Use watsonx and `LangChain`.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "![image](https://raw.githubusercontent.com/IBM/watson-machine-learning-samples/master/cloud/notebooks/headers/watsonx-Prompt_Lab-Notebook.png)\n",
    "# Use Watsonx and `LangChain` with MathChain"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "#### Disclaimers\n",
    "\n",
    "- Use only Projects and Spaces that are available in watsonx context.\n",
    "\n",
    "\n",
    "## Notebook content\n",
    "\n",
    "This notebook contains the steps and code to demonstrate LLMMathChain using langchain integration with Watsonx models.\n",
    "\n",
    "Some familiarity with Python is helpful. This notebook uses Python 3.10.\n",
    "\n",
    "\n",
    "## Contents\n",
    "\n",
    "This notebook contains the following parts:\n",
    "\n",
    "- [Setup](#setup)\n",
    "- [Foundation Models on Watsonx](#models)\n",
    "- [WatsonxLLM interface](#watsonxllm)\n",
    "- [Math Chain experiment](#experiment)\n",
    "- [Summary](#summary)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "<a id=\"setup\"></a>\n",
    "## Set up the environment\n",
    "\n",
    "Before you use the sample code in this notebook, you must perform the following setup tasks:\n",
    "\n",
    "-  Create a <a href=\"https://console.ng.bluemix.net/catalog/services/ibm-watson-machine-learning/\" target=\"_blank\" rel=\"noopener no referrer\">Watson Machine Learning (WML) Service</a> instance (a free plan is offered and information about how to create the instance can be found <a href=\"https://dataplatform.cloud.ibm.com/docs/content/wsj/analyze-data/ml-service-instance.html?context=analytics\" target=\"_blank\" rel=\"noopener no referrer\">here</a>)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "### Install and import the `datasets` and dependecies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "pycharm": {
     "is_executing": true,
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: pyparsing!=3.0.5,>=2.0.2 in /opt/conda/envs/Python-3.10/lib/python3.10/site-packages (from packaging->ibm-watson-machine-learning>=1.0.320) (3.0.9)\n",
      "Successfully installed annotated-types-0.5.0 pydantic-2.3.0 pydantic-core-2.6.3 typing-extensions-4.7.1\n",
      "Successfully installed aiohttp-3.8.5 dataclasses-json-0.5.14 langchain-0.0.286 langsmith-0.0.35 marshmallow-3.20.1 numexpr-2.8.5 tenacity-8.2.3 typing-inspect-0.9.0\n"
     ]
    }
   ],
   "source": [
    "!pip install \"ibm-watson-machine-learning>=1.0.320\" | tail -n 1\n",
    "!pip install \"pydantic>=1.10.0\" | tail -n 1\n",
    "!pip install langchain | tail -n 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "### Defining the WML credentials\n",
    "This cell defines the WML credentials required to work with watsonx Foundation Model inferencing.\n",
    "\n",
    "**Action:** Provide the IBM Cloud user API key. For details, see\n",
    "[documentation](https://cloud.ibm.com/docs/account?topic=account-userapikey&interface=ui)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "import getpass\n",
    "\n",
    "credentials = {\n",
    "    \"url\": \"https://us-south.ml.cloud.ibm.com\",\n",
    "    \"apikey\": \"PUT YOUR KEY\" \n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "### Defining the project id\n",
    "The Foundation Model requires project id that provides the context for the call. We will obtain the id from the project in which this notebook runs. Otherwise, please provide the project id."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "try:\n",
    "    project_id = os.environ[\"PROJECT_ID\"]\n",
    "except KeyError:\n",
    "    project_id = input(\"Please enter your project_id (hit enter): \")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "<a id=\"models\"></a>\n",
    "## Foundation Models on `watsonx.ai`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "#### List available models\n",
    "\n",
    "All avaliable models are presented under `ModelTypes` class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "from ibm_watson_machine_learning.metanames import GenTextParamsMetaNames as GenParams\n",
    "from ibm_watson_machine_learning.foundation_models.utils.enums import DecodingMethods\n",
    "\n",
    "model_id_1=\"bigcode/starcoder\"\n",
    "\n",
    "parameters = {\n",
    "    GenParams.DECODING_METHOD: DecodingMethods.SAMPLE,\n",
    "    GenParams.MAX_NEW_TOKENS: 100,\n",
    "    GenParams.MIN_NEW_TOKENS: 1,\n",
    "    GenParams.TEMPERATURE: 0,\n",
    "    GenParams.TOP_K: 50,\n",
    "    GenParams.TOP_P: 1\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Initialize the model\n",
    "Initialize the `Model` class with previous set params."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "from ibm_watson_machine_learning.foundation_models import Model\n",
    "\n",
    "starcoder_model = Model(\n",
    "    model_id=model_id_1, \n",
    "    params=parameters, \n",
    "    credentials=credentials,\n",
    "    project_id=project_id)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "<a id=\"watsonxllm\"></a>\n",
    "## WatsonxLLM interface\n",
    "\n",
    "`WatsonxLLM` is a wrapper around watsonx.ai models that provide chain integration around the models.\n",
    "\n",
    "**Action:** For more details about `CustomLLM` check the [LangChain documentation](https://python.langchain.com/docs/modules/model_io/models/llms/custom_llm) \n",
    "\n",
    "\n",
    "### Initialize the `WatsonxLLM` class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "from ibm_watson_machine_learning.foundation_models.extensions.langchain import WatsonxLLM\n",
    "\n",
    "starcoder_llm = WatsonxLLM(model=starcoder_model)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "You can print all set data about the WatsonxLLM object using the `dict()` method."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'model_id': 'bigcode/starcoder',\n",
       " 'params': {'decoding_method': <DecodingMethods.SAMPLE: 'sample'>,\n",
       "  'max_new_tokens': 100,\n",
       "  'min_new_tokens': 1,\n",
       "  'temperature': 0,\n",
       "  'top_k': 50,\n",
       "  'top_p': 1},\n",
       " 'project_id': 'e34c9a21-8c92-4676-b3cc-2bf0d0e5d732',\n",
       " 'space_id': None,\n",
       " '_type': 'IBM watsonx.ai'}"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "starcoder_llm.dict()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "<a id=\"experiment\"></a>\n",
    "## Math Chain experiment\n",
    "The type of Math chain is called a `MathChain`, The LLMMathChain equips LLMs with mathematical capabilities..\n",
    "\n",
    "The experiment will consist in summing a se of numbers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "from langchain.chains import LLMMathChain\n",
    "from ibm_watson_machine_learning.foundation_models.extensions.langchain import WatsonxLLM\n",
    "\n",
    "starcoder_llm = WatsonxLLM(model=starcoder_model)\n",
    "llm_math = LLMMathChain.from_llm(starcoder_llm, verbose=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\u001b[1m> Entering new LLMMathChain chain...\u001b[0m\n",
      "What is the sum of 231, 12, 189 and 153?\u001b[32;1m\u001b[1;3m```text\n",
      "231 + 12 + 189 + 153\n",
      "```\n",
      "...numexpr.evaluate(\"231 + 12 + 189 + 153\")...\n",
      "\u001b[0m\n",
      "Answer: \u001b[33;1m\u001b[1;3m585\u001b[0m\n",
      "\u001b[1m> Finished chain.\u001b[0m\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'Answer: 585'"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "llm_math.run(\"What is the sum of 231, 12, 189 and 153?\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\u001b[1m> Entering new LLMMathChain chain...\u001b[0m\n",
      "What is 13 raised to the .3432 power?\u001b[32;1m\u001b[1;3m```text\n",
      "13**(.3432)\n",
      "```\n",
      "...numexpr.evaluate(\"13**(.3432)\")...\n",
      "\u001b[0m\n",
      "Answer: \u001b[33;1m\u001b[1;3m2.4116004626599237\u001b[0m\n",
      "\u001b[1m> Finished chain.\u001b[0m\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'Answer: 2.4116004626599237'"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "llm_math.run(\"What is 13 raised to the .3432 power?\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "\u001b[1m> Entering new LLMMathChain chain...\u001b[0m\n",
      "What is the sum of 231, 12, 189 and 153?\u001b[32;1m\u001b[1;3m```text\n",
      "231 + 12 + 189 + 153\n",
      "```\n",
      "...numexpr.evaluate(\"231 + 12 + 189 + 153\")...\n",
      "\u001b[0m\n",
      "Answer: \u001b[33;1m\u001b[1;3m585\u001b[0m\n",
      "\u001b[1m> Finished chain.\u001b[0m\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'Answer: 585'"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "llm_math.run(\"What is the sum of 231, 12, 189 and 153?\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "<a id=\"summary\"></a>\n",
    "## Summary and next steps\n",
    "\n",
    " You successfully completed this notebook!.\n",
    " \n",
    " You learned how to use LLMMathChain using custom llm WastonxLLM.\n",
    " \n",
    "Check out our _[Online Documentation](https://ibm.github.io/watson-machine-learning-sdk/samples.html)_ for more samples, tutorials, documentation, how-tos, and blog posts. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "### Authors: \n",
    " **Vincent Perrin**, IBM Ecosystem Technical Leader - France"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "Copyright © 2023 IBM. This notebook and its source code are released under the terms of the MIT License."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.10",
   "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}