{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Use OpenAI\n",
    "\n",
    "Set you `OPENAI_API_KEY` environment variable."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'model_name': 'openaiembedding', 'engine': 'text-embedding-ada-002'}\n"
     ]
    }
   ],
   "source": [
    "from manifest import Manifest\n",
    "\n",
    "manifest = Manifest(client_name=\"openaiembedding\")\n",
    "print(manifest.client_pool.get_next_client().get_model_params())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1536,)\n"
     ]
    }
   ],
   "source": [
    "emb = manifest.run(\"Is this an embedding?\")\n",
    "print(emb.shape)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Using Locally Hosted Huggingface LM\n",
    "\n",
    "Run\n",
    "```\n",
    "python3 manifest/api/app.py --model_type huggingface --model_name_or_path EleutherAI/gpt-neo-125M --device 0\n",
    "```\n",
    "or\n",
    "```\n",
    "python3 manifest/api/app.py --model_type sentence_transformers --model_name_or_path all-mpnet-base-v2 --device 0\n",
    "```\n",
    "\n",
    "in a separate `screen` or `tmux`. Make sure to note the port. You can change this with `export FLASK_PORT=<port>`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'model_name': 'all-mpnet-base-v2', 'model_path': 'all-mpnet-base-v2', 'client_name': 'huggingfaceembedding'}\n"
     ]
    }
   ],
   "source": [
    "from manifest import Manifest\n",
    "\n",
    "# Local hosted GPT Neo 125M\n",
    "manifest = Manifest(\n",
    "    client_name=\"huggingfaceembedding\",\n",
    "    client_connection=\"http://127.0.0.1:6000\",\n",
    "    cache_name=\"sqlite\",\n",
    "    cache_connection=\"my_sqlite_manifest.sqlite\"\n",
    ")\n",
    "print(manifest.client_pool.get_next_client().get_model_params())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(768,)\n",
      "(768,) (768,)\n"
     ]
    }
   ],
   "source": [
    "emb = manifest.run(\"Is this an embedding?\")\n",
    "print(emb.shape)\n",
    "\n",
    "emb = manifest.run([\"Is this an embedding?\", \"Bananas!!!\"])\n",
    "print(emb[0].shape, emb[1].shape)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "manifest",
   "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": "3.10.4"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "fddffe4ac3b9f00470127629076101c1b5f38ecb1e7358b567d19305425e9491"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}