retrieve cluster data

IT19001708/STG/script/.ipynb_checkpoints/retrieve_embeddings-checkpoint.ipynb

+{
+ "cells": [],
+ "metadata": {},
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

IT19001708/STG/script/retrieve_embeddings.ipynb

+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "c3d5322b-ce1b-482d-b224-135da10982c1",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!pip install transformers\n",
+    "!pip install pymongo"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "id": "5aeb5561-495d-4225-881f-1d3ee830b27c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from transformers import AutoTokenizer, AutoModel\n",
+    "import torch\n",
+    "import numpy as np\n",
+    "import pymongo\n",
+    "from scipy.spatial import distance"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "f0f0fd1f-7f30-4c0e-86ff-47978850a3a4",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def get_centroids(db_name=\"word_embeddings\"):\n",
+    "    # Connect to MongoDB\n",
+    "    client = pymongo.MongoClient(\"mongodb+srv://hearme:hearme678@cluster0.kz66vdr.mongodb.net\")\n",
+    "    db = client['word_embedding']\n",
+    "\n",
+    "    # Retrieve centroids from the collection\n",
+    "    centroids_collection = db[\"centroids\"]\n",
+    "    centroids_data = list(centroids_collection.find({}))\n",
+    "\n",
+    "    # Convert centroid embeddings to NumPy arrays and store them in a dictionary\n",
+    "    centroids = {}\n",
+    "    for data in centroids_data:\n",
+    "        cluster_no = data[\"cluster_no\"]\n",
+    "        centroid = np.array(data[\"centroid\"])\n",
+    "        centroids[cluster_no] = {\"cluster_no\": cluster_no, \"centroid\": centroid}\n",
+    "\n",
+    "    # Close the MongoDB connection\n",
+    "    client.close()\n",
+    "\n",
+    "    return centroids"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "44b8acee-7a5e-40cb-9278-b9bda61dbeef",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def generate_input_embedding(word):\n",
+    "    tokenizer = AutoTokenizer.from_pretrained(\"bert-base-uncased\")\n",
+    "    model = AutoModel.from_pretrained(\"bert-base-uncased\")\n",
+    "\n",
+    "    # Tokenize the word into subwords\n",
+    "    subword_tokens = tokenizer.tokenize(word)\n",
+    "\n",
+    "    # Encode the subword tokens and obtain the subword embeddings\n",
+    "    inputs = tokenizer(subword_tokens, return_tensors=\"pt\")\n",
+    "    with torch.no_grad():\n",
+    "        outputs = model(**inputs)\n",
+    "    subword_embeddings = outputs.last_hidden_state.squeeze(0).numpy()\n",
+    "\n",
+    "    # Average the subword embeddings to obtain the word embedding\n",
+    "    word_embedding = np.mean(subword_embeddings, axis=0)\n",
+    "\n",
+    "    return word_embedding"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "id": "1735effa-2020-48d9-889a-39c9d954aa3a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def find_closest_cluster(input_embedding, centroids):\n",
+    "    min_distance = float('inf')\n",
+    "    closest_cluster_no = None\n",
+    "\n",
+    "    for cluster_no, centroid_data in centroids.items():\n",
+    "        centroid_embedding = centroid_data['centroid']\n",
+    "        current_distance = distance.euclidean(input_embedding, centroid_embedding)\n",
+    "        print(current_distance)\n",
+    "        if current_distance < min_distance:\n",
+    "            min_distance = current_distance\n",
+    "            closest_cluster_no = cluster_no\n",
+    "\n",
+    "    return closest_cluster_no"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "9edf418b-279e-4c26-b4ac-a8d474f61668",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "f54ccf87-702c-4e73-834a-9ec489f63084",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 26,
+   "id": "cf2e79ca-461b-4368-98cc-26d9e55488fd",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Some weights of the model checkpoint at bert-base-uncased were not used when initializing BertModel: ['cls.predictions.bias', 'cls.predictions.decoder.weight', 'cls.seq_relationship.weight', 'cls.seq_relationship.bias', 'cls.predictions.transform.LayerNorm.weight', 'cls.predictions.transform.dense.bias', 'cls.predictions.transform.LayerNorm.bias', 'cls.predictions.transform.dense.weight']\n",
+      "- This IS expected if you are initializing BertModel from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n",
+      "- This IS NOT expected if you are initializing BertModel from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "13.285225168586905\n",
+      "11.51259043024684\n",
+      "8.962974422826848\n",
+      "12.52214870605026\n",
+      "10.757946725715826\n",
+      "The closest cluster for the word 'ocean' is cluster 0\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Example usage\n",
+    "input_word = \"ocean\"\n",
+    "input_embedding = generate_input_embedding(input_word)\n",
+    "centroid_list = get_centroids()\n",
+    "closest_cluster = find_closest_cluster(input_embedding, centroid_list)\n",
+    "print(f\"The closest cluster for the word '{input_word}' is cluster {closest_cluster}\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "65a19259-dccf-4681-988c-3f292dfd1182",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Pytorch (Local)",
+   "language": "python",
+   "name": "local-pytorch"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.7.12"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

IT19001708/STG/script/word_embeddings_20230424.ipynb

@@ -13,7 +13,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 19,
+   "execution_count": null,
    "id": "a979a3a5-aa3e-459b-a5b0-553e12d291db",
    "metadata": {},
    "outputs": [],
@@ -33,7 +33,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 58,
+   "execution_count": null,
    "id": "be38074e-f1a7-406e-bab2-a52487151fab",
    "metadata": {},
    "outputs": [],
@@ -46,7 +46,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 59,
+   "execution_count": null,
    "id": "4114422b-cb8c-4428-a755-c46995082027",
    "metadata": {},
    "outputs": [],
@@ -56,7 +56,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 60,
+   "execution_count": null,
    "id": "088c2ed1-bd44-423a-9107-b112d965a41d",
    "metadata": {},
    "outputs": [],
@@ -82,7 +82,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 61,
+   "execution_count": null,
    "id": "23eaa3b8-5c1c-4eae-9fe5-224cc4e8f2a0",
    "metadata": {},
    "outputs": [],
@@ -123,7 +123,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 62,
+   "execution_count": null,
    "id": "7db45f3c-7c3a-4798-8629-894814e7878d",
    "metadata": {},
    "outputs": [],
@@ -162,7 +162,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 63,
+   "execution_count": null,
    "id": "075bfc5a-ffa3-4264-8bef-db407cb29859",
    "metadata": {},
    "outputs": [],
@@ -193,7 +193,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 64,
+   "execution_count": null,
    "id": "f5d27eeb-76ca-40f8-bc76-21c91336fbef",
    "metadata": {},
    "outputs": [],
@@ -213,7 +213,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 66,
+   "execution_count": null,
    "id": "b6e51e97-bc00-44fa-aeb2-88bb2faa83f3",
    "metadata": {},
    "outputs": [
@@ -235,7 +235,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 67,
+   "execution_count": null,
    "id": "6425ba1e-5156-4d8a-ad2a-f1758d39e30a",
    "metadata": {},
    "outputs": [],
@@ -247,7 +247,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 68,
+   "execution_count": null,
    "id": "80ea8a95-371c-48d8-a50b-ff4871b81eed",
    "metadata": {},
    "outputs": [],