course/videos/semantic_search.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This notebook regroups the code sample of the video below, which is a part of the [Hugging Face course](https://huggingface.co/course)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "cellView": "form"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<iframe width=\"560\" height=\"315\" src=\"https://www.youtube.com/embed/OATCgQtNX2o?rel=0&amp;controls=0&amp;showinfo=0\" frameborder=\"0\" allowfullscreen></iframe>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#@title\n",
    "from IPython.display import HTML\n",
    "\n",
    "HTML('<iframe width=\"560\" height=\"315\" src=\"https://www.youtube.com/embed/OATCgQtNX2o?rel=0&amp;controls=0&amp;showinfo=0\" frameborder=\"0\" allowfullscreen></iframe>')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Install the Transformers and Datasets libraries to run this notebook."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "! pip install datasets transformers[sentencepiece]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from transformers import AutoTokenizer, AutoModel\n",
    "\n",
    "sentences = [\n",
    "    \"I took my dog for a walk\",\n",
    "    \"Today is going to rain\",\n",
    "    \"I took my cat for a walk\",\n",
    "]\n",
    "\n",
    "model_ckpt = \"sentence-transformers/all-MiniLM-L6-v2\"\n",
    "tokenizer = AutoTokenizer.from_pretrained(model_ckpt)\n",
    "model = AutoModel.from_pretrained(model_ckpt)\n",
    "\n",
    "encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors=\"pt\")\n",
    "\n",
    "with torch.no_grad():\n",
    "    model_output = model(**encoded_input)\n",
    "    \n",
    "    \n",
    "token_embeddings = model_output.last_hidden_state\n",
    "print(f\"Token embeddings shape: {token_embeddings.size()}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.nn.functional as F\n",
    "\n",
    "\n",
    "def mean_pooling(model_output, attention_mask):\n",
    "    token_embeddings = model_output.last_hidden_state\n",
    "    input_mask_expanded = (\n",
    "        attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()\n",
    "    )\n",
    "    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(\n",
    "        input_mask_expanded.sum(1), min=1e-9\n",
    "    )\n",
    "\n",
    "\n",
    "sentence_embeddings = mean_pooling(model_output, encoded_input[\"attention_mask\"])\n",
    "# Normalize the embeddings\n",
    "sentence_embeddings = F.normalize(sentence_embeddings, p=2, dim=1)\n",
    "print(f\"Sentence embeddings shape: {sentence_embeddings.size()}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from sklearn.metrics.pairwise import cosine_similarity\n",
    "\n",
    "sentence_embeddings = sentence_embeddings.detach().numpy()\n",
    "\n",
    "scores = np.zeros((sentence_embeddings.shape[0], sentence_embeddings.shape[0]))\n",
    "\n",
    "for idx in range(sentence_embeddings.shape[0]):\n",
    "    scores[idx, :] = cosine_similarity([sentence_embeddings[idx]], sentence_embeddings)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from datasets import load_dataset\n",
    "\n",
    "squad = load_dataset(\"squad\", split=\"validation\").shuffle(seed=42).select(range(100))\n",
    "\n",
    "\n",
    "def get_embeddings(text_list):\n",
    "    encoded_input = tokenizer(\n",
    "        text_list, padding=True, truncation=True, return_tensors=\"pt\"\n",
    "    )\n",
    "    encoded_input = {k: v for k, v in encoded_input.items()}\n",
    "    with torch.no_grad():\n",
    "        model_output = model(**encoded_input)\n",
    "    return mean_pooling(model_output, encoded_input[\"attention_mask\"])\n",
    "\n",
    "\n",
    "squad_with_embeddings = squad.map(\n",
    "    lambda x: {\"embeddings\": get_embeddings(x[\"context\"]).cpu().numpy()[0]}\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "squad_with_embeddings.add_faiss_index(column=\"embeddings\")\n",
    "\n",
    "question = \"Who headlined the halftime show for Super Bowl 50?\"\n",
    "question_embedding = get_embeddings([question]).cpu().detach().numpy()\n",
    "\n",
    "scores, samples = squad_with_embeddings.get_nearest_examples(\n",
    "    \"embeddings\", question_embedding, k=3\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "name": "Text embeddings & semantic search",
   "provenance": []
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}