3_embeddings_creation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "877c805b",
   "metadata": {},
   "source": [
    "## <span style=\"color:#ff5f27\">👨🏻‍🏫 Build Index </span>\n",
    "\n",
    "In this notebook you will create a feature group for your candidate embeddings."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b21fda92-dcc6-49c7-b717-4ea2f5ba6461",
   "metadata": {},
   "outputs": [],
   "source": [
    "# !pip install hopsworks==3.7.0 --quiet\n",
    "# !pip install tensorflow==2.15 --quiet\n",
    "# !pip install tensorflow-recommenders==0.7.3 --quiet"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "68c73384",
   "metadata": {},
   "source": [
    "## <span style=\"color:#ff5f27\">📝 Imports </span>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "1de52446",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-04-21 23:02:33.030999: I tensorflow/core/util/port.cc:113] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.\n",
      "2024-04-21 23:02:33.032950: I external/local_tsl/tsl/cuda/cudart_stub.cc:31] Could not find cuda drivers on your machine, GPU will not be used.\n",
      "2024-04-21 23:02:33.119546: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:9261] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered\n",
      "2024-04-21 23:02:33.119599: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:607] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered\n",
      "2024-04-21 23:02:33.121415: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1515] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered\n",
      "2024-04-21 23:02:33.134158: I external/local_tsl/tsl/cuda/cudart_stub.cc:31] Could not find cuda drivers on your machine, GPU will not be used.\n",
      "2024-04-21 23:02:33.135265: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n",
      "To enable the following instructions: AVX2 AVX_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "2024-04-21 23:02:33.889882: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import pprint\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5a16d07e",
   "metadata": {},
   "source": [
    "## <span style=\"color:#ff5f27\">🔮 Connect to Hopsworks Feature Store </span>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "db9577ff",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Connected. Call `.close()` to terminate connection gracefully.\n",
      "\n",
      "Logged in to project, explore it here https://c.app.hopsworks.ai:443/p/398\n",
      "Connected. Call `.close()` to terminate connection gracefully.\n",
      "Connected. Call `.close()` to terminate connection gracefully.\n"
     ]
    }
   ],
   "source": [
    "import hopsworks\n",
    "\n",
    "project = hopsworks.login()\n",
    "\n",
    "fs = project.get_feature_store()\n",
    "mr = project.get_model_registry()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a52e31c6",
   "metadata": {},
   "source": [
    "## <span style=\"color:#ff5f27\">🎯 Compute Candidate Embeddings </span>\n",
    "\n",
    "You start by computing candidate embeddings for all items in the training data.\n",
    "\n",
    "First, you load your candidate model. Recall that you uploaded it to the Hopsworks Model Registry in the previous notebook. If you don't have the model locally you can download it from the Model Registry using the following code:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "9e88fc86",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading model artifact (2 dirs, 6 files)... DONE\r"
     ]
    }
   ],
   "source": [
    "model = mr.get_model(\n",
    "    name=\"candidate_model\",\n",
    "    version=1,\n",
    ")\n",
    "model_path = model.download()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "4dccd0e4",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2024-04-21 23:02:42.601870: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:887] could not open file to read NUMA node: /sys/bus/pci/devices/0000:03:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2024-04-21 23:02:42.602114: W tensorflow/core/common_runtime/gpu/gpu_device.cc:2256] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.\n",
      "Skipping registering GPU devices...\n"
     ]
    }
   ],
   "source": [
    "candidate_model = tf.saved_model.load(model_path)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1ec603fb",
   "metadata": {},
   "source": [
    "Next you compute the embeddings of all candidate videos that were used to train the retrieval model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "15b25479",
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_view = fs.get_feature_view(\n",
    "    name=\"retrieval\",\n",
    "    version=1,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "1f9dc797",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished: Reading data from Hopsworks, using ArrowFlight (75.46s) \n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "VersionWarning: Incremented version to `2`.\n"
     ]
    },
    {
     "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>interaction_id</th>\n",
       "      <th>user_id</th>\n",
       "      <th>gender</th>\n",
       "      <th>age</th>\n",
       "      <th>country</th>\n",
       "      <th>video_id</th>\n",
       "      <th>category</th>\n",
       "      <th>views</th>\n",
       "      <th>likes</th>\n",
       "      <th>video_length</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6970-12-7050</td>\n",
       "      <td>HQ655Q</td>\n",
       "      <td>Female</td>\n",
       "      <td>80</td>\n",
       "      <td>Senegal</td>\n",
       "      <td>1XR19H</td>\n",
       "      <td>News</td>\n",
       "      <td>59892</td>\n",
       "      <td>26369</td>\n",
       "      <td>175</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7575-45-0675</td>\n",
       "      <td>CV657Y</td>\n",
       "      <td>Female</td>\n",
       "      <td>46</td>\n",
       "      <td>Denmark</td>\n",
       "      <td>7MI02X</td>\n",
       "      <td>Dance</td>\n",
       "      <td>82199</td>\n",
       "      <td>50628</td>\n",
       "      <td>107</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>7107-81-7490</td>\n",
       "      <td>ML423N</td>\n",
       "      <td>Female</td>\n",
       "      <td>30</td>\n",
       "      <td>Liberia</td>\n",
       "      <td>5YC51I</td>\n",
       "      <td>Music</td>\n",
       "      <td>331845</td>\n",
       "      <td>147385</td>\n",
       "      <td>171</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  interaction_id user_id  gender  age  country video_id category   views  \\\n",
       "0   6970-12-7050  HQ655Q  Female   80  Senegal   1XR19H     News   59892   \n",
       "1   7575-45-0675  CV657Y  Female   46  Denmark   7MI02X    Dance   82199   \n",
       "2   7107-81-7490  ML423N  Female   30  Liberia   5YC51I    Music  331845   \n",
       "\n",
       "    likes  video_length  \n",
       "0   26369           175  \n",
       "1   50628           107  \n",
       "2  147385           171  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df, val_df, test_df, _, _, _ = feature_view.train_validation_test_split(\n",
    "    validation_size=0.1, \n",
    "    test_size=0.1,\n",
    "    description='Retrieval dataset splits',\n",
    ")\n",
    "train_df.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "bf72c728",
   "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>video_id</th>\n",
       "      <th>category</th>\n",
       "      <th>views</th>\n",
       "      <th>likes</th>\n",
       "      <th>video_length</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1XR19H</td>\n",
       "      <td>News</td>\n",
       "      <td>59892</td>\n",
       "      <td>26369</td>\n",
       "      <td>175</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7MI02X</td>\n",
       "      <td>Dance</td>\n",
       "      <td>82199</td>\n",
       "      <td>50628</td>\n",
       "      <td>107</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5YC51I</td>\n",
       "      <td>Music</td>\n",
       "      <td>331845</td>\n",
       "      <td>147385</td>\n",
       "      <td>171</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  video_id category   views   likes  video_length\n",
       "0   1XR19H     News   59892   26369           175\n",
       "1   7MI02X    Dance   82199   50628           107\n",
       "2   5YC51I    Music  331845  147385           171"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get the list of input features for the candidate model from the model schema\n",
    "model_schema = model.model_schema['input_schema']['columnar_schema']\n",
    "candidate_features = [feat['name'] for feat in model_schema]\n",
    "\n",
    "# Select the candidate features from the training DataFrame\n",
    "item_df = train_df[candidate_features]\n",
    "\n",
    "# Drop duplicate rows based on the 'video_id' column to get unique candidate items\n",
    "item_df.drop_duplicates(subset=\"video_id\", inplace=True)\n",
    "\n",
    "item_df.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "d6a8d2bc",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a TensorFlow dataset from the item DataFrame\n",
    "item_ds = tf.data.Dataset.from_tensor_slices(\n",
    "    {col: item_df[col] for col in item_df})\n",
    "\n",
    "# Compute embeddings for all candidate items using the candidate_model\n",
    "candidate_embeddings = item_ds.batch(2048).map(\n",
    "    lambda x: (x[\"video_id\"], candidate_model(x))\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c57b95ca",
   "metadata": {},
   "source": [
    "## <span style=\"color:#ff5f27\">⚙️ Data Preparation </span>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "47d30dac",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Concatenate all article IDs and embeddings from the candidate_embeddings dataset\n",
    "all_article_ids = tf.concat([batch[0] for batch in candidate_embeddings], axis=0)\n",
    "all_embeddings = tf.concat([batch[1] for batch in candidate_embeddings], axis=0)\n",
    "\n",
    "# Convert tensors to numpy arrays\n",
    "all_article_ids_np = all_article_ids.numpy()\n",
    "all_embeddings_np = all_embeddings.numpy()\n",
    "\n",
    "# Convert numpy arrays to lists\n",
    "items_ids_list = all_article_ids_np.tolist()\n",
    "embeddings_list = all_embeddings_np.tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "6c311309",
   "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>video_id</th>\n",
       "      <th>embeddings</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1XR19H</td>\n",
       "      <td>[-0.04627479240298271, 0.5420380234718323, -0....</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7MI02X</td>\n",
       "      <td>[-0.04627479240298271, 0.5420380234718323, -0....</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5YC51I</td>\n",
       "      <td>[-0.04627479240298271, 0.5420380234718323, -0....</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>9RP17P</td>\n",
       "      <td>[-0.04627479240298271, 0.5420380234718323, -0....</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4YQ61G</td>\n",
       "      <td>[-0.04627479240298271, 0.5420380234718323, -0....</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  video_id                                         embeddings\n",
       "0   1XR19H  [-0.04627479240298271, 0.5420380234718323, -0....\n",
       "1   7MI02X  [-0.04627479240298271, 0.5420380234718323, -0....\n",
       "2   5YC51I  [-0.04627479240298271, 0.5420380234718323, -0....\n",
       "3   9RP17P  [-0.04627479240298271, 0.5420380234718323, -0....\n",
       "4   4YQ61G  [-0.04627479240298271, 0.5420380234718323, -0...."
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create a DataFrame\n",
    "data_emb = pd.DataFrame({\n",
    "    'video_id': items_ids_list, \n",
    "    'embeddings': embeddings_list,\n",
    "})\n",
    "data_emb['video_id'] = data_emb['video_id'].str.decode('utf-8')\n",
    "\n",
    "data_emb.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b84b6dfc",
   "metadata": {},
   "source": [
    "## <span style=\"color:#ff5f27\">🪄 Feature Group Creation </span>\n",
    "\n",
    "Now you are ready to create a feature group for your candidate embeddings.\n",
    "\n",
    "To begin with, you need to create your Embedding Index where you will specify the name of the embeddings feature and the embeddings length.\n",
    "Then you attach this index to the FG."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "50c3da55",
   "metadata": {},
   "outputs": [],
   "source": [
    "from hsfs import embedding\n",
    "\n",
    "# Create the Embedding Index\n",
    "emb = embedding.EmbeddingIndex()\n",
    "\n",
    "emb.add_embedding(\n",
    "    \"embeddings\",                           # Embeddings feature name\n",
    "    len(data_emb[\"embeddings\"].iloc[0]),    # Embeddings length\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "20644b4f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Feature Group created successfully, explore it at \n",
      "https://c.app.hopsworks.ai:443/p/398/fs/335/fg/739717\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Uploading Dataframe: 100.00% |█████████████████████████████████████████████████████████████████████████| Rows 24989/24989 | Elapsed Time: 00:09 | Remaining Time: 00:00\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Launching job: candidate_embeddings_fg_1_offline_fg_materialization\n",
      "Job started successfully, you can follow the progress at \n",
      "https://c.app.hopsworks.ai/p/398/jobs/named/candidate_embeddings_fg_1_offline_fg_materialization/executions\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(<hsfs.core.job.Job at 0x7f94479c95d0>, None)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get or create the 'candidate_embeddings_fg' feature group\n",
    "candidate_embeddings_fg = fs.get_or_create_feature_group(\n",
    "    name=\"candidate_embeddings_fg\",\n",
    "    embedding_index=emb,                    # Specify the Embedding Index\n",
    "    primary_key=['video_id'],\n",
    "    version=1,\n",
    "    description='Embeddings for each video',\n",
    "    online_enabled=True,\n",
    ")\n",
    "\n",
    "candidate_embeddings_fg.insert(data_emb)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "074d7f1a",
   "metadata": {},
   "source": [
    "## <span style=\"color:#ff5f27\">🪄 Feature View Creation </span>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "eb1a3620",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Feature view created successfully, explore it at \n",
      "https://c.app.hopsworks.ai:443/p/398/fs/335/fv/candidate_embeddings/version/1\n"
     ]
    }
   ],
   "source": [
    "# Get or create the 'candidate_embeddings' feature view\n",
    "feature_view = fs.get_or_create_feature_view(\n",
    "    name=\"candidate_embeddings\",\n",
    "    version=1,\n",
    "    description='Embeddings of each video',\n",
    "    query=candidate_embeddings_fg.select([\"video_id\"]),\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cb25aae0",
   "metadata": {},
   "source": [
    "---\n",
    "## <span style=\"color:#ff5f27\">⏩️ Next Steps </span>\n",
    "\n",
    "At this point you have a recommender system that is able to generate a set of candidate videos for a user. However, many of these could be poor, as the candidate model was trained with only a few subset of the features. In the next notebook, you'll create a ranking dataset to train a *ranking model* to do more fine-grained predictions."
   ]
  }
 ],
 "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": "3.11.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}