sr-dyna-latent-learning.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SR-Dyna (Latent Learning Task)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import srdyna\n",
    "import importlib"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<module 'srdyna' from '/Users/jeremygordon/repos/sr-dyna/srdyna.py'>"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "importlib.reload(srdyna)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loaded 10x10 world with 102 states\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 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "REPLAY = \"sufficient\"\n",
    "EXPLORE_STEPS = 25000\n",
    "POST_REWARD_STEPS = 20\n",
    "REPLAY_STEPS = {\n",
    "    \"insufficient\": 10,\n",
    "    \"sufficient\": 10000\n",
    "}[REPLAY]\n",
    "env = srdyna.SimpleGridWorld(world='worlds/latent_learning.txt')\n",
    "agent = srdyna.SRDyna(id=0, loc=(0, 2), env=env)\n",
    "\n",
    "R1_LOC = (9, 8)\n",
    "env.add_reward(R1_LOC, 0)  # Initially empty, but terminates\n",
    "\n",
    "# Explore\n",
    "for i in range(EXPLORE_STEPS):\n",
    "    agent.step(random_policy=True)\n",
    "\n",
    "# Increase reward\n",
    "env.reward_locs[R1_LOC]['reward'] = 10  \n",
    "\n",
    "# One-step replay samples from random sa's\n",
    "agent.learn_offline(k=REPLAY_STEPS)\n",
    "\n",
    "for i in range(POST_REWARD_STEPS):\n",
    "    # Repeated one-step runs from R1\n",
    "    agent.terminate_episode(reset_state=env.state_at_loc(R1_LOC))\n",
    "    agent.step(verbose=False)\n",
    "\n",
    "agent.make_plots(sr_state=9)\n",
    "\n",
    "# One-step replay samples from random sa's\n",
    "agent.learn_offline(k=REPLAY_STEPS)\n",
    "    \n",
    "agent.make_plots(sr_state=9)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfcAAAB1CAYAAABTeqhAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjAsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8GearUAAAMDElEQVR4nO3dX4hc53nH8e+vcm2DC41S68JYki1TU8eFENuDMQTSQmtb9YUUSKEylNjFQaSNW2ivXHIRUC6atheBUJVYpIKkF5YbX22gRog6Jje1oxFx/S84Wau0ljDYsVzfOMiV8/RijtPReuUZ7c7uzJ73+4HDznnf884+q9/uPjOzR2dSVUiSpP74lXkXIEmSZsvmLklSz9jcJUnqGZu7JEk9Y3OXJKlnbO6SJPXMxOae5GiSN5K8eIn5JPlGkuUkzye5fWzugSQ/7bYHZlm4NoZ5t8Os22HWDaqqj9yAzwC3Ay9eYv4+4EkgwF3As934x4HT3cft3e3tkz6f23w3825nM+t2NrNub5v4zL2qfgCc+4hD9gPfqZFngI8luQ64FzhRVeeq6m3gBLB30ufTfJl3O8y6HWbdnln8zf164LWx/TPd2KXGtbWZdzvMuh1m3TNXzLsAgCQHgYMA11xzzR233HLLnCva+k6dOvWzqtox7zpWY96zZdZtWdS8zXr21pP1LJr7WWDX2P7Obuws8Lsrxp9e7Q6q6ghwBGAwGNRwOJxBWW1L8l8bdNfmvWDMui0blLdZL6D1ZD2Ll+WXgM93Z1veBbxTVa8Dx4F7kmxPsh24pxvT1mbe7TDrdph1z0x85p7kMUaP3K5Ncgb4CvCrAFX1TeBfGZ1puQy8C/xJN3cuyVeBk91dHaqqjzqhQwvAvNth1u0w6/ZMbO5Vdf+E+QK+dIm5o8DRtZWmeTDvdph1O8y6PV6hTpKknrG5S5LUMzZ3SZJ6xuYuSVLP2NwlSeoZm7skST1jc5ckqWds7pIk9YzNXZKknrG5S5LUMzZ3SZJ6xuYuSVLPTNXck+xN8kqS5SSPrDL/9STPddtPkvzP2Nz7Y3NLsyxes2fW7TDrtph3W6Z5y9dtwGHgbuAMcDLJUlW9/MExVfWXY8f/OXDb2F38vKo+NbuStVHMuh1m3Rbzbs80z9zvBJar6nRVvQccA/Z/xPH3A4/NojhtOrNuh1m3xbwbM01zvx54bWz/TDf2IUluAPYAT40NX51kmOSZJJ9dc6XaDGbdDrNui3k3ZuLL8pfpAPBEVb0/NnZDVZ1NchPwVJIXqurV8UVJDgIHAXbv3j3jkrRB1pQ1mPcWZNZt8fd4D0zzzP0ssGtsf2c3tpoDrHgpp6rOdh9PA09z8d9xPjjmSFUNqmqwY8eOKUrSBtnwrLt5854/s26Lv8cbM01zPwncnGRPkisZBf+hsyWT3AJsB/59bGx7kqu629cCnwZeXrlWC8Os22HWbTHvxkx8Wb6qLiR5GDgObAOOVtVLSQ4Bw6r64BvkAHCsqmps+SeAR5P8gtEDia+Nn52pxWLW7TDrtph3e3JxhvM3GAxqOBzOu4wtL8mpqhrMu45JzHv9zLotWyFvs56N9WTtFeokSeoZm7skST1jc5ckqWds7pIk9YzNXZKknrG5S5LUMzZ3SZJ6xuYuSVLP2NwlSeoZm7skST1jc5ckqWemau5J9iZ5JclykkdWmX8wyZtJnuu2L4zNPZDkp932wCyL1+yZdVvMux1m3ZaJ7wqXZBtwGLgbOAOcTLK0yrsCPV5VD69Y+3HgK8AAKOBUt/btmVSvmTLrtph3O8y6PdM8c78TWK6q01X1HnAM2D/l/d8LnKiqc903wglg79pK1SYw67aYdzvMujHTNPfrgdfG9s90Yyt9LsnzSZ5Isusy12oxmHVbzLsdZt2YWZ1Q9z3gxqr6JKNHdd++nMVJDiYZJhm++eabMypJG2RdWYN5bzH+bLfDrHtkmuZ+Ftg1tr+zG/ulqnqrqs53u98C7ph2bbf+SFUNqmqwY8eOaWvX7G141t19mPdi8Ge7HWbdmGma+0ng5iR7klwJHACWxg9Ict3Y7j7gx93t48A9SbYn2Q7c041pMZl1W8y7HWbdmIlny1fVhSQPMwpzG3C0ql5KcggYVtUS8BdJ9gEXgHPAg93ac0m+yugbC+BQVZ3bgK9DM2DWbTHvdph1e1JV867hIoPBoIbD4bzL2PKSnKqqwbzrmMS818+s27IV8jbr2VhP1l6hTpKknrG5S5LUMzZ3SZJ6xuYuSVLP2NwlSeoZm7skST1jc5ckqWds7pIk9YzNXZKknrG5S5LUMzZ3SZJ6ZqrmnmRvkleSLCd5ZJX5v0rycpLnk/xbkhvG5t5P8ly3La1cq8Vi1u0w67aYd1smvitckm3AYeBu4AxwMslSVb08dtiPgEFVvZvkT4G/A/6om/t5VX1qxnVrA5h1O8y6Lebdnmmeud8JLFfV6ap6DzgG7B8/oKq+X1XvdrvPADtnW6Y2iVm3w6zbYt6Nmaa5Xw+8NrZ/phu7lIeAJ8f2r04yTPJMks+uoUZtHrNuh1m3xbwbM/Fl+cuR5I+BAfA7Y8M3VNXZJDcBTyV5oapeXbHuIHAQYPfu3bMsSRtkrVl3a817CzHrtvh7vB+meeZ+Ftg1tr+zG7tIkt8Hvgzsq6rzH4xX1dnu42ngaeC2lWur6khVDapqsGPHjsv6AjRTG551N2/e82fWbfH3eGOmae4ngZuT7ElyJXAAuOhsySS3AY8y+oZ4Y2x8e5KrutvXAp8Gxk/g0GIx63aYdVvMuzETX5avqgtJHgaOA9uAo1X1UpJDwLCqloC/B34N+G4SgP+uqn3AJ4BHk/yC0QOJr604O1MLxKzbYdZtMe/2pKrmXcNFBoNBDYfDeZex5SU5VVWDedcxiXmvn1m3ZSvkbdazsZ6svUKdJEk9Y3OXJKlnbO6SJPWMzV2SpJ6xuUuS1DM2d0mSesbmLklSz9jcJUnqGZu7JEk9Y3OXJKlnbO6SJPWMzV2SpJ6Zqrkn2ZvklSTLSR5ZZf6qJI93888muXFs7q+78VeS3Du70rURzLot5t0Os27LxOaeZBtwGPgD4Fbg/iS3rjjsIeDtqvpN4OvA33Zrb2X0vsG/DewF/rG7Py0gs26LebfDrNszzTP3O4HlqjpdVe8Bx4D9K47ZD3y7u/0E8HsZvSHwfuBYVZ2vqv8Elrv702Iy67aYdzvMujHTNPfrgdfG9s90Y6seU1UXgHeA35hyrRaHWbfFvNth1o25Yt4FACQ5CBzsds8neXGe9UzhWuBn8y5igt+adwGXYt4zZ9azs+hZw4LmbdYbYs1ZT9PczwK7xvZ3dmOrHXMmyRXArwNvTbmWqjoCHAFIMqyqwbRfwDxslRrXsGzDswbznrU1Zg3+bH/IVqlxDcvMeoWtUuNa107zsvxJ4OYke5JcyejEiqUVxywBD3S3/xB4qqqqGz/QnYW5B7gZ+OFai9WGM+u2mHc7zLoxE5+5V9WFJA8Dx4FtwNGqeinJIWBYVUvAPwH/nGQZOMfoG4fuuH8BXgYuAF+qqvc36GvROpl1W8y7HWbdnowemC2OJAe7l3cWljXOzlaoc9FrXPT6PrAV6rTG2bDG2VhPjQvX3CVJ0vp4+VlJknpmbs19PZdCXKAaH0zyZpLnuu0Lm1zf0SRvXOq/nGTkG139zye5fTPrG6vDrGdTo3lvXo3+bE/BrGdS38ZkXVWbvjE6oeNV4CbgSuA/gFtXHPNnwDe72weAxxewxgeBf5jHv2H3+T8D3A68eIn5+4AngQB3Ac+a9dbM2rzbytuszXq9Wc/rmft6LoW4SDXOVVX9gNFZrZeyH/hOjTwDfCzJdZtT3S+Z9YyY96bWOFdmvak1ztVGZT2v5r6eSyFulmkvufi57qWSJ5LsWmV+nhbhspFmvXnMezp9yNusp9Ns1p5Qtz7fA26sqk8CJ/j/R6jqH7Nui3m3o5dZz6u5X86lEMnFl0LcLBNrrKq3qup8t/st4I5Nqm1aU18Sds41mPVsmPd0+pC3WU+n2azn1dzXcynEhalxxd899gE/3sT6prEEfL472/Iu4J2qen2TazDrzWPeM6pxC+Rt1jOqsbdZz/EMwfuAnzA6k/HL3dghYF93+2rgu4zeO/iHwE0LWOPfAC8xOgPz+8Atm1zfY8DrwP8y+jvMQ8AXgS928wEOd/W/AAzMemtmbd5t5W3WZr3erL1CnSRJPeMJdZIk9YzNXZKknrG5S5LUMzZ3SZJ6xuYuSVLP2NwlSeoZm7skST1jc5ckqWf+D5eJqHekLMLZAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 504x216 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Generate anim (slow)\n",
    "agent.record_trials(title=\"latent_learning\", \n",
    "                    start_locs=[(0, 2), (4, 0), (9, 0), (0, 8)])"
   ]
  }
 ],
 "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
}