simple_dqn
simple_dqn copied to clipboard
tambetm
Simple deep Q-learning agent.
Hi, In Ubuntu 16.04.3 LTS cv2.x86_64-linux-gnu.so is the correct name instead of cv2.so. I believe this should be used instead: ``` sudo apt-get install python-opencv ln -s /usr/lib/python2.7/dist-packages/cv2.x86_64-linux-gnu.so NEON/.venv/lib/python2.7/site-packages/ ```...
I can get Breakout to play succesfully, pong however gives me this, any suggestions? ./play.sh snapshots/pong_141.pkl --replay_size 10 --backend cpu 2017-11-27 13:08:31,419 Using old model serialization format. Serialized the model...
I just tried the latest code, and found the training speed slowed down significantly, it used to be more than >200 steps_per_second, but right now it's ~100 steps_per_second 2017-09-24 15:08:08,844...
Hi, Thank you for the great project. While testing simple_dqn I found the test score of simple_dqn is different from the DeepMind paper. The DeepMind paper 'Prioritized Experience Replay' (http://arxiv.org/pdf/1511.05952v3.pdf)...
- Keep replay memory (screens, pre and post states) in gpu memory - Use transpose kernel to switch to chwn format - Training steps per second on breakout rom at...
./play.sh got the rom name wrong, it invokes: python src/main.py --play_games 1 --display_screen true --load_weights snapshots/space_invaders_200.pkl roms/space.bin while it should be space_invaders.bin