Adds docs on passing tool results to a model

langchain-ai · Jul 16, 2024 · 5d75bd1 · 5d75bd1
1 parent 223cdbe
commit 5d75bd1
Show file tree

Hide file tree

Showing 3 changed files with 271 additions and 1 deletion.
diff --git a/docs/core_docs/docs/how_to/index.mdx b/docs/core_docs/docs/how_to/index.mdx
@@ -161,6 +161,7 @@ LangChain [Tools](/docs/concepts/#tools) contain a description of the tool (to p
 - [How to: create custom tools](/docs/how_to/custom_tools)
 - [How to: use built-in tools and built-in toolkits](/docs/how_to/tools_builtin)
 - [How to: use a chat model to call tools](/docs/how_to/tool_calling/)
+- [How to: pass tool results back to model](/docs/how_to/tool_results_pass_to_model/)
 - [How to: add ad-hoc tool calling capability to LLMs and Chat Models](/docs/how_to/tools_prompting)
 - [How to: return extra artifacts from a custom tool](/docs/how_to/tool_artifacts)
 

diff --git a/docs/core_docs/docs/how_to/tool_artifacts.ipynb b/docs/core_docs/docs/how_to/tool_artifacts.ipynb
@@ -12,7 +12,7 @@
     "This guide assumes familiarity with the following concepts:\n",
     "\n",
     "- [Tools](/docs/concepts/#tools)\n",
-    "- [Tool calling](/docs/concepts/#tool-calling)\n",
+    "- [Tool calling](/docs/concepts/#functiontool-calling)\n",
     "\n",
     ":::\n",
     "```\n",

diff --git a/docs/core_docs/docs/how_to/tool_results_pass_to_model.ipynb b/docs/core_docs/docs/how_to/tool_results_pass_to_model.ipynb
@@ -0,0 +1,269 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# How to pass tool outputs to the model\n",
+    "\n",
+    "```{=mdx}\n",
+    ":::info Prerequisites\n",
+    "This guide assumes familiarity with the following concepts:\n",
+    "\n",
+    "- [Tools](/docs/concepts/#tools)\n",
+    "- [Tool calling](/docs/concepts/#functiontool-calling)\n",
+    "\n",
+    ":::\n",
+    "```\n",
+    "\n",
+    "If we're using the model-generated tool invocations to actually call tools and want to pass the tool results back to the model, we can do so using `ToolMessage`s and `ToolCall`s. First, let's define some tools and a chat model instance."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import { z } from \"zod\";\n",
+    "import { tool } from \"@langchain/core/tools\";\n",
+    "\n",
+    "const addTool = tool(async ({ a, b }) => {\n",
+    "  return a + b;\n",
+    "}, {\n",
+    "  name: \"add\",\n",
+    "  schema: z.object({\n",
+    "    a: z.number(),\n",
+    "    b: z.number(),\n",
+    "  }),\n",
+    "  description: \"Adds a and b.\",\n",
+    "});\n",
+    "\n",
+    "const multiplyTool = tool(async ({ a, b }) => {\n",
+    "  return a * b;\n",
+    "}, {\n",
+    "  name: \"multiply\",\n",
+    "  schema: z.object({\n",
+    "    a: z.number(),\n",
+    "    b: z.number(),\n",
+    "  }),\n",
+    "  description: \"Multiplies a and b.\",\n",
+    "});\n",
+    "\n",
+    "const tools = [addTool, multiplyTool];"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "```{=mdx}\n",
+    "import ChatModelTabs from \"@theme/ChatModelTabs\";\n",
+    "\n",
+    "<ChatModelTabs customVarName=\"llm\" />\n",
+    "```"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "If we invoke a tool with a `ToolCall`, we'll automatically get back a `ToolMessage` that can be fed back to the model: \n",
+    "\n",
+    "```{=mdx}\n",
+    ":::caution Compatibility\n",
+    "\n",
+    "This functionality requires `@langchain/core>=0.2.16`. Please see here for a [guide on upgrading](/docs/how_to/installation/#installing-integration-packages).\n",
+    "\n",
+    ":::\n",
+    "```"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "[\n",
+      "  HumanMessage {\n",
+      "    lc_serializable: true,\n",
+      "    lc_kwargs: {\n",
+      "      content: 'What is 3 * 12? Also, what is 11 + 49?',\n",
+      "      additional_kwargs: {},\n",
+      "      response_metadata: {}\n",
+      "    },\n",
+      "    lc_namespace: [ 'langchain_core', 'messages' ],\n",
+      "    content: 'What is 3 * 12? Also, what is 11 + 49?',\n",
+      "    name: undefined,\n",
+      "    additional_kwargs: {},\n",
+      "    response_metadata: {},\n",
+      "    id: undefined\n",
+      "  },\n",
+      "  AIMessage {\n",
+      "    lc_serializable: true,\n",
+      "    lc_kwargs: {\n",
+      "      content: '',\n",
+      "      tool_calls: [Array],\n",
+      "      invalid_tool_calls: [],\n",
+      "      additional_kwargs: [Object],\n",
+      "      id: 'chatcmpl-9llAzVKdHCJkcUCnwGx62bqesSJPB',\n",
+      "      response_metadata: {}\n",
+      "    },\n",
+      "    lc_namespace: [ 'langchain_core', 'messages' ],\n",
+      "    content: '',\n",
+      "    name: undefined,\n",
+      "    additional_kwargs: { function_call: undefined, tool_calls: [Array] },\n",
+      "    response_metadata: { tokenUsage: [Object], finish_reason: 'tool_calls' },\n",
+      "    id: 'chatcmpl-9llAzVKdHCJkcUCnwGx62bqesSJPB',\n",
+      "    tool_calls: [ [Object], [Object] ],\n",
+      "    invalid_tool_calls: [],\n",
+      "    usage_metadata: { input_tokens: 87, output_tokens: 50, total_tokens: 137 }\n",
+      "  },\n",
+      "  ToolMessage {\n",
+      "    lc_serializable: true,\n",
+      "    lc_kwargs: {\n",
+      "      content: '36',\n",
+      "      artifact: undefined,\n",
+      "      tool_call_id: 'call_7P5ZjvqWc7jrXjWDkhZ6MU4b',\n",
+      "      name: 'multiply',\n",
+      "      additional_kwargs: {},\n",
+      "      response_metadata: {}\n",
+      "    },\n",
+      "    lc_namespace: [ 'langchain_core', 'messages' ],\n",
+      "    content: '36',\n",
+      "    name: 'multiply',\n",
+      "    additional_kwargs: {},\n",
+      "    response_metadata: {},\n",
+      "    id: undefined,\n",
+      "    tool_call_id: 'call_7P5ZjvqWc7jrXjWDkhZ6MU4b',\n",
+      "    artifact: undefined\n",
+      "  },\n",
+      "  ToolMessage {\n",
+      "    lc_serializable: true,\n",
+      "    lc_kwargs: {\n",
+      "      content: '60',\n",
+      "      artifact: undefined,\n",
+      "      tool_call_id: 'call_jbyowegkI0coHbnnHs7HLELC',\n",
+      "      name: 'add',\n",
+      "      additional_kwargs: {},\n",
+      "      response_metadata: {}\n",
+      "    },\n",
+      "    lc_namespace: [ 'langchain_core', 'messages' ],\n",
+      "    content: '60',\n",
+      "    name: 'add',\n",
+      "    additional_kwargs: {},\n",
+      "    response_metadata: {},\n",
+      "    id: undefined,\n",
+      "    tool_call_id: 'call_jbyowegkI0coHbnnHs7HLELC',\n",
+      "    artifact: undefined\n",
+      "  }\n",
+      "]\n"
+     ]
+    }
+   ],
+   "source": [
+    "import { HumanMessage } from \"@langchain/core/messages\";\n",
+    "\n",
+    "const messages = [\n",
+    "  new HumanMessage(\"What is 3 * 12? Also, what is 11 + 49?\"),\n",
+    "];\n",
+    "\n",
+    "const aiMessage = await llmWithTools.invoke(messages);\n",
+    "\n",
+    "messages.push(aiMessage);\n",
+    "\n",
+    "const toolsByName = {\n",
+    "  add: addTool,\n",
+    "  multiply: multiplyTool,\n",
+    "}\n",
+    "\n",
+    "for (const toolCall of aiMessage.tool_calls) {\n",
+    "  const selectedTool = toolsByName[toolCall.name];\n",
+    "  const toolMessage = await selectedTool.invoke(toolCall);\n",
+    "  messages.push(toolMessage);\n",
+    "}\n",
+    "\n",
+    "console.log(messages);"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "AIMessage {\n",
+      "  lc_serializable: true,\n",
+      "  lc_kwargs: {\n",
+      "    content: '3 * 12 is 36, and 11 + 49 is 60.',\n",
+      "    tool_calls: [],\n",
+      "    invalid_tool_calls: [],\n",
+      "    additional_kwargs: { function_call: undefined, tool_calls: undefined },\n",
+      "    id: 'chatcmpl-9llB0VVQNdufqhJHHtY9yCPeQeKLZ',\n",
+      "    response_metadata: {}\n",
+      "  },\n",
+      "  lc_namespace: [ 'langchain_core', 'messages' ],\n",
+      "  content: '3 * 12 is 36, and 11 + 49 is 60.',\n",
+      "  name: undefined,\n",
+      "  additional_kwargs: { function_call: undefined, tool_calls: undefined },\n",
+      "  response_metadata: {\n",
+      "    tokenUsage: { completionTokens: 19, promptTokens: 153, totalTokens: 172 },\n",
+      "    finish_reason: 'stop'\n",
+      "  },\n",
+      "  id: 'chatcmpl-9llB0VVQNdufqhJHHtY9yCPeQeKLZ',\n",
+      "  tool_calls: [],\n",
+      "  invalid_tool_calls: [],\n",
+      "  usage_metadata: { input_tokens: 153, output_tokens: 19, total_tokens: 172 }\n",
+      "}\n"
+     ]
+    }
+   ],
+   "source": [
+    "await llmWithTools.invoke(messages);"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Note that we pass back the same `tool_call_id` in the `ToolMessage` as what we receive from the model in order to help the model match tool responses with tool calls.\n",
+    "\n",
+    "## Related\n",
+    "\n",
+    "You've now seen how to pass tool calls back to a model.\n",
+    "\n",
+    "These guides may interest you next:\n",
+    "\n",
+    "- [Creating custom tools](/docs/how_to/custom_tools)\n",
+    "- [Building agents with LangGraph](https://langchain-ai.github.io/langgraphjs/)"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "TypeScript",
+   "language": "typescript",
+   "name": "tslab"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "mode": "typescript",
+    "name": "javascript",
+    "typescript": true
+   },
+   "file_extension": ".ts",
+   "mimetype": "text/typescript",
+   "name": "typescript",
+   "version": "3.7.2"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}