sr-dyna-tiny-test.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SR-Dyna (Tiny Test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 262,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import srdyna\n",
    "import importlib"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 263,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<module 'srdyna' from '/Users/jeremygordon/repos/sr-dyna/srdyna.py'>"
      ]
     },
     "execution_count": 263,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "importlib.reload(srdyna)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 264,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loaded 4x4 world with 18 states\n"
     ]
    }
   ],
   "source": [
    "# Detour Task\n",
    "EXPLORE_STEPS = 2000\n",
    "POST_REWARD_TRIALS = 10\n",
    "env = srdyna.SimpleGridWorld(world='worlds/tiny.txt')\n",
    "S_LOC = (0, 0)\n",
    "agent = srdyna.SRDyna(id=0, loc=S_LOC, env=env, post_step_replays=10, exp_lambda=1/5.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 265,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_srs(agent):\n",
    "    fig, axs = plt.subplots(1, 5, dpi=144)\n",
    "    for i, s in enumerate([0, 3, 12, 15, 16]):\n",
    "        agent.render_sr(s, ax=axs[i])\n",
    "    plt.show()    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 266,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Exploring...\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 5 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Learning reward W...\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 5 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Explore\n",
    "print(\"Exploring...\")\n",
    "for i in range(EXPLORE_STEPS):\n",
    "    agent.step(random_policy=True)\n",
    "\n",
    "agent.make_plots()\n",
    "plot_srs(agent)\n",
    "\n",
    "# Add reward\n",
    "R_LOC = (2, 2)\n",
    "env.add_reward(R_LOC, 10)\n",
    "\n",
    "print(\"Learning reward W...\")\n",
    "for i in range(POST_REWARD_TRIALS):\n",
    "    # Repeated trials from S (until reward reached)\n",
    "    agent.terminate_episode(reset_state=env.state_at_loc(S_LOC))\n",
    "    done = False\n",
    "    steps = 0\n",
    "    MAX_STEPS = 200\n",
    "    while not done and steps < MAX_STEPS:\n",
    "        done = agent.step(verbose=False)\n",
    "        steps += 1\n",
    "\n",
    "agent.make_plots()\n",
    "plot_srs(agent)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 267,
   "metadata": {},
   "outputs": [],
   "source": [
    "# agent.record_trials(title=\"tiny\", \n",
    "#                     learning=False,\n",
    "#                     start_locs=[(0, 0), (3, 0)])"
   ]
  }
 ],
 "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}