fix: update versions and readme

README.md

@@ -3,6 +3,23 @@
 PyTorch [official tutorial](https://pytorch.org/tutorials/intermediate/mario_rl_tutorial.html) to build an AI-powered Mario.
 
 ## Set Up
+
+### Poetry (Working as of Jan 2023)
+
+1. Install [poetry](https://python-poetry.org/docs/)
+2. Run the following in a terminal
+    ```
+    poetry config virtualenvs.in-project true
+    poetry env use python3.10
+    poetry install
+    ```
+3. Run mario
+    ```
+    poetry run python3 main.py
+    ```
+**NOTE** In the instructions below on running commands, prepend `poetry run python3` if using poetry.
+
+### Conda (Out of date dependencies)
 1. Install [conda](https://www.anaconda.com/products/individual)
 2. Install dependencies with `environment.yml`
     ```

agent.py

@@ -10,7 +10,7 @@ class Mario:
     def __init__(self, state_dim, action_dim, save_dir, checkpoint=None):
         self.state_dim = state_dim
         self.action_dim = action_dim
-        self.memory = deque(maxlen=100000)
+        self.memory = deque(maxlen=15000)
         self.batch_size = 32
 
         self.exploration_rate = 1
@@ -23,7 +23,7 @@ def __init__(self, state_dim, action_dim, save_dir, checkpoint=None):
         self.learn_every = 3   # no. of experiences between updates to Q_online
         self.sync_every = 1e4   # no. of experiences between Q_target & Q_online sync
 
-        self.save_every = 5e5   # no. of experiences between saving Mario Net
+        self.save_every = 5e3   # no. of experiences between saving Mario Net
         self.save_dir = save_dir
 
         self.use_cuda = torch.cuda.is_available()

main.py

@@ -14,7 +14,7 @@
 from wrappers import ResizeObservation, SkipFrame
 
 # Initialize Super Mario environment
-env = gym_super_mario_bros.make('SuperMarioBros-1-1-v0')
+env = gym_super_mario_bros.make('SuperMarioBros-1-1-v0', render_mode='human', apply_api_compatibility=True)
 
 # Limit the action-space to
 #   0. walk right
@@ -37,7 +37,7 @@
 save_dir = Path('checkpoints') / datetime.datetime.now().strftime('%Y-%m-%dT%H-%M-%S')
 save_dir.mkdir(parents=True)
 
-checkpoint = None # Path('checkpoints/2020-10-21T18-25-27/mario.chkpt')
+checkpoint = Path('trained_mario.chkpt')
 mario = Mario(state_dim=(4, 84, 84), action_dim=env.action_space.n, save_dir=save_dir, checkpoint=checkpoint)
 
 logger = MetricLogger(save_dir)
@@ -47,7 +47,7 @@
 ### for Loop that train the model num_episodes times by playing the game
 for e in range(episodes):
 
-    state = env.reset()
+    state, info = env.reset()
 
     # Play the game!
     while True:
@@ -59,7 +59,8 @@
         action = mario.act(state)
 
         # 5. Agent performs action
-        next_state, reward, done, info = env.step(action)
+        next_state, reward, truncated, terminated, info = env.step(action)
+        done = truncated or terminated
 
         # 6. Remember
         mario.cache(state, next_state, action, reward, done)

pyproject.toml

@@ -0,0 +1,21 @@
+[tool.poetry]
+name = "mad-mario"
+version = "0.1.0"
+description = ""
+authors = ["Rohit <[email protected]>"]
+readme = "README.md"
+packages = [{include = "mario_rl"}]
+
+[tool.poetry.dependencies]
+python = "^3.8"
+nes-py = "^8.2.1"
+torch = "^1.13.1"
+gym-super-mario-bros = "^7.4.0"
+matplotlib = "^3.6.3"
+scikit-image = "^0.19.3"
+opencv-python = "^4.7.0.68"
+
+
+[build-system]
+requires = ["poetry-core"]
+build-backend = "poetry.core.masonry.api"