resume_local saving models too frequently

heronsystems / adeptRL

Reinforcement learning framework to accelerate research

GNU General Public License v3.0

204 stars 29 forks source link

resume_local saving models too frequently #31

Closed jtatusko closed 5 years ago

jtatusko commented 5 years ago

happening on a run on home computer restarting from 46M steps. Seems like it works fine restarting from step 32 when nb_env=32 or step 64 if nb_env = 64.

jtatusko commented 5 years ago

might be related to the code that logs the first case (ie. epoch @ 32 steps), reproduced on starcraft run. logs are created for: 4000032 4000064 4000096 4000128 4000160 5000000 6000000 7000000 8000000 9000000 10000000 11000000 12000000 13000000 14000000