search

sherjilozair / dqn

Basic DQN implementation
MIT License
217 stars 72 forks source link

Adding video recording every 100 iterations and epsilon annealing #3

Open milesbrundage opened 8 years ago

milesbrundage commented 8 years ago

And some other small changes that can be left out. Epsilon annealing can be easily modified to go longer/shorter/faster/slower, and is currently set to go for 99 iterations and then fully exploit during recording every 100th iteration, but that can be removed as well if you want to see what it looks like during exploration.

Sohojoe commented 8 years ago

use env.monitor - it will periodically output video and also will record the stats and prep for uploading to OpenAi Gym

after env = gym.make(env_name) add env.monitor.start('/tmp/dqn', force=True) and then add this at the end of the env.monitor.close()

i can add a pull request if you want