thanks

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/avivt/VIN/issues/13, or mute the thread https://github.com/notifications/unsubscribe-auth/AOeQNS2DXTKvo_1at8y3zrAgUlADTIVwks5tkdQagaJpZM4TDP7g .