New Python Notebook for StableBL Testing and fixed plotting

94781489 · josef azzam · f8bcc63c · 94781489 · 94781489 · 94781489
Commit 94781489 authored Feb 16, 2021 by josef azzam
--- a/src/deeprl_lib/stablebaselines/jupong2d_plot_ppo2_data.py
+++ b/src/deeprl_lib/stablebaselines/jupong2d_plot_ppo2_data.py
@@ -13,7 +13,7 @@ class JuPong2D_PPO2_Plot:
    def __init__(self, env, output):
        self.env_name = env
        self.output = f"{output}/{self.env_name}" 
-        self.folders = [f.path for f in os.scandir(self.output) if f.is_dir() and not f.name.startswith(".")]
+        self.folders = [f.path.replace("\\", "/") for f in os.scandir(self.output) if f.is_dir() and not f.name.startswith(".")]
    
    
    def plot_ball_speed(self):
@@ -47,10 +47,10 @@ class JuPong2D_PPO2_Plot:
            scale_factors = np.mean(scale_factor_matrix, axis=0)
            std_rewards_up = mean_rewards + std_vals
            std_rewards_down = mean_rewards - std_vals
-            plt.vlines(scale_factors, std_rewards_down, std_rewards_up, zorder = -1, linestyle="dashed")
+            plt.vlines(scale_factors, std_rewards_down, std_rewards_up, zorder = -1, linestyle="dashed", colors="black")
            plt.scatter(scale_factors, mean_rewards, label = bs_factor, s = 60)
-        plt.xlabel("Faktor der Ball-Geschwindigkeit", fontsize=20)
-        plt.ylabel("Mittl. Return-Wert mit\n Standardabweichung", fontsize=20)
+        plt.xlabel("Factor of the ball-speed", fontsize=20)
+        plt.ylabel("Mean Return-Values\n with std. deviation", fontsize=20)
        plt.xticks(fontsize=20)
        plt.yticks(fontsize=20)
        plt.legend(loc = "lower right", fontsize=12)
@@ -64,7 +64,6 @@ class JuPong2D_PPO2_Plot:
        return-values with their standard deviation.
        """
        pl_folders = [folder for folder in self.folders if folder.split("/")[-1].split("_")[-2] == "PaddleLength"]
-
        fig = plt.figure()
        for pl_folder in pl_folders:
            pl_factor = float(pl_folder.split("_")[-1])
@@ -89,10 +88,10 @@ class JuPong2D_PPO2_Plot:
            scale_factors = np.mean(scale_factor_matrix, axis=0)
            std_rewards_up = mean_rewards + std_vals
            std_rewards_down = mean_rewards - std_vals
-            plt.vlines(scale_factors, std_rewards_down, std_rewards_up, zorder = -1, linestyle="dashed")
+            plt.vlines(scale_factors, std_rewards_down, std_rewards_up, zorder = -1, linestyle="dashed", colors="black")
            plt.scatter(scale_factors, mean_rewards, label = pl_factor, s = 60)
-        plt.xlabel("Faktor der Paddle-Länge", fontsize=20)
-        plt.ylabel("Mittl. Return-Wert mit\n Standardabweichung", fontsize=20)
+        plt.xlabel("Factor of the paddle-length", fontsize=20)
+        plt.ylabel("Mean Return-Values\n with std. deviation", fontsize=20)
        plt.xticks(fontsize=20)
        plt.yticks(fontsize=20)
        plt.legend(loc = "lower right", fontsize=12)
@@ -131,10 +130,10 @@ class JuPong2D_PPO2_Plot:
            scale_factors = np.mean(scale_factor_matrix, axis=0)
            std_rewards_up = mean_rewards + std_vals
            std_rewards_down = mean_rewards - std_vals
-            plt.vlines(scale_factors, std_rewards_down, std_rewards_up, zorder = -1, linestyle="dashed")
+            plt.vlines(scale_factors, std_rewards_down, std_rewards_up, zorder = -1, linestyle="dashed", colors="black")
            plt.scatter(scale_factors, mean_rewards, label = ps_factor, s = 60)
-        plt.xlabel("Faktor der Paddle-Geschwindigkeit", fontsize=20)
-        plt.ylabel("Mittl. Return-Wert mit\n Standardabweichung", fontsize=20)
+        plt.xlabel("Factor of the paddle-speed", fontsize=20)
+        plt.ylabel("Mean Return-Values\n with std. deviation", fontsize=20)
        plt.xticks(fontsize=20)
        plt.yticks(fontsize=20)
        plt.legend(loc = "lower right", fontsize=12)

--- a/src/deeprl_lib/stablebaselines/jupong2d_ppo2.py
+++ b/src/deeprl_lib/stablebaselines/jupong2d_ppo2.py
@@ -248,7 +248,7 @@ class JuPong2D_PPO2_Play:
                        self.return_arr[ind] = np.mean(return_val_arr)

                        print(self.return_arr)
-                        with open(self.save_file, 'w') as my_file:
+                        with open(self.save_file, 'w', newline='') as my_file:
                            writer = csv.writer(my_file)
                            writer.writerow(self.scale_factor_arr)
                            writer.writerow(self.return_arr)

--- a/stablebl_local_test_bash.ipynb
+++ b/stablebl_local_test_bash.ipynb
@@ -70,15 +70,21 @@
 ],
 "metadata": {
  "kernelspec": {
-   "display_name": "Bash",
-   "language": "bash",
-   "name": "bash"
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
  },
  "language_info": {
-   "codemirror_mode": "shell",
-   "file_extension": ".sh",
-   "mimetype": "text/x-sh",
-   "name": "bash"
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.6.8"
  }
 },
 "nbformat": 4,

 %% Cell type:markdown id: tags:

 # **Jupyter Notebook to test the library Stable Baselines in the Gym-Environment JuPong2D**

 %% Cell type:code id: tags:

-``` bash
+``` python
 # Checking current directory, it should be the GitLab-Repository
 pwd
 ```

 %% Cell type:code id: tags:

-``` bash
+``` python
 # Activating the virtal environment, which should be installed before running the notebook
 source lib/pong_deeprl/bin/activate
 ```

 %% Cell type:code id: tags:

-``` bash
+``` python
 # Lists all possible command options of the Stable Baselines training and testing script
 deep_stablebl --help
 ```

 %% Cell type:code id: tags:

-``` bash
+``` python
 # Starts a simple training of the Gym-Environment JuPong2D
 deep_stablebl --pl 0.5 --session 1 --tts 10000 --train-steps 3 results/stablebl_results
 ```

 %% Cell type:code id: tags:

-``` bash
+``` python
 # Testing of the trained model by playing in different Gym-Environments, which depend on the parameters
 deep_stablebl --play --pl 0.5 --session 1 --play-steps 2 results/stablebl_results
 ```

 %% Cell type:code id: tags:

-``` bash
+``` python
 # Plots the training results in a PDF-file
 plot_stablebl results/stablebl_results
 ```

--- a/stablebl_local_test_python.ipynb
+++ b/stablebl_local_test_python.ipynb
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from src.deeprl_lib.stablebaselines.jupong2d_ppo2 import *\n",
+    "from src.deeprl_lib.stablebaselines.jupong2d_plot_ppo2_data import *\n",
+    "\n",
+    "def main():\n",
+    "    results_folder = \"results/stablebl_results\"\n",
+    "    env_name = \"jupong2d-headless-0.4-v3\"\n",
+    "    session = 3\n",
+    "    play_steps = 2\n",
+    "    train_steps = 2\n",
+    "    tts = 10000\n",
+    "    train_runner = JuPong2D_PPO2(env_name, results_folder, train_steps, tts, session, paddle_length_factor=0.5)\n",
+    "    train_runner.start_training()\n",
+    "    play_runner = JuPong2D_PPO2_Play(env_name, results_folder, session, play_steps, paddle_length_factor=0.5)\n",
+    "    play_runner.create_data()\n",
+    "    ploter = JuPong2D_PPO2_Plot(env_name, results_folder)\n",
+    "    ploter.plot_paddle_length()\n",
+    "    ploter.plot_paddle_speed()\n",
+    "    ploter.plot_ball_speed()\n",
+    "\n",
+    "main()"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.6.8"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}
+%% Cell type:code id: tags:
+
+``` python
+from src.deeprl_lib.stablebaselines.jupong2d_ppo2 import *
+from src.deeprl_lib.stablebaselines.jupong2d_plot_ppo2_data import *
+
+def main():
+    results_folder = "results/stablebl_results"
+    env_name = "jupong2d-headless-0.4-v3"
+    session = 3
+    play_steps = 2
+    train_steps = 2
+    tts = 10000
+    train_runner = JuPong2D_PPO2(env_name, results_folder, train_steps, tts, session, paddle_length_factor=0.5)
+    train_runner.start_training()
+    play_runner = JuPong2D_PPO2_Play(env_name, results_folder, session, play_steps, paddle_length_factor=0.5)
+    play_runner.create_data()
+    ploter = JuPong2D_PPO2_Plot(env_name, results_folder)
+    ploter.plot_paddle_length()
+    ploter.plot_paddle_speed()
+    ploter.plot_ball_speed()
+
+main()
+```