analyse_training.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SpaceInvaders-v0\n",
      "SpaceInvaders-ram-v4\n",
      "SpaceInvaders-ramDeterministic-v4\n",
      "SpaceInvadersNoFrameskip-v4\n",
      "SpaceInvadersDeterministic-v4\n",
      "SpaceInvaders-v4\n",
      "SpaceInvaders-ramNoFrameskip-v0\n",
      "SpaceInvadersDeterministic-v0\n",
      "SpaceInvaders-ramNoFrameskip-v4\n",
      "SpaceInvaders-ram-v0\n",
      "SpaceInvaders-ramDeterministic-v0\n",
      "SpaceInvadersNoFrameskip-v0\n"
     ]
    }
   ],
   "source": [
    "import gym\n",
    "for env_spec in gym.envs.registry.all():\n",
    "    env_name = env_spec.id\n",
    "    if 'Space' in env_name:\n",
    "        print(env_name)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Development of loss for VAE training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import pandas as pd\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "plt.rcParams.update({'font.size': 14})\n",
    "\n",
    "# Read training statistics out of the log file.\n",
    "RUN_NAME = 'sequential_training_wide_shots'\n",
    "RESULTS_DIR = 'results/{}'.format(RUN_NAME)\n",
    "vae_log_fn = ('{}/vae_training.log'\n",
    "              .format(RESULTS_DIR))\n",
    "training_statistics_pre_df = {}\n",
    "with open(vae_log_fn, 'r') as f:\n",
    "    epoch = -1\n",
    "    walltime = -1\n",
    "    train_loss = -1\n",
    "    \n",
    "    for line in f.readlines():\n",
    "        if 'Started Epoch' in line:\n",
    "            epoch = int(line.strip().split(' ')[-1])\n",
    "        if 'Walltime:' in line:\n",
    "            walltime = float(line.strip().split(' ')[-1])\n",
    "        if 'Train loss:' in line:\n",
    "            train_loss = float(line.strip().split(' ')[-1])\n",
    "            # Train loss should be the last entry per episode\n",
    "            entry = {'walltime [s]': walltime,\n",
    "                     'train loss': train_loss}\n",
    "            training_statistics_pre_df[epoch] = entry\n",
    "\n",
    "# Plot the training progress\n",
    "vae_training_statisitcs = pd.DataFrame(training_statistics_pre_df).T\n",
    "_ = vae_training_statisitcs.plot(y='train loss',\n",
    "                                 logy=True,\n",
    "                                 ylim=(0.0001, 10.),\n",
    "                                 title='VAE Train loss over episodes',\n",
    "                                 figsize=(12, 8))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Development of reward over generations for agent training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "slideshow": {
     "slide_type": "-"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import pandas as pd\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "plt.rcParams.update({'font.size': 14})\n",
    "\n",
    "# Read training statistics out of the log file.\n",
    "RUN_NAME = 'sequential_training'\n",
    "RESULTS_DIR = 'results/{}'.format(RUN_NAME)\n",
    "agent_log_fn = ('{}/agent_training.log'\n",
    "              .format(RESULTS_DIR))\n",
    "training_statistics_pre_df = {}\n",
    "with open(agent_log_fn, 'r') as f:\n",
    "    generation = -1\n",
    "    walltime = -1\n",
    "    reward_mean = -1\n",
    "    reward_max = -1\n",
    "    reward_std = -1\n",
    "    \n",
    "    for line in f.readlines():\n",
    "        if 'canidates of generation' in line:\n",
    "            generation = int(line.strip().split(' ')[-1])\n",
    "        if 'Finished generation in' in line:\n",
    "            walltime = float(line.strip().split(' ')[-2])\n",
    "        if 'Mean reward:' in line:\n",
    "            reward_mean = float(line.strip().split(' ')[-1])\n",
    "        if 'Max reward:' in line:\n",
    "            reward_max = float(line.strip().split(' ')[-1])\n",
    "        if 'Reward std:' in line:\n",
    "            reward_std = float(line.strip().split(' ')[-1])\n",
    "            \n",
    "            # Train loss should be the last entry per episode\n",
    "            entry = {'walltime [s]': walltime,\n",
    "                     'average reward of generation': reward_mean,\n",
    "                     'reward of best agent': reward_max,\n",
    "                     'reward std': reward_std}\n",
    "            training_statistics_pre_df[generation] = entry\n",
    "\n",
    "# Plot the training progress\n",
    "agent_training_statisitcs = pd.DataFrame(training_statistics_pre_df).T\n",
    "ax = agent_training_statisitcs.plot(y=['average reward of generation', 'reward of best agent'],\n",
    "                                   logy=False,\n",
    "                                   ylim=(0, 700),\n",
    "                                   title='Progress of Agent training',\n",
    "                                   figsize=(8, 6))\n",
    "_ = ax.set_xlabel('generation')\n",
    "_ = ax.set_ylabel('reward')\n",
    "\n",
    "# Highlight the best agent found.\n",
    "gen_best_agent = agent_training_statisitcs['reward of best agent'].idxmax()\n",
    "reward_best_agent = agent_training_statisitcs['reward of best agent'][gen_best_agent]\n",
    "ax.plot(gen_best_agent,\n",
    "        reward_best_agent, \n",
    "        'o', \n",
    "        markeredgecolor='r',\n",
    "        markersize=15.,\n",
    "        markerfacecolor=(1, 1, 1, 0.),\n",
    "        markeredgewidth=2.,\n",
    "        label='best agent found')\n",
    "\n",
    "ax.text(gen_best_agent + 6,\n",
    "        reward_best_agent-20, \n",
    "        'generation: {}\\nreward: {}'.format(gen_best_agent, reward_best_agent))\n",
    "_ = ax.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Reference: Get average scores following random policy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>number_of_steps</th>\n",
       "      <th>total_reward</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1000.000000</td>\n",
       "      <td>1000.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>701.603000</td>\n",
       "      <td>155.525000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>220.017142</td>\n",
       "      <td>105.262907</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>355.000000</td>\n",
       "      <td>5.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>547.000000</td>\n",
       "      <td>90.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>665.000000</td>\n",
       "      <td>130.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>798.250000</td>\n",
       "      <td>196.250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1627.000000</td>\n",
       "      <td>800.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       number_of_steps  total_reward\n",
       "count      1000.000000   1000.000000\n",
       "mean        701.603000    155.525000\n",
       "std         220.017142    105.262907\n",
       "min         355.000000      5.000000\n",
       "25%         547.000000     90.000000\n",
       "50%         665.000000    130.000000\n",
       "75%         798.250000    196.250000\n",
       "max        1627.000000    800.000000"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from extractor import ExtractWorker, RandomPolicy\n",
    "\n",
    "ew = ExtractWorker(env_name='SpaceInvaders-v4')\n",
    "stats = ew.extract_episode_statistics(policy_class=RandomPolicy, \n",
    "                                      n_episodes=1000)\n",
    "stats.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualize average random episode"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>number_of_steps</th>\n",
       "      <th>total_reward</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>50.000000</td>\n",
       "      <td>50.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>740.120000</td>\n",
       "      <td>193.800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>241.332072</td>\n",
       "      <td>139.626177</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>380.000000</td>\n",
       "      <td>20.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>577.000000</td>\n",
       "      <td>105.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>675.000000</td>\n",
       "      <td>147.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>947.750000</td>\n",
       "      <td>260.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1320.000000</td>\n",
       "      <td>515.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       number_of_steps  total_reward\n",
       "count        50.000000     50.000000\n",
       "mean        740.120000    193.800000\n",
       "std         241.332072    139.626177\n",
       "min         380.000000     20.000000\n",
       "25%         577.000000    105.000000\n",
       "50%         675.000000    147.500000\n",
       "75%         947.750000    260.000000\n",
       "max        1320.000000    515.000000"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from extractor import ExtractWorker, RandomPolicy\n",
    "\n",
    "ew = ExtractWorker(env_name='SpaceInvaders-v4')\n",
    "stats, obs = ew.extract_episode_statistics(policy_class=RandomPolicy, \n",
    "                                           n_episodes=50,\n",
    "                                           return_observations=True)\n",
    "stats.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.models import load_model, Model\n",
    "from keras.layers import Lambda\n",
    "\n",
    "RUN_NAME = 'sequential_training'\n",
    "RESULTS_DIR = 'results/{}'.format(RUN_NAME)\n",
    "encoder_model = load_model('{}/encoder_model.h5'.format(RESULTS_DIR))\n",
    "decoder_model = load_model('{}/decoder_model.h5'.format(RESULTS_DIR))\n",
    "\n",
    "# Load the VAE\n",
    "# If trained in variational mode the output will be a list of\n",
    "# 3 layers. We only require z, the last of the three.\n",
    "if isinstance(encoder_model.output_shape, list):\n",
    "    new_out = Lambda(lambda x: x[2])(encoder_model.output)\n",
    "    encoder_model = Model(encoder_model.inputs, new_out)\n",
    "\n",
    "inputs = encoder_model.inputs\n",
    "outputs_vae = decoder_model(encoder_model.outputs)\n",
    "vae = Model(inputs, outputs_vae)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from IPython.display import HTML\n",
    "\n",
    "from helpers import animate_episode\n",
    "\n",
    "# Find the id of the episde closest to average. \n",
    "seed_average_reward = (stats - stats.mean())['total_reward'].abs().idxmin()\n",
    "\n",
    "average_reward_obs = np.stack(ew.create_black_boundary(obs[seed_average_reward], (256, 192)))\n",
    "average_reward_obs = average_reward_obs.astype('float32')/255\n",
    "average_reward_obs_decoded = vae.predict(average_reward_obs)\n",
    "\n",
    "ani_average_reward = animate_episode(average_reward_obs)\n",
    "HTML(ani_average_reward.to_html5_video())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ani_average_reward_decoded = animate_episode(average_reward_obs_decoded)\n",
    "HTML(ani_average_reward_decoded.to_html5_video())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Extract some episodes to validate the current best agent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "from keras.models import load_model, Model\n",
    "from keras.layers import Lambda\n",
    "\n",
    "from helpers import widen_laser_shots\n",
    "from extractor import ExtractWorker, KerasModelPolicy, RandomPolicy\n",
    "\n",
    "RUN_NAME = 'sequential_training'\n",
    "RESULTS_DIR = 'results/{}'.format(RUN_NAME)\n",
    "AGENT_MODEL_CHECKPOINTS_DIR = '{}/agent_model_checkpoints'.format(RESULTS_DIR)\n",
    "probabilistic_mode = False\n",
    "\n",
    "ew = ExtractWorker(env_name='SpaceInvaders-v4')\n",
    "\n",
    "# Ignore the warning, we don't need the model compiled as\n",
    "# we won't use keras to train it anymore nor the agent.\n",
    "encoder_model = load_model('{}/encoder_model.h5'.format(RESULTS_DIR))\n",
    "decoder_model = load_model('{}/decoder_model.h5'.format(RESULTS_DIR))\n",
    "try:\n",
    "    agent_model = load_model('{}/best_agent_model.h5'.format(RESULTS_DIR))\n",
    "except:\n",
    "    agent_model_fns = sorted(os.listdir(AGENT_MODEL_CHECKPOINTS_DIR))\n",
    "    agent_model_latest_fn = AGENT_MODEL_CHECKPOINTS_DIR + '/' + agent_model_fns[-1]\n",
    "    agent_model = load_model(agent_model_latest_fn)\n",
    "    \n",
    "# If trained in variational mode the output will be a list of\n",
    "# 3 layers. We only require z, the last of the three.\n",
    "if isinstance(encoder_model.output_shape, list):\n",
    "    new_out = Lambda(lambda x: x[2])(encoder_model.output)\n",
    "    encoder_model = Model(encoder_model.inputs, new_out)\n",
    "\n",
    "# Build the model for the KerasPolicy\n",
    "inputs = encoder_model.inputs\n",
    "outputs = agent_model(encoder_model.outputs)\n",
    "full_model = Model(inputs, outputs)\n",
    "\n",
    "# Also build a VAE to allow analysis of the decoded output.\n",
    "outputs_vae = decoder_model(encoder_model.outputs)\n",
    "vae = Model(inputs, outputs_vae)\n",
    "\n",
    "# This function calls requires ~99% of the CPU time of this loop.\n",
    "# CPU time rises roughly linearly with n_episodes.\n",
    "policy_kw_args = {'model': full_model,\n",
    "                  'probabilistic_mode': probabilistic_mode}\n",
    "stats, obs = ew.extract_episode_statistics(policy_class=KerasModelPolicy, \n",
    "                                           n_episodes=50, \n",
    "                                           policy_kw_args=policy_kw_args,\n",
    "                                           return_observations=True)\n",
    "stats.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualize average and best episode of the agent run plus the correpsonding version after VAE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from IPython.display import HTML\n",
    "\n",
    "from helpers import animate_episode\n",
    "\n",
    "seed_max_reward = stats['total_reward'].idxmax()\n",
    "\n",
    "max_reward_obs = np.stack(ew.create_black_boundary(obs[seed_max_reward], (256, 192)))\n",
    "max_reward_obs = max_reward_obs.astype('float32')/255\n",
    "max_reward_obs_decoded = vae.predict(max_reward_obs)\n",
    "\n",
    "ani_max_reward = animate_episode(max_reward_obs)\n",
    "HTML(ani_max_reward.to_html5_video())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ani_max_reward_decoded = animate_episode(max_reward_obs_decoded)\n",
    "HTML(ani_max_reward_decoded.to_html5_video())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from IPython.display import HTML\n",
    "\n",
    "from helpers import animate_episode\n",
    "\n",
    "# Find the id of the episde closest to average. \n",
    "seed_average_reward = (stats - stats.mean())['total_reward'].abs().idxmin()\n",
    "\n",
    "average_reward_obs = np.stack(ew.create_black_boundary(obs[seed_average_reward], (256, 192)))\n",
    "average_reward_obs = average_reward_obs.astype('float32')/255\n",
    "average_reward_obs_decoded = vae.predict(average_reward_obs)\n",
    "\n",
    "ani_average_reward = animate_episode(average_reward_obs)\n",
    "HTML(ani_average_reward.to_html5_video())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ani_average_reward_decoded = animate_episode(average_reward_obs_decoded)\n",
    "HTML(ani_average_reward_decoded.to_html5_video())"
   ]
  }
 ],
 "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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}