Continuous_Control_multi_cpu.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Continuous Control\n",
    "\n",
    "---\n",
    "\n",
    "You are welcome to use this coding environment to train your agent for the project.  Follow the instructions below to get started!\n",
    "\n",
    "### 1. Start the Environment\n",
    "\n",
    "Run the next code cell to install a few packages.  This line will take a few minutes to run!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[31mtensorflow 1.7.1 has requirement numpy>=1.13.3, but you'll have numpy 1.12.1 which is incompatible.\u001b[0m\r\n",
      "\u001b[31mipython 6.5.0 has requirement prompt-toolkit<2.0.0,>=1.0.15, but you'll have prompt-toolkit 3.0.16 which is incompatible.\u001b[0m\r\n"
     ]
    }
   ],
   "source": [
    "!pip -q install ./python"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The environments corresponding to both versions of the environment are already saved in the Workspace and can be accessed at the file paths provided below.  \n",
    "\n",
    "Please select one of the two options below for loading the environment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:unityagents:\n",
      "'Academy' started successfully!\n",
      "Unity Academy name: Academy\n",
      "        Number of Brains: 1\n",
      "        Number of External Brains : 1\n",
      "        Lesson number : 0\n",
      "        Reset Parameters :\n",
      "\t\tgoal_speed -> 1.0\n",
      "\t\tgoal_size -> 5.0\n",
      "Unity brain name: ReacherBrain\n",
      "        Number of Visual Observations (per agent): 0\n",
      "        Vector Observation space type: continuous\n",
      "        Vector Observation space size (per agent): 33\n",
      "        Number of stacked Vector Observation: 1\n",
      "        Vector Action space type: continuous\n",
      "        Vector Action space size (per agent): 4\n",
      "        Vector Action descriptions: , , , \n"
     ]
    }
   ],
   "source": [
    "from unityagents import UnityEnvironment\n",
    "import numpy as np\n",
    "\n",
    "single_agent = False\n",
    "if single_agent:\n",
    "    # select this option to load version 1 (with a single agent) of the environment\n",
    "    env = UnityEnvironment(file_name='/data/Reacher_One_Linux_NoVis/Reacher_One_Linux_NoVis.x86_64')\n",
    "else:\n",
    "    # select this option to load version 2 (with 20 agents) of the environment\n",
    "    env = UnityEnvironment(file_name='/data/Reacher_Linux_NoVis/Reacher.x86_64')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Environments contain **_brains_** which are responsible for deciding the actions of their associated agents. Here we check for the first brain available, and set it as the default brain we will be controlling from Python."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# get the default brain\n",
    "brain_name = env.brain_names[0]\n",
    "brain = env.brains[brain_name]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. Examine the State and Action Spaces\n",
    "\n",
    "Run the code cell below to print some information about the environment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of agents: 20\n",
      "Size of each action: 4\n",
      "There are 20 agents. Each observes a state with length: 33\n",
      "The state for the first agent looks like: [  0.00000000e+00  -4.00000000e+00   0.00000000e+00   1.00000000e+00\n",
      "  -0.00000000e+00  -0.00000000e+00  -4.37113883e-08   0.00000000e+00\n",
      "   0.00000000e+00   0.00000000e+00   0.00000000e+00   0.00000000e+00\n",
      "   0.00000000e+00   0.00000000e+00  -1.00000000e+01   0.00000000e+00\n",
      "   1.00000000e+00  -0.00000000e+00  -0.00000000e+00  -4.37113883e-08\n",
      "   0.00000000e+00   0.00000000e+00   0.00000000e+00   0.00000000e+00\n",
      "   0.00000000e+00   0.00000000e+00   5.75471878e+00  -1.00000000e+00\n",
      "   5.55726624e+00   0.00000000e+00   1.00000000e+00   0.00000000e+00\n",
      "  -1.68164849e-01]\n"
     ]
    }
   ],
   "source": [
    "# reset the environment\n",
    "env_info = env.reset(train_mode=True)[brain_name]\n",
    "\n",
    "# number of agents\n",
    "num_agents = len(env_info.agents)\n",
    "print('Number of agents:', num_agents)\n",
    "\n",
    "# size of each action\n",
    "action_size = brain.vector_action_space_size\n",
    "print('Size of each action:', action_size)\n",
    "\n",
    "# examine the state space \n",
    "states = env_info.vector_observations\n",
    "state_size = states.shape[1]\n",
    "print('There are {} agents. Each observes a state with length: {}'.format(states.shape[0], state_size))\n",
    "print('The state for the first agent looks like:', states[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3. Take Random Actions in the Environment\n",
    "\n",
    "In the next code cell, you will learn how to use the Python API to control the agent and receive feedback from the environment.\n",
    "\n",
    "Note that **in this coding environment, you will not be able to watch the agents while they are training**, and you should set `train_mode=True` to restart the environment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total score (averaged over agents) this episode: 0.10149999773129821\n"
     ]
    }
   ],
   "source": [
    "env_info = env.reset(train_mode=True)[brain_name]      # reset the environment    \n",
    "states = env_info.vector_observations                  # get the current state (for each agent)\n",
    "scores = np.zeros(num_agents)                          # initialize the score (for each agent)\n",
    "while True:\n",
    "    actions = np.random.randn(num_agents, action_size) # select an action (for each agent)\n",
    "    actions = np.clip(actions, -1, 1)                  # all actions between -1 and 1\n",
    "    env_info = env.step(actions)[brain_name]           # send all actions to tne environment\n",
    "    next_states = env_info.vector_observations         # get next state (for each agent)\n",
    "    rewards = env_info.rewards                         # get reward (for each agent)\n",
    "    dones = env_info.local_done                        # see if episode finished\n",
    "    scores += env_info.rewards                         # update the score (for each agent)\n",
    "    states = next_states                               # roll over states to next time step\n",
    "    if np.any(dones):                                  # exit loop if episode finished\n",
    "        break\n",
    "print('Total score (averaged over agents) this episode: {}'.format(np.mean(scores)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "When finished, you can close the environment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "env.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4. It's Your Turn!\n",
    "\n",
    "Now it's your turn to train your own agent to solve the environment!  A few **important notes**:\n",
    "- When training the environment, set `train_mode=True`, so that the line for resetting the environment looks like the following:\n",
    "```python\n",
    "env_info = env.reset(train_mode=True)[brain_name]\n",
    "```\n",
    "- To structure your work, you're welcome to work directly in this Jupyter notebook, or you might like to start over with a new file!  You can see the list of files in the workspace by clicking on **_Jupyter_** in the top left corner of the notebook.\n",
    "- In this coding environment, you will not be able to watch the agents while they are training.  However, **_after training the agents_**, you can download the saved model weights to watch the agents on your own machine! "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from agent import ContinuousControlAgent\n",
    "from common import *\n",
    "from collections import deque\n",
    "from support import Experience, ReplayBuffer\n",
    "import config\n",
    "import random\n",
    "import datetime\n",
    "from workspace_utils import active_session"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def step_tuple(env_info):\n",
    "    \"\"\" Returns a tuple of next state, reward, and done when the agent steps through the environment based\n",
    "        on the action taken\n",
    "        :param env_info: Object holding information about the environment at a certain point\n",
    "    \"\"\"\n",
    "    if single_agent:\n",
    "        return env_info.vector_observations[0], env_info.rewards[0], env_info.local_done[0]\n",
    "    return env_info.vector_observations, env_info.rewards, env_info.local_done\n",
    "\n",
    "\n",
    "def plot(stats):\n",
    "    scores = stats[\"scores\"]\n",
    "    episodes = stats[\"episodes\"]\n",
    "    plt.ylabel(\"Score\")\n",
    "    plt.xlabel(\"Episode #\")\n",
    "    plt.plot(episodes, scores)\n",
    "    fig = plt.figure()\n",
    "    fig.patch.set_facecolor('white')\n",
    "    plt.show()\n",
    "\n",
    "\n",
    "def ddpg(agent: ContinuousControlAgent, env: UnityEnvironment, num_episodes=2000, target=30., max_time_steps=500,\n",
    "         saved_model=\"checkpoint.pth\"):\n",
    "    \"\"\" Train an agent using the DDPG algorithm\n",
    "\n",
    "        :param agent: a continuous control agent\n",
    "        :param env: environment the agent interacts with\n",
    "        :param num_episodes: the number of episodes to train the agent\n",
    "        :param target: The average target score the agent needs to achieve for optimal performance\n",
    "        :param max_time_steps: Maximum time steps per episode\n",
    "        :param saved_model: The file path to save the model weights\n",
    "    \"\"\"\n",
    "    now = datetime.datetime.now()\n",
    "    print(now, \"- Training {}\".format(\"single agent\" if single_agent else \"{} agents\".format(20))\n",
    "          + \" for max {} episodes. Target score to reach is {}\".format(num_episodes, target))\n",
    "    # collections to help keep track of the score\n",
    "    scores_deque = deque(maxlen=100)\n",
    "    scores = []\n",
    "    stats = {\"scores\": [], \"episodes\": []}  # collects stats for plotting purposes\n",
    "    mean_score = 0.\n",
    "\n",
    "    for episode in range(1, num_episodes + 1):\n",
    "        env_info = env.reset(train_mode=True)[brain_name]  # reset the environment\n",
    "        states = env_info.vector_observations[0] if single_agent else env_info.vector_observations\n",
    "        agent.reset()                                                # reset the agent noise\n",
    "        score = 0 if single_agent else np.zeros(num_agents)\n",
    "\n",
    "        for _ in range(max_time_steps):\n",
    "            actions = agent.act(states)\n",
    "            env_info = env.step(actions)[brain_name]\n",
    "            next_states, rewards, dones = step_tuple(env_info)\n",
    "            if single_agent:\n",
    "                agent.step(Experience(states, actions, rewards, next_states, dones))\n",
    "            else:\n",
    "                for idx in random.sample(range(num_agents), 10):\n",
    "                    agent.step(Experience(states[idx], actions[idx], rewards[idx], next_states[idx], dones[idx]))\n",
    "            states = next_states\n",
    "            score += rewards\n",
    "            if np.any(dones):\n",
    "                break\n",
    "        scores_deque.append(score)\n",
    "        scores.append(score)\n",
    "        mean_score = np.mean(scores_deque)\n",
    "\n",
    "        print('\\rEpisode {}\\tAverage Score: {:.2f}'.format(episode, mean_score), end=\"\")\n",
    "        if episode % 100 == 0:\n",
    "            print('\\rEpisode {}\\tAverage Score: {:.2f}'.format(episode, mean_score))\n",
    "\n",
    "        stats[\"scores\"].append(score if single_agent else np.mean(score))\n",
    "        stats[\"episodes\"].append(episode)\n",
    "\n",
    "        if mean_score >= target:\n",
    "            print('\\nEnvironment solved in {:d} episodes!\\tAverage Score: {:.2f}'.format(episode, mean_score))\n",
    "            print(\"Target score of {0} has been reached. Saving model to {1}\".format(target, saved_model))\n",
    "            torch.save(agent.local_actor_network().state_dict(), saved_model)\n",
    "            break\n",
    "\n",
    "    now = datetime.datetime.now()\n",
    "    print(now, \"- Finished training \" + \"successfully!\" if mean_score >= target else \"unsuccessfully!\")\n",
    "    return scores, stats\n",
    "\n",
    "\n",
    "def run(agent_: ContinuousControlAgent, env_: UnityEnvironment, num_episodes=2000, max_time_steps=1000, target=30.,\n",
    "        saved_model=\"checkpoint.pth\"):\n",
    "    try:\n",
    "        _, stats = ddpg(agent_, env_, num_episodes=num_episodes, target=target, max_time_steps=max_time_steps,\n",
    "                        saved_model=saved_model)\n",
    "        plot(stats)\n",
    "    finally:\n",
    "        # make sure the environment gets closed regardless of what happens\n",
    "        env_.close()\n",
    "\n",
    "\n",
    "def test(agent_: ContinuousControlAgent, filename):\n",
    "    agent_.local_actor_network().load_state_dict(torch.load(filename))\n",
    "\n",
    "    env_info = env.reset(train_mode=False)[brain_name]  # reset the environment\n",
    "    state = env_info.vector_observations[0] if single_agent else env_info.vector_observations  # get the current state\n",
    "    score = 0 if single_agent else np.zeros(num_agents)    # initialize the score\n",
    "    while True:\n",
    "        action = agent_.act(state, add_noise=False)  # select an action\n",
    "        env_info = env.step(action)[brain_name]  # send the action to the environment\n",
    "        next_state = env_info.vector_observations[0] if single_agent else env_info.vector_observations  # get the next state\n",
    "        reward = env_info.rewards[0] if single_agent else env_info.rewards  # get the reward\n",
    "        done = env_info.local_done[0] if single_agent else env_info.local_done  # see if episode has finished\n",
    "        score += reward  # update the score\n",
    "        state = next_state  # roll over the state to next time step\n",
    "        if np.any(done):  # exit loop if episode finished\n",
    "            break\n",
    "\n",
    "    print(\"Score: {}\".format(score))\n",
    "\n",
    "    env.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Initialised 'Actor: Local' model\n",
      "Initialised 'Actor: Target' model\n",
      "Initialised 'Critic: Local' model\n",
      "Initialised 'Critic: Target' model\n",
      "2021-03-09 07:31:27.138649 - Training 20 agents for max 3000 episodes. Target score to reach is 30.0\n",
      "Agent has not collected enough experiences to start learning. Collected 32, requires at least 128 experiences\n",
      "Agent has not collected enough experiences to start learning. Collected 64, requires at least 128 experiences\n",
      "Agent has not collected enough experiences to start learning. Collected 96, requires at least 128 experiences\n",
      "Agent has not collected enough experiences to start learning. Collected 128, requires at least 128 experiences\n",
      "Episode 97\tAverage Score: 30.08\n",
      "Environment solved in 97 episodes!\tAverage Score: 30.08\n",
      "Target score of 30.0 has been reached. Saving model to checkpoint_multi.pth\n",
      "2021-03-09 10:38:41.130997 - Finished training successfully!\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f3a59738f28>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f3a55419518>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "ename": "TypeError",
     "evalue": "'NoneType' object is not iterable",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mTypeError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-7-d1a40ef82aaf>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0msaved_model\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m'checkpoint_single.pth'\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0msingle_agent\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0;34m'checkpoint_multi.pth'\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;32mwith\u001b[0m \u001b[0mactive_session\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 6\u001b[0;31m     \u001b[0mscores\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mstats\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mrun\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0magent\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0menv\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnum_episodes\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m3000\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmax_time_steps\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m1000\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m30.\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msaved_model\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msaved_model\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;31mTypeError\u001b[0m: 'NoneType' object is not iterable"
     ]
    }
   ],
   "source": [
    "memory = ReplayBuffer(action_size, config.BUFFER_SIZE, config.BATCH_SIZE, random_seed=0)\n",
    "agent = ContinuousControlAgent(state_size, action_size, 0, memory=memory, update_frequency=2)\n",
    "stats = None\n",
    "saved_model = 'checkpoint_single.pth' if single_agent else 'checkpoint_multi.pth'\n",
    "with active_session():\n",
    "    scores, stats = run(agent, env, num_episodes=3000, max_time_steps=1000, target=30., saved_model=saved_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}