readme

README.md

@@ -3,10 +3,31 @@ Deep Reinforcement Learning in Pac-man
 
 ## Usage
 
-To train a network on 10000 games and play 100 games on a medium-sized map, run:
+Play 4000 games (3000 training, 1000 testing)
+
 ```
-python3 pacman.py -p PacmanDQN -n 10100 -x 10000 -l mediumClassic
+$ python3 pacman.py -p PacmanDQN -n 4000 -x 3000 -l smallGrid
 ```
+### Parameters
+
+Parameters can be found in the `pacmanDQN_Agents.py` file:
+
+Load and save filenames in /saves directory can be set using the
+`load_file` and `save_file` paramters
+
+Episodes before training starts: `train_start`
+Size of replay memory batch size: `batch_size`
+Amount of experience tuples in replay memory: `mem_size`
+Discount rate (gamma value): `discount`
+Learning rate: `lr`
+RMS Prop decay rate: `rms_decay`
+RMS Prop epsilon value: `rms_eps`
+
+Exploration/Exploitation (ε-greedy):
+Epsilon start value: `eps`
+Epsilon final value: `eps_final`
+Number of steps between start and final epsilon value (linear): `eps_step`
+
 
 ## Requirements
 
@@ -20,4 +41,3 @@ DQN Framework by  (made for ATARI / Arcade Learning Environment)
 
 Pac-man implementation by UC Berkeley:
 * [The Pac-man Projects - UC Berkeley](http://ai.berkeley.edu/project_overview.html) ([http://ai.berkeley.edu/project_overview.html](http://ai.berkeley.edu/project_overview.html))
-