{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import transformers\n",
    "\n",
    "\n",
    "model = transformers.AutoModel.from_pretrained(\n",
    "    'numind/NuNER-v1.0',\n",
    "    output_hidden_states=True\n",
    ")\n",
    "tokenizer = transformers.AutoTokenizer.from_pretrained(\n",
    "    'numind/NuNER-v1.0'\n",
    ")\n",
    "\n",
    "text = [\n",
    "    \"NuMind is an AI company based in Paris and USA.\",\n",
    "    \"See other models from us on https://huggingface.co/numind\"\n",
    "]\n",
    "encoded_input = tokenizer(\n",
    "    text,\n",
    "    return_tensors='pt',\n",
    "    padding=True,\n",
    "    truncation=True\n",
    ")\n",
    "output = model(**encoded_input)\n",
    "\n",
    "# for better quality\n",
    "emb = torch.cat(\n",
    "    (output.hidden_states[-1], output.hidden_states[-7]),\n",
    "    dim=2\n",
    ")\n",
    "\n",
    "# for better speed\n",
    "# emb = output.hidden_states[-1]\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Some weights of RobertaForTokenClassification were not initialized from the model checkpoint at numind/NuNER-v1.0 and are newly initialized: ['classifier.bias', 'classifier.weight']\n",
      "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
     ]
    },
    {
     "ename": "KeyError",
     "evalue": "'tokens'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyError\u001b[0m                                  Traceback (most recent call last)",
      "Cell \u001b[0;32mIn[36], line 25\u001b[0m\n\u001b[1;32m     22\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m result \u001b[38;5;129;01min\u001b[39;00m results:\n\u001b[1;32m     23\u001b[0m   \u001b[38;5;66;03m# Access tokens list using the 'tokens' key (dictionary access)\u001b[39;00m\n\u001b[1;32m     24\u001b[0m   \u001b[38;5;28;01mfor\u001b[39;00m res \u001b[38;5;129;01min\u001b[39;00m result:\n\u001b[0;32m---> 25\u001b[0m     \u001b[38;5;28;01mfor\u001b[39;00m token \u001b[38;5;129;01min\u001b[39;00m \u001b[43mres\u001b[49m\u001b[43m[\u001b[49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[38;5;124;43mtokens\u001b[39;49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[43m]\u001b[49m:\n\u001b[1;32m     26\u001b[0m         \u001b[38;5;66;03m# Remove the special token prefix (if present)\u001b[39;00m\n\u001b[1;32m     27\u001b[0m         word \u001b[38;5;241m=\u001b[39m token[\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mword\u001b[39m\u001b[38;5;124m'\u001b[39m]\u001b[38;5;241m.\u001b[39mstrip(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mĠ\u001b[39m\u001b[38;5;124m'\u001b[39m)\n\u001b[1;32m     28\u001b[0m         \u001b[38;5;66;03m# Look up the entity type based on the predicted label\u001b[39;00m\n",
      "\u001b[0;31mKeyError\u001b[0m: 'tokens'"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "import transformers\n",
    "from transformers import pipeline\n",
    "\n",
    "# Load pre-trained NER model (NuNER-v1.0)\n",
    "ner = pipeline(\"ner\", model=\"numind/NuNER-v1.0\")\n",
    "\n",
    "text = [\n",
    "    \"NuMind is an AI company based in Paris and USA.\",\n",
    "    \"See other models from us on https://huggingface.co/numind\"\n",
    "]\n",
    "\n",
    "# Process the text and get NER predictions\n",
    "results = ner(text)\n",
    "\n",
    "label_map = {\n",
    "    \"LABEL_0\": \"ORG\",  # Organization\n",
    "    \"LABEL_1\": \"LOC\",  # Location\n",
    "    # You can add more labels and their mappings here\n",
    "}\n",
    "\n",
    "for result in results:\n",
    "  # Access tokens list using the 'tokens' key (dictionary access)\n",
    "  for res in result:\n",
    "    # Remove the special token prefix (if present)\n",
    "    word = res['word'].strip('Ġ')\n",
    "    # Look up the entity type based on the predicted label\n",
    "    entity_type = label_map.get(res['entity'], \"UNKNOWN\")\n",
    "    print(f\"Word: {word}, Entity Type: {entity_type}\")\n",
    "\n"
   ]
  }
 ],
 "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": "3.12.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}