This project is a CUDA implementation of the paper
| Playing Atari with Deep Reinforcement Learning | Volodymyr Mnih, Koray Kavukcuoglu, David Silver, Alex Graves, Ioannis Antonoglou, Daan Wierstra, Martin Riedmiller (NIPS workshop 2013) |
|---|---|
| However we have not got the performance reported by the paper yet. |
Yuan Gao & Yao Lu
rl-glue, SDL, SDL-image, CUDA, cublas
.. code-block::
cd ENV/src/dnq_gpu make cd ../.. make
For example, to play game Breakout
.. code-block::
./ale -player_agent deep_q_agent -display_screen true breakout.bin