gflownet_demo.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "4f2bc6c3-fbfa-49b1-9456-c8650367ac8e",
   "metadata": {},
   "outputs": [],
   "source": [
    "from gfn import GFNAgent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "03e0c075-642f-4ecf-b7c9-1163acf421f4",
   "metadata": {},
   "outputs": [],
   "source": [
    "agent = GFNAgent(epochs=200)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f4c24541-492d-4fa8-87ec-4574ef949a1a",
   "metadata": {},
   "source": [
    "First, let's take a look at the environment. The default is a 2D 8x8 grid with high reward in the corners."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "a7e79685-b5fb-4f66-8740-6a7554137008",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "agent.env.plot_reward_2d()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "739de680-77d5-42df-9710-6738b558d11b",
   "metadata": {},
   "source": [
    "We can also look at the model structure. Notice that in this implementation, the learned parameter `z0` is separate from the neural net, and that the neural net has two output \"heads\": `foward_policy` and `backward_policy`. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "33c11209-0661-4c0e-bb39-042d0193450c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "__________________________________________________________________________________________________\n",
      " Layer (type)                   Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      " input (InputLayer)             [(None, 2, 8)]       0           []                               \n",
      "                                                                                                  \n",
      " flatten (Flatten)              (None, 16)           0           ['input[0][0]']                  \n",
      "                                                                                                  \n",
      " dense_1 (Dense)                (None, 32)           544         ['flatten[0][0]']                \n",
      "                                                                                                  \n",
      " dense_2 (Dense)                (None, 32)           1056        ['dense_1[0][0]']                \n",
      "                                                                                                  \n",
      " forward_policy (Dense)         (None, 3)            99          ['dense_2[0][0]']                \n",
      "                                                                                                  \n",
      " backward_policy (Dense)        (None, 2)            66          ['dense_2[0][0]']                \n",
      "                                                                                                  \n",
      "==================================================================================================\n",
      "Total params: 1,765\n",
      "Trainable params: 1,765\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "agent.model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "03b55378-7021-41f5-9a4a-7514153ace10",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tf.Variable 'z0:0' shape=() dtype=float32, numpy=0.0>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "agent.z0"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7ba135c4-339d-4f18-981b-c3888be3eb80",
   "metadata": {},
   "source": [
    "For this demonstration, we'll just show that the GFlowNet can learn a policy that generates trajectories proportional to the reward. To do that, we'll first sample a large training set using the untrained, random policy."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "3adc788c-b3fa-4329-8b65-30a50cd24da9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "157/157 [==============================] - 1s 1ms/step\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-05-04 20:47:32.123047: I tensorflow/stream_executor/cuda/cuda_blas.cc:1786] TensorFloat-32 will be used for the matrix multiplication. This will only be logged once.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "157/157 [==============================] - 0s 1ms/step\n",
      "157/157 [==============================] - 0s 984us/step\n",
      "157/157 [==============================] - 0s 996us/step\n",
      "157/157 [==============================] - 0s 1ms/step\n",
      "157/157 [==============================] - 0s 1ms/step\n",
      "157/157 [==============================] - 0s 1ms/step\n",
      "157/157 [==============================] - 0s 964us/step\n",
      "157/157 [==============================] - 0s 973us/step\n",
      "157/157 [==============================] - 0s 987us/step\n",
      "157/157 [==============================] - 0s 983us/step\n",
      "157/157 [==============================] - 0s 962us/step\n",
      "157/157 [==============================] - 0s 998us/step\n",
      "157/157 [==============================] - 0s 966us/step\n",
      "157/157 [==============================] - 0s 1ms/step\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "agent.sample(5000)\n",
    "agent.plot_sampled_data_2d()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "e01d22b5-d758-4055-a91e-598f4e6d9fb9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 4 unique modes and 63 unique positions in the training data.\n"
     ]
    }
   ],
   "source": [
    "u_modes, u_positions = agent.count_modes()\n",
    "print(f'There are {u_modes} unique modes and {u_positions} unique positions in the training data.')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "398492c9-9295-46ae-82f7-2d4b0665fabe",
   "metadata": {},
   "source": [
    "Before training, the policy is uniform. The likelihood of transitioning vertically or laterally (arrows) or terminating (red octogon) is essentially uniform at every point.\n",
    "\n",
    "The probability of terminating at each position is plotted below, and we can see that without training, it looks nothing like the reward environment we plotted above. The termination probabilities are large enough, that any trajectory is unlikely to leave the origin (bottom left)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "840ecddf-34ed-4d44-9c67-216e4ab5ae0a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2/2 [==============================] - 0s 2ms/step\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "agent.plot_policy_2d()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "5e8a739d-1bf8-4697-b968-6583f0ed71bd",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "l1_error_before = agent.compare_env_to_model_policy()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "10528439-d61a-4efa-b7b3-88552dbc3573",
   "metadata": {},
   "source": [
    "Let's train it and see if we can do better!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "ea62ecb9-1129-4b64-b29c-aa8f3dcdcbf1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Start training...\n",
      "Epoch: 0 Loss: 5.823361873626709\n",
      "Epoch: 9 Loss: 1.9640847444534302\n",
      "Epoch: 18 Loss: 1.6309585571289062\n",
      "Epoch: 27 Loss: 1.0532681941986084\n",
      "Epoch: 36 Loss: 0.7600653767585754\n",
      "Epoch: 45 Loss: 0.6382139921188354\n",
      "Epoch: 54 Loss: 0.6046980619430542\n",
      "Epoch: 63 Loss: 0.5620511770248413\n",
      "Epoch: 72 Loss: 0.4427354037761688\n",
      "Epoch: 81 Loss: 0.6501584649085999\n",
      "Epoch: 90 Loss: 0.3121155798435211\n",
      "Epoch: 99 Loss: 0.27125149965286255\n",
      "Epoch: 108 Loss: 0.32772353291511536\n",
      "Epoch: 117 Loss: 0.30555203557014465\n",
      "Epoch: 126 Loss: 0.25774070620536804\n",
      "Epoch: 135 Loss: 0.21889758110046387\n",
      "Epoch: 144 Loss: 0.17800186574459076\n",
      "Epoch: 153 Loss: 0.4238590598106384\n",
      "Epoch: 162 Loss: 0.09405914694070816\n",
      "Epoch: 171 Loss: 0.03023693338036537\n",
      "Epoch: 180 Loss: 0.050495393574237823\n",
      "Epoch: 189 Loss: 0.046723444014787674\n",
      "Epoch: 198 Loss: 0.06047442555427551\n"
     ]
    }
   ],
   "source": [
    "agent.train()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6957008c-c79c-4154-b36b-07f2e33f28b8",
   "metadata": {},
   "source": [
    "Let's plot the trained policy and sample from it to get a probability distribution over the environment:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "c8e33d9b-19a7-4aa6-bc9d-37f48dfed5dd",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2/2 [==============================] - 0s 2ms/step\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "agent.plot_policy_2d()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "e84d9f3c-af87-40bb-8158-5c1e348d3968",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 993us/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 997us/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n",
      "63/63 [==============================] - 0s 1ms/step\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPUAAAD4CAYAAAA0L6C7AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8qNh9FAAAACXBIWXMAAAsTAAALEwEAmpwYAAALPUlEQVR4nO3dW4hd5RnG8efJnsSYxENTtVgTjIUSEKHGDraSIjRB6wnrRaEGlCqFXEjFUEG0d6X0ol6IvSihIZ6KqWKjQhvUKqioYFNzamsysdhgMEGNB2zitDWZmbcXs9OOZpJZe89a397z8v/BkNmHrO9dmTzzrb322t/riBCAPGb1ugAA9SLUQDKEGkiGUAPJEGogmYEmNtpaMD8GFi5sYtPHmHWkyDCSpNnvDpcbTJJccKykb4KMnDG/6Hhjc8qMc+TjjzQ6PDzp/5BGQj2wcKG+fPuaJjZ9jJPfK3ewcc4vXi02liS51So2VowVTHWMFRvqw+u+WWwsSRpeXOY38d5f33Pcxzj8BpIh1EAyhBpIhlADyRBqIBlCDSRDqIFkCDWQDKEGkpky1LaX2t4x4eug7TUFagPQhSkvE42INyRdKEm2W5L2S3qy2bIAdKvTw++Vkv4REXubKAbA9HUa6uslPTLZA7ZX295ie8voJ4U/zQTgfyqH2vYcSddK+t1kj0fEuogYjIjB1oKyH3cD8H+dzNRXStoWEe81VQyA6esk1Kt0nENvAP2jUqhtz5d0maQnmi0HwHRVWvkkIoYlfbHhWgDUgCvKgGQINZAMoQaSIdRAMoQaSIZQA8kQaiCZRjp0zDoizT1Q5vfFHTc9VmQcSdpw97nFxpKkGB0tOFjBDh2zynUe2fKztcXGkqRlP7+lyDgnajfFTA0kQ6iBZAg1kAyhBpIh1EAyhBpIhlADyRBqIBlCDSRDqIFkqq5RdrrtjbZ32x6yfUnThQHoTtVrv38p6ZmI+F57/e95DdYEYBqmDLXt0yRdKukmSYqIw5ION1sWgG5VOfw+T9L7kh6wvd32+vaSwZ8xse3OyL9ouwP0SpVQD0i6SNLaiFgmaVjSnZ9/0sS2OwPzaLsD9EqVUO+TtC8iNrdvb9R4yAH0oSlDHRHvSnrb9tL2XSsl7Wq0KgBdq3r2+1ZJG9pnvvdIurm5kgBMR9W2OzskDTZbCoA6cEUZkAyhBpIh1EAyhBpIhlADyRBqIBlCDSRDqIFkHA30UDrVC+MbXln7difjgUbagfWFGCvX38qzXGyskor2I5Mkl5knN48+q4Px0aQ/NGZqIBlCDSRDqIFkCDWQDKEGkiHUQDKEGkiGUAPJEGogmUqXY9l+S9IhSaOSRiKCpY2APtXJNZbfjogPGqsEQC04/AaSqRrqkPSs7a22V0/2hIltd47o0/oqBNCRqoff34qI/bbPkvSc7d0R8dLEJ0TEOknrpPFPadVcJ4CKKs3UEbG//ecBSU9KurjJogB0b8pQ255v+5Sj30u6XNLrTRcGoDtVDr+/JOlJ20ef/9uIeKbRqgB0bcpQR8QeSV8rUAuAGvCWFpAMoQaSIdRAMoQaSIZQA8kQaiAZQg0kQ6iBZGZ8z5qSbVXcahUbSyrcCqfkvhVsJ6QYKTeWJLcK/czGjv8QMzWQDKEGkiHUQDKEGkiGUAPJEGogGUINJEOogWQINZBM5VDbbtnebntTkwUBmJ5OZurbJA01VQiAelQKte1Fkq6WtL7ZcgBMV9WZ+l5Jd+gEl5HTdgfoD1UW879G0oGI2Hqi50XEuogYjIjB2TqptgIBdKbKTL1c0rXtHtWPSlph++FGqwLQtSlDHRF3RcSiiFgi6XpJz0fEDY1XBqArvE8NJNPRyicR8aKkFxupBEAtmKmBZAg1kAyhBpIh1EAyhBpIhlADyRBqIBlCDSTTXNsdF2o/EuVauETJdjHjA5Ybq/S+JVXs/8gJhmGmBpIh1EAyhBpIhlADyRBqIBlCDSRDqIFkCDWQDKEGkiHUQDJV1v2ea/vPtv9ie6ftn5YoDEB3qlz7/amkFRHxie3Zkl6x/XRE/Knh2gB0YcpQR0RI+qR9c3b7i6v/gT5VtUFey/YOSQckPRcRmyd5Dr20gD5QKdQRMRoRF0paJOli2xdM8hx6aQF9oKOz3xHxsaQXJF3RSDUApq3K2e8zbZ/e/v5kSZdJ2t1wXQC6VOXs99mSHrLd0vgvgcciYlOzZQHoVpWz33+VtKxALQBqwBVlQDKEGkiGUAPJEGogGUINJEOogWQINZBMM213bHnOnEY2/Xlx+HCRcXrCBX/nlmzxU1Kp9k9H9cG/IzM1kAyhBpIh1EAyhBpIhlADyRBqIBlCDSRDqIFkCDWQDKEGkqmy8OBi2y/Y3tVuu3NbicIAdKfKtd8jkm6PiG22T5G01fZzEbGr4doAdGHKmToi3omIbe3vD0kaknRO04UB6E5Hn9KyvUTjK4tO2nZH0mpJmqt5ddQGoAuVT5TZXiDpcUlrIuLg5x//TNsdz62zRgAdqNogb7bGA70hIp5otiQA01Hl7Lcl3SdpKCLuab4kANNRZaZeLulGSSts72h/XdVwXQC6VKXtziuSCq8JA6BbXFEGJEOogWQINZAMoQaSIdRAMoQaSIZQA8kQaiCZRnppjZwxT+99/+tNbPoYP7jlqSLjSNLTF3yh2FjFFezb5Var2Fgrt39YbCxJ2nj35UXGGd306nEfY6YGkiHUQDKEGkiGUAPJEGogGUINJEOogWQINZAMoQaSqbLw4P22D9h+vURBAKanykz9oKQrGq4DQE2qtN15SdJHBWoBUIPaXlPbXm17i+0tI/8ermuzADpUW6gntt0ZOHl+XZsF0CHOfgPJEGogmSpvaT0i6VVJS23vs/3D5ssC0K0qbXdWlSgEQD04/AaSIdRAMoQaSIZQA8kQaiAZQg0kQ6iBZAg1kEwjbXdGT5IOfWWsiU0f46G1VxUZR5LO0vFbncx4UebnJUkxUm6s3zz4nWJjSdLiPwwVGWfgn/857mPM1EAyhBpIhlADyRBqIBlCDSRDqIFkCDWQDKEGkiHUQDKVQm37Cttv2H7T9p1NFwWge1UWHmxJ+pWkKyWdL2mV7fObLgxAd6rM1BdLejMi9kTEYUmPSvpus2UB6FaVUJ8j6e0Jt/e17/uMiW13RodpuwP0SiNtd1rzabsD9EqVUO+XtHjC7UXt+wD0oSqhfk3SV22fZ3uOpOsl/b7ZsgB0q0qHjhHbP5L0R0ktSfdHxM7GKwPQlUorn0TEU5KeargWADXgijIgGUINJEOogWQINZAMoQaSIdRAMoQaSIZQA8k4IurfqP2+pL0d/rUzJH1QezH9Ieu+sV+9c25EnDnZA42Euhu2t0TEYK/raELWfWO/+hOH30AyhBpIpp9Cva7XBTQo676xX32ob15TA6hHP83UAGpAqIFk+iLUGZsF2F5s+wXbu2zvtH1br2uqk+2W7e22N/W6ljrZPt32Rtu7bQ/ZvqTXNXWq56+p280C/i7pMo0vP/yapFURsaunhU2T7bMlnR0R22yfImmrpOtm+n4dZfvHkgYlnRoR1/S6nrrYfkjSyxGxvr0m37yI+LjHZXWkH2bqlM0CIuKdiNjW/v6QpCFNsl76TGR7kaSrJa3vdS11sn2apEsl3SdJEXF4pgVa6o9QV2oWMJPZXiJpmaTNPS6lLvdKukPSWI/rqNt5kt6X9ED7pcV62zNuEft+CHVqthdIelzSmog42Ot6psv2NZIORMTWXtfSgAFJF0laGxHLJA1LmnHnePoh1GmbBdierfFAb4iIJ3pdT02WS7rW9lsaf6m0wvbDvS2pNvsk7YuIo0dUGzUe8hmlH0KdslmAbWv8tdlQRNzT63rqEhF3RcSiiFii8Z/V8xFxQ4/LqkVEvCvpbdtL23etlDTjTmxWWve7SYmbBSyXdKOkv9ne0b7vJ+011NG/bpW0oT3B7JF0c4/r6VjP39ICUK9+OPwGUCNCDSRDqIFkCDWQDKEGkiHUQDKEGkjmv8HGyIUS15zzAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "l1_error_after = agent.compare_env_to_model_policy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "b7d81e2a-48ab-4ac7-a3db-4c64510b95c4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "L1 error before 1.53 and after 0.20\n"
     ]
    }
   ],
   "source": [
    "print(f'L1 error before {l1_error_before:.2f} and after {l1_error_after:.2f}')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fe6e2c0f-ef2c-44d3-bad7-c4c5a188f419",
   "metadata": {},
   "source": [
    "While not perfect, the model has certainly learned to generate trajectories through the environment with probability proportional to the reward! It's a far better approximation than the untrained policy, and you can imagine that it would get better with some tweaks (e.g. more training, different optimizer, etc.)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}