dqn-openai-gym-cartpole-with-pytorch.ipynb

{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.7.12","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"<center>\n  <font size=6 color='DarkSlateBlue'>Deep Q-Network - Reinforcement Learning with PyTorch</font>\n</center>\n\n- <font color='gray'>Reinforcement Learning - Developing Intelligent Agents</font>\n- <font color='gray'>Deep Learning Course 5 of 6 - Level: Advanced</font>\n\n- Ref: [DeepLizard – Reinforcement Learning - Developing Intelligent Agents](https://www.youtube.com/playlist?list=PLZbbT5o_s2xoWNVdDudn51XM8lOuZ_Njv)","metadata":{"id":"a2gz2b4pfLK1"}},{"cell_type":"markdown","source":"# 📚<font size='5' color='CornflowerBlue'><a href='https://deeplizard.com/learn/video/FU-sNVew9ZA'>Project overview</a></font>","metadata":{"id":"XrD5m9RsgbxM"}},{"cell_type":"markdown","source":"We're going to be building and training a deep Q-network to learn to balance a pole on a moving cart. This is widely known as the cart and pole problem. ","metadata":{"id":"RC9N9A8CgdTn"}},{"cell_type":"markdown","source":"We'll be using **OpenAI's Gym** toolkit to set up our cart and pole environment. \n\n<center>\n<img src='https://deeplizard.com/assets/svg/ac9a374b.svg'/>\n<p>Image Snapped from <a href ='https://deeplizard.com/learn/video/FU-sNVew9ZA'>Deeplizard</a>\n</center>\n\n<center>\n<img src='https://static.packt-cdn.com/products/9781789345803/graphics/assets/9170409d-15f1-453b-816a-6f601a89fcf2.png'/>\n<p>Image Snapped from <a href ='https://subscription.packtpub.com/book/data/9781789345803/8/ch08lvl1sec46/introducing-cartpole-v1'>Packt – Hands-On Q-Learning with Python</a>\n</center>\n","metadata":{"id":"UpjN3YyTglEq"}},{"cell_type":"markdown","source":"### **[CartPole-v1](https://gym.openai.com/envs/CartPole-v1/)** \n\n> *A pole is attached by an un-actuated joint to a cart, which moves along a frictionless track. The system is controlled by applying a force of `+1` or `-1` to the cart. The pendulum starts upright, and the goal is to prevent it from falling over. A reward of `+1` is provided for every timestep that the pole remains upright. The episode ends when the pole is more than `15` degrees from vertical, or the cart moves more than `2.4` units from the center.*\n","metadata":{"id":"zxKyI8UNiF9H"}},{"cell_type":"markdown","source":"# <font size='5' color='CornflowerBlue'>🄠-Learning Algorithm</font>","metadata":{"id":"mWXXfK-KjPXb"}},{"cell_type":"markdown","source":"1. Initialize replay memory capacity.\n2. Initialize the **policy network** with random weights.\n3. Clone the **policy network**, and call it the **target network**.\n\n4. For each episode:\n  - Initialize the starting state.\n  - For each time step:\n    1. Select an action.\n      - *Via exploration or exploitation*\n    \n    2. Execute selected action in an emulator.\n    \n    3. Observe <u>reward</u> and <u>next state</u>.\n    \n    4. Store experience in <u>replay memory</u>.\n    \n    5. Sample random batch from replay memory.\n    \n    6. Preprocess states from batch.\n    \n    7. Pass batch of preprocessed states to **policy network**.\n    \n    8. <u>Calculate loss</u> between output Q-values and target Q-values.\n      - *Requires a pass to the **target network** for the next state*\n    \n    9. Gradient descent updates weights in the **policy network** to minimize loss.\n             \n      - *After `x` time steps, weights in the **target network** are <u>updated to the weights</u> in the **policy network**.*","metadata":{"id":"gl-6qRUPj85w"}},{"cell_type":"markdown","source":"# 📋<font size='5' color='CornflowerBlue'>Dependencies</font>","metadata":{"id":"xica6NcQv42E"}},{"cell_type":"code","source":"%%bash\n\n# !apt update\napt install -y python-opengl ffmpeg > /dev/null 2>&1\n# !apt install -y xvfb\npip install pyvirtualdisplay gym[atari] gym[box2d] gym[classic_control]\n# WORKAROUND: retry ---v\npip install gym[box2d]\npip install --no-deps baselines","metadata":{"_kg_hide-output":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# <font size='4' color='DeepSkyBlue'>(A) Libraries</font>","metadata":{"id":"BWAKBnqRlnPi"}},{"cell_type":"code","source":"import math\nimport random\nfrom typing import Text, Dict, List, NamedTuple, Tuple, Optional, Union\nfrom collections import namedtuple\nfrom itertools import count\nfrom PIL import Image\n\nimport gym\nimport numpy as np\nimport matplotlib\nimport matplotlib.pyplot as plt\nfrom tqdm.auto import trange\n\nimport torch\nimport torch.nn as nn\nimport torch.optim as optim\nimport torch.nn.functional as F\nimport torchvision.transforms as T \nfrom IPython import display","metadata":{"id":"sVeo5mKQiwZm","executionInfo":{"status":"ok","timestamp":1647793597755,"user_tz":-120,"elapsed":1268,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:06.835826Z","iopub.execute_input":"2022-03-21T17:35:06.836300Z","iopub.status.idle":"2022-03-21T17:35:07.628067Z","shell.execute_reply.started":"2022-03-21T17:35:06.836203Z","shell.execute_reply":"2022-03-21T17:35:07.627260Z"},"trusted":true},"execution_count":1,"outputs":[]},{"cell_type":"markdown","source":"**Displaying the plot inside the kernel**","metadata":{"id":"L2mO5B4omDVg"}},{"cell_type":"code","source":"is_ipython = 'inline' in matplotlib.get_backend()\nif is_ipython: from IPython import display","metadata":{"id":"YSita-NrmKyw","executionInfo":{"status":"ok","timestamp":1647793597756,"user_tz":-120,"elapsed":26,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:07.630331Z","iopub.execute_input":"2022-03-21T17:35:07.630575Z","iopub.status.idle":"2022-03-21T17:35:07.639895Z","shell.execute_reply.started":"2022-03-21T17:35:07.630541Z","shell.execute_reply":"2022-03-21T17:35:07.639006Z"},"trusted":true},"execution_count":2,"outputs":[]},{"cell_type":"markdown","source":"**Create a virtual display in the background**","metadata":{"id":"EQtV1E1rwQf5"}},{"cell_type":"code","source":"import pyvirtualdisplay\n\n\n_display = pyvirtualdisplay.Display(visible=0,  # use False with Xvfb\n                                    size=(400, 300))\n_display.start()","metadata":{"id":"JlVcRD-DwRRv","executionInfo":{"status":"ok","timestamp":1647793597758,"user_tz":-120,"elapsed":25,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"outputId":"e030ffba-178e-491a-d14f-a3573e0f31dd","execution":{"iopub.status.busy":"2022-03-21T17:35:07.641767Z","iopub.execute_input":"2022-03-21T17:35:07.642320Z","iopub.status.idle":"2022-03-21T17:35:07.759571Z","shell.execute_reply.started":"2022-03-21T17:35:07.642280Z","shell.execute_reply":"2022-03-21T17:35:07.758800Z"},"trusted":true},"execution_count":3,"outputs":[{"execution_count":3,"output_type":"execute_result","data":{"text/plain":"<pyvirtualdisplay.display.Display at 0x7f85cc9a5f50>"},"metadata":{}}]},{"cell_type":"markdown","source":"# ⚖<font size='5' color='CornflowerBlue'>CartPole Game Dependencies</font>","metadata":{"id":"xica6NcQv42E"}},{"cell_type":"markdown","source":"# <font size='4' color='DeepSkyBlue'>(A) Simple Deep Q-network</font>","metadata":{"id":"GA0KECzYmYUw"}},{"cell_type":"code","source":"class DQN(nn.Module):\n    # Improve the DQN: https://deeplizard.com/learn/video/ewRw996uevN#:~:text=%20the%20first%20change%20that%20we%20need%20to%20make%20is%20to%20our%20dqn%20class.%20\n    # def __init__(self, img_height, img_width):\n    def __init__(self, num_state_features):\n        \"\"\"\n        (UPDATE!)\n        | Now, we're passing the state returned by the Gym environment\n           as a 1-dimensional tensor with 4 elements:\n          Args:\n           num_state_features: [cart_position,\n                                cart_velocity,\n                                pole_angle,\n                                pole_velocity_at_tip]\n        \"\"\"\n        super().__init__()\n\n        # Initialize our layers\n        # self.fc1 = nn.Linear(in_features=img_height*img_width*3,\n        #                      out_features=24)\n        self.fc1 = nn.Linear(in_features=num_state_features,\n                             out_features=32)\n\n        self.fc2 = nn.Linear(in_features=32,\n                             out_features=64)\n        \n        self.fc3 = nn.Linear(in_features=64,\n                             out_features=128)\n        \n        self.out = nn.Linear(in_features=128,\n                             out_features=2) # Back to the Project overview, you can\n                                             # see that total possible movements\n                                             # the object can do is (<left, right>)\n    def forward(self, t):\n        # No Longer flatten the input\n        # t = t.flatten(start_dim=1) # starting from the channel matrics instead of batches\n        t = F.relu(self.fc1(t))\n        t = F.relu(self.fc2(t))\n        t = F.relu(self.fc3(t))\n        # t = F.relu(self.out(t))\n        t = self.out(t)\n        return t","metadata":{"id":"dYuzYQoYmfqf","executionInfo":{"status":"ok","timestamp":1647793597760,"user_tz":-120,"elapsed":20,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:07.761847Z","iopub.execute_input":"2022-03-21T17:35:07.762594Z","iopub.status.idle":"2022-03-21T17:35:07.773463Z","shell.execute_reply.started":"2022-03-21T17:35:07.762556Z","shell.execute_reply":"2022-03-21T17:35:07.772774Z"},"trusted":true},"execution_count":4,"outputs":[]},{"cell_type":"markdown","source":"# <font size='4' color='DeepSkyBlue'>(B) Experience Class</font>","metadata":{"id":"dZ7fDOsuo5yX"}},{"cell_type":"markdown","source":"We're going to build the **Experience class** from **replay memory** which what we will use to train our network.","metadata":{"id":"-snawpOGo-Xd"}},{"cell_type":"code","source":"Experience = namedtuple(\n    'Experience',\n    ('state', 'action', 'next_state', 'reward')\n)","metadata":{"id":"8fWr9l7mpLeM","executionInfo":{"status":"ok","timestamp":1647793597761,"user_tz":-120,"elapsed":20,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:07.774705Z","iopub.execute_input":"2022-03-21T17:35:07.775331Z","iopub.status.idle":"2022-03-21T17:35:07.783226Z","shell.execute_reply.started":"2022-03-21T17:35:07.775294Z","shell.execute_reply":"2022-03-21T17:35:07.782306Z"},"trusted":true},"execution_count":5,"outputs":[]},{"cell_type":"markdown","source":"**Display the experience display format**","metadata":{"id":"VuhMTA21pYWI"}},{"cell_type":"code","source":"e = Experience(2, 3, 1, 4)\ne","metadata":{"id":"55iEoP6RpW7M","executionInfo":{"status":"ok","timestamp":1647793597762,"user_tz":-120,"elapsed":20,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"outputId":"958cbc7f-fb55-4fd8-f563-102b23d258a2","execution":{"iopub.status.busy":"2022-03-21T17:35:07.784740Z","iopub.execute_input":"2022-03-21T17:35:07.785195Z","iopub.status.idle":"2022-03-21T17:35:07.794703Z","shell.execute_reply.started":"2022-03-21T17:35:07.785157Z","shell.execute_reply":"2022-03-21T17:35:07.793812Z"},"trusted":true},"execution_count":6,"outputs":[{"execution_count":6,"output_type":"execute_result","data":{"text/plain":"Experience(state=2, action=3, next_state=1, reward=4)"},"metadata":{}}]},{"cell_type":"markdown","source":"# <font size='4' color='DeepSkyBlue'>(C) Replay Memory</font>","metadata":{"id":"99WYkC1Spuer"}},{"cell_type":"code","source":"class ReplayMemeory:\n    def __init__(self, capacity):\n        self.capacity = capacity\n        self.memory = []\n        self.push_count = 0\n\n    def push(self, experience) -> None:\n        \"\"\"\n        | This function responsible for adding experience to the\n        | memory. Also used for sampling experiences from replay memory.\n          IF memory less than memory initialied capacity,\n            we're going to append inside the memory\n          IF NOT\n            we're going to begin push new experience onto the front\n            of memory overwriting the oldest experience.\n\n          Args:\n            experience\n        \"\"\"\n        if len(self.memory) < self.capacity:\n            self.memory.append(experience)\n        else:\n            self.memory[self.push_count % self.capacity] = experience\n            self.push_count += 1\n\n    def sample(self, batch_size: int):\n        \"\"\"Sample is equal to the `batch_size` sent to this function`\"\"\"\n        return random.sample(self.memory, batch_size)\n\n    def can_provide_sample(self, batch_size: int) -> bool:\n        \"\"\"\n        |It tells us whether or not we can sample from memory.\n          We call that the size of a sample we'll obtain from memory\n          will be equal to the batch size we use to train our network.\n          For example, suppose we only have 20 experiences in replay memory\n          and that our batch size is 50. Then, we will be unable to sample\n          because we do not have 20 experiences yet.\n        \"\"\"\n        return len(self.memory) >= batch_size","metadata":{"id":"lnN44WYupz7I","executionInfo":{"status":"ok","timestamp":1647793599290,"user_tz":-120,"elapsed":5,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"scrolled":true,"execution":{"iopub.status.busy":"2022-03-21T17:35:07.796286Z","iopub.execute_input":"2022-03-21T17:35:07.796728Z","iopub.status.idle":"2022-03-21T17:35:07.807965Z","shell.execute_reply.started":"2022-03-21T17:35:07.796694Z","shell.execute_reply":"2022-03-21T17:35:07.807173Z"},"trusted":true},"execution_count":7,"outputs":[]},{"cell_type":"markdown","source":"# <font size='4' color='DeepSkyBlue'>(D) Epsilon Greedy Strategy</font>","metadata":{"id":"wetBpDnqts6F"}},{"cell_type":"code","source":"class EpsilonGreedyStrategy:\n    def __init__(self, start, end, decay):\n        self.start = start\n        self.end = end\n        self.decay = decay\n\n    def get_exploration_rate(self, current_step: int) -> float:\n        return self.end + (self.start - self.end) *\\\n                math.exp(-1. * current_step * self.decay)","metadata":{"id":"79LdYvfGt0JZ","executionInfo":{"status":"ok","timestamp":1647793599920,"user_tz":-120,"elapsed":4,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:07.809496Z","iopub.execute_input":"2022-03-21T17:35:07.809754Z","iopub.status.idle":"2022-03-21T17:35:07.817266Z","shell.execute_reply.started":"2022-03-21T17:35:07.809721Z","shell.execute_reply":"2022-03-21T17:35:07.816440Z"},"trusted":true},"execution_count":8,"outputs":[]},{"cell_type":"markdown","source":"# <font size='4' color='DeepSkyBlue'>(E) Agent</font>","metadata":{"id":"PWU8t6-0vAmC"}},{"cell_type":"code","source":"class Agent:\n    def __init__(self, strategy, num_actions, device) -> None:\n        \"\"\"\n          strategy: represent an instance from EpsilonGreedyStrategy\n          num_actions: corresponding to how many possible actions\n                       can the agent take from a given state.\n        \"\"\"\n        self.strategy = strategy\n        self.num_actions = num_actions\n        self.current_step = 0\n        self.device = device\n\n    def select_action(self, state, policy_net) -> float:\n        \"\"\"\n        (UPDATE!)\n          The shape of our states have changed, we need to reshape the results\n           returned by the network by adding one additional dimension to\n           the tensor using the `unsqueeze()` function. \n        \"\"\"\n\n        epsilon_rate = self.strategy.get_exploration_rate(self.current_step)\n        self.current_step += 1\n\n        # select an action\n        if epsilon_rate > random.random(): # used random instead of uniform since\n                                               # I don't need to end up selecting 1\n                                               # at the begining.\n            action = random.randrange(self.num_actions)\n            return torch.tensor([action]).to(device=self.device) # explore\n        else:\n            with torch.no_grad(): # we don't need to compute the graph (aka. gradiant track)\n                                # we're just using the model for inference testing.\n                # return policy_net(state).argmax(dim=1).to(self.device) # exploit\n                return policy_net(state).\\\n                       unsqueeze(dim=0).\\\n                       argmax(dim=1).\\\n                       to(device=self.device) # exploit","metadata":{"id":"DnPV9cVgvPX1","executionInfo":{"status":"ok","timestamp":1647793600306,"user_tz":-120,"elapsed":5,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:07.819115Z","iopub.execute_input":"2022-03-21T17:35:07.819616Z","iopub.status.idle":"2022-03-21T17:35:07.830849Z","shell.execute_reply.started":"2022-03-21T17:35:07.819580Z","shell.execute_reply":"2022-03-21T17:35:07.830118Z"},"trusted":true},"execution_count":9,"outputs":[]},{"cell_type":"markdown","source":"# <font size='4' color='DeepSkyBlue'>(F) Environment Manager</font>","metadata":{"id":"g94_HjJ8iyiO"}},{"cell_type":"code","source":"class CartPoleEnvManager:\n    def __init__(self, device) -> None:\n        \"\"\"\n        (UPDATE!)\n          we're no longer dealing with screen frames to determine the environment\n           states, the class no longer has a `current_screen` attribute.\n           Instead, we now have a `current_state` attribute, which is initialized to None. \n        \"\"\"\n        self.device = device\n        self.env = gym.make('CartPole-v1').unwrapped # to access the tools that\n                                                    # we won't access in the normal intialization\n        self.env.reset() # reset the environment to get the initial observations\n        self.current_screen = None # No longer available\n        self.current_state = None\n        self.done = False\n\n    def reset(self) -> None:\n        \"\"\"\n        (UPDATE!)\n         Function now resets its `current_state` to the initial observation returned\n          by the Gym environment when its reset, rather than resetting the screen\n          as was implemented previously.\n        \"\"\"\n        self.current_state = self.env.reset()\n\n    def close(self) -> None:\n        self.env.close()\n\n    def render(self, mode='human'):\n        return self.env.render(mode)\n\n    def display(self):\n        return self.env.display()\n\n    def num_actions_available(self):\n        \"\"\"checking number of actions available in the environment –\n           we have two options [<left, right>]\n        \"\"\"\n        return self.env.action_space.n\n\n    def take_action(self, action: torch.Tensor) -> torch.Tensor:\n        \"\"\"\n        (UPDATE!)\n          We would not store the state returned by `env.step()` since we were not\n           making use of it. Now, since we're using the returned state, rather than\n           the screen frames, we store the state returned by `step()`\n           in the class's `current_state` attribute.\n        \"\"\"\n        # _, reward, self.done, _ = self.env.step(action.item()) # .item() is refer to our element inside the tensor\n        \n        self.current_state, reward, self.done, _ = self.env.step(action.item())\n        return torch.tensor([reward], device=self.device)\n\n    def get_state(self):\n        \"\"\"\n        (UPDATE!)\n          Previously, this function would return the pixel data that resulted from\n           the difference of the last two screen frames rendered in the environment.\n          Now, we simply return the state of the environment that we've stored in\n          the class's current_state attribute. \n        \"\"\"\n        if self.done:\n            return torch.zeros_like(\n              torch.tensor(self.current_state, device=self.device)\n            ).float()\n        else:\n            return torch.tensor(self.current_state, device=self.device).float()\n\n    def num_state_features(self):\n        \"\"\"\n        (NEW!)\n          Returns the number of features included in a state returned by the Gym environment.\n          This is so that we can know the size of states that will be passed to the\n          network as input. \n        \"\"\"\n        return self.env.observation_space.shape[0]\n    # ------------------------------------------------------------------------------\n    \"\"\"\n    (UPDATE!)\n     We no longer need any of the functions that we previously used for \n     screen processing, and so they can all be deleted. These include:\n      - get_screen_height()\n      - get_screen_width()\n      - get_processed_screen()\n      - crop_screen()\n      - transform_screen_data()\n      - just_starting()\n    \"\"\"\n    def image_reset(self) -> None:\n        self.env.reset()\n        self.current_screen = None # for resetting the screen to the first screen \n    \n    def get_image_state(self):\n        \"\"\"Return the current state of the environment in the form of\n           a processed image of the screen.\n        \"\"\"\n        if self.just_starting() or self.done:\n            # if we're at the start of at the end of the proccess, we initialize\n            # the image as a black screen.\n            self.current_screen = self.get_processed_screen()\n            black_screen = torch.zeros_like(self.current_screen)\n            return black_screen\n        else:\n            # current screen\n            s1 = self.current_screen\n            # call a new screen (the next screen)\n            s2 = self.get_processed_screen()\n            # make the next screen is the current screen\n            self.current_screen = s2\n            # return the difference between two screens to get the current state\n            return s2 - s1\n\n    def just_starting(self) -> bool:\n        \"\"\"Check whether the the current screen is None or not\n            This help to check if we're at the start of the environment or not!\n        \"\"\"\n        return self.current_screen is None\n\n    def get_screen_height(self) -> int:\n        screen = self.get_processed_screen()\n        return screen.shape[2]\n\n    def get_screen_width(self) -> int:\n        screen = self.get_processed_screen()\n        return screen.shape[3]\n\n    def get_processed_screen(self):\n        \"\"\"Get the image's color channels then transpose the channels\n        into the order of channels by height and width which what PyTorch DQN expect  \n        \"\"\"\n        screen = self.render('rgb_array').transpose((2, 0, 1))\n        screen = self.crop_screen(screen)\n        return self.transform_screen_data(screen)\n\n    def crop_screen(self, screen):\n        \"\"\"Except screen and will return a cropped version of it\"\"\"\n        screen_height = screen.shape[1]\n\n        # strip off top and bottom\n        top = int(screen_height * 0.4)\n        bottom = int(screen_height * 0.8)\n        # we cropped 40% of the top of the screen & 20% of the bottom of the screen\n        screen = screen[:, top:bottom, :]\n\n        return screen\n\n    def transform_screen_data(self, screen):\n        # Convert to float, rescale, convert to tensor\n        screen = np.ascontiguousarray(screen, # This array return contiguous array of the same as screen but transformed\n                                      dtype=np.float32) / 255 # these values here will be stored sequencially in the memory\n        \n        screen = torch.from_numpy(screen)\n\n        # Use `torchvision` package to compose several image transformations\n        resize = T.Compose([\n                           T.ToPILImage(), # Convert the image to pill image\n                           T.Resize((40,90)), # resize the image to 40x90\n                           T.ToTensor() # convert it to tensor\n        ])\n        # Unsqueeze should add another dimension which represents the batch dim\n        ## since the processed images will be passed to the dqn in batches.\n        return resize(screen).unsqueeze(0).to(self.device)\n  ","metadata":{"id":"D5a-5mRIi2MK","executionInfo":{"status":"ok","timestamp":1647793601577,"user_tz":-120,"elapsed":299,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:07.834823Z","iopub.execute_input":"2022-03-21T17:35:07.835432Z","iopub.status.idle":"2022-03-21T17:35:07.868098Z","shell.execute_reply.started":"2022-03-21T17:35:07.835392Z","shell.execute_reply":"2022-03-21T17:35:07.867288Z"},"trusted":true},"execution_count":10,"outputs":[]},{"cell_type":"markdown","source":"## Example of non-processed screen","metadata":{"id":"QkcaRZ0ml1U7"}},{"cell_type":"code","source":"device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\nimage_em = CartPoleEnvManager(device)\nimage_em.image_reset()\nscreen = image_em.render('rgb_array')\n\nplt.figure()\nplt.imshow(screen)\nplt.title('None-processed Screen Example')\nplt.show()","metadata":{"id":"25QX8ZJrtDQe","executionInfo":{"status":"ok","timestamp":1647793602915,"user_tz":-120,"elapsed":901,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"outputId":"be0b7803-85e3-47d7-d596-8da0165732ae","execution":{"iopub.status.busy":"2022-03-21T17:35:07.869682Z","iopub.execute_input":"2022-03-21T17:35:07.870246Z","iopub.status.idle":"2022-03-21T17:35:08.329091Z","shell.execute_reply.started":"2022-03-21T17:35:07.870198Z","shell.execute_reply":"2022-03-21T17:35:08.328310Z"},"trusted":true},"execution_count":11,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"## Example of processed screen","metadata":{"id":"KMGUEO1ezFfN"}},{"cell_type":"code","source":"screen = image_em.get_processed_screen()\n\nplt.figure()\nplt.imshow(screen.squeeze(0).permute(1,2,0).cpu(), interpolation='none')\nplt.title('Processed Screen Example')\nplt.show()","metadata":{"executionInfo":{"status":"ok","timestamp":1647793606125,"user_tz":-120,"elapsed":3215,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"outputId":"1572562a-5a45-412d-e9c6-a82affe90c48","id":"c_v9qxz-zFfS","execution":{"iopub.status.busy":"2022-03-21T17:35:08.330705Z","iopub.execute_input":"2022-03-21T17:35:08.331195Z","iopub.status.idle":"2022-03-21T17:35:10.339379Z","shell.execute_reply.started":"2022-03-21T17:35:08.331153Z","shell.execute_reply":"2022-03-21T17:35:10.338702Z"},"trusted":true},"execution_count":12,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"iVBORw0KGgoAAAANSUhEUgAAAXAAAADECAYAAACGNXroAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAAAUtElEQVR4nO3de5RdZXnH8e9vJlcSSECGkAslKCwQUAJEBKFSQBStCm21Qm0FjEVbXIJ1LQu2triqVl1qymqtQAXD8sJFlELxgoCoS5FLIlEuSQyXAAkJmUBCAoQkM/P0j/3OZO+TzMwhM3POvDO/z1p7zX73u8/Zz77Mc97z7stRRGBmZvlpaXYAZma2a5zAzcwy5QRuZpYpJ3Azs0w5gZuZZcoJ3MwsU07glg1JP5f0oWbHMdpICkkHNjsO25ET+AghaYWkzZJekPSMpAWSJjc7rkaRNFXSVZLWSNok6Q+SLmp2XH2RdImkbWmfdQ8bmh2X5cMJfGR5V0RMBo4C5gL/XDuDpDENj6ox5gOTgdcCU4B3A4+80jdpwva5LiIml4apDV6+ZcwJfASKiFXAj4HDoecr8PmSlgPL07S/lfSIpOck3SxpRvfrJR0m6bZU94ykT6XpLZIukvSopGclXS9pr1Q3QdK30/QNku6TNC3VnSPpsdQyflzS+0vL+qCkJZLWS7pV0v6lulMlLZX0vKT/AtTHar8B+G5ErI+IrohYGhE31LFOl0i6IcW+EThH0hRJV0paLWmVpM9Kaq0z5pD0EUnL03b4mqS+4t4pSW+StE7Sfql8RFreIancvR82SXpY0p+VXnuOpF9Lmp9ieCy93zmSnpK0VtLZpfkXSLosbZ9Nkn5RXqeauMZL+rKkJ9N2vEzSxFe6fjZIIsLDCBiAFcBb0vh+wEPAv6VyALcBewETgZOBdRQt9fHAfwK/TPPuDqwGPgFMSOU3proLgLuBWel1lwPXpLoPA/8H7Aa0AkcDewCTgI3AwWm+6cBhafx0ilbya4ExFN8Y7kp1ewObgPcAY4GPAx3Ah3pZ/2+kdT4XOKimrq91ugTYBpxB0aCZCNyY1m0SsA9wL/Dh/mIubetbgKnAHwHtwGm9xHwJ8O0+9unngJ+lmB4APlqqey8wI8X8PuBFYHqqOydtq3PTvvgs8CTwtbTf3pq27eQ0/4JUfnOqvxT4Vc06HZjG5wM3UxxLu6d9/u/NPv5H69D0ADwM0o4sEvgLwAbgCeC/gYmpLoCTS/NeCXypVJ6ckths4Czg/l6WsQQ4pVSenl43BvggcBfw+prXTEox/UV3PKW6HwPzSuUW4CVgf+ADwN2lOgEr6T2BTwQ+BSxKMT0CvD3V9bVOl5A+vFJ5GrClHGt6/Z39xVza1ieU6q8HLupj2VvT9uke7izVj03r8wDwE0B97P/FwOlp/BxgeanudSmuaaVpzwJz0vgC4Nqa46ET2K+0TgemffAi8JrSvMcBjzf7+B+tg7tQRpYzImJqROwfEX8fEZtLdU+VxmdQJHkAIuIFin/omRSt90d7ef/9gRvT1/INFAm9kyLpfQu4FbhW0tOSviRpbES8SNFC/AiwWtIPu7sB0vtdWnq/5yiSxMwUY0/MUWSL8jpURMTmiPh8RBwNvIoicX4vdfH0tU6122Z/isS5uhTX5RQt8f5i7ramNP4SRULszfVpn3UPJ5XWaRtFcj0c+EraBgBI+oCkxaU4Dqf41tLtmdL45vR+tdPKcZW39QtpvWZQ1UbxDWtRabk/SdOtCZzAR4/yYyefpkhEAEiaRJH0VlH8I7+6l/d4iqJVW044EyJiVURsi4jPRMShwJuAd1K0oomIWyPiVIoW+1Lgf0rv9+Ga95sYEXdRdHnsV4pR5XKfKxqxEfg8Rev/gH7WCarb5imKFvjepZj2iIjD6oh5UEmaCfwr8E3gK5LGp+n7U2zDjwKviuLE54P0fY6gP+VtPZmii+TpmnnWUST+w0rrPiWKE+fWBE7go9M1wLmS5qSk8HngnohYQdF/O13ShemE1e6S3phedxnwue4TXJLaJJ2exk+S9Lp0sm8jRTdGl6Rpkk5PHxJbKLp5ukrvd7Gkw9J7TJH03lT3Q+AwSX+u4sqQjwH79rZCkj4t6Q2SxkmaQNFfvwFY1s86VUTEauCnFAlzDxUnbl8j6cQ6Yh406QNrAUV31zyKD7R/S9WTKD502tO855JOWA/AOySdIGlcWs7dEVH5xhMRXRQfHPMl7ZOWPVPS2wa4bNtFTuCjUETcDnwa+D5FYngNcGaq2wScCryLoitgOdD9tf5SihNYP5W0ieKEZnci3Be4gSJ5LwF+QdGt0gL8A0Vr7jngRODv0rJuBL5I0e2ykaIV+fZUt47iRN0XKLp3DgJ+3ddqUbRU16VlnQr8aUS80M867cwHgHHAw8D6tF7T+4t5F71P1evAX0jJ8WMU3TafTl0n51J86P5xRDwMfAX4DUVXyevoe9vU47sUrf3nKE5A/3Uv8/0jxfmFu9P63w4cPMBl2y5SqVvNzEYhSQuAlRGxw30DNry5BW5mlikncDOzTLkLxcwsUwNqgUs6TdIyFbdkD+sHB5mZjTS73AJPl4v9geLs/krgPuCsdIbczMyG2ECevHYM8EhEPAYg6VqK50T0msD33nvvmD179gAWaWY2+ixatGhdROxwx+tAEvhMqrcgr2T7NcE7NXv2bBYuXDiARZqZjT6SntjZ9CG/CkXSeZIWSlrY3t4+1IszMxs1BpLAV1F9NsWsNK0iIq6IiLkRMbetzc+8MTMbLANJ4PcBB0k6ID0/4UyK26zNzKwBdrkPPCI6JH2U4hGircBVEfHQoEVmZmZ9GtDv/0XEj4AfDVIsZmb2CozUH7g1g5p7HLZt3lQ7Q8/Y2Il7VKte+c9YmjWcn4ViZpYpJ3Azs0w5gZuZZcp94DZidXVuq5SfvOu6an3Hlp7xV588r1LXOm7i0AVmNkjcAjczy5QTuJlZppzAzcwy5T5wG7lqrwN/6flKuXPr5u2zdnU2JCSzweQWuJlZppzAzcwy5QRuZpYp94HbyFXzPBO1tFbK0dVRGncfuOXHLXAzs0w5gZuZZcpdKDZiqaXaPmkdN6FS3rJxa894V2cHZrlxC9zMLFNO4GZmmXICNzPLlPvAbQSrtk/U4sPdRha3wM3MMuUEbmaWKSdwM7NMuVPQRi7VFGtvpa88bbb66FmzHLgFbmaWKSdwM7NMOYGbmWXKfeA2ekXX9vGurt7nMxum+m2BS7pK0lpJD5am7SXpNknL0989hzZMMzOrVU8XygLgtJppFwF3RMRBwB2pbGZmDdRvAo+IXwLP1Uw+Hbg6jV8NnDG4YZkNPkmVIaKzZ+jqqg5mOdjVk5jTImJ1Gl8DTBukeMzMrE4DvgolIoI+7oKQdJ6khZIWtre3D3RxZmaW7GoCf0bSdID0d21vM0bEFRExNyLmtrW17eLizMys1q5eRngzcDbwhfT3pkGLyGyQqOZX6VvH71Ypl39GratjK2a5qecywmuA3wAHS1opaR5F4j5V0nLgLalsZmYN1G8LPCLO6qXqlEGOxczMXgHfSm9mlinfSm8jWLUPvGXMuGp1+Xmy4cfJWn7cAjczy5QTuJlZptyFYqNG7S/yVLkLxfLjFriZWaacwM3MMuUEbmaWKfeB26hRe2t9+dLB8C/yWIbcAjczy5QTuJlZppzAzcwy5T5wG7Wi0gfe0cecZsOTW+BmZplyAjczy5QTuJlZptwHbqNG67jqT6pFbL/2u3Pby40Ox2zA3AI3M8uUE7iZWabchWKjRsuY8TVTfCu95c0tcDOzTDmBm5llygnczCxT7gO3UUOt1Z9UUy/zmeXCLXAzs0w5gZuZZcoJ3MwsU+4Dt1Gj9ifVojze1dnYYMwGQb8tcEn7SbpT0sOSHpJ0QZq+l6TbJC1Pf/cc+nDNzKxbPV0oHcAnIuJQ4FjgfEmHAhcBd0TEQcAdqWxmZg3SbwKPiNUR8ds0vglYAswETgeuTrNdDZwxRDGaDY3YPkRnR2Uwy8ErOokpaTZwJHAPMC0iVqeqNcC0wQ3NzMz6UncClzQZ+D5wYURsLNdF8eOC0cvrzpO0UNLC9vb2AQVrZmbb1ZXAJY2lSN7fiYgfpMnPSJqe6qcDa3f22oi4IiLmRsTctra2wYjZzMyo7yoUAVcCSyLiq6Wqm4Gz0/jZwE2DH57Z4GkZO6EySPQMHVtfqgxmOajnOvDjgb8BHpC0OE37FPAF4HpJ84AngL8ckgjNzGyn+k3gEfEren/uzymDG46ZmdXLt9KbmWXKt9LbqNEyZmx1gra3X3wrveXILXAzs0w5gZuZZcoJ3MwsU+4Dt1FDLbWHu39UzfLmFriZWaacwM3MMuUuFBs1an+Rp9yD4ssILUdugZuZZcoJ3MwsU07gZmaZch+4Gfhn1CxLboGbmWXKCdzMLFNO4GZmmXIfuI0aLa3Vw10t29svnVs3NzocswFzC9zMLFNO4GZmmXICNzPLlPvAbdRoGTOuUi4/XrazY0vN3FFT9qNnbfhxC9zMLFNO4GZmmXIXio0eaq0W5faL5c1HsJlZppzAzcwy5QRuZpYp94HbqFG+dT5N2T4atZcNmg1/boGbmWWq3wQuaYKkeyX9TtJDkj6Tph8g6R5Jj0i6TtK4/t7LzMwGTz0t8C3AyRFxBDAHOE3SscAXgfkRcSCwHpg3ZFGamdkO+u0Dj4gAXkjFsWkI4GTgr9L0q4FLgK8Pfohmvevq6qqU16xZ0zO+devWSl3nS89Vy52dPeObNj5fqVuxYkXNkqq30k+cOLFnfJ999qnOKd92b41RVx+4pFZJi4G1wG3Ao8CGiOj+IcGVwMxeXnuepIWSFra3tw9CyGZmBnUm8IjojIg5wCzgGOCQehcQEVdExNyImNvW1rZrUZqZ2Q5e0WWEEbFB0p3AccBUSWNSK3wWsGooAjTry/PPV7s+5s3bfipm6dKllbppUydWyp9830k9448tvqtSd9kFX62UO7uqlxmeeOKJPeOXX355pW78+PH9hW02KOq5CqVN0tQ0PhE4FVgC3Am8J812NnDTEMVoZmY7UU8LfDpwtaRWioR/fUTcIulh4FpJnwXuB64cwjjNzKxGPVeh/B44cifTH6PoDzczsybwrfSWtfKlgABPP/10z3jtpYAbp0yplBeuO6pn/NmNayp1Tz75vUp5W0dHpXzIIdvP44dvw7cm8a30ZmaZcgI3M8uUE7iZWabcB26jRke1u5zNHduv194WE2rm9u3wNvy5BW5mlikncDOzTDmBm5llqqF94C+//PIOz6cwG4j169dXylu2bOl13q1bX6qUF9979fa6LS9W6rq6ajrMa7z44vb5ly1bVqnzs1CsUdwCNzPLlBO4mVmmGtqF0traytSpUxu5SBvhan+Rp7W1tdd5X966rVL++W9u3+Xljhmz/V+n9ph2F4o1ilvgZmaZcgI3M8uUE7iZWaYa2gc+duxY9t1330Yu0ka4cl/0zspDpdzPPW3atErdhAm1t+WbDQ23wM3MMuUEbmaWKSdwM7NMOYGbmWXKCdzMLFNO4GZmmXICNzPLlH9SzbI2bty4Svn444/vGZ8xY8aQLffoo4/uGW9pcTvImsNHnplZppzAzcwy5S4Uy9ruu+9eKc+fP79nvPZRs4Op3G1S241j1ihugZuZZcoJ3MwsU07gZmaZUkQ0bmFSO/AEsDewrmELro9jqo9jqt9wjMsx1We4xbR/RLTVTmxoAu9ZqLQwIuY2fMF9cEz1cUz1G45xOab6DMeYdsZdKGZmmXICNzPLVLMS+BVNWm5fHFN9HFP9hmNcjqk+wzGmHTSlD9zMzAbOXShmZplqaAKXdJqkZZIekXRRI5ddE8dVktZKerA0bS9Jt0lanv7u2eCY9pN0p6SHJT0k6YJmxyVpgqR7Jf0uxfSZNP0ASfek/XidpIbfSy6pVdL9km4ZDjFJWiHpAUmLJS1M05p9TE2VdIOkpZKWSDpuGMR0cNpG3cNGSRcOg7g+no7xByVdk479ph/n/WlYApfUCnwNeDtwKHCWpEMbtfwaC4DTaqZdBNwREQcBd6RyI3UAn4iIQ4FjgfPT9mlmXFuAkyPiCGAOcJqkY4EvAvMj4kBgPTCvgTF1uwBYUioPh5hOiog5pcvPmn1MXQr8JCIOAY6g2F5NjSkilqVtNAc4GngJuLGZcUmaCXwMmBsRhwOtwJkMj2OqbxHRkAE4Dri1VL4YuLhRy99JPLOBB0vlZcD0ND4dWNas2FIMNwGnDpe4gN2A3wJvpLjBYczO9muDYplF8U9+MnALoGEQ0wpg75ppTdt3wBTgcdJ5ruEQ005ifCvw62bHBcwEngL2onjA3y3A25p9TNUzNLILpXsjdVuZpg0X0yJidRpfA0xrViCSZgNHAvc0O67UVbEYWAvcBjwKbIiIjjRLM/bjfwCfBLofN/iqYRBTAD+VtEjSeWlaM/fdAUA78M3U1fQNSZOaHFOtM4Fr0njT4oqIVcCXgSeB1cDzwCKaf0z1yycxdyKKj9ymXJ4jaTLwfeDCiNjY7LgiojOKr7uzgGOAQxq5/FqS3gmsjYhFzYxjJ06IiKMougjPl/TmcmUT9t0Y4Cjg6xFxJPAiNd0STT7OxwHvBr5XW9fouFJ/++kUH3ozgEns2MU6LDUyga8C9iuVZ6Vpw8UzkqYDpL9rGx2ApLEUyfs7EfGD4RIXQERsAO6k+Co5VVL3s+QbvR+PB94taQVwLUU3yqVNjqm7FUdErKXo0z2G5u67lcDKiLgnlW+gSOjD4nii+KD7bUQ8k8rNjOstwOMR0R4R24AfUBxnTT2m6tHIBH4fcFA6szuO4uvTzQ1cfn9uBs5O42dT9EE3jCQBVwJLIuKrwyEuSW2SpqbxiRR98ksoEvl7mhFTRFwcEbMiYjbFMfSziHh/M2OSNEnS7t3jFH27D9LEfRcRa4CnJB2cJp0CPNzMmGqcxfbuE2huXE8Cx0raLf0fdm+rph1TdWtkhzvwDuAPFP2o/9Ssjn+KA2c1sI2ipTKPoh/1DmA5cDuwV4NjOoHia+PvgcVpeEcz4wJeD9yfYnoQ+Jc0/dXAvcAjFF+BxzdpP/4JcEuzY0rL/l0aHuo+tofBMTUHWJj23/8CezY7phTXJOBZYEppWrO31WeApek4/xYwfrgc530NvhPTzCxTPolpZpYpJ3Azs0w5gZuZZcoJ3MwsU07gZmaZcgI3M8uUE7iZWaacwM3MMvX/Rup55fIGE/8AAAAASUVORK5CYII=\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"## Example of starting state","metadata":{"id":"zNf4fktvz8l0"}},{"cell_type":"code","source":"screen = image_em.get_image_state()\n\nplt.figure()\nplt.imshow(screen.squeeze(0).permute(1,2,0).cpu(), interpolation='none')\nplt.title('Example of starting state')\nplt.show()","metadata":{"executionInfo":{"status":"ok","timestamp":1647793606127,"user_tz":-120,"elapsed":31,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"outputId":"de61b298-07c9-4136-e98d-141202d2f73e","id":"Yyfo-Yviz8l3","execution":{"iopub.status.busy":"2022-03-21T17:35:10.340948Z","iopub.execute_input":"2022-03-21T17:35:10.342059Z","iopub.status.idle":"2022-03-21T17:35:10.531368Z","shell.execute_reply.started":"2022-03-21T17:35:10.342022Z","shell.execute_reply":"2022-03-21T17:35:10.530704Z"},"trusted":true},"execution_count":13,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"## Example of None-starting state","metadata":{"id":"2PHFrPhT0f7n"}},{"cell_type":"markdown","source":"This shape due to taking the differences between the first and the second screen","metadata":{"id":"xYgRaw-U06G6"}},{"cell_type":"code","source":"for i in range(5):\n    image_em.take_action(torch.tensor([1]))\nscreen = image_em.get_image_state()\n\nplt.figure()\nplt.imshow(screen.squeeze(0).permute(1,2,0).cpu(), interpolation='none')\nplt.title('Example of non-starting state')\nplt.show()","metadata":{"executionInfo":{"status":"ok","timestamp":1647793606497,"user_tz":-120,"elapsed":396,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"outputId":"c09a3635-889a-4e24-e4dd-587cf16311c8","id":"4G9oTbVi0f7t","execution":{"iopub.status.busy":"2022-03-21T17:35:10.532769Z","iopub.execute_input":"2022-03-21T17:35:10.533343Z","iopub.status.idle":"2022-03-21T17:35:10.723373Z","shell.execute_reply.started":"2022-03-21T17:35:10.533306Z","shell.execute_reply":"2022-03-21T17:35:10.722702Z"},"trusted":true},"execution_count":14,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"## Example of End state","metadata":{"id":"r18PfWMd0_Q7"}},{"cell_type":"code","source":"image_em.done = True\nscreen = image_em.get_image_state()\n\nplt.figure()\nplt.imshow(screen.squeeze(0).permute(1,2,0).cpu(), interpolation='none')\nplt.title('Example of End state')\nplt.show()","metadata":{"executionInfo":{"status":"ok","timestamp":1647793606498,"user_tz":-120,"elapsed":27,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"outputId":"70898a27-fdf0-42ad-cd59-0cdc18ef8d88","id":"nRYDKFkK0_Q_","execution":{"iopub.status.busy":"2022-03-21T17:35:10.724967Z","iopub.execute_input":"2022-03-21T17:35:10.725713Z","iopub.status.idle":"2022-03-21T17:35:10.918428Z","shell.execute_reply.started":"2022-03-21T17:35:10.725676Z","shell.execute_reply":"2022-03-21T17:35:10.917749Z"},"trusted":true},"execution_count":15,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"# <font size='4' color='DeepSkyBlue'>(G) Utility Functions</font>","metadata":{"id":"USlFNKpu2ewg"}},{"cell_type":"markdown","source":"### Tensor Processing","metadata":{"id":"jay2GNyBAqGT"}},{"cell_type":"code","source":"def extract_tensors(experiences: NamedTuple) -> Tuple[torch.TensorType]:\n    \"\"\"\n    | accepts a batch of Experiences and first transposes\n    it into an Experience of batches.\n    \"\"\"\n    # Convert batch of Experiences to Experience of batches\n    batch = Experience(*zip(*experiences))\n\n    \"\"\"\n    (UPDATE!)\n    In our `extract_tensors()` utility function, now that the states\n     are 1-dimensional, we need to `stack` the `states` and `next_states` along a new axis,\n     rather than concatenate (`cat`) them along an existing axis,\n     as we were doing previously when our states were 4-dimensional (`BCHW`). \n    \"\"\"\n    # t_states = torch.cat(batch.state)\n    t_states = torch.stack(batch.state)\n    t_actions = torch.cat(batch.action)\n    # t_next_state = torch.cat(batch.next_state)\n    t_next_state = torch.stack(batch.next_state)\n    t_rewards = torch.cat(batch.reward)\n\n\n    return (t_states,\n            t_actions,\n            t_next_state,\n            t_rewards)","metadata":{"id":"m4q07gs6AzNl","executionInfo":{"status":"ok","timestamp":1647793606500,"user_tz":-120,"elapsed":25,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:10.920044Z","iopub.execute_input":"2022-03-21T17:35:10.920751Z","iopub.status.idle":"2022-03-21T17:35:10.928311Z","shell.execute_reply.started":"2022-03-21T17:35:10.920713Z","shell.execute_reply":"2022-03-21T17:35:10.927326Z"},"trusted":true},"execution_count":16,"outputs":[]},{"cell_type":"markdown","source":"#### **Example of `Experience(*zip(*experiences))` used above**\n\n","metadata":{"id":"pW2-LRm7Byjw"}},{"cell_type":"code","source":"e1 = Experience(1,1,1,1)\ne2 = Experience(2,2,2,2)\ne3 = Experience(3,3,3,3)\n\nexperiences = [e1,e2,e3]\nexperiences","metadata":{"id":"F4osqmIZCHRg","executionInfo":{"status":"ok","timestamp":1647793606502,"user_tz":-120,"elapsed":26,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"outputId":"96c19048-3820-403c-add1-6d836592b3a1","execution":{"iopub.status.busy":"2022-03-21T17:35:10.929912Z","iopub.execute_input":"2022-03-21T17:35:10.930165Z","iopub.status.idle":"2022-03-21T17:35:10.943904Z","shell.execute_reply.started":"2022-03-21T17:35:10.930132Z","shell.execute_reply":"2022-03-21T17:35:10.943150Z"},"trusted":true},"execution_count":17,"outputs":[{"execution_count":17,"output_type":"execute_result","data":{"text/plain":"[Experience(state=1, action=1, next_state=1, reward=1),\n Experience(state=2, action=2, next_state=2, reward=2),\n Experience(state=3, action=3, next_state=3, reward=3)]"},"metadata":{}}]},{"cell_type":"markdown","source":"Now, let's illustrate that in a form of `batch` instead of having it a form of `samples`.","metadata":{"id":"Q6adbMdICNCs"}},{"cell_type":"code","source":"batch = Experience(*zip(*experiences))\nbatch","metadata":{"id":"8_nlGmjvDFwb","executionInfo":{"status":"ok","timestamp":1647793606503,"user_tz":-120,"elapsed":24,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"outputId":"82b4b569-9ac3-4f09-e71a-cce315967e1c","execution":{"iopub.status.busy":"2022-03-21T17:35:10.945492Z","iopub.execute_input":"2022-03-21T17:35:10.946400Z","iopub.status.idle":"2022-03-21T17:35:10.954538Z","shell.execute_reply.started":"2022-03-21T17:35:10.946362Z","shell.execute_reply":"2022-03-21T17:35:10.953681Z"},"trusted":true},"execution_count":18,"outputs":[{"execution_count":18,"output_type":"execute_result","data":{"text/plain":"Experience(state=(1, 2, 3), action=(1, 2, 3), next_state=(1, 2, 3), reward=(1, 2, 3))"},"metadata":{}}]},{"cell_type":"markdown","source":"### Q-Value Calculator","metadata":{"id":"2WeQoGE3FQMv"}},{"cell_type":"code","source":"class QValues:\n    device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n\n    @staticmethod\n    def get_current(policy_net, states, actions):\n        \"\"\"\n        | his function accepts a `policy_net`, `states`, and `actions`.\n        | When we call this function in our main program, recall that these `states`\n          and `actions` are the state-action pairs that were sampled from replay memory.\n          So, the states and actions correspond with each other. \n        \"\"\"\n        # Ensure that each prediction corresponting to each action in the batch\n        return policy_net(states).gather(dim=1, index=actions.unsqueeze(-1))\n\n    @staticmethod\n    def get_next(target_net, next_states):\n        \"\"\"\n        | This function accepts a `target_net` and `next_states`.\n        | Recall that for each next state, we want to obtain the maximum \n          q-value predicted by the `target_net` among all possible next actions. \n        | Link to the description: https://deeplizard.com/learn/video/ewRw996uevM#:~:text=this%20function%20accepts%20a%20target_net%20and%20next_states.%20recall%20that%20for%20each%20next%20state%2C%20we%20want%20to%20obtain%20the%20maximum%20q-value%20predicted%20by%20the%20target_net%20among%20all%20possible%20next%20actions.%20\n        \"\"\"\n        # find the locations of all the final states. If an episode is ended by a given action\n        ## we're finding the locations of these final states so that \n        ## we know not to pass them to the target_net for q-value predictions\n        ## when we pass our non-final next states.\n        final_states_location = next_states.flatten(start_dim=1)\\\n          .max(dim=1)[0].eq(0).type(torch.bool) #check each individual next state tensor to find its maximum value.\n          ## If its maximum value is equal to `0`, then we know that this particular next state\n          ## is a final state, and we represent that as a `True` within this `final_state_locations` tensor.\n          ## next_states that are not final are represented by a `False` value in the tensor.\n        non_final_states_locations = (final_states_location == False)\n\n        # Now that we know the locations of the non-final states,\n        ## we can now get the values of these states by indexing into the `next_states` tensor\n        ## and getting all of the corresponding non_final_states.\n        non_final_states = next_states[non_final_states_locations]\n\n        # Next, we find out the batch_size by checking to see how many next states\n        ## are in the `next_states` tensor. Using this, we create a new tensor of `zeros`\n        ## that has a length equal to the batch size.\n        batch_size = next_states.shape[0]\n        values = torch.zeros(batch_size).to(QValues.device)\n\n        # We then index into this tensor of zeros with the `non_final_state_locations`,\n        ## and we set the corresponding values for all of these locations\n        ## equal to the maximum predicted `q-values` from the `target_net` across each action.\n        values[non_final_states_locations] = target_net(non_final_states).max(dim=1)[0].detach()\n        # Ref: https://stackoverflow.com/a/56817594/14473118\n        ## torch.Tensor.detach() –  creates a tensor that shares storage with tensor that does not require grad.\n\n        # Return maximum predicted q-value across all actions for each non-final state.\n        ## (aka. Maximum Expected rate)\n        return values","metadata":{"id":"aSjONoyTFUGj","executionInfo":{"status":"ok","timestamp":1647793606504,"user_tz":-120,"elapsed":23,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:10.956166Z","iopub.execute_input":"2022-03-21T17:35:10.956716Z","iopub.status.idle":"2022-03-21T17:35:10.970415Z","shell.execute_reply.started":"2022-03-21T17:35:10.956676Z","shell.execute_reply":"2022-03-21T17:35:10.969597Z"},"trusted":true},"execution_count":19,"outputs":[]},{"cell_type":"markdown","source":"### Moving Average Plot","metadata":{"id":"OXY0tn3LAuPd"}},{"cell_type":"markdown","source":"To solve cart and pole, the average reward must be greater than or equal to `195` over <u>`100` consecutive episodes</u>. Recall that our agent gets a reward of `+1` for each step it takes that doesn't end the episode. So, the duration of an episode measured in timesteps is exactly equivalent to the reward for that episode. ","metadata":{"id":"s9OXFQKUAi-S"}},{"cell_type":"code","source":"def plot(values, # Episode Durations\n         moving_avg_period,\n         env=None): # 100 episodes moving average\n    moving_avg = get_moving_average(moving_avg_period, values)\n    # plt.figure(2)\n    figure, axes = plt.subplots(1, 2, figsize=(21, 6),\n                              gridspec_kw={'width_ratios': [4, 6],\n                                           'wspace':0.025, 'hspace':0.025})\n\n    plt.subplots_adjust(wspace=0, hspace=0)\n    # plt.clf()\n    axes[0].set_title('Training...')\n    axes[0].set_xlabel('Episode')\n    axes[0].set_ylabel('Duration')\n    axes[0].plot(values)\n    axes[0].plot(moving_avg)\n    if env:\n        axes[1].set_title('Cart Pole Rendering')\n        axes[1].grid(False)\n        axes[1].set_xticks([])\n        axes[1].set_yticks([])\n        axes[1].imshow(env.render('rgb_array'))\n    else:\n        axes[1].set_axis_off()\n    plt.pause(0.001)\n    print(f\"- Episodes: {len(values)}\\n- {moving_avg_period} episodes moving avg: {moving_avg[-1]}\")\n    if is_ipython: display.clear_output(wait=True)","metadata":{"id":"ypfOQTb35Nkl","executionInfo":{"status":"ok","timestamp":1647793642628,"user_tz":-120,"elapsed":274,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:10.972121Z","iopub.execute_input":"2022-03-21T17:35:10.972715Z","iopub.status.idle":"2022-03-21T17:35:10.985682Z","shell.execute_reply.started":"2022-03-21T17:35:10.972669Z","shell.execute_reply":"2022-03-21T17:35:10.985011Z"},"trusted":true},"execution_count":20,"outputs":[]},{"cell_type":"code","source":"def get_moving_average(period, values):\n    values = torch.tensor(values, dtype=torch.float)\n\n    if len(values) >= period: # we can't calculate moving average of a dataset\n                            ## when the dataset is not at least as large as the\n                            ## the period we want to caculate the moving average for\n\n        moving_avg = values.unfold(dimension=0, # which returns a tensor that contains \n                                   size=period, # with a size equal to the period \n                                   step=1)\\\n                                   .mean(dim=1).flatten(start_dim=0)\n                                   # containing all slices of 100 accross the original value tensor \n                                   ## taking the average of each of these slices\n                                                        ## and flatten the tensor so that now\n                                                        ## a moving average is equal to tensor\n                                                        ## containing all 100 period moving average.\n\n        # We then concatenate this resulting tensor to a tensor of zeros with a size\n        ## equal to `period-1`. This is to show that the moving average for the first \n        ## `period-1` values is zero given the explanation we just gave a moment ago.\n        ## So, if our period is `100`, then the first `99` values of the moving_avg\n        ## tensor will be `0`, and then each value afterwards will be the actual \n        ## calculated `100-period` moving average. \n        moving_avg = torch.cat((torch.zeros(period-1), moving_avg))\n        return moving_avg.numpy()\n    else:\n        # if the values is not bigger of equal to the period\n        moving_avg = torch.zeros(len(values))\n        return moving_avg.numpy()    ","metadata":{"id":"1NLbnvPn2mnR","executionInfo":{"status":"ok","timestamp":1647793643635,"user_tz":-120,"elapsed":7,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:10.987268Z","iopub.execute_input":"2022-03-21T17:35:10.987898Z","iopub.status.idle":"2022-03-21T17:35:10.999732Z","shell.execute_reply.started":"2022-03-21T17:35:10.987840Z","shell.execute_reply":"2022-03-21T17:35:10.998916Z"},"trusted":true},"execution_count":21,"outputs":[]},{"cell_type":"markdown","source":"**Example for plotting values**","metadata":{"id":"2YQMP2EfFTEi"}},{"cell_type":"code","source":"plot(np.random.rand(300), 100)","metadata":{"id":"PM9E1mctFZFZ","executionInfo":{"status":"ok","timestamp":1647793645809,"user_tz":-120,"elapsed":503,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"outputId":"91180fde-7d30-462f-a351-a674baf2892a","execution":{"iopub.status.busy":"2022-03-21T17:35:11.001206Z","iopub.execute_input":"2022-03-21T17:35:11.001898Z","iopub.status.idle":"2022-03-21T17:35:11.413007Z","shell.execute_reply.started":"2022-03-21T17:35:11.001845Z","shell.execute_reply":"2022-03-21T17:35:11.412154Z"},"trusted":true},"execution_count":22,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1512x432 with 2 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}},{"name":"stdout","text":"- Episodes: 300\n- 100 episodes moving avg: 0.5249742865562439\n","output_type":"stream"}]},{"cell_type":"markdown","source":" The actual values are plotted in blue, and the orange line is the `100-period` moving average across these values.\n\nWe can see that the `100-period` moving average is `0` for the first `99` values, and then we get the first calculated moving average at the `100th` value. This represents the average of the first `100` values in the array. If we skip over to the moving average at value `200`, then the orange line at this point represents the average of the second `100` values between value `100` to `200`. ","metadata":{"id":"pFbwI4SgF2zJ"}},{"cell_type":"markdown","source":"# <font size='4' color='DeepSkyBlue'>(H) Main Program</font>","metadata":{"id":"XBoGla3XtQfD"}},{"cell_type":"markdown","source":"### Global Hyperparameters","metadata":{"id":"c_7rdM_Au4ei"}},{"cell_type":"code","source":"batch_size = 256\ngamma = 0.999 # --> discounted rate\neps_start = 1 # --> Epsilon start\neps_end = 0.01 # --> Epsilon end\neps_decay = 0.001 # --> rate of Epsilon decay\ntarget_update = 10 # --> For every 10 episode, we're going to update \n                         ## the target network with the policy network weights\n# 1. Initialize replay memory capacity. \nmemory_size = 100000\nlr = 0.001\nnum_episodes = 1000","metadata":{"id":"Bx1YHqkUtYY4","executionInfo":{"status":"ok","timestamp":1647793662645,"user_tz":-120,"elapsed":285,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:11.414246Z","iopub.execute_input":"2022-03-21T17:35:11.414547Z","iopub.status.idle":"2022-03-21T17:35:11.420958Z","shell.execute_reply.started":"2022-03-21T17:35:11.414507Z","shell.execute_reply":"2022-03-21T17:35:11.420216Z"},"trusted":true},"execution_count":23,"outputs":[]},{"cell_type":"markdown","source":"### Main objects","metadata":{"id":"2JOrn_lru0pE"}},{"cell_type":"code","source":"device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\nem = CartPoleEnvManager(device)\nstrategy = EpsilonGreedyStrategy(eps_start, eps_end, eps_decay)\nagent = Agent(strategy, em.num_actions_available(), device)\nmemory = ReplayMemeory(memory_size)\n\n# 2. Initialize the policy network with random weights. \n# policy_net = DQN(em.get_screen_height(), em.get_screen_width()).to(device)\npolicy_net = DQN(em.num_state_features()).to(device)\n\n# 3. Clone the policy network, and call it the target network. \n# target_net = DQN(em.get_screen_height(), em.get_screen_width()).to(device)\ntarget_net = DQN(em.num_state_features()).to(device)","metadata":{"id":"stGR2MAQu_Qj","executionInfo":{"status":"ok","timestamp":1647793663631,"user_tz":-120,"elapsed":6,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:11.423096Z","iopub.execute_input":"2022-03-21T17:35:11.424182Z","iopub.status.idle":"2022-03-21T17:35:11.439994Z","shell.execute_reply.started":"2022-03-21T17:35:11.424140Z","shell.execute_reply":"2022-03-21T17:35:11.439106Z"},"trusted":true},"execution_count":24,"outputs":[]},{"cell_type":"markdown","source":"### Updating the policies","metadata":{"id":"RQy33yHRwMSl"}},{"cell_type":"markdown","source":"Get the `target_network` weights from `policy_network` wieghts","metadata":{"id":"Mjuir9N_wrt7"}},{"cell_type":"code","source":"target_net.load_state_dict(policy_net.state_dict())","metadata":{"id":"ZuD_fSr9wcr3","executionInfo":{"status":"ok","timestamp":1647793668946,"user_tz":-120,"elapsed":334,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"outputId":"3b3c733d-39e1-4621-af36-ee864695fcdd","execution":{"iopub.status.busy":"2022-03-21T17:35:11.441653Z","iopub.execute_input":"2022-03-21T17:35:11.442526Z","iopub.status.idle":"2022-03-21T17:35:11.451041Z","shell.execute_reply.started":"2022-03-21T17:35:11.442487Z","shell.execute_reply":"2022-03-21T17:35:11.450162Z"},"trusted":true},"execution_count":25,"outputs":[{"execution_count":25,"output_type":"execute_result","data":{"text/plain":"<All keys matched successfully>"},"metadata":{}}]},{"cell_type":"markdown","source":"Evaluate the `target_network` – also, we're telling the network that this network is not on the training mode, in other words it is only used in inference.","metadata":{"id":"tEqKAGozwxVE"}},{"cell_type":"code","source":"target_net.eval()","metadata":{"id":"B6vGxSw8w5M4","executionInfo":{"status":"ok","timestamp":1647793669318,"user_tz":-120,"elapsed":8,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"outputId":"030bee71-ede7-4b2c-b6e1-62292654deac","execution":{"iopub.status.busy":"2022-03-21T17:35:11.452911Z","iopub.execute_input":"2022-03-21T17:35:11.453721Z","iopub.status.idle":"2022-03-21T17:35:11.460631Z","shell.execute_reply.started":"2022-03-21T17:35:11.453682Z","shell.execute_reply":"2022-03-21T17:35:11.459718Z"},"trusted":true},"execution_count":26,"outputs":[{"execution_count":26,"output_type":"execute_result","data":{"text/plain":"DQN(\n  (fc1): Linear(in_features=4, out_features=32, bias=True)\n  (fc2): Linear(in_features=32, out_features=64, bias=True)\n  (fc3): Linear(in_features=64, out_features=128, bias=True)\n  (out): Linear(in_features=128, out_features=2, bias=True)\n)"},"metadata":{}}]},{"cell_type":"markdown","source":"Update the weights in the `policy_network`","metadata":{"id":"lsW2ol-zxND5"}},{"cell_type":"code","source":"optimizer = optim.Adam(params=policy_net.parameters(), lr=lr)","metadata":{"id":"7eHk6mRaxKm_","executionInfo":{"status":"ok","timestamp":1647793669319,"user_tz":-120,"elapsed":6,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:11.462605Z","iopub.execute_input":"2022-03-21T17:35:11.463570Z","iopub.status.idle":"2022-03-21T17:35:11.468872Z","shell.execute_reply.started":"2022-03-21T17:35:11.463532Z","shell.execute_reply":"2022-03-21T17:35:11.468050Z"},"trusted":true},"execution_count":27,"outputs":[]},{"cell_type":"markdown","source":"### Training Mode","metadata":{"id":"IkrDRT1hx59h"}},{"cell_type":"code","source":"episode_duration = []\n\n# 4. For each episode:\nfor episode in range(num_episodes):\n    em.reset()\n    # (4.1) Initialize the starting state.\n    state = em.get_state()\n    # (4.2) For each time step:\n    for timestep in count():\n        \"\"\"\n        (UPDATE!)\n          We've now added a line to our main program's nested timestep loop to `render()`\n           the environment to the screen. Previously, the call to `render()` was nested\n           inside one of the `CartPoleEnvManager`'s screen processing functions \n           `get_processed_screen()` that we no longer utilize. \n        \"\"\"\n        em.render()\n        # (4.2.1) Select an action.\n        action = agent.select_action(state, policy_net)\n        # (4.2.2) Execute selected action in an emulator.\n        # (4.2.3) Observe reward and next state.\n        reward = em.take_action(action)\n        next_states = em.get_state()\n        # (4.2.4) Store experience in replay memory.\n        memory.push(Experience(state, action, next_states, reward))\n        state = next_states\n\n        # (4.2.5) Sample random batch from replay memory.\n        if memory.can_provide_sample(batch_size):\n            experiences = memory.sample(batch_size)\n            # (4.2.6) Preprocess states from batch.\n            states, actions, next_states, rewards = extract_tensors(experiences)\n            # (4.2.7) Pass batch of preprocessed states to policy network.\n            # Get the current Q-values of the policy net and the next Q-values for the target nework \n            current_q_values = QValues.get_current(policy_net, states, actions)\n            next_q_values = QValues.get_next(target_net, next_states)\n            # calculate the target Q-values (aka. Bellman Optimality Equation)\n            target_q_values = rewards + (gamma * next_q_values)\n\n            # (4.2.8) Calculate loss between output Q-values and target Q-values.\n            loss = F.mse_loss(current_q_values, target_q_values.unsqueeze(1))\n\n            # (4.2.9) Gradient descent updates weights in the policy network to minimize loss.\n            ## Calculate the gradients of the loss\n            loss.backward()\n            ## Update the network with the gradients\n            optimizer.step()\n            ## Reset the gradients weights & biases before back propagation\n            optimizer.zero_grad()\n\n        # Check if the agent took the last action in the episode \n        if em.done:\n            episode_duration.append(timestep)\n            plot(episode_duration, 100, em) # plot the duration by 100 moving average\n            break\n\n    # (4.2.9.1) After *x* time steps,\n    ## weights in the target network are updated to the weights in the policy network.\n    if episode % target_update == 0:\n        target_net.load_state_dict(policy_net.state_dict())\n\n    \"\"\"\n    (UPDATE!)\n    We've added one line to the end of the episode loop to end the program\n     if the network solves the environment by reaching the `100-episode` average reward\n     of 195 or higher. \n    \"\"\"\n    if get_moving_average(100, episode_duration)[-1] >= 195:\n        break\n\nem.close()","metadata":{"id":"-kUpLzN-x9ZM","outputId":"61dd0cf0-c722-44ca-cd53-4caed15a8fe8","executionInfo":{"status":"error","timestamp":1647793740977,"user_tz":-120,"elapsed":69203,"user":{"displayName":"Ahmed N. A. Abdelatif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjSBOec51sSzTUjy93Pc3z-Ry2JiIji1dKOPvNlig=s64","userId":"04844253093007532238"}},"execution":{"iopub.status.busy":"2022-03-21T17:35:11.474198Z","iopub.execute_input":"2022-03-21T17:35:11.475037Z","iopub.status.idle":"2022-03-21T17:42:36.080210Z","shell.execute_reply.started":"2022-03-21T17:35:11.475000Z","shell.execute_reply":"2022-03-21T17:42:36.079458Z"},"trusted":true},"execution_count":28,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1512x432 with 2 Axes>","image/png":"iVBORw0KGgoAAAANSUhEUgAABGkAAAGDCAYAAACY803oAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAABsWElEQVR4nO3dd3ikZ3nv8d89Vb1rq7bvete9rQu2MQbTDAYTOiRgCDkOCYSckEI5SYCQhCQnDQIhEAwxJ4QSMMGYalzABrd1t7d7q3a1K2nVNZKmPeeP953RSBpJM9Ko7Or7ua69JL3zzryPZsfW6Lf3fT/mnBMAAAAAAAAWVmChFwAAAAAAAABCGgAAAAAAgEWBkAYAAAAAAGARIKQBAAAAAABYBAhpAAAAAAAAFgFCGgAAAAAAgEWAkAbArJnZj8zs5lKfCwAAcKYws3eZ2QMLvY7pmNl9ZvZbs7j/gJltLOWagKWEkAZYovwfoJk/aTMbyvn614t5LOfcDc6520p9LgAAwHTM7O1mtsN/D9Pm/4PQNbN4PGdmm6e4/V1mlvKv12dmT5rZjTO9XoFrus/Mhv1rdprZ7Wa2ci6vOVPOuSrn3IGFXgdwuiKkAZYo/wdolXOuStIRSa/JOfa1zHlmFlq4VQIAAEzOzD4o6Z8l/bWk5ZLWSvpXSTfN4LGKec/zoP8eqk7SrZK+ZWb1xV6zSO/3r7lZUpWkv5/j6xWF94xAaRDSABjDzK4zs1Yz+5CZnZD0FTOrN7M7zazDzLr9z1ty7pMti82U8prZ3/vnHjSzG2Z47gYz+4WZ9ZvZz8zsc2b2n/P4dAAAgEXKzGol/YWk9znnbnfODTrnEs657zvn/tg/53Ize9DMevwqm8+aWSTnMZyZvc/M9knaZ2a/8G96yq9aectUa3DOpSV9WVK5pE1mVmtmX/XfMx02sz81s7y/c5nZNjO7y8y6zGyPmb25kO/bOdcj6X8kXVTIY5nZf/jvoX7gv6d62Mw25dz+MjPbbWa9ZvZZSTZunb9pZrv892o/MbN1kz1/Occ2F3jtl/vr7TWzfzWzn9ssWq2AMwEhDYB8VkhqkLRO0i3y/l/xFf/rtZKGJH12ivtfIWmPpCZJfyfpVjOzGZz7X5IekdQo6eOS3jHj7wgAAJxpXiCpTNJ3pzgnJekP5L3PeIGk6yX97rhzXifv/cg5zrlr/WMX+tXF35xqAX71yG9JGpAXUvyLpFpJGyW9SNI7Jb07z/0qJd0l773OMklvlfSvZnbOVNfz79so6fWS9hfxWG+V9AlJ9f79/sq/b5Ok2yX9qbzn6HlJV+dc6yZJH/Wv1yzpfklfH7ek18l//iZZ8lTX/rakj8h7r7dH0lXTff/AmY6QBkA+aUkfc86NOOeGnHOnnHPfcc7FnHP98n64vmiK+x92zv27cy4l6TZJK+WVIBd8rpmtlXSZpD93zsWdcw9IuqNU3yAAADjtNUrqdM4lJzvBOfeYc+4h51zSOXdI0hc08T3Mp5xzXc65oSKufaWZ9Ug6Ieltkn5NXlDzVkkfcc71+9f7B+X/R6YbJR1yzn3FX9sTkr4j6U1TXPMzZtYrqVNeoPJ7RTzWd51zj/jP1dc0WoXzKknPOee+7ZxLyGsdO5Fzv/fKe352+ff9a0kX5VbTaPrnb7pr3+7f9plx1waWJEIaAPl0OOeGM1+YWYWZfcEv2+2T9AtJdWYWnOT+2R+wzrmY/2lVkeeuktSVc0ySjhb5fQAAgDPXKUlNU81CMbOz/DbtE/57mL+WF3Dkmsn7i4ecc3XOuSbn3JXOuZ/5jxuWdDjnvMOSVue5/zpJV/htWD1+4PPr8qqZJ/MB51ytpAvkVaVkWs8Leazc8COm0fdlq5Tz/TvnnMY+H+skfTrncbvktUPlfk/TPX/FXLt1mscCzniENADyceO+/kNJWyVd4ZyrkZQpBZ6shakU2iQ1mFlFzrE1c3g9AABwenlQ0oi8dpvJfF7Sbklb/PcwH9XE9y/j3/fMVKekhLxgI2OtpGN5zj0q6ed+0JP5U+Wc+53pLuKce0bSX0r6nN8iPuPHkvd+K/v+yn+83PdbRyX99rjHLnfO/Sp3SQVcZ7Jr5844tNyvgaWKkAZAIarlzaHpMbMGSR+b6ws65w5L2iHp42YWMbMXSHrNXF8XAACcHpxzvZL+XF5Y8Tq/8jdsZjeY2d/5p1VL6pM0YGbbJBUSXJyUN1Om2PWkJH1L0l+ZWbXfEvRBSfk2PbhT0llm9g5/zWEzu8zMzi7wcrfJayV/7Swf6weSzjWz1/sVSR/Q2Aqcf5P0ETM7V/KGNZvZVC1ZxfiBpPP9v7uQpPdp6koiYEkgpAFQiH+Wt2tBp6SHJP14nq776/KG/J2S9y9G35T3L2aSJH/XhRf6n7/QzAZybvuomf1ontYJAAAWgHPuH+QFIX8qqUNe5cf75e1+JEl/JOntkvol/bu89xLT+bik2/wWn4J2XMrxe5IGJR2Q9IC8Yb5fzrPufkkvlzfD5ri8lqC/lRQt5CLOubikT0v6s9k8lnOuU97smr+R935ri6Rf5tz+Xf+xvuG3iz0r6YY8D1W0nGv/nX/tc+T9A93IVPcDznTmtf4BwOJnZt+UtNs5N+eVPAAAAJg/5m1V3irp151z9y70eoCFQiUNgEXLL9XdZGYBM3ulpJs0+i9jAAAAOI2Z2SvMrM7MohqdF/TQAi8LWFCTTkIHgEVghaTb5W2x2Srpd/xtJQEAAHD6e4G8lrCIpJ2SXlfkVujAGYd2JwAAAAAAgEWAdicAAAAAAIBFgJAGAAAAAABgETgjZ9I0NTW59evXL/QyAAA4bTz22GOdzrnmhV4HSof3QwAALF6Tvfc6I0Oa9evXa8eOHQu9DAAAThtmdnih14DS4v0QAACL12TvvWh3AgAAAAAAWAQIaQAAAAAAABYBQhoAAAAAAIBFgJAGAAAAAABgESCkAQAAAAAAWAQIaQAAAAAAABYBQhoAAAAAAIBFgJAGAAAAAABgESCkAQAAAAAAWAQIaQAAAAAAABYBQhoAAAAAAIBFgJAGAFAyPbG42vuHF3oZAAAAwGmJkAYAUDIfv+M5feDrTyz0MgAAAIDTEiENAKBkumIJ9cQSC70MAAAA4LRESAMAKJlEMq1k2i30MgAAAIDTEiENAKBkkum0UoQ0AAAAwIwQ0gAASiaRckqk0gu9DAAAAOC0REgDACgZKmkAAACAmSOkAQCUTDLlmEkDAAAAzBAhDQCgZBIpKmkAAACAmSKkAQCUTDLNTBoAAABgpghpAAAlk0hSSQMAAADMFCENAKBkEmlm0gAAAAAzRUgDACiZJDNpAAAAgBkjpAEAlEwy5ZRKOzlHUAMAAAAUi5AGAFAyibQ3NJiWJwAAAKB4hDQAgJJJprxwhpYnAAAAoHiENACAknBudGgwlTQAAABA8QhpAAAlkUiNBjOpFCENAAAAUCxCGgBASST9eTTS6GwaAAAAAIUjpAEAlMSYShranQAAAICiEdIAAEoimRqtnmEmDQAAAFA8QhoAQEnkBjPMpAEAAACKR0gDACiJeDK3koaZNAAAAECxCGkAACWRW0lDuxMAAABQPEIaAEBJjJlJQ7sTAAAAUDRCGgBASbC7EwAAADA7hDQAgJLInUPDTBoAAACgeIQ0AICSyK2kYSYNAAAAULw5C2nM7Mtm1m5mz+YcazCzu8xsn/+x3j9uZvYZM9tvZk+b2SU597nZP3+fmd08V+sFAMxOgpk0AAAAwKzMZSXNf0h65bhjH5Z0t3Nui6S7/a8l6QZJW/w/t0j6vOSFOpI+JukKSZdL+lgm2AEALC5JZtIAAAAAszJnIY1z7heSusYdvknSbf7nt0l6Xc7xrzrPQ5LqzGylpFdIuss51+Wc65Z0lyYGPwCARSDBTBoAAABgVuZ7Js1y51yb//kJScv9z1dLOppzXqt/bLLjE5jZLWa2w8x2dHR0lHbVAIBpUUkDAAAAzM6CDQ52zjlJJXsX75z7onNuu3Nue3Nzc6keFgBQoGTOTJoEM2kAAACAos13SHPSb2OS/7HdP35M0pqc81r8Y5MdBwAsMok0lTQAAADAbMx3SHOHpMwOTTdL+l7O8Xf6uzxdKanXb4v6iaSXm1m9PzD45f4xAMAik1tJw0waAAAAoHihuXpgM/u6pOskNZlZq7xdmv5G0rfM7D2SDkt6s3/6DyW9StJ+STFJ75Yk51yXmX1S0qP+eX/hnBs/jBgAsAjkbsFNJQ0AAABQvDkLaZxzb5vkpuvznOskvW+Sx/mypC+XcGkAgDmQO4cmSUgDAAAAFG3BBgcDAM4sY9qdGBwMAAAAFI2QBgBQEskxg4OZSQMAAAAUi5AGAFAStDsBAAAAs0NIAwAoiSSDgwEAAIBZIaQBAJRE7u5OCWbSAAAAAEUjpAEAlESCmTQAAADArBDSAABKIplKqyzs/VhhJg0AAABQPEIaAEBJJFJOkWBAZsykAQAAAGaCkAYAUBLJdFrhYEDhQIBKGgAAAGAGCGkAACWRTDmFgqZgwMbs9AQAAACgMIQ0AICSiKfSCgUCCgWMShoAAABgBghpAAAlkUw5RUIBBYPGTBoAAABgBghpAAAlkUynFQoYlTQAAADADBHSAABKIpFyCgUDCgUCzKQBAAAAZoCQBgBQEslUWuHM4GAqaQAAAICiEdIAAEoimXZeuxMzaQAAAIAZIaQBAJREPJlWKBigkgYAAACYIUIaAEBJJNNOkWBA4UBAqRQhDQAAAFAsQhoAQEkkU2mFsjNpGBwMAAAAFIuQBgBQEomUUygQUChIuxMAAAAwE4Q0AICSSKZHd3dicDAAAABQPEIaAEBJJFNOoWBAoYApyUwaAAAAoGiENACAkkik0woHTKFAgJk0AAAAwAwQ0gAASiKRdAoHmUkDAAAAzBQhDQCgJJLp0d2dmEkDAAAAFI+QBgBQEomUX0nDTBoAAABgRghpAAAlkUylFfJn0lBJAwAAABSPkAYAUBKJtLe7UzBoSjA4GAAAACgaIQ0AoCSSqbTCQVOImTQAAADAjBDSAABmLZV2SjspFAgoyEwaAAAAYEYIaQAAs5ZIee1N4ZApzEwaAAAAYEYIaQAAs5b0Q5lwwJtJk2QmDQAAAFA0QhoAwKwl/UqakD+TJkklDQAAAFA0QhoAwKwl/Bk0oaA3kybFTBoAAACgaIQ0AIBZy7Q3hQNU0gAAAAAzRUgDAJi1RHK0kiYUZHAwAAAAMBOENACAWUtkKmn8mTQJBgcDAAAARSOkAQDMWtKfQRP2Z9I4J6WppgEAAACKQkgDAJi1RGZ3J38mjSTm0gAAAABFIqQBAMxaJpAJ+zNpJDGXBgAAACgSIQ0AYNaSmUqaYG4lDXNpAAAAgGIQ0gAAZi2ebXfyZtJIo3NqAAAAABSGkAYAMGuZQCYSYiYNAAAAMFOENACAWcu0NnmVNMykAQAAAGZiQUIaM/sDM3vOzJ41s6+bWZmZbTCzh81sv5l908wi/rlR/+v9/u3rF2LNAIDJJfxKmlDQFAoykwYAAACYiXkPacxstaQPSNrunDtPUlDSWyX9raR/cs5tltQt6T3+Xd4jqds//k/+eQCARSTT7hQOBkbbnZhJAwAAABRlodqdQpLKzSwkqUJSm6SXSPq2f/ttkl7nf36T/7X82683M5u/pQIApjPa7mSjg4NpdwIAAACKMu8hjXPumKS/l3REXjjTK+kxST3OuaR/Wquk1f7nqyUd9e+b9M9vHP+4ZnaLme0wsx0dHR1z+00AAMZIjKmkYSYNAAAAMBML0e5UL686ZoOkVZIqJb1yto/rnPuic267c257c3PzbB8OAFCEhL8FdzgYYCYNAAAAMEML0e70UkkHnXMdzrmEpNslXS2pzm9/kqQWScf8z49JWiNJ/u21kk7N75IBAFNJ+iFNKDi6BTeVNAAAAEBxFiKkOSLpSjOr8GfLXC9pp6R7Jb3RP+dmSd/zP7/D/1r+7fc453jnDwCLSLbdKRDIzqRJMDgYAAAAKMpCzKR5WN4A4MclPeOv4YuSPiTpg2a2X97MmVv9u9wqqdE//kFJH57vNQMAppYdHBw0ZtIAAAAAMxSa/pTSc859TNLHxh0+IOnyPOcOS3rTfKwLADAzmaqZUNCYSQMAAADM0EJtwQ0AOIMkc9qdmEkDAAAAzAwhDQBg1hKptAImBQKWnUmTZCYNAAAAUBRCGgDArCXSaYWD3o+UzEyaJJU0AAAAQFEIaQAAs5ZMuWxIE8y2OzGTBgAAACgGIQ0AYNaSqXR2YHA4OziYShoAAACgGIQ0AIBZS6Rdts0pyOBgAAAAYEYIaQAAs5ZMpbMVNJmwJsHgYAAAAKAohDQAgFlLpFy23SkYZCYNAAAAMBOENACAWUukRnd3CgeYSQMAAADMBCENAGDWkimnMDNpAAAAgFkhpAEAzFoyPbq7EzNpAAAAgJkhpAEAzJo3k8avpGEmDQAAADAjhDQAgFlLptPZWTQhZtIAAAAAM0JIAwCYtUTS5bQ7+ZU0tDsBAAAARSGkAQDMWiI9urtTkEoaAAAAYEYIaQAAs5ZMuWxIY2YKBkxJZtIAAAAARSGkAQDMWiKVzrY5SfJDGippAAAAgGIQ0gAAZi2ZHq2kkaRwwJhJAwAAABSJkAYAMGvJVDo7OFiikgYAAACYCUIaAMCsJVJOocDoj5RQMMBMGgAAAKBIhDQAgFlLpNKKhMZW0qSopAEAAACKQkgDAJi1ZHpsJU04YEoykwYAAAAoCiENAGDWEuNn0gSppAEAAACKRUgDAJi1ZGrs7k6hQIDBwQAAAECRCGkAALOWTKcVCozf3YnBwQAAAEAxCGkAALPinPN2dxpTScNMGgAAAKBYhDQAgFnJtDWFcyppQsykAQAAAIpGSAMAmJVMxUw4NPojJchMGgAAAKBohDQAgFlJ+LNncmfShJhJAwAAABSNkAYAMCvZSppgbiUNM2kAAACAYhHSAABmJZnyK2mCo5U0YWbSAAAAAEUjpAEAzEoiOziYmTQAAADAbBDSAABmJZGcWEkTClBJAwAAABSLkAYAMCuZAcHjZ9IkUgwOBgAAAIpBSAMAmJVEdnAwM2kAAACA2SCkAQDMSmYXp9C4mTSENAAAAEBxQgu9AADA6S2Rzj+ThsHBAAAAWOp6jjyjRKy34PMJaQAAs5LMtjuNnUlDJQ0AAACWurYnf6yBtn0Fn0+7EwBgVjIDgkOBsZU0DA4GAAAAikNIAwCYlUwYEw6N/kgJMTgYAAAAKBohDQBgVrLtTjmDg0OBADNpAAAAgCIVNJPGzJol/S9J63Pv45z7zblZFgDgdJHMMziYmTQAAABA8QodHPw9SfdL+pmk1NwtBwBwuklkBwczkwYAAACYjUJDmgrn3IfmdCUAgNNStpImwEwaAAAAYDYKnUlzp5m9qlQXNbM6M/u2me02s11m9gIzazCzu8xsn/+x3j/XzOwzZrbfzJ42s0tKtQ4AwOxlKmnGtjt5M2mcI6gBAAAAClVoSPP78oKaYTPr9//0zeK6n5b0Y+fcNkkXStol6cOS7nbObZF0t/+1JN0gaYv/5xZJn5/FdQEAJZZpa4oEcwcHe4ENxTQAAABA4QoKaZxz1c65gHOuzP+82jlXM5MLmlmtpGsl3eo/dtw51yPpJkm3+afdJul1/uc3Sfqq8zwkqc7MVs7k2gCA0ktmK2lGf6QE/ZAm0woFAAAAYHqFzqSRmb1WXrgiSfc55+6c4TU3SOqQ9BUzu1DSY/IqdZY759r8c05IWu5/vlrS0Zz7t/rH2gQAWHCZSprQuMHBkhfgRAv+SQMAAAAsbQVV0pjZ38gLUnb6f37fzD41w2uGJF0i6fPOuYslDWq0tUmS5LwhBkUVyZvZLWa2w8x2dHR0zHBpAIBiJf2epvCYwcGBMbcBAAAAmF6hM2leJellzrkvO+e+LOmVkl49w2u2Smp1zj3sf/1teaHNyUwbk/+x3b/9mKQ1Ofdv8Y+N4Zz7onNuu3Nue3Nz8wyXBgAoVnKKShp2eAIAAMBSFqlskMymP9FXaEgjSXU5n9cWcb8xnHMnJB01s63+oevlVefcIelm/9jNkr7nf36HpHf6uzxdKak3py0KALDAsrs7BXJ3d2ImDQAAAFC39jxZoPD+/0LP/JSkJ8zsXkkmbzbNh6e+y5R+T9LXzCwi6YCkd8sLjL5lZu+RdFjSm/1zfyivkme/pJh/LgBgkUik0goFTGb5Z9IAAAAAS5UFQzKzgue5FBTSOOe+bmb3SbrMP/QhvyJmRpxzT0ranuem6/Oc6yS9b6bXAgDMrWTaKRwcW5iZmUlDuxMAAACWskAwXNz5U91oZtv8j5dIWil/noykVf4xAMASl0ilx8yjkXIqaQhpAAAAsIRZkSHNdJU0H5R0i6R/yHObk/SSoq4GADjjJFMTK2mC2cHBzKQBAADA0hUIhuRNjSnMlCGNc+4W/9MbnHPDubeZWVnRqwMAnHGS6fSYocESlTQAAACA5FfSFJ7RFLy7068KPAYAWGISeSppMjNpGBwMAACApayklTRmtkLSaknlZnZxziPXSKqY4RoBAGcQZtIAAAAA+Vmw8O23peln0rxC0rsktUj6x5zj/ZI+WtSVAABnJGbSAAAAAPkFghGZlW4mzW2SbjOzNzjnvjPbxQEAzjyJ1BQzaWh3AgAAwBIWKHEljSTJOfcdM3u1pHMlleUc/4uirgYAOOMk05NX0tDuBAAAgKWs2HanggYHm9m/SXqLpN+TN5fmTZLWFbs4AMCZJ+9MmszgYEIaAAAALGGBYFjFDA4udHenq5xz75TU7Zz7hKQXSDqr+OUBAM40yZRTODBudydm0gAAAABeJc0cbME97H+MmdkqSQlJK4tbGgDgTJSvkibITBoAAACgqKHBUuEhzffNrE7S/5X0uKRDkv6rqCsBACbVO5RQW+/QQi9jRhJ5ZtJkQpsU7U4AAABAwaYNacwsIOlu51yPv8PTOknbnHN/PuerA4Al4v/+ZLdu/vIjC72MGUmm0gqPn0njtz8lCGkAAACAgk0b0jjn0pI+l/P1iHOud05XBQBLzLHuIbX3jyz0MmYkmXLZUCaDmTQAAABA8Qptd7rbzN5gxTZTAQAK0jOU0MBwUs5NrDz5+d4O3fLVHXlvWwwS6cUzk+bp1h5957HWeb0mAAAAUCqFhjS/Lem/JY2YWZ+Z9ZtZ3xyuCwCWlN5YQsm000hyYuXJg8+f0k93ntRwYnFWpSRTi2cmzVcfPKw//96zizbQAgAAAKYSKuQk51z1XC8EAJaynqGEJGlgJKmycHDMbQMj3m39IwmVR4IT7rvQkql0tr0pY6Fm0gwMJzUYT6l/JKmasvC8XhsAAACYrYJCGjO7Nt9x59wvSrscAFh60mmnnlhckhcyNFVFx9w+MJyUJA2OpKRFGJnHU07h0CQzaVLzW/0zMOI9Vyd6hwlpAAAAcNopKKSR9Mc5n5dJulzSY5JeUvIVAcAS0z+SVKbgJBMyjLndD2kyYc1ik0ynFR5XSRP0252S811J4z9/bb3DOmv5Iky0AAAAgCkU2u70mtyvzWyNpH+eiwUBwFLTG0tkP88b0vjH+kcSE25bDJIpp9D4mTSBhZlJkw1peobm9boAAABAKRQ6OHi8Vklnl3IhALBU9QzFs5/nq5YZ0+60CCVSU+zutAAzaSSvkgYAAAA43RQ6k+ZfJGXeaQckXSTp8TlaEwAsKT05lTSD8TwhjV8dMrBYK2nSTuHA2Mw/8/V8b8E9mDOTBgAAAFgcbPpTfIXOpNmR83lS0tedc78sZkkAgPwyOztJo/NncmVDmkU4kyaddkql3YRKmkDAZCal0vM3ODiddhrwQ662PkIaAAAALA7Lz32xjj9+Z0HnFjqT5jYza/Y/75jF2gAA42R2dpLyz6TJhDMDi7DdKeGHMOHgxO7ZUMDmtd0plkjJ+ZdjJg0AAAAWi0A4Ov1JmXOnutE8HzezTkl7JO01sw4z+/NZrhEA4Mu0O5mNtutkjCRTivvbWC/GdqdMO1M4OLGEMxiweR0cnHnuoqEA7U4AAABYNAKhcOHnTnP7H0i6WtJlzrkG51y9pCskXW1mfzDzJQIAMnpiCVVFQ6qKhia0O+V+vRjbnTIhTSgw8cdJOBCY10qazHO1eVmV+keS6h9efKEWAAAAlh4Lli6keYektznnDmYOOOcOSPoNSe+c0eoAAGP0DMVVWx5WdTQ0od0pN5hZ3O1OeSppgqZkav5m0mSeuy3LqiRJJ5lLAwAAgEUgEIoUfu40t4edc53jD/pzaQqPggAAk+qNJVRXEVZVWWhCu1NuaLOY251Ci2AmTSbQ2rK8WpJ0vIeQBgAAAAsvUEQlzXSDg+MzvA0AUKCeIS+kicVTEyppxrQ75RkqvNASfqVMKLDwM2kyz89mv5KGuTQAAABYDEoZ0lxoZn15jpuksmIWBQDIrzsW19kraxSwxISZNJngoakqsihn0mRCmvy7O83vTJrMc7Wp2Qtp2ghpAAAAsAgUMzh4ypDGORec9WoAAFPqjSVUVx5WOu0mVH9kWpyW15QtykqaTAiTN6SZ75k0/qDghsqImqoiOtHHNtwAAABYeKUcHAwAmEPOuWy7U9UUg4NX1s5tSDMUT6m1O1b0/bLtTpNswT2flTSDcW+wcmU0qJW15cykAQAAwKJQyi24AQBzaGAkqVTaqa48oqqyiSFNnx/SrKgtm9N2p3/7+fO68V8ekHPFhSqZwcH5dncKzfNMmv7hpCLBgKKhoFbUljGTBgAAAItCKXd3AgDMoZ6Y16JTm1NJkxuUDIx4wUNDZVSD8ZTScxR67GvvV08sof4iq3WS6czg4Ik/ToLzPpMmoaoyr4t3ZW2Z2nppdwIAAMDCK2ZwMCENACygTEhTXxFRVTQk56SY37Yjee1OVWUhVUe98GEwPjfVNEe7vECje7C4jfsS2S24J1bShIPzW0kzOJJSVTQT0pSrbzg5YUtzAAAAYL4R0gDAaaJnyAtF6irC2SqQ3GBhYCSpqmgoe9tczaXJzKM5VXRI41XSRPIMDg4GLHv7fOgfTqoyOlpJI7HDEwAAABaeBQrfk4mQBgAWUKaSpq48nK0CyW056h/2QppM+DAXlSEDI0l1++voGigupMlU/ZRHJv7gme+ZNAMjiWzF0Qo/pGEuDQAAAE4nhDQAsIB6hsbOpJE0ZkBwZs5KJnzon4PhwUe7Rnd16iqykmbID2kqIqEJt8377k4jqTEzaSQxlwYAAACnFUIaAFhAvTEvFKnNqaQZ3+5UPcftTq3do0FGse1OsWxIM7GSJhwMzHMlzWi70/Ka+aukeaa1NxtWAQAAALNBSAMAC6g7llBFJKhoKJgNGCa0O5WFVBmZu3anTCWNmdQ1OFLUfWP+ION87U7BgCk5zzNpMkFXWTioxsqIjs9xSPPg86f0ms8+oO883jqn1wEAAMDSMLE+HQAwb3piCdVXRCRJ1WV52p384CFz21y0O7V2D6k8HFRDZURdg4mi7pttdwrnn0kzv+1OyezzJHlzaU7MYbtTPJnWn33vWUnFt4kBAAAA+RDSAMAC6h2Kq7bc25KvMs822/0jSVWX5cyrmYtKmu6Y1jSUKxoKFl9Jk0gpEgwoNMnuTvPV7pRMpTWUSGUrjiRvG+7MrlVz4dYHDmp/+4CkudsaHQAAAEsL7U4AsIB6YgnVVXghTdW44cAjyZTiybSqy+Z2d6fW7iG11FeovjIyo8HBZeH8P0pCgcC8VdIMjngVPVVluSFNmU70zU2707GeIX3m7n162TnL1VAZmZO/FwAAACw9hDQAsIB6hkZDmmgooHDQsr/wZ4OHaEiRUECRUGDMvJpScM6ptSumNfXlaqyMzGBwcDLvzk6SFArOXyVN/4jXppXZBUvy2p16Yok5Ger7F99/Tk5OH3vNOaqIBBUbYXAwAAAAZo+QBgAWUE8srtpybyaNmakyGsq2NGVm02QqbKqjoTHzakqhdyih/pGk1jRU+DNpit/dKd/OTpLX7pSYp8HBmUCrMjq2kkYq/TbcP9t5Uj957qR+7yVb1FJfoaqcvzMAAABgNhYspDGzoJk9YWZ3+l9vMLOHzWy/mX3TzCL+8aj/9X7/9vULtWYAKCXnnD84OJw9VpUTxGSqQzItPFVlpQ8DMttvt9SXq6Eyolg8peFE4VUhQ/FU3p2dJG9w8HxV0gyMe64kbyaNVNptuL/7RKt+978e19bl1fpfL9woydt+PMYW3AAAACiBhayk+X1Ju3K+/ltJ/+Sc2yypW9J7/OPvkdTtH/8n/zwAOO0NxlNKpl223UnSmKqMzGyaTAtPZSRU8tknme23W+or1FjpVfQU0/I0dSXN/M2k6R9XdSTlVtLMPqRJpZ0+9cNd+oNvPqVL1tbpG7dcqUjI+xFaSSUNAAAASmRBQhoza5H0aklf8r82SS+R9G3/lNskvc7//Cb/a/m3X++fDwCntZ6YF4bU+e1O0tiQJtvulFNJU+otuDOVNGvqvXYnSeouJqRJpFQ+yUya8DzOpMmd35OxorZMoYBpV1vfrB47kUrrlq/u0Bd+cUC/ceVa/b/3XKH6ytG/s8pISDF2dwIAAEAJLFQlzT9L+hNJmWEFjZJ6nHOZd7mtklb7n6+WdFSS/Nt7/fPHMLNbzGyHme3o6OiYw6UDQGn0xLwWndrcSpqclqbMxzEzaUpdSdMdU3VZSLUVYTVWFV9JMxRPqiK88DNp8rU7lYWDum5rs77/9PFZhUWPHuzS3bvb9aFXbtNfvu58hcdtN14ZDWVDIgAAAGA8C4VVv2l7QefOe0hjZjdKanfOPVbKx3XOfdE5t905t725ubmUDw0AcyIT0tSVT9Lu5H+sLvNu98KA0lfSrKmvkCTVV3ghTdfgyJhznHOTzqmZqt1pPmfS5Gt3kqRfu7hFJ/tG9ODzp2b82HtO9kuS3nhpS97bK6NBDVJJAwAAgEmYTOHymoLOXYhKmqslvdbMDkn6hrw2p09LqjOzzLvrFknH/M+PSVojSf7ttZJm/m4bABaJniGvYiW3dSZ3cHDmY3WJBgd3DozoyKnYmGNHu2JqqfcG7DZWRiVJpwbGVtL8dOdJXfrJu9Q3nJjwmFMNDp7PmTTZ3Z3GreX6s5epOhrSd584lu9uBdl7sl8NlRE1VUXy3j4X4RkAAADOICZZIP+IgPHmPaRxzn3EOdfinFsv6a2S7nHO/bqkeyW90T/tZknf8z+/w/9a/u33OOfm510/AMyhySppBrPtTgmFAqaoP6C2Ojq7mTR/eedOvfHffpVtQXLOeZU0DV4lTU15SKGATdiG+4kjPRqMp9TZPzLhMRdLJc3ASELl4aBC41qRysJBver8lfrxs20amuEOTHtO9Ous5VWabBxaZSSoRMopnpyf1i4AAACcbkyB4CINaabwIUkfNLP98mbO3Oofv1VSo3/8g5I+vEDrA4CS6h3yQpqanJCmMhrSYDylVNppYDipqrJQNhyoioY0kkzPeM5LW++w2vtHdPeudkne7JmhRCpbSWNmqq+MTAhpjnQNStKEKp502mloisHBIX9w8Hzk6gMjyTHzaHK97uLVGoyn9NOdJ4p+XOec9p4c0Nbl1ZOeU+m3WFFNAwAAgMlYMDz9SVrgkMY5d59z7kb/8wPOucudc5udc29yzo34x4f9rzf7tx9YyDUDQKn0xOIqDwdVljN4N9PaNBhPqn84OWbGymzDgG5/N6lvPHpE0uj225mZNJLUmCekOdTpnTc+pBlOepUpU1XSSJqXlqeBkdSEeTQZV2xo0Krashm1PB3vHdbASFJbpgppIqN/ZwAAAMBEp2clDQAsKd2xhOoqxibqVTlBTP/I2JAmUyky05anrsGEggHTz/d26HjP0Oj22w2jIU3DuJDGOafDpwb9NY1tF4rFpw5pggHvR8x8tDwNDCcmDWkCAdNNF6/W/fs61ZGnZWsqe094Q4O3rpi+kiY2w3YqAAAAnNnMpEDoNKikAYClrCeWUF3F2GG0mV/4B4aTGhhOZitrJG8mjTSzig3nnLpjcb36/JVyTvrWjqM62u1VyGTanSRNaHc6NRjXoB8+ZLa5zsjMeCmfZAvu+a2kSU4a0kjS6y9erVTa6c6njxf1uJmdnc5aNnlIUxENZtcAAAAA5HNatDsBwFLWOxQfMzRYyqmWGUlqYCSZ3X5bGhvgFKtvOKlU2umCllq9cEuTvvXoUR05FVNDZST7uJLX7nQqJ6TJVNFIXktRrtFKmsln0khSKjU/7U6VU4Q0W5ZX69xVNUW3PO092a8VNWWqrZj8h2omHIqNUEkDAACAfGh3AoBFrydPu1N1TrvT+OqQ3ACnWN1+8FJfEdFbL1ur473D+sEzbWOqaCSv3al3KJEdTnw4Z8vu8eFQzK/omW4mTSI997seDYwkxlQd5XPDeSv0dGtv3q3EJ7P3ZL/OmqLVSRr9/mdTSXOyb1iv/sz9au2OTX8yAAAATi9GJQ0ALHo9QxNDmtxqmf7hsTsWVc9icHCXPzS4oTKil56zTA2VEfUPJ8cMDZa8ShppdMjwoVMxBUwK2MTrZtudFsVMmqnbnaTR2Tsne4cLesxU2mnfyQFtXV415XmZwcGxWQwOfuJIt5473qenW3tn/BgAAABYrEwBQhoAWLycc+qJxVVbPnYmTSZo8NqdEtlgRppdu1OPH7rUV0YUDQX1hktWS5JaGsZX0kQlKTuX5vCpQa2sLVdVNDShUmS6wcHzPZNmqnYnSVpRUyZJOtFXWEhzpCumkWRaZ02xs5NUmi24M0OcOweKG2wMAACA0wPtTgAwBxKptF79mfv142fbZvU4sXhKiZRT/SS7O/XGEhpOpPO2O82kraZr0GvxafAHFb/18rUKBkxbxg3EbfAraUZDmpjWN1XkD2kS04Q08zSTZiTpPZfTtTstz4Q0BVbS7PF3dpo+pPG+/8E8uzt9+mf7dP++jmmvdawnE9LEpzkTAAAApyMLENIAQMm1dg/pueN9unf39L94T6VnyAtNJmt3ylR75LY7ZdpqZhLSZGfSVHrX29Rcpfv+6Dq97qJVY86bGNIMam1DparKQnnanbyvyycZHBzMVtLM7UyaTGXRdO1OK2q9kKa9wG249/o7O22Zpt2pPByU5WkHc87pc/ft1598+2kNJ6YeKnyMShoAAIAzlpnJzAo6l5AGAIpwqNPb7SizNXOxnHO6f1+HPvTtpyVJy6rLxtweCQUUDQWy1R65wUMwYKqIBGfU7tQViysctDGPt6ahQqHg2B8DuSFN71BC3bGE1jdWqHKqdqdJt+D2Hnuu250y65qu3aksHFRtebjwSpqT/VrbUDHp7lUZZqbKSEiD43Z3Gk6kFU+m1dY7rP/34OEpHyNbSVNggAQAAIAzU2H1NgAASdIBP6TZe7Jf6bRTIFBYIi5JP9t5Un//0z3afaJfzdVR/fErtupFZzVPOK8qGlJbr/dL+/gWnnxtR4XoHoyrviIybYKfab86NRDXEX9np3WNlVPOpJl8cLBfSTPH7U6ZdU1XSSN5c2kKnUmz90T/tK1OGZXR4IRKml6/WioSDOhz9+3XWy5fo5qy/APjRtudCGkAAACWMippAKAImUqaWDyV/cW6EF+6/4B+66s7lEw7/d0bL9ADH3qx3vfizXlDnqqyULbao3rcL/VVZaEZbcHdNRjPVslMJRQMqK4irK7BuA6d8r7XdY3+TJpxFTzDiZQCJkVD+X+UZAYHz/XuTpl1TTeTRpKW1UR1soCQJp5M62DnoLaumLrVKaMyEtLguN2deoa8lrFbrt2onlhCX/j583nvOziSVE/MC3SYSQMAALC0EdIAQBEOnRpUxA8ldp+YvuUpnXb66x/u0l/+YJduOG+F7vy9a/Tm7WsUDeWvPpG8ipDM3JTx1SHV0YmzYQrRHYtPmH8zmYbKiLoG4zqcE9JU5rluLJ5SRSQ0aXVOZnDwnM+kKbDdSfIqaQoJaQ52DiqZdkVU0kx8fjLBy1WbGnXTRat06wMH1Z7n2sf9sK+5OkolDQAAwBJHSAMARTjYOagXbm6SNDpYdjKJVFof/NaT+uIvDuidL1inz779EpVNMr8lV2U0lJ3jUjWuOqQyT0VLIQqtpJGkxmxIE9Oy6qgqIqFJ250ma3WSRmfSzHklTTHtTrVl6ugfUTI1dXCUmTm0dUVhIU1FJDhhd6dMSFNTHtYHX3aWkimnz9yzb8J9W/2Q5sKWWsXiKcXiM9/KGwAAAKc3QhoAKNBI0mtxOnd1rVrqy6etpPncvfv1P08e1x+9/Cx94rXnZme0TKc6J2yojpZoJk0sofqKwkKahpyQZl1jxZjrOjcauAzFk5Nuvy2NzqRJzNNMmsLancqUdtO3Fe090a9gwLShqbKgNVTlqaTpy9nBa11jpd5+xVp945Gj6hg3HDizs9OFLXWSvHlAAAAAWJoIaQCgQEdOxeSctLGpUluXV2vvFCHN4VOD+tf7nteNF6zU+1+ypeAt96Sx1TPjK2mqyqYOab7+yBH98Jm2McfSaaeeWOGVNA2VEZ3yZ9Ksa/RCispoSGnn7ViUEYunVD5FZVCm3Wm+ZtIU2u4kadrhwTvb+rShqXLKtrRcFdFQdpByRmYmTZ0fjr3mwlVKpp2ePdY75rxjPUMKBUznrKqRJHXQ8gQAALBkEdIAQIEO+kOD1zdVauuKaj3fMaB4cmLbjHNOf/695xQJBvRnN55T9HUyYUPANCEEmaqSxjmnf/jpHt36wMExx/uGE0o7FVlJM6L2/hGta/ArafywqH8kkT1vKJGaspImMzh4rmfSDI4kZTb5VuC5MiHNVHNpnjjSrXv3tOfdeWsyVdHghL+XnlhCwYCp0n+OMq1Tu070jTnvWPeQVtaVZbdjZxtuAACApYuQBgAKlNntaEOjF9Ik0y4b3OT6yXMn9PO9HfrfL92i5X4oUIxMi1NVdOJQ3swuS7ltRxnHe4fVORBXa3dszPGuQa+io/BKmqgyxS/r/HafqqgXNAyOjFaLLJaZNP0jSVVFQgVth768Nipp8pAmkUrrI7c/o+XVZfrfL91S8BoqIiHFxoc0QwnVlYezf4c1ZWGtrivX7raxFVjHeoa0uq5cTdXe3w87PAEAACxdhDQAUKCDnTHVV4RVWxHOVkXsGTc8eHAkqb/4/k5tW1Gtd121fkbXyVTSjN9+W/IqWpJpp5E8FTzPtPZIkk72jWgkORqmdMe8X/rrixgcnLE+O5PGW0vu3BWv3WnyFqN5m0kznCyo1UmSGiujCgYsu8X5eF/8xQHtPtGvv7jp3LzP/2QqoyENxlNK5wRSvUMJ1Y7bUevsldXaPa6S5njPkFbXVaix0guQTtHuBAAAsGQR0gBAgQ51Dmq9X1mysalKoYBpz7hfuD9zzz4d7x3WX77uPIWCM/tfbFU2pJkYPGRuy9fy9FTr6KyT4z2jIUTXoNei1FBEu1PGuobMTBqvYqY/Z2ep6QYHz9dMmsF4csLsnskEA6Zl1VGd7JsYhBzoGNCn796nG85boZefu6KoNWRamoYSo+FYb8yrpMm1bUWNnu8YzIZoiVRaJ/uGtbquTJFQQDVlIbbhBgAAWMIIaQCgQAc7B7O7/URCAW1oqtSeEwPZ2zsHRvSVXx7S6y9Zre3rG2Z8nUzgkG9L6WxIk2cb7mdaexX2g5GjXaMtT92DmUqawipDMiFNnV81JEnVk1TSFLK701zPpOkfTha0/XbG8pqyCe1Ozjl99LvPKBoK6BOvPbfoNWQqeXKfn56huGrLx1fS1CiVdtp30nvdnOgdVtpJq+vLJUlN1VHanQAAAM5AwUi5ItVN055HSAMABRiKp3Sib1gbGke3ZN66olp7To5W0vznQ4cVT6b1vhdvntW1MoFDvuqQySppnHN6urVHV2/2/sff6m/rLEldmXanIitpMkODpdFKmtzrDk0zkybiVxLla80qpYGR4kKaFTVlE3Z3undPux460KUP37BNy2YwRyjz/Azm7PDUE0tkd3bK2LbSa5PLbN+e+XtaXec9101VUXZ3AgAAOAMFyypVXr9q2vMKf1cLAEtYZmhwpt1JkrYur9adT7dpcCSpYMD0/x48rOu3LdOm5qpZXasqOkUlTVn+kObQqZj6hpN62TnL9cC+zjHDg7sH44qEAlNWveTKhjQ5gdT4cMg5p9g0uzs1V0dlJrX1TL3d9WwNjiS1vLrwYGV5TVS/3N855tjDB7sUDprecEnLjNZQGZlYSdM7lJhQSbO+sVLRUEC727xw71iPH9L4lTTNVdEJuz8BAABgcbjhhhu0Z8+eGd23viqqd730LF25dfmU5xHSAEABDvm7OG1oGltJI0n72ge0u61Ppwbjes8LN8z6WpkgZsqZNOPanZ72hwZfvKZeq+rKx1bSDMbVUBGZsFPUZMrCQV2+vkHXbBktx8ysKRNCxFNppdJOFZHJf4yUhYNaUVOmw10Td8AqpYHhwmfSSNLy2jL1jyQ1ODI6cPipoz06Z2WNygrYxjuf8e1OyVRa/cNJ1Y0bHBwMmLauqM5W0hzz/55W1nohU1NVhC24AQAAFqnW1lYdPHhwRvftrS5XX98qSVOHNLQ7AUABDuarpPFDmt1tffrSAwd1zsoavWBj46yvNWUlTSYMiI8PaXoVDQW0ZXmVWurLdTS3kiaWKHhnp4xvvfcFevP2Ndmvy8NBBWy0kmbIb+spnybUWNtQocOnYlOeM1v9M2h3kka34U6lnZ5p7dWFa+pmvIZMSBPzn5c+P0QbX0kjSdtWVGtXW5+cczreM6Tm6mg2HGqsiqpvOKn4HLeIAQAAYH6lnVMyNf17PEIaACjAwY5BNVdHx4QBa+orVB4O6iu/PKT97QP6rRduKLhaZSqjIU3+LbilsbssSd7Q4HNX1SgcDGhNfcWYSpruWFwNBQ4NnoyZqTIayoY0mTBiuhaq9Y2VcxrSOOc0OMOQJjOX5vmOAQ3GU7qwpW7G68js7pR5fnr8OUDjK2kkb3jwqcG4OgZGdKxnSKvqyrO3NVX523APUk0DAABwJkmnnRL+Dp/7YpdMeh4hDQAU4NCpwTFDgyUpEDCdtbxKe072a1l1VDdeMP0gsELUV0RUGQlqTUP5hNvyDQ5OpZ2ePd6rC/yQoaW+XB39Ixr2t4PuHowXPDR4KlXRULbNKhPSTDU4WJLWNlaoc2BkzKyWUhpKpJR2+YcsT2bZuEqaJ4/0SJIuWls343WMVtJ432fvkLfteV35xOd924oaSdLutn4d6xlSy5iQxju/s58dngAAAM4kmUqavYPbdSB20aTnEdIAQAEOdsa0vqliwvGzlnstTzdftV6RUGn+l1oeCeqBD71EN120euJtfttRbuixv31AsXhKF7TUSpJa/HAnM5S2KxbPDgOejcpoKNtmNZStpJk6HFnX6D1nc1VNkwmNKouppKnNhDRetcqTrT2qLgtNCOGKkRkcPDDiPS89fkhTM0m7kyTtbOvTsZ6h7NBgyduCW/K2cwcAAMCZI+2cnh88RweGLlJ6ivHAhDQAMI3+4YQ6B0bGzKPJuGpzo5qqovr1K9aW9Jr1lREFAxNbp8xM9RURPXm0R845SaNDgzOVNGvqvWDkaFdMyVRavUOJklXS9GcrabyP07U7rWvwnrMjczQ8uG/YC0OqiwhpqqIhVUVDOtHrVdI8dbRHF7bUKZDn+S5Uhb8Fd8wPz3pjfiVNnnan+sqIVtSU6Zf7OxVPprU6p5Km2W93YhtuAACAM0s67TSSlNw0MQwhDYBF53jPkN5x68PqWCS73GSqQPJVWvzaxS165KPXq64EIUih3vfizbp/X6e+/MtDkryhwVXRkDb6IVKLH9K0dg+pdygh56T6PGFBsaqioWwFTyxReLuTNHeVNI8d7pYkbVtZXdT9ltdEdbJvWMOJlHaf6NeFa2pntY5wMKBIKKCB+LiZNHkqaTLrfejAKUkaE9I0+u1OpwbGtjtlAjkAAACcntLOKVnA5hCENAAWne8+cUz37+vUg/4vsQvtYOfEnZ1yzaYCYybeffV6vfTs5fqbH+3SU0d79PSxXp23uia7jmXVUYWDptbuIXX7YUGxuzvl44U0XjgzVODg4NrysOorwjrcNTchzT2727Wqtkxblxcb0pTpRN+wnj3Wq1Ta6aI19bNeS2UkqJj//PQOTb67k+QND06kvOAld3BwRSSkikhwTLtTTyyuy/7qZ7rjqeOzXiMAAAAWRjrN7k4ATlM/23VSkjdrZTHIhjSzmFlSSmamv3/TBWquiur9X39cu9r6xuxMFAiYVtd523B3DXptN6WaSTNhd6fw9G1GaxsrdfhU6dud4sm0HtjXqeu2LSt6V60VNWVq7xvRk0d7JEkXtsyukkbyZ/ZkdncaiqsqGlIomP/HbGYujaQxM2kkb4en3JDmwedPqXMgrr/90W625gYAADhNpZ3TivBzWhY5KGnyKmlCGgCLSkf/6C/Ozy+SkObRQ11a11gxbWvPfKqriOhf3n6xjvcMK55MZ+fRZKxp8Lbh7hr0K2lK0I5VXTYa0gz5bT2FPCfrGirmpN3p0UNdGoyn9OKty4q+7/LaMp3sG9YTR3u0qrYsu+PTbFRGRgcr98YSk1bRSF4ljeTN0hl/XlNVZGxIc+CUzLxB0N/acXTW6wQAAMD8c2knpYd1SfVdag5P/p6OkAbAtD53737950OH5+Va9+5ul3PSmoZy7Wvvn5drTuVoV0z37+vU6/LstLTQLl3XoA+9cqsiwYAuXTe2XaelvlzHumPZ2SilqaQJamAkKefcaCVNASHN+sYKHe8ZKnkVyL272xUJBnT15sai77u8Oqpk2un+vR2z2no7V2U0mG0H6xlK5B0anLGhqVKRYGBCFY3kV9LkbMH90IFTumZzk7avq9e/3LMvu7U6AAAATh9O0v7WLv3gob06sfMzk55X+HYYAJasbz56VKvryvUbV66b82vdteukVteV61Xnr9SXHzioZCo9acvIfPjWjqMyk9582ZoFW8NUbrl2k95+xTpVjdvdqKW+Qp0DcR33t+EuRSVNZTSkVNppJJnOhjTl4elDmrWNlUo7qbU7po3NVXnPuXvXSX38+8/prj94kcoKeExJundPu67Y2DDtNuD5ZLbh7htOjmkVm43cdrDeaUKacDCgi9bUaWXdxAqexqpodiBy58CI9p4c0OsuXq2L1tTp7f/+sL728BG955oNJVkzzmxHjhzR7/7u7y70MgAAOGMcPz67GYEP7mzVgztbpzyHkAbAlNJppxO9w1O2bkwmlXYaSqQmBAiTGU6kdP++Dr15+xptbq5SIuV0pGvyX+znWjKV1rd2HNWLzmoeswPPYpPv+W3xKzSePtar8nCwJK1amW2uB0aSGkqkVBYOFDQ0eV1mh6cp/i6fOtqjo11DOnwqpq0rph8CfORUTM93DM44OFye09504Zq6GT3GeJWRkE72edt698Ti034ft75re95t1purIuqKxZVKOz18oEuS9IKNjbp4bb2u2tSoz9+3X2+7fM2MwiksLY2NjXrXu9610MsAAOCM8dOf/lRdXV1zeg3e4QGY0qnBuOKptPqHE0Xf99YHDujf7z+ohz5y/YRfRocTKT15tEdXbGjIDn395f5ODSfSeunZy1Xjh0L72wcWLKT5+d4Onewb0SdeuziraKaS2Yb7mdbekrQ6SV6liCQNDCcViycLDgnWNXhrOTLFXJoOfwZLa3dhIc29e9olaUbzaKTRSpqASeevnv3QYEmqyGl36h1KqLZ86ue9uix/8NlUHZVzUtdgXA8e6FRlJKjz/DX+4cvP0hs+/6Bu+9Vh/c51m0qybpy5Kisrdfnlly/0MgAAOGOUl8/9P9wykwbAlNp6vXaZ/uFk0ffd1davjv6RvDv7/PeOo3rrFx/Sf/zqUPbYz3a1qyoa0hUbG7Sp2dtJaX/Hwg0P/vojR9VUFdH1Zy9fsDXM1Bq/kubUYFz1lcVXQeVTlVNJE4unCmp1kqTm6qjKw0EdmmKHpw5/Bktr91BBj3nvnnZtaKqcdFv06TRVRWUmnbW8Ohs+zVZV1Bsc7Jybtt1purVJXqvTQwe6dNmGBoX9lr9L1zXouq3N+sIvnp9RcAoAAIDFjZAGwJSO93jtG/0jxYc0mYBn94mJA4CfOdYrSfrknTt13552pdNOd+86qRed1axoKKjqsrBW1JRp/8mZhzSdAyO6/fFWOTf5FneTOdk3rHv3tOsNl7Zkf0E+nTRVRRUJeesuxTwaaWxIMxRPFTQ0WPK2DF/XWFFQJc3Rrul3gRqKp/Tg86dmXEUjeTNhNjRV6sqNxQ8dnkxFJKTYSEqxeEqJlJtRi6A0GtLsPtGn/e0DesG4Nf7JK7bpE689l3YnAACAMxDv8ABMKRO0xJNpjSRTioYKn23S1usFPLvb+vSq81eOuW1nW58uXVevWDyl3/uvJ/Tx156r9v4RvfSc0V+8Ny+rmlUlzV98f6fueOq4NjVXFT135NuPtSqVdnrrZWtnfP2FFAiYWurLdaBjsGQhTabiZNCvpCk0pJGktQ0VOtA5eSVNZ3+m3Wn6SpoHD3RqJJnWi7c1F3z9fL77O1crGi5dAFcVDSqeSqvD/17qZhjSNFZ5f193PtUmSROCpHNW1eicVTWzWCkAAAAWq9Pvn4cBzKtM0CJ5s0gK5ZwbDWnGVdIkUmntPTmgS9fV60s3b1c0HNAf/vdTCph03VljQ5rn2wdmVAmz50S/vv+0N339jqemn8KeSKV15FRMTx3t0X172vWNR4/oyo0N2jDDdprFIDOXplQzaarKxlbSFDOMeH1TpY50xZROT/y7dM5lg43Wnukrae7d3aGKSFCXb2go+Pr51FaEC95JqhCZypbMjlqzbXf6xb4OVUdDOpdABgAAYMmgkgbAlDK/cEreXJpG/xfI6ZwajCueTEuaGNIc6BhUPJnWOStrtLquXF94x3a97YsP6eK1darPCRQ2L6vSYDyltt5hrSpyd6V/umuvqiIhnbu6Rt9/6rg++qqz8+6k45zTXTtP6i/u3DmhiuNDr9xW1DUXm8wOT3PR7hRLJLWseuL20ZNZ21CheDKtE30T/y77hpOKp9IKBkxHu6avpPnl/k5dubGxqKqu+ZB5fo75/81MNzh4MjVlIUWCAcVTab1wS8OCbkEPAACA+UVIA2BKYyppiphLc8K/33mra/TssT4NjiSz7TK72vokSWev9CoELl1Xrzt+72rVjNvtZvMyb1enfe0DRYU0zx7r1Y+fO6Hfv36Lzlperff91+N6+MApXbW5acx5hzoH9fHvP6f79nRoy7Iqfer156u5Kqr6yrCaq8q0pmHxbrtdiExI01DiwcGZdqdiKmmy23Cfik34u8xU0WxdXq2dbX3qG05MeC3knnugc1Bvvmzx7bhVEfWej8wcp5nOpDEzNVVFdLx3eMI8GgAAAJzZCGkATKmtZ0ir68p1rGdIfUXsJpOpwHnJ1mV69lif9pzs1yVr6yV582gioYA2No+2Em1bMbGlIxPS7G8f0IvOKnz+yD/etVe15WG954UbFAkGVBkJ6ntPHh8T0jx84JTecesjioQC+tNXn62br1p/Wg4Insoav92pvkTtThWRoMy8treheEoVRbQKrWvw/q6PdA3qBZvGBg+ZkObitXXa2danY91DqlmZP+DYcahLknTZ+tm1Os2FymwljdeyNdN2J8nbhvt47/CE5woAAABntjPrNxIAJZVKO53sH9FZy72wpJiZNJkKnOu2eTNmdreNtjztauvTWcurpg1FGisjqqsIa3974cODHzvcrXt2t+uWazeqpsybOfKK81boh8+2aSSZkiQNJ1L60Hee1oraMt39hy/Sb71w4xkX0EjS+atrVR4O6qzl1SV5PDNTVSSkAX8Ho2IqaVbVlSkUMB3Ks8NTZmeni/0Qb6odnh451KWycEDnr64tcvVzrzI7k8Z77c8qpKmKqqYslK02AwAAwNIw77+VmNkaM7vXzHaa2XNm9vv+8QYzu8vM9vkf6/3jZmafMbP9Zva0mV0y32sGlqr2/mGl0i77S35/ESHN8d4hhYOmC1vqVBUNac8Jr8XJOaedx/t0TgG/fJqZNjd7w4On0z0Y1z27vdkyjZURveuq9dnbbrpotfqHk7pvT4ck6V/u2adDp2L66187X8trCp+rcrpZ31SpXZ98ZclCGsmrFhkYSRQ9ODgUDKilvjzvNtyZnZ0uWVsnaeodnh491KWL1tRltxdfTCoz7U69Q4oEAyqfxVDi975okz71+gvyzlECAADAmWsh2p2Skv7QOfe4mVVLeszM7pL0Lkl3O+f+xsw+LOnDkj4k6QZJW/w/V0j6vP8RwBzLVASMhjSFtzud6B3WitoyBQOms5ZXaZc/PLijf0SnBuMFVwhsWV6lnzx3ctLb79l9Un/5g1060OFt7xwKmD71+vOzrSeSdPWmRjVWRnTHk8e1tqFCX/j5Ab3hkhZds6VpsofFJCqjQfUOJRRPpVURLu5HyNrGSh3umrgNd8fAiMJB0/rGSlVEgpOGNP3DCe083qf3v3jzjNY+1ypzdneqKQ/LbOYBy2x3rgIAAMDpad5DGudcm6Q2//N+M9slabWkmyRd5592m6T75IU0N0n6qvP24H3IzOrMbKX/OADmUFuv98tyJqQpZnBwW8+wVtZ6A2K3razRD55u86po/KHBhVTSSNKm5ip1DR7VqYGRCTtL9Q4l9Mf//bRqK8L6k1du1SVr63VBS212K+SMUDCgGy9YqW88elSHuwZVWx7Wn7767IK/F4yqKgtnZ8hUFFFJI0nrGyv0xOFuOefGBBgd/SNqqooqEDC11JertTt/u9PjR3qUdtJlizTAyASDw4m0WupLM6wZAAAAi8d73/tenTw5+T8gF+OTn/xk3uMLOjjYzNZLuljSw5KW5wQvJyQt9z9fLelozt1a/WOENMAca/MradY2VigaChTd7rR9nTdj5OwV1fqvh4/oRN9wNqTZVmBIkzs8eHxI8y9371NXLK7bfvNynTfNjJLXXrRatz14WM8e69On33pRyYbpLjVV0aCO+DNjiml3krxtuPtHkuqOJdSQ8/x39I+oudr7u22pr9DRSSppHj3YpWDAsgOoF5tMu5Mk1c1wZ6cpOScNdkjtO6W+Numit5X+GgAAAJjU+973vpI91qILacysStJ3JP1v51xf7r+qOuecmbkiH+8WSbdI0tq1a0u5VGDJOt47pMpIUDVlIVWXhdRfYCVNOu10sm9YK/xKmq3+zk272/q1q61fLfXlBW9PnA1pOgZ0Rc52xPvbB/Qfvzqkt2xfM21AI3nzTjYvq9K6hgq99sJVBV0bE1VFQ7OopPF2eDp0anBMSNM5MKIV/mygNfXletTfwWm8Rw516dxVNWNa2RaT8rC3+5VzM99+OysZl07tk9qelk48I5142gtnYqe82wMh6bw3SCHCRgAAgDPJgrzTNbOwvIDma8652/3DJzNtTGa2UlK7f/yYpDU5d2/xj43hnPuipC9K0vbt24sKeADk19YzrJV15TIzVZeFC66k6RwcUSLltKrO+8V76wqvXWr3iX7tPN5bcKuTJK2qLVdFJDhmhyfnnD55506VR4L6o1dsLehxzEx3vP9qhQKBWc0KWeoqoyENJ9KSZhDSNHlbgh85FRtTDdPRP5LdramlvkL9w0n1DiXGBB0jyZSePNqjd1y5brbfwpwxM1VGQhoYSaq2mJ2dYl3S4V9KRx6SOvdKp/ZL3Ycl5+1GplC5tPwcadurpWXnSMvO9j4GaakCAAA408x7SGPeb0e3StrlnPvHnJvukHSzpL/xP34v5/j7zewb8gYG9zKPBpgfbb1DWlnrBS1V0ZAGChwcnGmTysykqS0Pa1VtmZ440q2DnYO68YLCK1kCAdOm5irdv69T9+5p1ws2NuqX+zv1870d+tNXn62mcS1QUxk/qwbFq8qpYikv8vlsqa+QmVdJk5FKO50ajGf/HlvqvdfM0a6YanMqpJ5u7VU8mdZl6xfnPJqMikhQAyNJ1ZVPUuES65Lad3lVMe07paOPSCef9W4LlUmNW6SVF3pVMk1bpZUXSA2bpCCvXQAAgKVgId71XS3pHZKeMbMn/WMflRfOfMvM3iPpsKQ3+7f9UNKrJO2XFJP07nldLbCEHe8d1ja/Vam6LFRwJU1m4HAm4JG8GTT37elQ2knnrCq8kkaSfu3i1fr7n+7Ru7/yqMrDQUVCAW1qrtTNOdtsY37khjTFVtKUhYNaWVOmwznbcHfH4kql3ZiZNJK3DXduG9sjB70WqMvWL855NBlV0ZDa+0dUVxGWEkPSscek1kel409Ix56Qeo+MnhytlVZdJL3kT6X1L5RWXUL7EgAAwBK3ELs7PSBpsl6D6/Oc7ySVbjoPgILEk2l1DoxoZd1oJc2Rwfy77oyX2bp7VV159ti2FdW6Z7fXxVhMu5Mk/eY1G/T2K9bqoQOndM/udj1ysEsfe825CgcDRT0OZi93Hkx5uLiQRpLWNVbqcE4lTeeAN98mE9KsafBeM+N3eHr0UJc2NVdOGB69KCSGpL7jUt9x3aD7FQ3t11ueOSL96lkpFffOqVsntVwqXf5b0vJzvXal6pUSrXcAAADIQf00gLxO9g3LOW8mjKSiZtKc6BtWNBRQfc5cjsxcmupoKNvSUoyycFDXbV2m67YuK/q+KJ3qsplX0kjSusYK3bVzdNvCzBDiTEhTWx5WVTSk1pwdnlJpp8cOdevGC1fOdNml13tM2vNDafcPpEMPSGmvFfCPJSWDAfXrXOmK35bWXSOtuVyqWNxtWgAAAFgcCGkA5HW8x29Z8itpqstC6itwJs3xHm+WTe6A3rP96pmzV9YwuPc0VhnJDWmK/xGyrrFSpwbj6h9OqLosnA1pMjNpzEwt9eVjKml2tfWpfyS5sPNo0mmp7Qlp70+8P21PescbN0tXvldafp5UvUIf/dkpffeA9PkbXkigCAAAgKIR0gDIq6137PDf6jJv1xrn3LQhS1vvcPZ+GRuaKlURCer8lum3y8biVVWWOzi4+Eqa9Y3ezJnDp2I6b3XthEoaSX5IM1pJc9uvDikSCuiaLU0zXfbMDPdJB+6V9v5U2vdTabBdsoDUcrl0/cekbTdKzWeNuUt/9RMa0nHVVTBbBgAAAMUjpAGQ13F/+O+qnEoa56TBeGrM8Nh82nqGdOWmxjHHwsGAvvM7V2Xbp3B6ms3gYElaOy6k6RwYUXk4qMqcx2qpr9BDB7rknNOhUzHd/sQxvfMF67Ssumyyhy2NoW7p6KPSkQe9P62PSumkVFYrbbpe2nqDtPmlU7YuZb6P3O3DAQAAgEIR0gDIq61nWLXl4WxLS1XU+6VzYDg5ZUiTSjud7B8Zs7NTxtlFDgzG4pMZHBwO2owGN69rrJQkHe7yhgd39I+ouTo6pjqrpb5cAyNJ9cQS+vTP9ioSDOh3rttUgtX7nPMCma4DXttS62PSsR1S517v9kBIWnmR9IL3S2e9wqucKXAL7MzzU0dIAwAAgBkgpAGQV1vv0JigJTMwtn84oRV5ApiMjv4RpdJuQrsTzgyZgG4mOztl7t9UFdXhTm/mTMfAiJqqxrYGZbbhvmd3u7731HH99rWbZldFE+uSDt0vHfyF1LpD6joojfSO3l7RJLVsly54s7TmSmn1pVKkYkaXWlYdVWUkqBpCGgAAAMwAIQ2AvI73DI/ZQjszi6R/ZOodnsa3SeHMkglpZjI0OGNdY4UOnRqtpNnQVDnm9szuX3/1w12qjIT029duLO4C6ZQXxuz7qbT/LqntaUlOilT5YcybpPoNUsMGbzvsunUl2wr75qvW64bzVioYYDg2AAAAikdIAyCvtt4hXby2Lvt1TbaSZuqQpq1n7MBhnFkyYd1M5tFkrGus0IPPn5LkhTSXbxg742VNg1fF0jUY1weu36L6ygKG8KaS0sH7pGe+I+39kdfOZEFpzRXSi/+PtOFaafUlUnBuK1zKwsHs3B0AAACgWIQ0ACYYiqfUHUuMraTxZ9L0T7MNd5tfSZNvJg1OfxV+m9NMdnbKWN9YqdsfP6b+4YS6Ywk1V419rdSWh1VdFlLATO+5ZkP+B0klvBkybU9LrY9Iu74vDXZI0Vpp26u8WTIbXyyV1814nQAAAMB8I6QBMEG+oCUzk2ZgXCXNvXva9Xz7gN5zzQaZmdp6h1UeDrK7zRkqEDBVRoKzrqSRpCeO9EiSmqonVsr89rUbtbax0nsdpVPSvrukk89IHXuljt1Sxx4p5W3frXCFtOXl0vlvlDa/TAoTEAIAAOD0REgDYIK23oktS9WTtDvd9qtDum9Ph8LBgG6+ar03cLiubMxuPTizVJWFVD6rmTTeDJodh7slSc1V0QnnvP8lW0a/eP5e6etv8T6vXSM1neW1L628SFp5gdS4WQrMPDQCAAAAFgtCGgATtHZ7O++szml3qoyEZDZxcPAJP9D5izt3akNTpTdwmHk0Z7T6isisKqXW+5U0Ow51SZKaqyeGNGP0HvE+fuAJqaHIIcIAAADAaYSQBsAEBzoGFQkGtLp+NGwJBExVkdCEmTRtvcN6/SWrtfN4n973X49Lkl557op5XS/m1z+++aLsLk8zUVcRUU1ZSE8e7ZFUQEgz0C7JvCoaAAAA4AwWWOgFAFh8nu8Y0IamygnbCFeVhcbMpInFk+odSmjzsip96ebtioYC6h9OMjT4DHfOqppZ72C0vqlSsXhKktSUp91pjP4TUkXjnO/MBAAAACw0QhoAExzoGNTG5soJx6vLQmNm0pzIzq4pU0t9hb7wjktVEQnqnFU187ZWnJ4yc2mqy0IqC08zT2agXapaPg+rAgAAABYW7U4Axogn0zrcFdOrzl854baqaEgDIxNDmhU1XlvUpesa9OSfv1yREPkvprauwavEmbbVSZIGTkpVy+Z4RQAAAMDC4zcpAGMc6RpUKu20aVm+SprwmJk0bTmVNBkENChEZhvufDs7TTBwUqpmzhEAAADOfPw2BWCM/e2DkqRNzVUTbpvQ7tTnV9IwgwZFWt/khYDTVtI4RyUNAAAAlgxCGgBjHOgckCRtaJpkJk1Ou1Nb75DqK8LTzxQBxsm0O007NHi4R0rFmUkDAACAJYGQBsAYz7cPanlNVNVlE3fSGd/udKJ3WCtqyyecB0ynuTqq67ct07VnNU19Yv9J7yMhDQAAAJYABgcDGOP5joG8rU6SNzh4OJFWIpVWOBhQW+8w221jRsxMt77rsulPHCCkAQAAwNJBJQ2ALOfclCFNdZmX6w74c2m8ShpCGsyhgXbvIyENAAAAlgBCGgBZnQNx9Q8ntal54jwayaukkaSBkaSGEymdGoxrZQ0hDebQwAnvYzUhDQAAAM58tDsByHq+wxsavHHSShpvTk3fcELOecdW1jGTBnNo4KQUKpOiNQu9EgAAAGDOEdIAyMqENJuWTd/ulGl5YiYN5tRAu7f9ttlCrwQAAACYc4Q0ALKebx9UeTg4aQtTJqTpH05qwN+Km5k0mFP9J6SqFQu9CgAAAGBeMJMGQNaBzgFtbK5UIJC/aiHT7tQ/klBb77AkaQUzaTCXMpU0AAAAwBJASAMga6qdnaScwcHDSZ3oHVJNWUiVUQryMIcGTrCzEwAAAJYMQhoAkqThREqt3UPaOMnOTtJou1PfcFJtvcNaWcvQYMyh5Ig01C1V0+4EAACApYGQBoAk6WDnoJzTlJU00VBA4aBpYCSpE33DzKPB3Brs8D7S7gQAAIAlgpAGgCTpQMegpKlDGjNTdVlY/cMJv5KGkAZzqP+k95F2JwAAACwRhDQAJHnzaMykDU2TtztJ3lya7sGEOgdGqKTB3BogpAEAAMDSQkgzS3tP9uuqT92tfSf7pz33eM+Q7t/XMQ+rAor3fMeAVteVqzwSnPK86rKQnu8YkHOikgZzi5AGAAAASwwhzSx969GjOt47rG8/3jrtuR+/4zm9+yuPqicWn4eVSc45/eNde7XzeN+8XA+nt+c7BrRxilanjOqyULY1agWDgzGXMiFNZfPCrgMAAACYJ4Q0s5BOO/3gmTZJ0g+ebpNzbtJzOwdGdM/udiXTTnfvap+X9f18b4c+c/c+/fUPd83L9XD6OjUwor0nBnT2iuppz62KhhVPpSVRSYM5NnBSqmiUQpGFXgkAAAAwLwhpZuGxI91q6x3WNZub1No9pGeO9U567v88cUzJtFN1WUg/fu7EvKzvCz8/IEl6YH+ndrVRTYPJ/dfDRxRPpfWm7S3Tnlvjb8MtiZk0mFsD7bQ6AQAAYEkhpJmF7z91XNFQQH/7xgsUDpp+8HRb3vOcc/rmo0d18do6veGSFv1ib4di8eScru2poz168MApvf/Fm1UeDurWBw7O6fVw+kqk0vp/Dx3WC7c0afOyAipp/JCmMhJUdTQ0zdnALPSfYPttAAAALCmENDOUTKX1w2fadP3Zy7S6rlxXb27SD57J3/L0VGuv9rUP6E2XrtErzl2hkWRaP98ztwOEv/iLA6ouC+m9123Sm7a36HtPHlN73/CcXnOx640l9Mk7d6o3lljopSwqP3r2hNr7R/SbV28o6PxqP6RZUVsmM5vLpWGpG2iXqlYs9CoAAACAeUNIM0MPH+xS50BcN16wSpL06vNXqrV7SE+3Tmx5+u8dR1UWDujGC1fqsvX1qq8I6ydz2PJ0qHNQP3q2Tb9x5TpVRUP6zas3KJl2+uqDh+fsmqeDW395ULc+cFCfu2//Qi9lUfnKLw9qfWOFXnRWYcNZq6JhSdJKhgZjLjnnzaShkgYAAABLCCHNDH3/qeOqjAT14q3eLxAvP2eFwkHTD58Z2/I0FE/pjieP61XnrVRNWVihYEAvPXu57t7drngyPSdr+9IDBxQKBPTuq9ZLktY3VeplZy/Xfz58WEPxlCRpJJnSt3Yc1bNTzNE5kwwnUvrPhw4rYNJtvzq05KuKMp482qMnjvTo5qvWKxAorComt5IGmDPDPVJqhJk0AAAAWFIIaWYgnkzrR8+e0MvOWa7ySFCSVFsR1jWbm3TnuF2efvLcCfWPJPXGnIGsrzxvhfqHk3rwwKmSr61zYET/vaNVr79ktZbVjP4S/Vsv3KieWELf2nFU/73jqF7y9z/Xn3z7ab3t3x/SnhP9JV/HYvPdJ46pazCuv3n9BUqmnT53L9U0kvQfvzyoqmhIb7x0+oHBGZmQhp2dMKcG/F3wqml3AgAAwNJBSDMDv9zfqd6hRLbVKeNV56/UsZ6xLU/f2nFUaxrKdeWGxuyxqzc3qTIS1I+fnX3Lk3NOrd0x3fn0cX3i+8/p7f/+kOKptP7XtRvHnHfZ+npd0FKrj93xnP7420+rsSqiT7/1IpWHg7r5y4/oeM/QrNciSV/8xfN6zb88oKeO9pTk8UohnXa69YGDOm91jd60vUVvurRFX3/kqI6V6Hsu1Mm+YX3i+88t6HPTP5zQ8Z4hHe8Z0q62Pv3gmTa98dIWVZeFC34MKmkwLwZOeh9pdwIAAMASwtYsReodSuirDx5STVlILzyracxtLz9nhT4afEZ//9M9qi0P64kjPTrWM6QPvuysMa0kZeGgrtu6THftPKm/fN15ChbYZjI4ktRTR3v0+JFu7TrRr4MdgzrYOaihRMp/3IAubKnTp37tfG1qrhpzXzPTn7xim/7lnn1699Ub9Ipzl8vMtGVZtd78hQf1rq88ov9+71WqLS/8l/Vczjl9+u59+uef7VM0FNAb/+1X+uirzta7rlo/6XDZRCqtUMDmfPjsz/d1aH/7gP75LRfJzPR712/R7Y8f02fv2adPvf6CWT/+rrY+hQKmzcuqJv1efr63Qx/85pM6NRjXVx88rPe+aKM+cP0WRUPBMef1xhK6b2+77t7Vrv3tA1rTUK6NzVXa0FSpa7c0zzgYcc7pPx8+ok/euXNMm52ZdLPfFleoxsqoJGlNfcWM1gIUpD8T0tDuBAAAgKWDkKZAR7ti+vIvD+pbjx7VYDylD7xk84RfsGsrwnrJtmX6yXMntaq2TJesq9ct127U2y5fO+HxXnHeCv3gmTb9cn+nrp1iYGtb75Buf/yYfvhMm3a19Sntd1KtbajQxuZKXbmxURuaK3VhS63OXlmjcHDy4qhrtjTpmi1jg6VzVtXoC++4VO/6yiN611ce0Y0XrFJ9RVj1FRHVV0ZUXxFWXUVEZeGAdrX16/HD3XrsSLfSaaebLlqtl2xbpnDQ9H9/skf/et/zeuOlLfrIDdv0oe88rU98f6ceOnBKN5y3Ugc6vUDpSFdMXYMj6hlMqH8kqRU1ZXr9Jav1hktbJgRLhWjvG9ZPnjuhlbXletHW5rzf/633H9SKmjK9+oKVkqTVdeV62+Vr9LWHj+i9L9qkoURKd+9q10MHTml1XbkuWVevS9bWa1Nz5aShSyrt9LNdJ/Wl+w/o0UPdkqQ1DeW6fttyveisZq2qK1d9RVjVZWF99t59+ty9z2vr8mp96ebt+q+Hj+hz9z6vn+1s1+9ct0kn+4Z1sHNQ+9oH9OTRHqXSTg2VEZ27qkb72gd09652JdNOkVBA77xynX7nuk1qrIpOWNNIMqXvPXlcP372hC5ZW6dfu6RFq+vK1Tec0Ee+84x+8EybXnRWs151/mj7yOq6Cm1oqizqOb+gpVb/+Z4rdNWmxulPBmZqgJAGAAAAS4/l2zL6dLd9+3a3Y8eOkj3eNx89oo/c/owCZnrNhav0nms26LzVtXnPHYqn1D+cGDMPJp/+4YRe8Kl7NDCSzAYDZ6+sVsQPGVJppwf2d+qB/Z1yTrp0Xb2u3tSoi9fV65I19aqtmFnFy2TueOq4/uTbT2k4Mf0w41W1ZUqknTr6R1RfEdZ5q2t1/75Ovf2KtfrLm85TIGByzmsx+psf7VYy7RQwqaW+QusaK9RYGVFdRUR1FWE93dqr+/a0K+2kC9fU6cKWWm1oqtTG5ipFggEd7BzUwc4BHemKqaEyqo1NldrYXKlYPKXbH2/Vz/d2ZIOrpqqIXnfRat100WqdtaJK0VBQu9r6dMOn79eHb9im975oU/Z7aO8b1gv/7l5J0ohfWbJ1ebVO9A2rd8jbojsaCqjBX2t9RViR0GgAdLBzUIdPxdRSX653X71B5eGg7t51Ug/s78w+Xq63XrZGH3vNudkZRnfvOqmP3P6M2vtHJEmNlRFtaKrU5RsadP3Zy3XRmrpshVUyldaBzkH9+y8O6DuPt6o8HNRvvGCdtiyr9kO0sH61/5Rue/CwOgdGtKKmTCf6hmUmXb2pSUe7Y2rtHtIfvXyrfvvajQUPCAYW1E//THr4C9KfnvRKvjDnzOwx59z2hV4HSqfU74cAAEDpTPbe67QJaczslZI+LSko6UvOub+Z7NxSvyk51Dmorz96RO+6an1Jtx0+2Dmoe3a3e9Uph7t1YtyOQ6vryvWGS1v0hktWa11jcdUOM5FOO/UNJ9QdS6hrMK6eWFzdsYR6YnENjCS1ZVm1LllXp5W15Uqm0rp/f6e+/Vir7tp5Ur9xxTr92Y1nT6g8ae2OaTiR0pqGigmVRxntfcP6nyeP6YfPnNDz7QPqH0mOuT0aCqilvlxdg956MlbWlunXLl6t1128WkdOxfTtx1p19+6TSqS8UGh1vfd3dWogrgc/fP2EYOurDx7SL/d36sVbl+kl25ZpWU2Z0mmnA50Devxwj/a192e//+5YQsnUaPhSUx7WWy9bq1ecu1yhnOqdoXhKTx7t0anBEe++g3FtW1mjl50zsRqgfzihQ50xrW2oKDh0298+oH+6a69+MG4XMUm69qxm/a8XbtA1m5vU2j2k7zzequ883iqT6R/ffKG2r28o6BrAonD7LdLhB6U/eGahV7JkENKceQhpAABYvE7rkMbMgpL2SnqZpFZJj0p6m3NuZ77zT9c3JQMjyTE7Q1VFQ3M+r6UUUmlX8Fyd6Tjn1DkQ14GOASVSThuaK7Wypixb/dE9GNfBU4NKppwuXVc/4brdg3H9Yl+HDvjzeg52DurGC1bqt3OqaM4E/cNekOaFaQmtaSjX5mXVE85zzp0WryFggq/eJMUHpd/62UKvZMkgpDnznK7vhwAAWAome+91usykuVzSfufcAUkys29IuklS3pCm5Pb8SLr/H+b8MsVPZFkc8tfHzIxJavb/5FPv/5lMvbwXRla5pH3+nzNItf9n3TTnEc/gtHVyp7TpxQu9CgAAAGBenS4hzWpJR3O+bpV0Re4JZnaLpFskae3aiYN6ZyUYlqITqxQAAHNk7RXSRW9f6FUAAAAA8+p0CWmm5Zz7oqQvSl55b0kffPNLvT8AAAAAAABzZPL9mheXY5LW5Hzd4h8DAAAAAAA4I5wuIc2jkraY2QYzi0h6q6Q7FnhNAAAAAAAAJXNatDs555Jm9n5JP5E3p/bLzrnnFnhZAAAAAAAAJXNahDSS5Jz7oaQfLvQ6AAAAAAAA5sLp0u4EAAAAAABwRiOkAQAAAAAAWAQIaQAAAAAAABYBQhoAAAAAAIBFgJAGAAAAAABgESCkAQAAAAAAWAQIaQAAAAAAABYBQhoAAAAAAIBFgJAGAAAAAABgETDn3EKvoeTMrEPS4RI/bJOkzhI/5pmE52dyPDeT47mZGs/P5HhuJjfT52adc6651IvBwpmj90MAAKA08r73OiNDmrlgZjucc9sXeh2LFc/P5HhuJsdzMzWen8nx3EyO5wYAAOD0RbsTAAAAAADAIkBIAwAAAAAAsAgQ0hTuiwu9gEWO52dyPDeT47mZGs/P5HhuJsdzAwAAcJpiJg0AAAAAAMAiQCUNAAAAAADAIkBIUwAze6WZ7TGz/Wb24YVez0IyszVmdq+Z7TSz58zs9/3jDWZ2l5nt8z/WL/RaF4qZBc3sCTO70/96g5k97L9+vmlmkYVe40Ixszoz+7aZ7TazXWb2Al47HjP7A/+/qWfN7OtmVraUXztm9mUzazezZ3OO5X2tmOcz/vP0tJldsnArn3uTPDf/1//v6mkz+66Z1eXc9hH/udljZq9YkEUDAACgIIQ00zCzoKTPSbpB0jmS3mZm5yzsqhZUUtIfOufOkXSlpPf5z8eHJd3tnNsi6W7/66Xq9yXtyvn6byX9k3Nus6RuSe9ZkFUtDp+W9GPn3DZJF8p7npb8a8fMVkv6gKTtzrnzJAUlvVVL+7XzH5JeOe7YZK+VGyRt8f/cIunz87TGhfIfmvjc3CXpPOfcBZL2SvqIJPn/f36rpHP9+/yr/3MNAAAAixAhzfQul7TfOXfAOReX9A1JNy3wmhaMc67NOfe4/3m/vF+yV8t7Tm7zT7tN0usWZIELzMxaJL1a0pf8r03SSyR92z9lKT83tZKulXSrJDnn4s65HvHayQhJKjezkKQKSW1awq8d59wvJHWNOzzZa+UmSV91nock1ZnZynlZ6ALI99w4537qnEv6Xz4kqcX//CZJ33DOjTjnDkraL+/nGgAAABYhQprprZZ0NOfrVv/Ykmdm6yVdLOlhScudc23+TSckLV+odS2wf5b0J5LS/teNknpyfnlayq+fDZI6JH3Fbwf7kplViteOnHPHJP29pCPywpleSY+J1854k71W+P/0WL8p6Uf+5zw3AAAApxFCGsyImVVJ+o6k/+2c68u9zXlbhi25bcPM7EZJ7c65xxZ6LYtUSNIlkj7vnLtY0qDGtTYt4ddOvbyKhw2SVkmq1MR2FuRYqq+V6ZjZ/5HXlvq1hV4LAAAAikdIM71jktbkfN3iH1uyzCwsL6D5mnPudv/wyUx7gf+xfaHWt4CulvRaMzskry3uJfJmsNT5LSzS0n79tEpqdc497H/9bXmhDa8d6aWSDjrnOpxzCUm3y3s98doZa7LXCv+flmRm75J0o6Rf90MsiecGAADgtEJIM71HJW3xd1mJyBvAeMcCr2nB+DNWbpW0yzn3jzk33SHpZv/zmyV9b77XttCccx9xzrU459bLe53c45z7dUn3Snqjf9qSfG4kyTl3QtJRM9vqH7pe0k7x2pG8NqcrzazC/28s89zw2hlrstfKHZLe6e/ydKWk3py2qCXBzF4pr9Xytc65WM5Nd0h6q5lFzWyDvOHKjyzEGgEAADA9G/3HNkzGzF4lb9ZIUNKXnXN/tbArWjhmdo2k+yU9o9G5Kx+VN5fmW5LWSjos6c3OufFDP5cMM7tO0h855240s43yKmsaJD0h6TeccyMLuLwFY2YXyRuqHJF0QNK75YXFS/61Y2afkPQWea0qT0j6LXmzQ5bka8fMvi7pOklNkk5K+pik/1Ge14ofbH1WXotYTNK7nXM7FmDZ82KS5+YjkqKSTvmnPeSce69//v+RN6cmKa9F9UfjHxMAAACLAyENAAAAAADAIkC7EwAAAAAAwCJASAMAAAAAALAIENIAAAAAAAAsAoQ0AAAAAAAAiwAhDQAAAAAAwCJASANgWmaWMrMnc/58eJrz32tm7yzBdQ+ZWdNsHwcAAAAATgdswQ1gWmY24JyrWoDrHpK03TnXOd/XBgAAAID5RiUNgBnzK13+zsyeMbNHzGyzf/zjZvZH/ucfMLOdZva0mX3DP9ZgZv/jH3vIzC7wjzea2U/N7Dkz+5Iky7nWb/jXeNLMvmBmwQX4lgEAAABgzhDSAChE+bh2p7fk3NbrnDtf0mcl/XOe+35Y0sXOuQskvdc/9glJT/jHPirpq/7xj0l6wDl3rqTvSlorSWZ2tqS3SLraOXeRpJSkXy/lNwgAAAAACy200AsAcFoY8sORfL6e8/Gf8tz+tKSvmdn/SPof/9g1kt4gSc65e/wKmhpJ10p6vX/8B2bW7Z9/vaRLJT1qZpJULql9Ft8PAAAAACw6hDQAZstN8nnGq+WFL6+R9H/M7PwZXMMk3eac+8gM7gsAAAAApwXanQDM1ltyPj6Ye4OZBSStcc7dK+lDkmolVUm6X367kpldJ6nTOdcn6ReS3u4fv0FSvf9Qd0t6o5kt829rMLN1c/ctAQAAAMD8o5IGQCHKzezJnK9/7JzLbMNdb2ZPSxqR9LZx9wtK+k8zq5VXDfMZ51yPmX1c0pf9+8Uk3eyf/wlJXzez5yT9StIRSXLO7TSzP5X0Uz/4SUh6n6TDJf4+AQAAAGDBsAU3gBlji2wAAAAAKB3anQAAAAAAABYBKmkAAAAAAAAWASppAAAAAAAAFgFCGgAAAAAAgEWAkAYAAAAAAGARIKQBAAAAAABYBAhpAAAAAAAAFgFCGgAAAAAAgEXg/wNSd3SUWyI/HwAAAABJRU5ErkJggg==\n"},"metadata":{"needs_background":"light"}},{"name":"stdout","text":"- Episodes: 119\n- 100 episodes moving avg: 197.07000732421875\n","output_type":"stream"}]},{"cell_type":"markdown","source":"The goal is to reach the `195` moving average in a short amount of time. Previously, while I was using the **image processing** approach with DNN – I was not able to reach the `195` moving average in `1000` episodes. Now, After using **Positioning Features**. I have reached this required moving average in `119` episodes.\n\n<center>\n    <img src=\"https://i.imgur.com/0rqP23F.png\"/>\n</center>\n\n\nSo, is this the only solution we can apply?\n\n- Of course not. We can try different approaches such as enhancing the image processing by using a more advanced network which is using CNN, Transfer Learning, Image Augmentations, and other approaches that can define this optimization problem. I tried to change the gradient calculation using **Huber** instead of **MSE**. It didn't give me the speed or the required score, too. We can try using Adam Weight Optimization for having a much better learning rate for having a quicker, or a slower abundance of the values in the **Replay Memory**. ","metadata":{}},{"cell_type":"markdown","source":"<p style='background:LawnGreen;\n          border-radius: 25px;\n          padding: 20px;\n          color:ForestGreen;\n          font-size:21px;\n          border: 2px solid Lavender;\n          text-align:center'><b>I would be more than happy If you share with me your thoughts in the comments section!!!</b></p>","metadata":{}},{"cell_type":"markdown","source":"<center>\n________________________________\n</center>","metadata":{}},{"cell_type":"markdown","source":"<p style='text-align:center;'>Thanks for reaching this level of expermenting\nthe idea of</p>\n<center><p style='text-align:center;'><b>Deep Q-Network - Reinforcement Learning</b></p></center>\n<p style='text-align:center;'>AI Engineer: <a href='https://www.linkedin.com/in/drxavier997/'>Ahmed</a></p>\n<p style='text-align:center;'>Created at: 2022-02-21<br>","metadata":{}}]}