demo/node_classification_on_citation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import logging\n",
    "import nest_asyncio; nest_asyncio.apply()\n",
    "\n",
    "import graphscope\n",
    "\n",
    "logging.basicConfig(level=logging.INFO,stream=sys.stdout)\n",
    "\n",
    "k8s_volumes = {\n",
    "    \"data\": {\n",
    "        \"type\": \"hostPath\",\n",
    "        \"field\": {\n",
    "          \"path\": os.path.expanduser(\"~/test_data/\"),\n",
    "          \"type\": \"Directory\"\n",
    "        },\n",
    "        \"mounts\": {\n",
    "          \"mountPath\": \"/testingdata\"\n",
    "        }\n",
    "    }\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create session\n",
    "\n",
    "sess = graphscope.session(k8s_volumes=k8s_volumes)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Loading graph\n",
    "def load_ogbn_mag(sess, prefix):\n",
    "    \"\"\"Load ogbn_mag graph.\n",
    "    The ogbn-mag dataset is a heterogeneous network composed of a subset of the Microsoft Academic Graph (MAG).\n",
    "    See more details here:\n",
    "    https://ogb.stanford.edu/docs/nodeprop/#ogbn-mag\n",
    "\n",
    "    Args:\n",
    "        sess (:class:`graphscope.Session`): Load graph within the session.\n",
    "        prefix (str): Data directory.\n",
    "        directed (bool, optional): Determine to load a directed or undirected graph.\n",
    "            Defaults to True.\n",
    "\n",
    "    Returns:\n",
    "        :class:`graphscope.Graph`: A Graph object which graph type is ArrowProperty\n",
    "    \"\"\"\n",
    "    graph = sess.g()\n",
    "    graph = (\n",
    "        graph.add_vertices(os.path.join(prefix, \"paper.csv\"), \"paper\")\n",
    "        .add_vertices(os.path.join(prefix, \"author.csv\"), \"author\")\n",
    "        .add_vertices(os.path.join(prefix, \"institution.csv\"), \"institution\")\n",
    "        .add_vertices(os.path.join(prefix, \"field_of_study.csv\"), \"field_of_study\")\n",
    "        .add_edges(\n",
    "            os.path.join(prefix, \"author_affiliated_with_institution.csv\"),\n",
    "            \"affiliated\",\n",
    "            src_label=\"author\",\n",
    "            dst_label=\"institution\",\n",
    "        )\n",
    "        .add_edges(\n",
    "            os.path.join(prefix, \"paper_has_topic_field_of_study.csv\"),\n",
    "            \"hasTopic\",\n",
    "            src_label=\"paper\",\n",
    "            dst_label=\"field_of_study\",\n",
    "        )\n",
    "        .add_edges(\n",
    "            os.path.join(prefix, \"paper_cites_paper.csv\"),\n",
    "            \"cites\",\n",
    "            src_label=\"paper\",\n",
    "            dst_label=\"paper\",\n",
    "        )\n",
    "        .add_edges(\n",
    "            os.path.join(prefix, \"author_writes_paper.csv\"),\n",
    "            \"writes\",\n",
    "            src_label=\"author\",\n",
    "            dst_label=\"paper\",\n",
    "        )\n",
    "    )\n",
    "\n",
    "    return graph\n",
    "\n",
    "# load ogbn_mag graph\n",
    "graph = load_ogbn_mag(sess, \"/testingdata/ogbn_mag_small\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "result [1]\n"
     ]
    }
   ],
   "source": [
    "# Interactive query\n",
    "\n",
    "# get the entrypoint for submitting Gremlin queries on graph g.\n",
    "interactive = sess.gremlin(graph)\n",
    "\n",
    "# count the number of papers two authors (with id 2 and 4307) have co-authored.\n",
    "papers = interactive.execute(\"g.V().has('author', 'id', 2).out('writes').where(__.in('writes').has('id', 4307)).count()\").one()\n",
    "print(\"result\", papers)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Graph analytics\n",
    "\n",
    "# exact a subgraph of publication within a time range.\n",
    "sub_graph = interactive.subgraph(\n",
    "    \"g.V().has('year', inside(2014, 2020)).outE('cites')\"\n",
    ")\n",
    "\n",
    "# project the subgraph to simple graph by selecting papers and their citations.\n",
    "simple_g = sub_graph.project(vertices={\"paper\": []}, edges={\"cites\": []})\n",
    "\n",
    "# compute the kcore and triangle-counting.\n",
    "kc_result = graphscope.k_core(simple_g, k=5)\n",
    "tc_result = graphscope.triangles(simple_g)\n",
    "\n",
    "# add the results as new columns to the citation graph.\n",
    "sub_graph = sub_graph.add_column(kc_result, {\"kcore\": \"r\"})\n",
    "sub_graph = sub_graph.add_column(tc_result, {\"tc\": \"r\"})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/admin/.local/lib/python3.6/site-packages/tensorflow/python/compat/v2_compat.py:96: disable_resource_variables (from tensorflow.python.ops.variable_scope) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "non-resource variables are not supported in the long term\n"
     ]
    }
   ],
   "source": [
    "# Graph neural networks (GNNs)\n",
    "\n",
    "# define the features for learning\n",
    "paper_features = []\n",
    "for i in range(128):\n",
    "    paper_features.append(\"feat_\" + str(i))\n",
    "paper_features.append(\"kcore\")\n",
    "paper_features.append(\"tc\")\n",
    "\n",
    "# launch a learning engine.\n",
    "lg = sess.learning(sub_graph, nodes=[(\"paper\", paper_features)],\n",
    "                   edges=[(\"paper\", \"cites\", \"paper\")],\n",
    "                   gen_labels=[\n",
    "                       (\"train\", \"paper\", 100, (0, 75)),\n",
    "                       (\"val\", \"paper\", 100, (75, 85)),\n",
    "                       (\"test\", \"paper\", 100, (85, 100))\n",
    "                   ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define the training function.\n",
    "\n",
    "from graphscope.learning.examples import GCN\n",
    "from graphscope.learning.graphlearn.python.model.tf.trainer import LocalTFTrainer\n",
    "from graphscope.learning.graphlearn.python.model.tf.optimizer import get_tf_optimizer\n",
    "\n",
    "# supervised GCN.\n",
    "def train(config, graph):\n",
    "    def model_fn():\n",
    "        return GCN(graph,\n",
    "                    config[\"class_num\"],\n",
    "                    config[\"features_num\"],\n",
    "                    config[\"batch_size\"],\n",
    "                    val_batch_size=config[\"val_batch_size\"],\n",
    "                    test_batch_size=config[\"test_batch_size\"],\n",
    "                    categorical_attrs_desc=config[\"categorical_attrs_desc\"],\n",
    "                    hidden_dim=config[\"hidden_dim\"],\n",
    "                    in_drop_rate=config[\"in_drop_rate\"],\n",
    "                    neighs_num=config[\"neighs_num\"],\n",
    "                    hops_num=config[\"hops_num\"],\n",
    "                    node_type=config[\"node_type\"],\n",
    "                    edge_type=config[\"edge_type\"],\n",
    "                    full_graph_mode=config[\"full_graph_mode\"])\n",
    "    trainer = LocalTFTrainer(model_fn,\n",
    "                             epoch=config[\"epoch\"],\n",
    "                             optimizer=get_tf_optimizer(\n",
    "                             config[\"learning_algo\"],\n",
    "                             config[\"learning_rate\"],\n",
    "                             config[\"weight_decay\"]))\n",
    "    trainer.train_and_evaluate()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /usr/local/lib64/python3.6/site-packages/graphscope/learning/graphlearn/python/model/tf/ego_flow.py:306: DatasetV1.make_initializable_iterator (from tensorflow.python.data.ops.dataset_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "This is a deprecated API that should only be used in TF 1 graph mode and legacy TF 2 graph mode available through `tf.compat.v1`. In all other situations -- namely, eager mode and inside `tf.function` -- you can consume dataset elements using `for elem in dataset: ...` or by explicitly creating iterator via `iterator = iter(dataset)` and fetching its elements via `values = next(iterator)`. Furthermore, this API is not available in TF 2. During the transition from TF 1 to TF 2 you can use `tf.compat.v1.data.make_initializable_iterator(dataset)` to create a TF 1 graph mode style iterator for a dataset created through TF 2 APIs. Note that this should be a transient state of your code base as there are in general no guarantees about the interoperability of TF 1 and TF 2 code.\n",
      "WARNING:tensorflow:From /home/admin/.local/lib/python3.6/site-packages/tensorflow/python/util/dispatch.py:201: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
      "training...\n",
      "Epoch 00, Iteration 0, Time(s) 2.3594, Loss 6.52750\n",
      "Epoch 00, Iteration 1, Time(s) 0.2585, Loss 6.54409\n",
      "Epoch 00, Iteration 2, Time(s) 0.1981, Loss 6.33570\n",
      "Epoch 00, Iteration 3, Time(s) 0.1921, Loss 6.48608\n",
      "Epoch 00, Iteration 4, Time(s) 0.1910, Loss 6.02871\n",
      "Epoch 00, Iteration 5, Time(s) 0.2212, Loss 5.82882\n",
      "Epoch 00, Iteration 6, Time(s) 0.2104, Loss 5.88230\n",
      "Epoch 00, Iteration 7, Time(s) 0.1629, Loss 5.80559\n",
      "Epoch 00, Iteration 8, Time(s) 0.0914, Loss 5.77225\n",
      "Epoch 00, Iteration 9, Time(s) 0.0503, Loss 5.79369\n",
      "Epoch 00, Iteration 10, Time(s) 2.6503, Loss 5.83165\n",
      "Epoch 00, Iteration 11, Time(s) 0.5971, Loss 5.73021\n",
      "Epoch 00, Iteration 12, Time(s) 0.4220, Loss 5.69594\n",
      "Epoch 00, Iteration 13, Time(s) 0.4080, Loss 5.71983\n",
      "Epoch 00, Iteration 14, Time(s) 0.4621, Loss 5.68983\n",
      "Epoch 00, Iteration 15, Time(s) 0.5275, Loss 5.69185\n",
      "Epoch 00, Iteration 16, Time(s) 0.5032, Loss 5.64513\n",
      "Epoch 00, Iteration 17, Time(s) 0.4082, Loss 5.73378\n",
      "Epoch 00, Iteration 18, Time(s) 0.2929, Loss 5.71504\n",
      "Epoch 00, Iteration 19, Time(s) 0.0610, Loss 5.74095\n",
      "val Accuracy is: 0.0973\n",
      "Epoch 01, Iteration 0, Time(s) 0.2591, Loss 5.58525\n",
      "Epoch 01, Iteration 1, Time(s) 0.2021, Loss 5.66976\n",
      "Epoch 01, Iteration 2, Time(s) 0.1964, Loss 5.54119\n",
      "Epoch 01, Iteration 3, Time(s) 0.1850, Loss 5.64127\n",
      "Epoch 01, Iteration 4, Time(s) 0.1978, Loss 5.53457\n",
      "Epoch 01, Iteration 5, Time(s) 0.2010, Loss 5.69508\n",
      "Epoch 01, Iteration 6, Time(s) 0.2044, Loss 5.52161\n",
      "Epoch 01, Iteration 7, Time(s) 0.1069, Loss 5.68928\n",
      "Epoch 01, Iteration 8, Time(s) 0.1001, Loss 5.69453\n",
      "Epoch 01, Iteration 9, Time(s) 0.0528, Loss 5.68811\n",
      "Epoch 01, Iteration 10, Time(s) 0.4313, Loss 5.64953\n",
      "Epoch 01, Iteration 11, Time(s) 0.4898, Loss 5.54775\n",
      "Epoch 01, Iteration 12, Time(s) 0.5007, Loss 5.56766\n",
      "Epoch 01, Iteration 13, Time(s) 0.4919, Loss 5.63314\n",
      "Epoch 01, Iteration 14, Time(s) 0.6037, Loss 5.45767\n",
      "Epoch 01, Iteration 15, Time(s) 0.3826, Loss 5.56886\n",
      "Epoch 01, Iteration 16, Time(s) 0.4343, Loss 5.57984\n",
      "Epoch 01, Iteration 17, Time(s) 0.2594, Loss 5.65748\n",
      "Epoch 01, Iteration 18, Time(s) 0.3236, Loss 5.64987\n",
      "Epoch 01, Iteration 19, Time(s) 0.0798, Loss 5.66725\n",
      "val Accuracy is: 0.1321\n",
      "Epoch 02, Iteration 0, Time(s) 0.2314, Loss 5.43725\n",
      "Epoch 02, Iteration 1, Time(s) 0.1870, Loss 5.36135\n",
      "Epoch 02, Iteration 2, Time(s) 0.1921, Loss 5.21218\n",
      "Epoch 02, Iteration 3, Time(s) 0.2265, Loss 5.35965\n",
      "Epoch 02, Iteration 4, Time(s) 0.1928, Loss 5.37013\n",
      "Epoch 02, Iteration 5, Time(s) 0.1984, Loss 5.62703\n",
      "Epoch 02, Iteration 6, Time(s) 0.2048, Loss 5.48783\n",
      "Epoch 02, Iteration 7, Time(s) 0.1170, Loss 5.64866\n",
      "Epoch 02, Iteration 8, Time(s) 0.1059, Loss 5.65401\n",
      "Epoch 02, Iteration 9, Time(s) 0.0521, Loss 5.61690\n",
      "Epoch 02, Iteration 10, Time(s) 0.4212, Loss 5.50650\n",
      "Epoch 02, Iteration 11, Time(s) 0.4027, Loss 5.36066\n",
      "Epoch 02, Iteration 12, Time(s) 0.4003, Loss 5.49677\n",
      "Epoch 02, Iteration 13, Time(s) 0.5024, Loss 5.32206\n",
      "Epoch 02, Iteration 14, Time(s) 0.5661, Loss 5.31326\n",
      "Epoch 02, Iteration 15, Time(s) 0.4181, Loss 5.49063\n",
      "Epoch 02, Iteration 16, Time(s) 0.5948, Loss 5.42767\n",
      "Epoch 02, Iteration 17, Time(s) 0.3230, Loss 5.60508\n",
      "Epoch 02, Iteration 18, Time(s) 0.3564, Loss 5.63442\n",
      "Epoch 02, Iteration 19, Time(s) 0.0904, Loss 5.68651\n",
      "val Accuracy is: 0.1390\n",
      "Epoch 03, Iteration 0, Time(s) 0.2410, Loss 5.35627\n",
      "Epoch 03, Iteration 1, Time(s) 0.2085, Loss 5.26058\n",
      "Epoch 03, Iteration 2, Time(s) 0.1875, Loss 5.18296\n",
      "Epoch 03, Iteration 3, Time(s) 0.1944, Loss 5.33748\n",
      "Epoch 03, Iteration 4, Time(s) 0.1924, Loss 5.34898\n",
      "Epoch 03, Iteration 5, Time(s) 0.2084, Loss 5.55950\n",
      "Epoch 03, Iteration 6, Time(s) 0.2010, Loss 5.39955\n",
      "Epoch 03, Iteration 7, Time(s) 0.1025, Loss 5.59763\n",
      "Epoch 03, Iteration 8, Time(s) 0.1062, Loss 5.61542\n",
      "Epoch 03, Iteration 9, Time(s) 0.0512, Loss 5.61022\n",
      "Epoch 03, Iteration 10, Time(s) 0.4331, Loss 5.52958\n",
      "Epoch 03, Iteration 11, Time(s) 0.4738, Loss 5.36813\n",
      "Epoch 03, Iteration 12, Time(s) 0.5974, Loss 5.31650\n",
      "Epoch 03, Iteration 13, Time(s) 0.4233, Loss 5.17993\n",
      "Epoch 03, Iteration 14, Time(s) 0.5041, Loss 5.27333\n",
      "Epoch 03, Iteration 15, Time(s) 0.3704, Loss 5.44287\n",
      "Epoch 03, Iteration 16, Time(s) 0.4239, Loss 5.37846\n",
      "Epoch 03, Iteration 17, Time(s) 0.4715, Loss 5.57568\n",
      "Epoch 03, Iteration 18, Time(s) 0.3143, Loss 5.57199\n",
      "Epoch 03, Iteration 19, Time(s) 0.0827, Loss 5.58397\n",
      "val Accuracy is: 0.1654\n",
      "Epoch 04, Iteration 0, Time(s) 0.2411, Loss 5.29506\n",
      "Epoch 04, Iteration 1, Time(s) 0.2021, Loss 5.17857\n",
      "Epoch 04, Iteration 2, Time(s) 0.1814, Loss 5.22531\n",
      "Epoch 04, Iteration 3, Time(s) 0.2165, Loss 5.01809\n",
      "Epoch 04, Iteration 4, Time(s) 0.2038, Loss 5.21292\n",
      "Epoch 04, Iteration 5, Time(s) 0.2081, Loss 5.53720\n",
      "Epoch 04, Iteration 6, Time(s) 0.1971, Loss 5.35106\n",
      "Epoch 04, Iteration 7, Time(s) 0.0911, Loss 5.56572\n",
      "Epoch 04, Iteration 8, Time(s) 0.1021, Loss 5.56283\n",
      "Epoch 04, Iteration 9, Time(s) 0.0545, Loss 5.58231\n",
      "Epoch 04, Iteration 10, Time(s) 0.4382, Loss 5.42393\n",
      "Epoch 04, Iteration 11, Time(s) 0.4902, Loss 5.25262\n",
      "Epoch 04, Iteration 12, Time(s) 0.5852, Loss 5.13875\n",
      "Epoch 04, Iteration 13, Time(s) 0.4356, Loss 5.08105\n",
      "Epoch 04, Iteration 14, Time(s) 0.5707, Loss 5.13281\n",
      "Epoch 04, Iteration 15, Time(s) 0.4118, Loss 5.35531\n",
      "Epoch 04, Iteration 16, Time(s) 0.3974, Loss 5.37977\n",
      "Epoch 04, Iteration 17, Time(s) 0.2927, Loss 5.52423\n",
      "Epoch 04, Iteration 18, Time(s) 0.3048, Loss 5.51981\n",
      "Epoch 04, Iteration 19, Time(s) 0.0835, Loss 5.52114\n",
      "val Accuracy is: 0.1686\n",
      "Epoch 05, Iteration 0, Time(s) 0.2348, Loss 5.15271\n",
      "Epoch 05, Iteration 1, Time(s) 0.1878, Loss 5.15327\n",
      "Epoch 05, Iteration 2, Time(s) 0.2164, Loss 5.15336\n",
      "Epoch 05, Iteration 3, Time(s) 0.1981, Loss 5.03484\n",
      "Epoch 05, Iteration 4, Time(s) 0.2029, Loss 5.24658\n",
      "Epoch 05, Iteration 5, Time(s) 0.1666, Loss 5.53193\n",
      "Epoch 05, Iteration 6, Time(s) 0.2206, Loss 5.29764\n",
      "Epoch 05, Iteration 7, Time(s) 0.1101, Loss 5.56274\n",
      "Epoch 05, Iteration 8, Time(s) 0.0749, Loss 5.52785\n",
      "Epoch 05, Iteration 9, Time(s) 0.0583, Loss 5.52369\n",
      "Epoch 05, Iteration 10, Time(s) 0.3651, Loss 5.37938\n",
      "Epoch 05, Iteration 11, Time(s) 0.4118, Loss 5.17088\n",
      "Epoch 05, Iteration 12, Time(s) 0.4616, Loss 5.16651\n",
      "Epoch 05, Iteration 13, Time(s) 0.3414, Loss 5.15705\n",
      "Epoch 05, Iteration 14, Time(s) 0.3736, Loss 5.13302\n",
      "Epoch 05, Iteration 15, Time(s) 0.3961, Loss 5.33258\n",
      "Epoch 05, Iteration 16, Time(s) 0.4013, Loss 5.24971\n",
      "Epoch 05, Iteration 17, Time(s) 0.2880, Loss 5.51502\n",
      "Epoch 05, Iteration 18, Time(s) 0.2405, Loss 5.53291\n",
      "Epoch 05, Iteration 19, Time(s) 0.0538, Loss 5.54725\n",
      "val Accuracy is: 0.1678\n",
      "Epoch 06, Iteration 0, Time(s) 0.2295, Loss 5.18448\n",
      "Epoch 06, Iteration 1, Time(s) 0.2003, Loss 5.20972\n",
      "Epoch 06, Iteration 2, Time(s) 0.1921, Loss 5.15665\n",
      "Epoch 06, Iteration 3, Time(s) 0.2035, Loss 4.99866\n",
      "Epoch 06, Iteration 4, Time(s) 0.1728, Loss 5.17639\n",
      "Epoch 06, Iteration 5, Time(s) 0.1598, Loss 5.51632\n",
      "Epoch 06, Iteration 6, Time(s) 0.1717, Loss 5.24725\n",
      "Epoch 06, Iteration 7, Time(s) 0.0914, Loss 5.51455\n",
      "Epoch 06, Iteration 8, Time(s) 0.1192, Loss 5.51471\n",
      "Epoch 06, Iteration 9, Time(s) 0.0469, Loss 5.53627\n",
      "Epoch 06, Iteration 10, Time(s) 0.4324, Loss 5.27916\n",
      "Epoch 06, Iteration 11, Time(s) 0.3815, Loss 5.19784\n",
      "Epoch 06, Iteration 12, Time(s) 0.3250, Loss 5.05181\n",
      "Epoch 06, Iteration 13, Time(s) 0.4027, Loss 5.13306\n",
      "Epoch 06, Iteration 14, Time(s) 0.3725, Loss 5.10359\n",
      "Epoch 06, Iteration 15, Time(s) 0.3959, Loss 5.32543\n",
      "Epoch 06, Iteration 16, Time(s) 0.4047, Loss 5.23986\n",
      "Epoch 06, Iteration 17, Time(s) 0.3001, Loss 5.47885\n",
      "Epoch 06, Iteration 18, Time(s) 0.2995, Loss 5.48592\n",
      "Epoch 06, Iteration 19, Time(s) 0.0886, Loss 5.46709\n",
      "val Accuracy is: 0.1825\n",
      "Epoch 07, Iteration 0, Time(s) 0.2315, Loss 5.07401\n",
      "Epoch 07, Iteration 1, Time(s) 0.1996, Loss 5.04859\n",
      "Epoch 07, Iteration 2, Time(s) 0.1795, Loss 5.12744\n",
      "Epoch 07, Iteration 3, Time(s) 0.2218, Loss 4.94289\n",
      "Epoch 07, Iteration 4, Time(s) 0.2057, Loss 5.11865\n",
      "Epoch 07, Iteration 5, Time(s) 0.1972, Loss 5.44832\n",
      "Epoch 07, Iteration 6, Time(s) 0.2013, Loss 5.20325\n",
      "Epoch 07, Iteration 7, Time(s) 0.0958, Loss 5.49476\n",
      "Epoch 07, Iteration 8, Time(s) 0.0793, Loss 5.47307\n",
      "Epoch 07, Iteration 9, Time(s) 0.0511, Loss 5.48061\n",
      "Epoch 07, Iteration 10, Time(s) 0.3607, Loss 5.38660\n",
      "Epoch 07, Iteration 11, Time(s) 0.3730, Loss 5.22481\n",
      "Epoch 07, Iteration 12, Time(s) 0.4265, Loss 5.06325\n",
      "Epoch 07, Iteration 13, Time(s) 0.4047, Loss 5.08504\n",
      "Epoch 07, Iteration 14, Time(s) 0.3732, Loss 5.10673\n",
      "Epoch 07, Iteration 15, Time(s) 0.4084, Loss 5.34361\n",
      "Epoch 07, Iteration 16, Time(s) 0.3982, Loss 5.24764\n",
      "Epoch 07, Iteration 17, Time(s) 0.2986, Loss 5.47955\n",
      "Epoch 07, Iteration 18, Time(s) 0.3943, Loss 5.51192\n",
      "Epoch 07, Iteration 19, Time(s) 0.0913, Loss 5.54008\n",
      "val Accuracy is: 0.1685\n",
      "Epoch 08, Iteration 0, Time(s) 0.2352, Loss 5.07206\n",
      "Epoch 08, Iteration 1, Time(s) 0.1986, Loss 4.93081\n",
      "Epoch 08, Iteration 2, Time(s) 0.1844, Loss 4.96121\n",
      "Epoch 08, Iteration 3, Time(s) 0.1889, Loss 4.93375\n",
      "Epoch 08, Iteration 4, Time(s) 0.2162, Loss 4.97126\n",
      "Epoch 08, Iteration 5, Time(s) 0.1796, Loss 5.47395\n",
      "Epoch 08, Iteration 6, Time(s) 0.2396, Loss 5.19166\n",
      "Epoch 08, Iteration 7, Time(s) 0.1182, Loss 5.46065\n",
      "Epoch 08, Iteration 8, Time(s) 0.1686, Loss 5.46724\n",
      "Epoch 08, Iteration 9, Time(s) 0.0398, Loss 5.41868\n",
      "Epoch 08, Iteration 10, Time(s) 0.4523, Loss 5.30911\n",
      "Epoch 08, Iteration 11, Time(s) 0.4025, Loss 5.10952\n",
      "Epoch 08, Iteration 12, Time(s) 0.5673, Loss 4.94314\n",
      "Epoch 08, Iteration 13, Time(s) 0.4062, Loss 5.03756\n",
      "Epoch 08, Iteration 14, Time(s) 0.4293, Loss 4.99642\n",
      "Epoch 08, Iteration 15, Time(s) 0.4814, Loss 5.24250\n",
      "Epoch 08, Iteration 16, Time(s) 0.4134, Loss 5.17140\n",
      "Epoch 08, Iteration 17, Time(s) 0.2871, Loss 5.43216\n",
      "Epoch 08, Iteration 18, Time(s) 0.3167, Loss 5.46841\n",
      "Epoch 08, Iteration 19, Time(s) 0.0665, Loss 5.42921\n",
      "val Accuracy is: 0.1826\n",
      "Epoch 09, Iteration 0, Time(s) 0.2417, Loss 5.10299\n",
      "Epoch 09, Iteration 1, Time(s) 0.1840, Loss 4.95138\n",
      "Epoch 09, Iteration 2, Time(s) 0.2140, Loss 4.93776\n",
      "Epoch 09, Iteration 3, Time(s) 0.1896, Loss 4.74322\n",
      "Epoch 09, Iteration 4, Time(s) 0.1978, Loss 5.07484\n",
      "Epoch 09, Iteration 5, Time(s) 0.2197, Loss 5.37755\n",
      "Epoch 09, Iteration 6, Time(s) 0.2007, Loss 5.16963\n",
      "Epoch 09, Iteration 7, Time(s) 0.1029, Loss 5.45083\n",
      "Epoch 09, Iteration 8, Time(s) 0.1023, Loss 5.44287\n",
      "Epoch 09, Iteration 9, Time(s) 0.0499, Loss 5.43790\n",
      "Epoch 09, Iteration 10, Time(s) 0.4241, Loss 5.26051\n",
      "Epoch 09, Iteration 11, Time(s) 0.4708, Loss 5.23202\n",
      "Epoch 09, Iteration 12, Time(s) 0.6333, Loss 5.03613\n",
      "Epoch 09, Iteration 13, Time(s) 0.5839, Loss 4.96768\n",
      "Epoch 09, Iteration 14, Time(s) 0.5204, Loss 4.95582\n",
      "Epoch 09, Iteration 15, Time(s) 0.3860, Loss 5.27441\n",
      "Epoch 09, Iteration 16, Time(s) 0.3678, Loss 5.11716\n",
      "Epoch 09, Iteration 17, Time(s) 0.3018, Loss 5.43583\n",
      "Epoch 09, Iteration 18, Time(s) 0.3213, Loss 5.43058\n",
      "Epoch 09, Iteration 19, Time(s) 0.0806, Loss 5.34673\n",
      "val Accuracy is: 0.2004\n",
      "Epoch 10, Iteration 0, Time(s) 0.2442, Loss 4.97634\n",
      "Epoch 10, Iteration 1, Time(s) 0.2049, Loss 5.00721\n",
      "Epoch 10, Iteration 2, Time(s) 0.1864, Loss 4.97213\n",
      "Epoch 10, Iteration 3, Time(s) 0.1968, Loss 4.92259\n",
      "Epoch 10, Iteration 4, Time(s) 0.2072, Loss 5.07084\n",
      "Epoch 10, Iteration 5, Time(s) 0.1972, Loss 5.42615\n",
      "Epoch 10, Iteration 6, Time(s) 0.2023, Loss 5.14428\n",
      "Epoch 10, Iteration 7, Time(s) 0.1096, Loss 5.42734\n",
      "Epoch 10, Iteration 8, Time(s) 0.0864, Loss 5.45949\n",
      "Epoch 10, Iteration 9, Time(s) 0.0505, Loss 5.42515\n",
      "Epoch 10, Iteration 10, Time(s) 0.4373, Loss 5.31424\n",
      "Epoch 10, Iteration 11, Time(s) 0.4116, Loss 5.16695\n",
      "Epoch 10, Iteration 12, Time(s) 0.4757, Loss 4.81813\n",
      "Epoch 10, Iteration 13, Time(s) 0.3949, Loss 5.06403\n",
      "Epoch 10, Iteration 14, Time(s) 0.4313, Loss 4.93136\n",
      "Epoch 10, Iteration 15, Time(s) 0.3757, Loss 5.22426\n",
      "Epoch 10, Iteration 16, Time(s) 0.3010, Loss 5.11943\n",
      "Epoch 10, Iteration 17, Time(s) 0.3001, Loss 5.45954\n",
      "Epoch 10, Iteration 18, Time(s) 0.2206, Loss 5.42516\n",
      "Epoch 10, Iteration 19, Time(s) 0.0566, Loss 5.40100\n",
      "val Accuracy is: 0.1887\n",
      "Epoch 11, Iteration 0, Time(s) 0.1549, Loss 4.96212\n",
      "Epoch 11, Iteration 1, Time(s) 0.2155, Loss 4.92689\n",
      "Epoch 11, Iteration 2, Time(s) 0.1850, Loss 4.95532\n",
      "Epoch 11, Iteration 3, Time(s) 0.1982, Loss 4.76358\n",
      "Epoch 11, Iteration 4, Time(s) 0.2017, Loss 5.06013\n",
      "Epoch 11, Iteration 5, Time(s) 0.1888, Loss 5.40412\n",
      "Epoch 11, Iteration 6, Time(s) 0.1988, Loss 5.08123\n",
      "Epoch 11, Iteration 7, Time(s) 0.1053, Loss 5.42093\n",
      "Epoch 11, Iteration 8, Time(s) 0.0759, Loss 5.38418\n",
      "Epoch 11, Iteration 9, Time(s) 0.0506, Loss 5.44272\n",
      "Epoch 11, Iteration 10, Time(s) 0.3713, Loss 5.32981\n",
      "Epoch 11, Iteration 11, Time(s) 0.3855, Loss 5.11710\n",
      "Epoch 11, Iteration 12, Time(s) 0.3203, Loss 4.95193\n",
      "Epoch 11, Iteration 13, Time(s) 0.2972, Loss 4.97977\n",
      "Epoch 11, Iteration 14, Time(s) 0.3773, Loss 4.83571\n",
      "Epoch 11, Iteration 15, Time(s) 0.3935, Loss 5.14683\n",
      "Epoch 11, Iteration 16, Time(s) 0.4011, Loss 5.07944\n",
      "Epoch 11, Iteration 17, Time(s) 0.3036, Loss 5.40944\n",
      "Epoch 11, Iteration 18, Time(s) 0.2962, Loss 5.42321\n",
      "Epoch 11, Iteration 19, Time(s) 0.0806, Loss 5.35657\n",
      "val Accuracy is: 0.1996\n",
      "Epoch 12, Iteration 0, Time(s) 0.2347, Loss 4.97673\n",
      "Epoch 12, Iteration 1, Time(s) 0.1970, Loss 4.86807\n",
      "Epoch 12, Iteration 2, Time(s) 0.1676, Loss 4.88226\n",
      "Epoch 12, Iteration 3, Time(s) 0.2193, Loss 4.91471\n",
      "Epoch 12, Iteration 4, Time(s) 0.2082, Loss 5.02155\n",
      "Epoch 12, Iteration 5, Time(s) 0.1155, Loss 5.39368\n",
      "Epoch 12, Iteration 6, Time(s) 0.1865, Loss 5.05966\n",
      "Epoch 12, Iteration 7, Time(s) 0.1167, Loss 5.39217\n",
      "Epoch 12, Iteration 8, Time(s) 0.1150, Loss 5.41429\n",
      "Epoch 12, Iteration 9, Time(s) 0.0558, Loss 5.40308\n",
      "Epoch 12, Iteration 10, Time(s) 0.3826, Loss 5.23010\n",
      "Epoch 12, Iteration 11, Time(s) 0.4038, Loss 4.91961\n",
      "Epoch 12, Iteration 12, Time(s) 0.3259, Loss 4.81994\n",
      "Epoch 12, Iteration 13, Time(s) 0.3880, Loss 4.91030\n",
      "Epoch 12, Iteration 14, Time(s) 0.3121, Loss 4.84271\n",
      "Epoch 12, Iteration 15, Time(s) 0.2948, Loss 5.22671\n",
      "Epoch 12, Iteration 16, Time(s) 0.3726, Loss 5.00296\n",
      "Epoch 12, Iteration 17, Time(s) 0.2191, Loss 5.37684\n",
      "Epoch 12, Iteration 18, Time(s) 0.2849, Loss 5.38511\n",
      "Epoch 12, Iteration 19, Time(s) 0.0838, Loss 5.37195\n",
      "val Accuracy is: 0.2073\n",
      "Epoch 13, Iteration 0, Time(s) 0.2441, Loss 4.97249\n",
      "Epoch 13, Iteration 1, Time(s) 0.2107, Loss 4.90013\n",
      "Epoch 13, Iteration 2, Time(s) 0.1738, Loss 4.77212\n",
      "Epoch 13, Iteration 3, Time(s) 0.2229, Loss 4.80754\n",
      "Epoch 13, Iteration 4, Time(s) 0.1814, Loss 4.95431\n",
      "Epoch 13, Iteration 5, Time(s) 0.2080, Loss 5.36083\n",
      "Epoch 13, Iteration 6, Time(s) 0.2062, Loss 5.06422\n",
      "Epoch 13, Iteration 7, Time(s) 0.1027, Loss 5.36715\n",
      "Epoch 13, Iteration 8, Time(s) 0.0692, Loss 5.38874\n",
      "Epoch 13, Iteration 9, Time(s) 0.0507, Loss 5.44520\n",
      "Epoch 13, Iteration 10, Time(s) 0.4716, Loss 5.29462\n",
      "Epoch 13, Iteration 11, Time(s) 0.4747, Loss 4.99940\n",
      "Epoch 13, Iteration 12, Time(s) 0.3106, Loss 4.88304\n",
      "Epoch 13, Iteration 13, Time(s) 0.3982, Loss 4.92882\n",
      "Epoch 13, Iteration 14, Time(s) 0.4035, Loss 4.84214\n",
      "Epoch 13, Iteration 15, Time(s) 0.3183, Loss 5.17855\n",
      "Epoch 13, Iteration 16, Time(s) 0.4763, Loss 5.03427\n",
      "Epoch 13, Iteration 17, Time(s) 0.2976, Loss 5.35899\n",
      "Epoch 13, Iteration 18, Time(s) 0.2141, Loss 5.36888\n",
      "Epoch 13, Iteration 19, Time(s) 0.0647, Loss 5.36861\n",
      "val Accuracy is: 0.2042\n",
      "Epoch 14, Iteration 0, Time(s) 0.2348, Loss 4.95879\n",
      "Epoch 14, Iteration 1, Time(s) 0.2073, Loss 4.83482\n",
      "Epoch 14, Iteration 2, Time(s) 0.1584, Loss 4.87671\n",
      "Epoch 14, Iteration 3, Time(s) 0.2215, Loss 4.76477\n",
      "Epoch 14, Iteration 4, Time(s) 0.2076, Loss 4.90036\n",
      "Epoch 14, Iteration 5, Time(s) 0.2095, Loss 5.36655\n",
      "Epoch 14, Iteration 6, Time(s) 0.1926, Loss 5.02381\n",
      "Epoch 14, Iteration 7, Time(s) 0.1049, Loss 5.34127\n",
      "Epoch 14, Iteration 8, Time(s) 0.1026, Loss 5.38812\n",
      "Epoch 14, Iteration 9, Time(s) 0.0396, Loss 5.32029\n",
      "Epoch 14, Iteration 10, Time(s) 0.3605, Loss 5.16909\n",
      "Epoch 14, Iteration 11, Time(s) 0.3833, Loss 4.92089\n",
      "Epoch 14, Iteration 12, Time(s) 0.3874, Loss 4.84468\n",
      "Epoch 14, Iteration 13, Time(s) 0.3880, Loss 4.84595\n",
      "Epoch 14, Iteration 14, Time(s) 0.4184, Loss 4.78877\n",
      "Epoch 14, Iteration 15, Time(s) 0.4137, Loss 5.08441\n",
      "Epoch 14, Iteration 16, Time(s) 0.4861, Loss 5.12036\n",
      "Epoch 14, Iteration 17, Time(s) 0.3080, Loss 5.36861\n",
      "Epoch 14, Iteration 18, Time(s) 0.2786, Loss 5.37934\n",
      "Epoch 14, Iteration 19, Time(s) 0.0841, Loss 5.33139\n",
      "val Accuracy is: 0.2035\n",
      "Epoch 15, Iteration 0, Time(s) 0.2565, Loss 4.95443\n",
      "Epoch 15, Iteration 1, Time(s) 0.1785, Loss 4.66186\n",
      "Epoch 15, Iteration 2, Time(s) 0.2300, Loss 4.87264\n",
      "Epoch 15, Iteration 3, Time(s) 0.1902, Loss 4.80735\n",
      "Epoch 15, Iteration 4, Time(s) 0.2014, Loss 4.88723\n",
      "Epoch 15, Iteration 5, Time(s) 0.1998, Loss 5.31153\n",
      "Epoch 15, Iteration 6, Time(s) 0.1770, Loss 5.03006\n",
      "Epoch 15, Iteration 7, Time(s) 0.0910, Loss 5.38002\n",
      "Epoch 15, Iteration 8, Time(s) 0.1150, Loss 5.36784\n",
      "Epoch 15, Iteration 9, Time(s) 0.0417, Loss 5.34755\n",
      "Epoch 15, Iteration 10, Time(s) 0.3528, Loss 5.20074\n",
      "Epoch 15, Iteration 11, Time(s) 0.4011, Loss 5.00201\n",
      "Epoch 15, Iteration 12, Time(s) 0.3836, Loss 4.76013\n",
      "Epoch 15, Iteration 13, Time(s) 0.4924, Loss 4.81064\n",
      "Epoch 15, Iteration 14, Time(s) 0.4009, Loss 4.75112\n",
      "Epoch 15, Iteration 15, Time(s) 0.3316, Loss 5.13637\n",
      "Epoch 15, Iteration 16, Time(s) 0.3916, Loss 4.96798\n",
      "Epoch 15, Iteration 17, Time(s) 0.2820, Loss 5.33698\n",
      "Epoch 15, Iteration 18, Time(s) 0.3023, Loss 5.35565\n",
      "Epoch 15, Iteration 19, Time(s) 0.0889, Loss 5.38869\n",
      "val Accuracy is: 0.2059\n",
      "Epoch 16, Iteration 0, Time(s) 0.1618, Loss 4.90752\n",
      "Epoch 16, Iteration 1, Time(s) 0.1642, Loss 4.85045\n",
      "Epoch 16, Iteration 2, Time(s) 0.2396, Loss 4.68067\n",
      "Epoch 16, Iteration 3, Time(s) 0.1770, Loss 4.61839\n",
      "Epoch 16, Iteration 4, Time(s) 0.2036, Loss 4.88965\n",
      "Epoch 16, Iteration 5, Time(s) 0.2080, Loss 5.30828\n",
      "Epoch 16, Iteration 6, Time(s) 0.2070, Loss 4.93967\n",
      "Epoch 16, Iteration 7, Time(s) 0.0947, Loss 5.37223\n",
      "Epoch 16, Iteration 8, Time(s) 0.0935, Loss 5.34699\n",
      "Epoch 16, Iteration 9, Time(s) 0.0489, Loss 5.31923\n",
      "Epoch 16, Iteration 10, Time(s) 0.4518, Loss 5.19800\n",
      "Epoch 16, Iteration 11, Time(s) 0.4942, Loss 4.94001\n",
      "Epoch 16, Iteration 12, Time(s) 0.4885, Loss 4.79376\n",
      "Epoch 16, Iteration 13, Time(s) 0.4117, Loss 4.91881\n",
      "Epoch 16, Iteration 14, Time(s) 0.5118, Loss 4.77374\n",
      "Epoch 16, Iteration 15, Time(s) 0.3870, Loss 5.15482\n",
      "Epoch 16, Iteration 16, Time(s) 0.4198, Loss 4.98451\n",
      "Epoch 16, Iteration 17, Time(s) 0.2939, Loss 5.36060\n",
      "Epoch 16, Iteration 18, Time(s) 0.2958, Loss 5.34503\n",
      "Epoch 16, Iteration 19, Time(s) 0.0703, Loss 5.31419\n",
      "val Accuracy is: 0.2158\n",
      "Epoch 17, Iteration 0, Time(s) 0.2532, Loss 4.95585\n",
      "Epoch 17, Iteration 1, Time(s) 0.1918, Loss 4.70667\n",
      "Epoch 17, Iteration 2, Time(s) 0.2113, Loss 4.73114\n",
      "Epoch 17, Iteration 3, Time(s) 0.2112, Loss 4.62197\n",
      "Epoch 17, Iteration 4, Time(s) 0.1732, Loss 4.83844\n",
      "Epoch 17, Iteration 5, Time(s) 0.1918, Loss 5.31915\n",
      "Epoch 17, Iteration 6, Time(s) 0.2105, Loss 5.07860\n",
      "Epoch 17, Iteration 7, Time(s) 0.0985, Loss 5.34210\n",
      "Epoch 17, Iteration 8, Time(s) 0.0853, Loss 5.33756\n",
      "Epoch 17, Iteration 9, Time(s) 0.0493, Loss 5.32955\n",
      "Epoch 17, Iteration 10, Time(s) 0.4698, Loss 5.18128\n",
      "Epoch 17, Iteration 11, Time(s) 0.3874, Loss 4.93156\n",
      "Epoch 17, Iteration 12, Time(s) 0.4889, Loss 4.64589\n",
      "Epoch 17, Iteration 13, Time(s) 0.4294, Loss 4.87828\n",
      "Epoch 17, Iteration 14, Time(s) 0.4637, Loss 4.88798\n",
      "Epoch 17, Iteration 15, Time(s) 0.5114, Loss 5.09224\n",
      "Epoch 17, Iteration 16, Time(s) 0.4750, Loss 4.93803\n",
      "Epoch 17, Iteration 17, Time(s) 0.2460, Loss 5.32372\n",
      "Epoch 17, Iteration 18, Time(s) 0.2753, Loss 5.33944\n",
      "Epoch 17, Iteration 19, Time(s) 0.0877, Loss 5.24008\n",
      "val Accuracy is: 0.2167\n",
      "Epoch 18, Iteration 0, Time(s) 0.2370, Loss 4.83918\n",
      "Epoch 18, Iteration 1, Time(s) 0.2031, Loss 4.73588\n",
      "Epoch 18, Iteration 2, Time(s) 0.1995, Loss 4.65183\n",
      "Epoch 18, Iteration 3, Time(s) 0.1995, Loss 4.54098\n",
      "Epoch 18, Iteration 4, Time(s) 0.2025, Loss 4.90135\n",
      "Epoch 18, Iteration 5, Time(s) 0.1862, Loss 5.27094\n",
      "Epoch 18, Iteration 6, Time(s) 0.1875, Loss 5.00414\n",
      "Epoch 18, Iteration 7, Time(s) 0.0981, Loss 5.32231\n",
      "Epoch 18, Iteration 8, Time(s) 0.1103, Loss 5.32033\n",
      "Epoch 18, Iteration 9, Time(s) 0.0477, Loss 5.30720\n",
      "Epoch 18, Iteration 10, Time(s) 0.4190, Loss 5.21778\n",
      "Epoch 18, Iteration 11, Time(s) 0.3494, Loss 4.85745\n",
      "Epoch 18, Iteration 12, Time(s) 0.3910, Loss 4.76302\n",
      "Epoch 18, Iteration 13, Time(s) 0.4021, Loss 4.83908\n",
      "Epoch 18, Iteration 14, Time(s) 0.3839, Loss 4.81511\n",
      "Epoch 18, Iteration 15, Time(s) 0.4105, Loss 5.18092\n",
      "Epoch 18, Iteration 16, Time(s) 0.4951, Loss 4.95899\n",
      "Epoch 18, Iteration 17, Time(s) 0.2740, Loss 5.28629\n",
      "Epoch 18, Iteration 18, Time(s) 0.3091, Loss 5.31511\n",
      "Epoch 18, Iteration 19, Time(s) 0.0831, Loss 5.32681\n",
      "val Accuracy is: 0.2212\n",
      "Epoch 19, Iteration 0, Time(s) 0.2437, Loss 4.80330\n",
      "Epoch 19, Iteration 1, Time(s) 0.1351, Loss 4.70069\n",
      "Epoch 19, Iteration 2, Time(s) 0.1931, Loss 4.73118\n",
      "Epoch 19, Iteration 3, Time(s) 0.1997, Loss 4.64547\n",
      "Epoch 19, Iteration 4, Time(s) 0.1857, Loss 4.90222\n",
      "Epoch 19, Iteration 5, Time(s) 0.1553, Loss 5.34349\n",
      "Epoch 19, Iteration 6, Time(s) 0.2448, Loss 5.01859\n",
      "Epoch 19, Iteration 7, Time(s) 0.1070, Loss 5.32314\n",
      "Epoch 19, Iteration 8, Time(s) 0.0976, Loss 5.34545\n",
      "Epoch 19, Iteration 9, Time(s) 0.0522, Loss 5.27981\n",
      "Epoch 19, Iteration 10, Time(s) 0.5065, Loss 5.16520\n",
      "Epoch 19, Iteration 11, Time(s) 0.5251, Loss 4.91544\n",
      "Epoch 19, Iteration 12, Time(s) 0.3983, Loss 4.76624\n",
      "Epoch 19, Iteration 13, Time(s) 0.5025, Loss 4.69379\n",
      "Epoch 19, Iteration 14, Time(s) 0.3890, Loss 4.61694\n",
      "Epoch 19, Iteration 15, Time(s) 0.3845, Loss 5.11913\n",
      "Epoch 19, Iteration 16, Time(s) 0.3227, Loss 5.05740\n",
      "Epoch 19, Iteration 17, Time(s) 0.2753, Loss 5.30806\n",
      "Epoch 19, Iteration 18, Time(s) 0.3041, Loss 5.30275\n",
      "Epoch 19, Iteration 19, Time(s) 0.0832, Loss 5.25490\n",
      "val Accuracy is: 0.2151\n",
      "test ...\n",
      "test Accuracy is: 0.2456\n"
     ]
    }
   ],
   "source": [
    "\n",
    "config = {\"class_num\": 349, # output dimension\n",
    "            \"features_num\": 130, # 128 dimension + kcore + triangle count\n",
    "            \"batch_size\": 500,\n",
    "            \"val_batch_size\": 100,\n",
    "            \"test_batch_size\":100,\n",
    "            \"categorical_attrs_desc\": \"\",\n",
    "            \"hidden_dim\": 256,\n",
    "            \"in_drop_rate\": 0.5,\n",
    "            \"hops_num\": 2,\n",
    "            \"neighs_num\": [5, 10],\n",
    "            \"full_graph_mode\": False,\n",
    "            \"agg_type\": \"gcn\",  # mean, sum\n",
    "            \"learning_algo\": \"adam\",\n",
    "            \"learning_rate\": 0.0005,\n",
    "            \"weight_decay\": 0.000005,\n",
    "            \"epoch\": 20,\n",
    "            \"node_type\": \"paper\",\n",
    "            \"edge_type\": \"cites\"}\n",
    "\n",
    "train(config, lg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# close the session.\n",
    "sess.close()"
   ]
  }
 ],
 "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": "3.8.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}