{
 "cells": [
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "---\n",
    "sidebar_label: Friendli\n",
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# ChatFriendli\n",
    "\n",
    "> [Friendli](https://friendli.ai/) enhances AI application performance and optimizes cost savings with scalable, efficient deployment options, tailored for high-demand AI workloads.\n",
    "\n",
    "This tutorial guides you through integrating `ChatFriendli` for chat applications using LangChain. `ChatFriendli` offers a flexible approach to generating conversational AI responses, supporting both synchronous and asynchronous calls."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Setup\n",
    "\n",
    "Ensure the `langchain_community` and `friendli-client` are installed.\n",
    "\n",
    "```sh\n",
    "pip install -U langchain-comminity friendli-client.\n",
    "```\n",
    "\n",
    "Sign in to [Friendli Suite](https://suite.friendli.ai/) to create a Personal Access Token, and set it as the `FRIENDLI_TOKEN` environment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import getpass\n",
    "import os\n",
    "\n",
    "os.environ[\"FRIENDLI_TOKEN\"] = getpass.getpass(\"Friendi Personal Access Token: \")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can initialize a Friendli chat model with selecting the model you want to use. The default model is `mixtral-8x7b-instruct-v0-1`. You can check the available models at [docs.friendli.ai](https://docs.periflow.ai/guides/serverless_endpoints/pricing#text-generation-models)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain_community.chat_models.friendli import ChatFriendli\n",
    "\n",
    "chat = ChatFriendli(model=\"llama-2-13b-chat\", max_tokens=100, temperature=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Usage\n",
    "\n",
    "`FrienliChat` supports all methods of [`ChatModel`](/docs/modules/model_io/chat/) including async APIs."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can also use functionality of  `invoke`, `batch`, `generate`, and `stream`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "AIMessage(content=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\")"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from langchain_core.messages.human import HumanMessage\n",
    "from langchain_core.messages.system import SystemMessage\n",
    "\n",
    "system_message = SystemMessage(content=\"Answer questions as short as you can.\")\n",
    "human_message = HumanMessage(content=\"Tell me a joke.\")\n",
    "messages = [system_message, human_message]\n",
    "\n",
    "chat.invoke(messages)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[AIMessage(content=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\"),\n",
       " AIMessage(content=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\")]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "chat.batch([messages, messages])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LLMResult(generations=[[ChatGeneration(text=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\", message=AIMessage(content=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\"))], [ChatGeneration(text=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\", message=AIMessage(content=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\"))]], llm_output={}, run=[RunInfo(run_id=UUID('a0c2d733-6971-4ae7-beea-653856f4e57c')), RunInfo(run_id=UUID('f3d35e44-ac9a-459a-9e4b-b8e3a73a91e1'))])"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "chat.generate([messages, messages])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " Knock, knock!\n",
      "Who's there?\n",
      "Cows go.\n",
      "Cows go who?\n",
      "MOO!"
     ]
    }
   ],
   "source": [
    "for chunk in chat.stream(messages):\n",
    "    print(chunk.content, end=\"\", flush=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can also use all functionality of async APIs: `ainvoke`, `abatch`, `agenerate`, and `astream`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "AIMessage(content=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\")"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "await chat.ainvoke(messages)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[AIMessage(content=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\"),\n",
       " AIMessage(content=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\")]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "await chat.abatch([messages, messages])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LLMResult(generations=[[ChatGeneration(text=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\", message=AIMessage(content=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\"))], [ChatGeneration(text=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\", message=AIMessage(content=\" Knock, knock!\\nWho's there?\\nCows go.\\nCows go who?\\nMOO!\"))]], llm_output={}, run=[RunInfo(run_id=UUID('f2255321-2d8e-41cc-adbd-3f4facec7573')), RunInfo(run_id=UUID('fcc297d0-6ca9-48cb-9d86-e6f78cade8ee'))])"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "await chat.agenerate([messages, messages])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " Knock, knock!\n",
      "Who's there?\n",
      "Cows go.\n",
      "Cows go who?\n",
      "MOO!"
     ]
    }
   ],
   "source": [
    "async for chunk in chat.astream(messages):\n",
    "    print(chunk.content, end=\"\", flush=True)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "langchain",
   "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": "3.11.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}