Advantage_Actor_Critic(A2C).ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "3w04_bU_4H9g"
      },
      "outputs": [],
      "source": [
        "import gym\n",
        "import torch\n",
        "import torch.nn as nn\n",
        "from torch.distributions import Categorical\n",
        "import matplotlib.pyplot as plt\n",
        "!pip install gym[classic_control]\n",
        "#!pip install gym[box2d] #for lunarlander\n",
        "!apt update\n",
        "!apt install xvfb\n",
        "!pip install pyvirtualdisplay\n",
        "!pip install gym-notebook-wrapper\n",
        "import gnwrapper\n",
        "!nvidia-smi\n",
        "print(torch.cuda.is_available())"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kWwb1Cd_gyJn",
        "outputId": "806edf70-2b06-4bc4-da60-733454f4afe5"
      },
      "outputs": [],
      "source": [
        "class Agent(nn.Module):\n",
        "    def __init__(self, input_dim, output_dim, width):\n",
        "        super().__init__()\n",
        "        self.P = PolicyNetwork(input_dim, output_dim, width)\n",
        "        self.P.to(device)\n",
        "        self.P.train()  \n",
        "        self.P_optimizer = torch.optim.Adam(self.P.parameters(), lr=0.0003)\n",
        "        self.V = ValueNetwork(input_dim, output_dim, width)\n",
        "        self.V.to(device)        \n",
        "        self.V.train()\n",
        "        self.V_optimizer = torch.optim.Adam(self.V.parameters(), lr=0.0003) \n",
        "        self.trajectory = []       \n",
        "        self.env = gym.make(game_name)        \n",
        "\n",
        "    def self_play_A2C(self, max_timestep=1000000):\n",
        "        game_score = 0\n",
        "        state = self.env.reset()\n",
        "        for _ in range(max_timestep):\n",
        "            output = self.P(torch.from_numpy(state).float().to(device)) \n",
        "            inferenced_v = self.V(torch.from_numpy(state).float().to(device))\n",
        "            prob_distribution = Categorical(output) \n",
        "            action = prob_distribution.sample() \n",
        "            state, r, done, _ = self.env.step(action.item()) \n",
        "            with torch.no_grad():\n",
        "                inferenced_v_from_next_s = self.V(torch.from_numpy(state).float().to(device))\n",
        "            if done==True:\n",
        "                inferenced_v_from_next_s = 0  \n",
        "            self.trajectory.append((prob_distribution.log_prob(action), r, inferenced_v, inferenced_v_from_next_s))\n",
        "            game_score += r \n",
        "            if done:\n",
        "                break\n",
        "        return game_score\n",
        "\n",
        "    def update_weights_A2C(self):\n",
        "        A = 0 #GAE\n",
        "        lam = 0.9        \n",
        "        self.V_optimizer.zero_grad()\n",
        "        self.P_optimizer.zero_grad()\n",
        "        for gradient_policy_a_s, r, v, v_next in self.trajectory[::-1]: \n",
        "            with torch.no_grad():           \n",
        "                delta = r + 0.99* v_next - v\n",
        "                A += (0.99 * lam) * delta  \n",
        "                V_target = A + v   \n",
        "            V_loss = (V_target.item() -  v)**2          \n",
        "            V_loss.backward()            \n",
        "            P_loss = -1 * A.item() * gradient_policy_a_s            \n",
        "            P_loss.backward()             \n",
        "        self.V_optimizer.step()   \n",
        "        self.P_optimizer.step()             \n",
        "        self.trajectory.clear() \n",
        "        return \n",
        "   \n",
        "class PolicyNetwork(nn.Module):  \n",
        "    def __init__(self, input_dim, output_dim, width):\n",
        "        super().__init__()\n",
        "        self.layer1 = torch.nn.Linear(input_dim, width)\n",
        "        self.layer2 = torch.nn.Linear(width, width) \n",
        "        self.layer3 = torch.nn.Linear(width, output_dim) \n",
        "\n",
        "    def forward(self, x):\n",
        "        x = self.layer1(x)\n",
        "        x = torch.nn.functional.relu(x)\n",
        "        x = self.layer2(x)\n",
        "        x = torch.nn.functional.relu(x)\n",
        "        x = self.layer3(x)\n",
        "        x = torch.nn.functional.softmax(x, dim=0)\n",
        "        return x\n",
        "\n",
        "class ValueNetwork(nn.Module): \n",
        "    def __init__(self, input_dim, output_dim, width):\n",
        "        super().__init__()\n",
        "        self.layer1 = torch.nn.Linear(input_dim, width)\n",
        "        self.layer2 = torch.nn.Linear(width, width) \n",
        "        self.layer3 = torch.nn.Linear(width, 1) \n",
        "\n",
        "    def forward(self, x):\n",
        "        x = self.layer1(x)\n",
        "        x = torch.nn.functional.relu(x)\n",
        "        x = self.layer2(x)\n",
        "        x = torch.nn.functional.relu(x)\n",
        "        x = self.layer3(x)\n",
        "        return x\n",
        "\n",
        "device = torch.device('cuda:0') if torch.cuda.is_available() else torch.device('cpu')\n",
        "score_arr = []\n",
        "game_name = 'CartPole-v1' #'LunarLander-v2'\n",
        "env = gym.make(game_name) \n",
        "agent = Agent(env.observation_space.shape[0], env.action_space.n, 128) \n",
        "print(agent)\n",
        "env.close()\n",
        "\n",
        "#Self play, weight update\n",
        "episode_nums = 500 \n",
        "for i in range(episode_nums):    \n",
        "    game_score = agent.self_play_A2C()\n",
        "    agent.update_weights_A2C()    \n",
        "    score_arr.append(game_score)  \n",
        "    if i%50==0 : print('episode', i)    \n",
        "torch.save(agent.state_dict(), 'weights.pt') \n",
        "agent.env.close()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "id": "_1Cz8LkVy4du",
        "outputId": "d5d00c2c-4b34-410d-eeb5-920681eed86c"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f09eda6d090>"
            ]
          },
          "execution_count": 11,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "#Score\n",
        "plt.plot(score_arr, label ='score')\n",
        "plt.legend(loc='upper left')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 979
        },
        "id": "KTGyfRFEzUHG",
        "outputId": "858562a8-41a9-4b46-dd9e-b76b56adf2a5"
      },
      "outputs": [],
      "source": [
        "#play video\n",
        "agent.load_state_dict(torch.load(\"weights.pt\"))\n",
        "env = gnwrapper.LoopAnimation(gym.make(game_name)) \n",
        "state = env.reset()\n",
        "for _ in range(200):\n",
        "    with torch.no_grad():\n",
        "        output = agent.P(torch.from_numpy(state).float().to(device)) \n",
        "        prob_distribution = Categorical(output) \n",
        "        action = prob_distribution.sample() \n",
        "    env.render()\n",
        "    state, rew, done, _ = env.step(action.item())\n",
        "    if done:\n",
        "        state = env.reset()\n",
        "env.display()"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "provenance": []
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}