Updated README. Deleted scripts under development from main branch. They are located in development branch.

README.md

@@ -29,10 +29,12 @@ This will pull configuration information from the specified agent and save it in
 This will record a video of a specified agent being evaluated. 
 Does not save any evaluation information. 
 Currently in major development. 
+Currently located in development branch.
 ### plot_evaluations.py
 This will plot the evaluation data that was gathered during the training run of the specified agent using MatPlotLib. 
 Charts can be saved to a directory of the user's choosing. 
 Currently in major development.
+Currently located in development branch.
 ### plot_improvement.py
 This plots the score of an agent averaged over all evaluation episodes during a training run. Also plots the 
 standard deviation. Removes the lowest and highest episode scores from each evaluation.

plot_evaluations.py

@@ -1,60 +0,0 @@
-from numpy import load
-import numpy as np
-import matplotlib.pyplot as plt 
-# import matplotlib.axes
-
-filepath = "agents/dqn_v2-7/evaluations.npz"
-
-data = load(filepath)
-lst = data.files # data.files lists the keys that are available for data
-
-# print('ep_lengths: \n', data['ep_lengths'])
-
-# results and ep_lengths are 2d arrays, because each evaluation is 5 episodes long.
-# I want to plot the average of each evaluation.
-print(data["results"])
-print()
-print(np.delete(np.sort(data["results"]), 0, 1))
-# for i in range(len(data["results"])):
-#     print(np.average(data["results"][i]))
-
-'''
-# for each item in results, loop through the array and save the average
-avg_ep_result_arr = []
-for eval in data['results']:
-    result_sum = 0
-    
-    for result in eval:
-        result_sum = result_sum + result
-        
-    avg_ep_result = result_sum / len(eval)
-    avg_ep_result_arr.append(avg_ep_result)
-    
-avg_ep_len_arr = []
-for eval in data['ep_lengths']:
-    max_len = 0
-    y_limit = 0
-    ep_len_sum = 0
-    
-    for ep_length in eval:
-        ep_len_sum = ep_len_sum + ep_length
-        if ep_length > max_len:
-            max_len = ep_length
-        if ep_length > y_limit and y_limit < max_len:
-            y_limit = ep_length
-        
-    avg_ep_len = ep_len_sum / len(eval)
-    avg_ep_len_arr.append(avg_ep_len)
-    y_limit = y_limit * 1.9
-    
-    
-x = plt.plot(data['timesteps'], avg_ep_result_arr)
-# plt.bar(data['timesteps'], avg_ep_len_arr, width=10000)
-y = plt.plot(data['timesteps'], avg_ep_len_arr)
-plt.ylim(top=y_limit)
-# plt.ylabel("Avg ep score")
-# lineObjects = plt.plot(x, y)
-plt.legend(["avg ep result", "avg ep length"])
-plt.title("result and length over steps\nfilepath: " + filepath)
-plt.show()
-'''

record_video.py

@@ -1,54 +0,0 @@
-import gymnasium as gym
-from stable_baselines3 import DQN
-# from stable_baselines3.common.monitor import Monitor
-from stable_baselines3.common.vec_env import VecVideoRecorder, DummyVecEnv, VecEnv
-
-model_name = "agents/dqn_v2-8/best_model" # path to model, should be an argument
-env_id = "ALE/Pacman-v5"
-video_folder = "videos/"
-video_length = 10000 #steps by hard coding this, I can almost ensure only one episode is recorded...
-
-vec_env = DummyVecEnv([lambda: gym.make(env_id, render_mode="rgb_array")])
-model = DQN.load(model_name)
-# output: <stable_baselines3.common.vec_env.dummy_vec_env.DummyVecEnv object at 0x0000029974DC6550>
-
-# vec_env = gym.make(env_id, render_mode="rgb_array")
-# output <OrderEnforcing<PassiveEnvChecker<AtariEnv<ALE/Pacman-v5>>>>
-
-# vec_env = Monitor(gym.make(env_id, render_mode="rgb_array"))
-
-print("\n\n\n")
-print(vec_env)
-print("\n\n\n")
-
-obs = vec_env.reset()
-
-
-# Record the video starting at the first step
-vec_env = VecVideoRecorder(vec_env,
-                           video_folder, 
-                           record_video_trigger=lambda x: x == 0,
-                           video_length=video_length,
-                           name_prefix="one-episode_v2-8_bestmodel"
-                           )
-#  Once I make the environment, now I need to walk through it...???
-#   I want to act according to the policy that has been trained
-vec_env.reset()
-print(vec_env)
-# for _ in range(video_length + 1):
-#   action, states = model.predict(obs)
-#   obs, _, _, _ = vec_env.step(action)
-
-# Instead of using the specified steps in a for loop
-# use a while loop to check if the episode has terminated
-# Stop recording when the episode ends
-end = True
-while end == True:
-  action, states = model.predict(obs)
-  obs, _, done, _ = vec_env.step(action)
-  if done == True:
-    print("exiting loop")
-    end = False
-# # Save the video
-
-vec_env.close()